ML20148L054
| ML20148L054 | |
| Person / Time | |
|---|---|
| Issue date: | 02/24/1988 |
| From: | NRC |
| To: | |
| References | |
| NUDOCS 8804010123 | |
| Download: ML20148L054 (429) | |
Text
-__
UNITED STATES NUCLEAR REGULATORY COMMISSION In the Matter of:
BWR MARK I CONTAINMENT PERFORMANCE INFORMATION EXCHANGE WORKSHOP l
CORRECTED COPY l
I MORNING SESSION Pages: 1 through 161 Place: Baltimore, Maryland Date: February 24, 1988
.. . . ....... .as . .... . . . . . .... . . .. .i HERITAGE REPORTING CORPORATION OfkielReporters g 1220 L Street, N.W., Suke 646 Washinston, D.C. 20005 (202) 628-4888 8804010123 880224 PDR TOPRP EMVGENE C PDR
1 1
NUCLEAR REGULATORY COMMISSION O . )
In the Matter of: )
)
BWR MARK I CONTAINMENT )
INFORMATION EXCHANGE )
WORKSHOP )
)
12th Floor Ballroom The Belvedere Hotel 1 East Chase Street
. Baltimore, Maryland Wednesday, February 24, 1988 The above-entitled matter came on for hearing, '
pursuant to notice, at 12:43 p.m.
APPEARANCES:
On behalf of the Nuclear Regulatory Commission:
- DR. ERIC BECKJORD -
DR. WAYNE HOUSTON DR. THEMIS SPEIS MR. JERRY HULMAN MR. LEONARD SOFFER C
8 Ileritage Reporting Corporation (202) 628-4888
2 O C0NTENTS STATEMENT OF: PAGE:
ERIC BECKJOLD 4 THEMIS SPEIS, DEPUTY DIRECTOR FOR GENERIC ISSUES, OFFICE OF NUCLEAR REGULATORY RESEARCH 9 BILL RASIN, TECHNICAL DIRECTOR,. NUCLEAR MANAGEMENT AND RESOURCES COUNCIL 23 TERRY PICKENS, SENIOR NUCLEAR SAFETY SERVICES ENGINEER, NORTHERN STATES POWER 28 STEVE SHOLLY, MHB ASSOCIATES 35 JERRY HULMAN, CHIEF, SEVERE ACCIDENT ISSUES BRANCH, RES 40 MOHSEN KHATIB-RAHBAR, DEPARTMENT OF NUCLEAR ,
) ENERGY, BROOKHAVEN NATIONAL LABS 48 PAUL HILL, PENNSYLVANIA POWER AND LIGHT 57 l
I l
l l
l I
Heritage Re prting Corporation (202) 628-4888 l
--_. _ ._ -- ___ , _ .. . _ , _ _ _ _ - .l
I l
3 1 PROCEEDINGS
.- / 2 DR. SPEIS: Gentlemen, can you please take your 1
3 seats?
l 4 (Pause) l 1
5 DR. SPEIS: Can you please take your seats back i
6 there? l I
7 (Pause) 8 DR. SPEIS: Hello. Okay. Can you hear me back 9 there? No? You cannot hear me? Who is the electrician? I 10 guess most of you are experts, so there,must be an electrician 11 among you.
12 How about now? Is that better or worse? Oh, boy.
13 We're really going to hav,e -- I guess we'll have to be quiet, 14 and I'll try to speak as loud as I can and maybe with.my heavy 15 accent, you might be able to pay some attention.
16 My name is Themis spels. I'm from the NRC, and I 17 want to welcome you to the BWR Mark I Containment Performance l
\
18 Workshop.
19 I am very happy to see so many people here.
20 Initially, we thought we were going to have a meeting of the 21 twenty experts, but I never realized that there were so many l
l 22 experts in this field, but we welcome you anyhow.
23 You cannot hear me? How.about now? Boy. I think ,
24 Jerry will have to get the --
0 25 MR. HULMAN: The P.A. system. Until then, would you i Heritage Rep rting Corporation I (202) 628-4888 l
r 4
1 all mind moving up and filling some of the seats in front and Q
\' 2 try to keep down the side conversations? We will try and get 3 it fixed.
4 DR. SPEIS: We have a lot of work to do for the next 5 few days. We're here to discuss a number of technical issues 6 dealing with the performance of Mark I Containment. I'll have 7 more to say in a few minutes, but before I start my 8 introduction, I would like to introduce some of the people here .
9 who have been preparing for this.
10 Jerry Hulman is the Branch Chief who is responsible 11 for organizing this workshop. He is the Division Director.
12 Wayne Houston is here, who will participate in the
() 13 deliberations for the nekt few days. And we have with us the 14 Director of the Office of Research, Eric Beckjold, who is going 15 to open the meeting with some remarks, and then I'll follow him 16 later on.
17 Eric.
18 STATEMENT OF ERIC BECKJOLD, DIRECT, OFFICE OF NUCLEAR 19 REGULATORY RESEARCH, USNRC 20 DR. BECKJOLD: Hello.
21 MR. HULMAN: Eric, it does for him.
22 DR. BECKJOLD: Well, I'll speak as loudly as I can.
23 I want to welcome each and every one of you here, s 24 ladies and gentlemen, to this Mark I Containment Performance 25 Workshop.
. Heritage Reporting Corporation (202) 628-4888 w-r-- ---n-,
1 5
1 Can you hear me back there? Okay.
() 2 I'm very pleased to take part in this meeting. It's 3 an important one, and I want to thank Dr. Speis for delaying 4 the' start for me because I was a few minutes late. I guess 5 that implies that I should have something important to say, and 6 I hope to fulfill that.
7 The attendance at this meeting and the wide 1 8 representation which I can see before me already from some .
9 faces that I recognize is a clear indication of the importance 10 that you are putting on this subject.
11 The purpose of the meeting is to consider the 12 performance of Mark I Containments under severe accident
() 13 14 conditions, including those which extend beyond design basis accidents.
15 The events of this class were analyzed in the 16 Rasmussen Wash 1400 Report and are being analyzed again in 17 NUREG 1150, the draft study that is still underway.
18 Further than that, the accident at Chernobyl has 1
19 focused attention on containment performance questions, even 20 though the Chernobyl reactor type is quite different from the 21 reactors elsewhere, with very different safety characteristics, I 22 and despite the fact that it did not have a containment in the 1
23 sense that we understand that. l l
l 24 The meeting is especially important now because of O 25 the evaluation which is underway at the NRC on the Mark I Heritage Re orting Corporation (20 ) 628-4888
t 6
1 Containment Performance.
(-)s 2 This evaluation was foreseen in the presentation to 3 the NRC Commissioners in July of last year, and it was 4 described in some. detail in a Commission Paper, dated December 1
5 8th of 1987, and now we're involved in a gathering and a 1
6 consideration of information and the identification of key 7 issues, and thic is one of several very important steps in the !
8 process.
9 The next step after this meeting is over and done 10 with will be the evaluation of the issues by'the NRC staff and 11 consideration of potential improvements and, then, the staff 12 will make a preliminary report to the Commission in April, a
' I 13 couple of months from now, and will make a final report and
)
14 recommendations in August of 1988.
15 Dr. Speis and Mr. Hulman are going to describe this 16 process in a definitive way, and, so, I won't say much more 17 about the process itself.
18 I do want to talk for a minute about where we're 19 going from here. This is, as you know, the Mark I Containment 20 Study, but I note that the Mark I Containments for the BWR are 21 the first to be evaluated, and we will conduct a similar 22 process as soon as we possibly can for the other containment 23 types. l l
l 24 I regard the Mark I evaluation which is underway as a i 25 model for the consideration of the remaining containments, and Heritage Reporting Corporation I (202) 628-4888 l
7 1 a very big step forward in the larger process of resolution and
<3 ij 2 closure of severe accident issues for the operating reactors 3 generally.
4 We expect this process, that is this larger process, 5 of severe accident closure to take place over the next two to 6 three years, and I wish to emphasize our intention regarding 7 the schedule for severe accident closure.
8 Then, I want to say a bit about the regulatory agenda ,
9 and the research agenda. As we move through the Mark I 10 Containment process, we anticipate that the Commission will 11 make final regulatory decisions.
12 As has been the case in the past, there is always 13 some incomplete information, some parts of technical
)
14 uncertainty that need to be clarified. The task of 15 clarification and of obtaining further data is called 16 Confirmatory Research, and I expect that Mark I decisions may 17 entail an agenda of these confirmatory research problems to be 10 carried out over in the next several years or so.
19 I think the example of this that I would cite is the 20 loss of coolant accident, the design basis accidents for the 21 light water reactors. That work is now essentially complete, 22 and by analogy, I can say that.the same thing -- is likely to l
23 be the same kind of thing in connection with the Mark I !
24 question and, further, the same similar kind of agenda in the l
C} 25 case of other containments, if need be. \
1 lieritage Reprting Corporation (202) 628-4888 .
~ .n
8 1 I'd like to say just a word also about another kind
(~T g_) 2 of research that is underway, which I will call Anticipatory 3 Research. That also plays a key role in the process that I'm 4 describing.
5 Anticipatory research has been and is being carried 6 out tc understand, for one thing, severe accident phenomenon; 7 that is to say, the progressive damage that would envelope an 8 uncool core, generation of hydrogen, fission product release, ,
9 and source air composition, that type of thing, to name a few 10 examples.
11 In addition to research on the phenomena of severe 12 accidents, we are this year initiating a couple of new 13 prograr,s. - One on accident management, and one on human
) ,
14 factors.
15 Accident management, in brief, the objective there is 16 to develop strategies of success to regain control of a core-17 damaged reactor, quenching the core within the reactor vessel, 18 and thus assuring containment integrity.
19 If core retention in the vessel is not possible, the 20 next stage of accident management would be a strategy of action 21 aimed at maintaining containment integrity, despite a melting 22 core.
23 Human factors, in brief. 'The activity that we have 24 in mind there includes the study of human capabilities as O 25 individuals and in organizations with the objective of Heritage Rewrting Corporation (202) 628-4888
I I
9 1 improving the methods of reliable nuclear power plant j
() 2 operations in all modes of operation, normal and accident.
3 I expect that all three of these research programs, 4 the severe accident phenomenon, the accident management, and 5 the human factors, wi.11 be parts of the research agenda which I 6 described briefly a minute ago. )
7 I think that completes what I had to say about how I 8 see the research agenda ahead, and I think now it's appropriate .
9 to move into the program itself, and Dr. Speis will take back -
10 the podium.
11 DR. SPEIS: Thank you, Eric.
12 STATEMENT OF DR. THEMIS SPEIS, DEPUTY DIRECTOR FOR 13 GENERIC ISSUES, OFFICE OF NUCLEAR REGULATORY RESEARCH 14 RES 15 DR. SPEIS: I will continue with some introductory 16 remarks.
17 I will start my co-presentation by trying to put this 18 whole effort into some perspective because, as hopefully most 19 of you know, we at the NRC, the primary effort that we put in 20 the licensing and regulation of reaptors is to prevent 21 accidents.
22 But we all recognize that accidents can happen and J
l 23 even though we put the primary emphasis and focus on accid'ent !
l 24 prevention, we would like to see and to understand what happens j 25 if, indeed, something more severe damage accident takes place Heritage ReW rting Corporation .
(202) 628-4888
10 1 and hear what -- try to understand what the residual risks are
(~%
\_j 2 and see if it makes sense to pursue any appropriate measures in 3 that area.
4 As you see from the first viewgraph that I have put 5 up there, the licensing regulatory approach starts with 6 defense-in-depth, which involves multiple successive barriers 7 to fission product release. The main effort has been put in 8 making sure that the fuel rod contains its integrity under .
9 transients or design basis accidents.
10 Then, tremendous effort has been put to making sure 11 that the primary coolant also retains its integrity and is as 12 reliable as possible, and last but not least, to cause -- we
() 13 want to make sure that there are margins in case some of the 14 design basis accidents take place. We will have a big 15 containment which prevents any adverse consequences as a result 16 of that.
17 The design basis events again are both in terms of 18 accidents and the design basis accident itself, as you know, is 19 the large break water or the steam line break, whichever is 20 culminating. Also, containments are designed for external 21 events.
22 In addition to that, some consideration has been 23 given to the existing framework as far as accidents are 24 concerned. We postulate in an arbitrary fashion that a large O 25 fission product release takes place in the containment and then Heritage ReDorting Corp 6 ration (202) 628-4888
11 1 the containment is designed not to leak those fission products 2 or to meet some of those guidelines.
3 The only difficulty is that this fission product that 4 is postulated to be released into the containment and then the 5 containment is designed to accommodate it does not include all 6 the attributes of a severe accident. It only. includes the 7 radioactive source, the radioactivity itself. It doesn't 8 include the pressure a'nd temperature that are associated with ,
9 the severe accident.
10 Let's go to the next.viewgraph. In this viawgraph, I 11 show some of the things that have taken place after TMI. I'm 12 trying to impress upon you here that the licensing regulatory 13 process has not been a staggered one. It has been revised as a 14 result of the TMI accident, as a result of operational 15 experience, and we are at this point now, the last part of the 16 slide, where we want to give some more consideration to the 17 severe accidents to see, as I said earlier, what are the l
18 residual risks in the area and see whether this involves steps 19 that can be taken.
20 If you will recall, the TMI experience pointed out 21 that single failures and a design approach is okay, but there 22 are things that can happen that deviate from that narrow 23 framework. Errors can take place, operators can make mistakes, 24 and as a result of that, we paid a lot of attention to 25 considering those multi-failures from human error or from fleritage Remrting Corporation (202) 628-4888
l I
I 12 1 equipment itself. I
()
I 2 As a result of that, we revised the emergency
. i 3 operating procedures to consider multi-failure events and put 4 together something called symptom-oriented emergency operating 5 proce!ures. Those proce'dures are when all the earlier design 6 basis accidents and transients and look at the plant and the 7 accident in an integral way and see whether additional things I 8 could happen and what additional things could operators take ,
9 advantage of at the plant.
10 For example, even though the auxiliary water system 11 is supposed to be there to remove the decay, in the existing 12 emergency operating procedures, assumptions are made that,
/\ 13 indeed, you can lose the auxiliary water system and then what O
14 additional things can the operator do to make sure that the 15 core is cooled and the decay is removed.
16 For example, procedures like take and bleed have been 17 developed for BWRs and related ones for boiling water reactors.
18 Also, as a result of the TMI accident, the so-called
)
l 19 weaker containments, severe accident, hydrogen releases, must 20 monitor your quantities and were considered originally in the 21 design basis. For example, the Mark I's and II's are inerted, 22 the ice condensers, and Mark III's would have put hydrogen 23 ignited to make sure that the hydrogen action evolves from the 24 core following the degree of accident is burning and, i 25 therefore, there is no test for it, to accumulate in the l
lieritage Rewrting Corporation (202) 628-4888 1
l 13 1 containment and lead to some type of detonation.
-s l
(_,) _2 This hydrogen design.for the hydrogen release was put 3 into the regulations in 10 CFR 5044 for those of you who are l 4 going to pursue this further, 5 Another thing that we will pay special and continuous ,
6 attention and, of course, that is the primary function of the 7 Office of Nuclear Reactor Regulations, is to continuously l
\
8 follow the operating experience. As a result of that, we quite ,
9 often in the past and the present, we change regulations and 10 suppositions.
11 For example, all of you are familiar with the ATWS 12 rule and also at the present time, we just have sent to the l
f)
V 13 Commission the so-called Black-out rule, which further enhances 14 your liability of power supplies, electric power assistance.
15 Again, this is something that is taught from 16 operational experience. We found out that power supply systems 17 can be enhanced and all of us know that from PRAs and from 18 other considerations, loss of power, both off-site and on-site, 19 is a very important risk. Therefore, as a result of that, we 20 are purs'uing this final rule which will deal with this black-21 out issue.
22 Again, as we proceed and we find out about other 23 things that can be enhanced and improved, we do so on a 24 continuing basis.
25 Coming down to the last part of the slide, the area Heritage Rep rting Corporation (202) 628-4888
14 1 of seve-o accident considerations, this is the area that we
( 2 will ta. 1.ng today and tomorrow and Friday.
3 Again, this is an area that we will put into the 4 residual risk category. We will put all our effort into the 5 prevention, into making sure that containments don't leak, 6 making sure that if something happens, the consequences are as 7 much minimized as possible, but, again, we cannot go through 8 this area. Severe accidents have taken place, TMI has taken .
9 place, and Chernobyl has taken place, and as I said two or t
10 three times already, we would like to understand what the 11 residual risks are in this area and see what meaningful things 12 can be done into this area.
() 13 The Commission issu,ed a policy statement two years 14 ago or so, I guess two or three years ago maybe, two and a half 15 years ago, and there is the confusion that even though risk 16 from severe accidents were acceptable, there were things that 17 could be done in individual plants to further reduce the i 18 residual risk which could both prevent an accident from l l
19 happening as well as possibly enhancing the containment given a 20 severe accident.
i 21 We are in the process at the present of initiating 22 the implementation of the policy statement, and Mr. Beckjold 23 will talk a little bit about it. One of the key parts of'it j 24 will involve an individual plant examination to look for 8 25 improvements that can be made to both preventing core lleritage Rew rting Corporation (202) 628-4888
_ _ _ _ _. . - - -_ . . _ . . ~ . -
e 15 1 degradation or core melt event as well as enhancing the I -
) 2 containment performance, and in parallel with it, we are 3 looking at generic issues associated with the containment and, 4 of course, we will be talking in the next few days generic 5 issues associated with containment performance and we're 6 starting the process with the Mark I Containment.
7 Let's go to the next viewgraph. I think the next one 8 says more about severe accidents. I think I have talked about .
9 it. Again, the issue here is defense-in-depth, whether it 10 makes sense to do more in the' prevention area or in the 11 mitigation.
12 We think that the individual plant examination
( ) 13 process and our studies generically on the containments a
14 hopefully will put all the information on the table and some 15 reasonable judgments can be made where we should be -- what are 16 the more fruitful areas to be pursuing our efforts.
17 Again, we want to take a look at whether there is a 18 reasonable source of mitigation probability for some of the 19 more dominant threats to containment, and we think that a key 20 there to have a pretty reasonable understanding of the 21 potential containment failure modes and their importance.
22 Failure modes, of course, one has to know what are 23 the loads given the severe accident, how a containment is 24 threatened, what are the processes and what are the phenomena, 9 25 and, more important, to be able to shift those processes and lieritage Rep rting Corporation (202) 628-4888 0
16 1 phenomena and see whether the more credible ones, even though t
O) g_ 2 we are in the severe accident region, again, you know, you have 3 to separate what are the more credible ones from the more --
4 less credible ones.
5 One thing'I want to say here, and let's go to the 6 next viewgraph, is that containments, even though they were 7 designed for design basis accidents, for locas, for 8 earthquakes, for external events, for the so-called severe' ,
9 accident source term,-also from the utilization of conservative 10 course and standards, they have quite a bit of margin which 11 goes beyond the design probability.
12 For example, we are talking about the Mark I. I 13 think the design of the Mark I is something like sixty pounds
[)
14 per square inch, Jerry.
15 MR. HULMAN: Forty-five to sixty.
16 DR. SPEIS: Forty-five to sixty, both the wet well 17 and the dry well.
18 But as you see later on, I have a viewgraph which 19 addresses specifically the margins of elbow in the Mark I.
20 There are margins of both these design basis.
21 Now, this margins are normally are containment 22 specific. They depend on the voir e of the containment, the 23 material, the configuration, how tar the b6undary of the
- 24 containment is, vis-a-vis the core itself.
(/ 25 For example, basud on experiments that have been done 1
l l
Ileritage Rep rting Corporation (202) 628-4888
17 1 at Sandia, from analyses that have been done in a number of
() 2 places, we are finding out that as far as pressure is 3 concerned, these margins can range anywhere from 2.5 to three 4 times the design level.
5 Again, you know, there are plant-specific things that 6 have to be considered to make sure that-those numbers are 7 relevant for specific designs.
8 Coming back to the question af residual risk from ,
9 severe accidents, even though there are margins on these 10 containments, again based on analysis and experiments, for each 11 containment type, there are main failure mechanisms which could 12 lead to containment failure.
/
b 13 Now, the key question that we are addressing right 14 now and we will be focusing our attention in the next days, we 15 like to get as reasonable as possible our understanding of what 16 are the challenges to containments, and I think I mentioned 17 this already, but I want to stress this point because that will 18 be the focus of our work the next few days, and your views and 19 the discussions that will take place.
20 What are these loads, what are the failure modes, i
21 and, again, you know, what type of understanding we can put 22 together as far as the probability of those failure modes and 23 the challenges to the containment in this severe accident 24 arena.
O 25 Let's go to the next viewgraph. Now, once we have Heritage Re@ rting Corporation (202) 628-4888 4
18 1 all this information, we have to, as Eric said already, we have O
(_j 2 to be recommending actions to our Commission. We have to 3 consider the rules that we go by. We have a safety goal that we 4 have to consider. We have to consider, you know, what is more 5 appropriate, to spend more money here or there.. "e have to i
6 consider the cost effectiveness and weigh that appropriately 7 with the risk reduction characteristics of some of the fixes 8 that we might be recommending. ,
9 So, these are the things that we will be doing the 10 next few months, but, again, I want to say maybe for the fourth 11 or fifth time that this is a technical meeting and we are 12 looking forward to getting some good input, getting your
{} 13 wisdom, ,your insights into some of the technical areas.
14 Again, the bottom line that we would like to gain 15 from you people in the next few days is, you know, what are the 16 most probable threats to these containments from a thermal 17 hydraulics standpoint. How do they fail. What are the ranges.
18 What things can be done to ameliorate those things and when 19 Jerry comes to. talk, I will make you try and summarize some of i
20 these things.
21 Let's go to the next viewgraph. I promised you 22 earlier that I will show you -- summarize what we know as far i
23 Mark I containments, as far as their reserve capability to 24 pressures.
25 This is something that was put together by Mr. Loto Heritage Re rting Corporation (20 ) 628-4838
19 1 from our Division of Engineering based on his work that has
() 2 been_done here and in some other places.
3 You see here that even though the Mark I containment 4 is designed to something like forty-five to sixty pounds per 5- square inch, if the challenge comes from a pressure or 6 temperature and this happens before the vessel fails, the 7 failure mode has been postulated to take place in the Torus, 8 and you can see that the estimated failure pressure ranges ,
9 anywhere from a 104 to a 108 pounds per square inch gauged.
10 Another failure mode manifests itself by leakage to ,
11 the dry well head flange, gives you an estimated failure 12 pressure which ranges somewhere between 120 and 180 pounds per 13 square inch.
) ,
14 Now, if there is significant call for complets 15 interaction which provides more energy to the containment, you 1
16 are talking about some more higher temperatures, and then you j 17 can see that this range, that this failure pressure range is 18 reduced and it varies somewhere between 125 and 150 pounds per 19 square inch, and if you are postulating higher temperatures, 20 then this range goes down.
21 So, you see that the scenarios are very important, i 22 but, in any event, I want to make the point that these 23 containments, even though they were designed for design basis s 24 accidents, because of the conservatism, they have margins that
%.)
25 can be utilized in the severe accident arena.
Heritage Re wrting Corporation (202) 628-4888
20 1 In Part I of the important things that we have to
( 2 understand in this arena is what are the margins.not only for 3 the Mark I but for other containments, and how we can best 4 utilize those margins either in the management of an accident 5- framework and some other ways to make sure that we can 6 ameliorate or prevent containment failure, and even if it takes 7 place, that the activity that gets out is minimized.as much as 8 possible. .
9 Let's go to the next one. The severe accident 10 closure includes a consideration of a number of things. I 11 think both Eric and I have alluded to them. The individual 12 plant determinations that we will be pursuing shortly, the
() 13 14 containment performance that we're talking now and we're starting with Mark I. I 15 We are at the midst of deciding how to go forward and 16 implement some parts of the safety goal that will tell us -- '
17 give us some steep limits and some guidance about how far we l 18 can go into these areas. We have to decide what to do with 19 external evcnts. The accident management is an important part !
20 of this process. j 21 The effort that we will be pursuing is more than 22 taking extreme consequences or events and see what we can do 23 about them, but we look at this area in a very mechanistic and 24 orderly fashion and see what things we can do to further 25 prevent accidents by further enhancing the emergency operating Heritage Reprting Corporation (202) 628-4888
I 21 1 procedures, what further things can-be done to, even if'you ,
( 2 have some core degradation, what further things can be done to 3 keep the degraded or molten core in the vessel and, of course, 4 what further things can be done once that entails to further 5 enhance containment performance.
6 So, accident management is a key part'of the severe 7 acc-ident ef fort.
8 The next viewgraph shows some examples of.what I mean .
9 by being more specific. One of the things that -- I guess the 10 example that I have in mind in the BWR Mark I or maybe this is 11 also applicable to the other BWR reactors, if you assume that 12 you lose all cooling, you know, the HPCI, the RCIC, the low 13 pressure injection, including the residual heat removal,
(
14 including the core sprays, you can probably uncover the core 15 very fast, maybe in half an hour, and you-can melt the core 16 maybe in one and a half hours, probably you cnn fill the vessel 17 in two to three hours.
18 But if you are able to only retain water which is
-l 19 available to cool the control rod drives, which isn't much, you 20 can delay core damage or, I don't know, three to four hours 21 based on calculations at Oak Ridge and for most sequences, cote 22 melting can be prevented.
23 So, one of the things that we have to focus on in -
1 24 accident management is how to make sure that all the sources of C 25 water can be made available, you know, even if you lose your l
Heritage Re w rting Corporation ;
(202) 628-4888 j
22 ,
1 outside power.
2 So, with that introduction and that perspective, I 3 would like to have Jerry Hulman, who has been the organizer and 4 the force behind this workshop and focused on these specific 5 issues that we will be discussing today and tomorrow.
6 Thank you.
7 Maybe after Jerry is through, if there are any 8 general questions, we can take them at that time and then we ,
9 can proceed with the more detailed presentations.
10 MR. HULMAN: John, I don't want to go through the 11 introduction yet. According to the agenda, there are a few 12 changes that I have to make based on what the speakers have 13 requested.
14 Firs.t, at the top, after Themis spoke, Eric spoke.
15 Then, shown on your agenda at.12:55 is Bob Janecek. Filling in 16 for him will be Terry Pickens from Northern States Power.
17 From 1:45 to 2:45, Roj Sehgal from EPRI is shown. He l
18 will not speak on that subject.
19 The next one, next slide, please. The session 20 tomorrow morning from 8:30 till 10:30, we are adding Steve
- 21 Hodge from Oak Ridge first and deleting Raj Sehgal from that l 22 brief.
23 In the sessi'on from 2:00 to 2:45, the individual from 24 GE that was to be announced won't show up, whoever that was.
25 So, 'you can throw that one out.
. Heritage Reporting Corporation (202) 628-4888
23 1 Those are the only changes I have to the agenda'now. '
'(yk 2 I will be back after Mr. Sholly speaks.
3 Right now, I would like to ask Bill Rasin to dome up 4 and'make some remarks.
5 Bill.
6 STATEMENT OF BILL RASIN, TECHNICAL DIRECTOR, NUCLEAR 7 MANAGEMENT AND RESOURCES COUNCIL (NUMARC) 8 MR. RASIN: Thank you, Jerry. ,
9 I would like to just make a few remarks to give a 10 nuclear industry management perspective to the subjects of 11 severe accidents and particularly the subject of Mark I 12 Containments and proposed modifications under discussion at 13 this meeting.
( ,
14 The nuclear industry, along with the NRC, has 15 sponsored quite a lot of work, both in research and in ;
i 16 analytical studies, over the past eight years in response to 17 the concerns of the Commission after the Three Mile Island 18 accident.
19 During that period of time, we had a lot.of technical !
l 20 exchange in public meetings somewhat like this one, except I i
21 don't think quite so large, between the staff and the industry 22 to explore the many issues that came before us.
23 During those discussions, we were able to come to 24 agreement on many, many issues. We finally came down to a set 25 of issues on which essentially we agreed to disagree, and I Heritage Reprting Corporation (202) 628-4888 -
1 ~ _
l l
24 j 1 think concluded on both sides that the only way to answer those 2 questions was not through further analysis with different 3 assumptions, but thro' ugh some hard research carefully chosen to ;
)
4 address the important issues, so that further analysis could be 5 done without a wide range of uncertain assumptions.
6 Having come to this point, the Commission was able to i
7 make the severe accident policy statement that's already been-8 mentioned, and that statement concluded that at this time, .
9 further regulatory action in regard to severe accidents for ,
10 present plants was not required, unless some new information 11 came to light that called the results we had come to to date I
12 into question.
() 13 Is it not working? Can you not hear me back there?
14 Now? Well, I'll just have to try. I don't know if I I
l 15 can talk as loud as you, Jerry. I'll do my best. j J
16 (Pause) 17 MR. RASIN: Those in the back, just wrinkle your brow
- 18 and look confused if you can't hear me, and I'll try to adjust.
19 We had come to the point where the Commission could i
20 issue a severe accident policy statement that said that j j 21 regulatory actions for existing plants were not required unless l 22 new information came to light.
t 23 However, that statement also said that what we had rg 24 learned showed us that there were significant effects of V
25 particular designs, plant specific designs, and that we should
! Heritage Reprting Corporation (202) 628-4888
1 proceed to do an individual plant evaluation to try to identify
) 2 any plant specific concerns which may remain.
3 The industry,~ with the guidance of the staff and 4 their subsequent review, went on to produce a methodology to 5 allow us to proceed with these individual plant evaluations, 6 and we hope that these evaluations will be underway shortly 7 with the issuance of a generic- letter to that ef fect.
8 Now, about the time we were completing the .
9 methodology for individual plant evaluations, a proposal came 10 forward from the staff for some modifications to Mark I 11 Containments. There was a meeting between the staff, the Id 12 Corps Management and Contractors, and the BWR Owners Group, l
() 13 held to discuss those potential modifications.
14 Out of that meeting came a concern on the part of 15 industry management, both because of the importance of the 16 issue and because of the regard we had for the staff members 17 proposing those modifications, that we felt particular 18 attention was due to the study of those issues.
19 To that end, industry formed a NUMARC Integrity 20 Working Group consisting of senior management representatives 21 to oversee the technical efforts to address the proposed 22 modifications. That working group reviewed information 23 developed from the IDCOR program, from the NRC Research and 24 Analytical Programs and from work done by the BWR Owners Group.
l 25 The conclusions of that working group were completed lleritage ReWrting Corporation (202) 628-4888
26 1 last fall and reviewed by NUMARC management and were just !
( 2 transmitted to the NRC staff'in January, which is last month.
3 The conclusion of the working group are that all of 4 the information reviewed shows that the risk from BWR Mark I's 5 is very low. The safety of the plant is well within th? smiety 6 goals established by the Commission, even given very 7 pessimistic assumptions used in.the studies, and this is trua 8 of the conclusions of the NRC studies as well as the industry .
9 studies.
10 Furthermore, we felt that the issues giving r..se to 11 the proposed modifications at this time were not new 12 information. The information was well in ha,nd w'en thw studies j 1
13 were comp 1bted and, the.refore, the results of the st., dies 14 showing the low level of risk were not in jeopandy because gf 15 these issues.
16 Furthermore, in looking at the e'Tsuts ot the l l
17 proposed modifications, we found there to be a very small l
18 impact on overall risk of the plants. We, therefore, concluded ;
l
, 19 that to make these modifications on a voluntary basis, aheac) pf
]
20 individual plant evaluations was not warranted by the industri !
21 and has so concluded and informed the NRC.
22 Now, IDCOR is winding down. However, severe accident 1
23 research proceed within the industry. There is still a larga 24 effort at the Electric Power Research Institute to address O 25 severe accident issues. We realize that what we have learned Heritage Rep rting Corporation 1 (202) 628-4888 ,
, _ . . _-- .o - - - ._. _ - - - _ _ _ , ,
27 1 over the last seven or eight years has not been fully taken F1
( 2 advantage of and put into practice across the industry, and, w f 3 so, we intend to continue to work on severe accident issues.
4 A very important area already alluded to is that of 5 accident management strategies, and we hope to interact with 6 the staff and trade ideas and concerns as we proceed in that' 7 effort.
8 We also continue to take initiatives as an industry ,
9 and with the staff to further improve the prevention of -
10 accidents. An example of that is the station black-out 11 initiatives that we're working on, 12 However, we believi at this point that the most
), 13 important action we can take to maintain and improve the safe'ty -
14 of existing plants is to proceed with the plant specific 15 evaluations called for by the policy statement.
16 The issues that must go on into further analysis and 17 research are certainly important issues and they should 18 Continue. However, we feel it's very important that we not 19 allow the study of those issues and the speculations and 20 concerns that arise as we proceed through that research to 21 divert important indus;ry and NRC rescurces away from the 22 mainacream effort of doing the individual plant evalutations and 4
22 bring.'Dg into play acress the industry all of the informatlon 14 we nave learned over *.be loot seven or eight years of intensive 0 25 study of this area.
Heritago Reporting Corporation -
(202) 638-4888
~ _. , , - .
28 1 Thank you very much.
l 2 MR. HULMAN: Terry. Terry.
3 STATEMENT OF TERRY PICKENS, SENIOR NUCLEAR SAFETY 4 SERVICES ENGINEER, NORTHERN STATES POWER 5 MR. Pickens. For those of you that don't know me, 6 my name is Terry Pickens. I'm from Northern States Power, and 7 when I talk about Janecek, who couldn't make it here today, for 8 those of you that know him, not many people can probably really ,
9 speak for Bob Janecek, I guess what I wanted -- when he said 10 that he couldn't make it, he talked to me about possibly making 11 this presentation, I thought that it would probably be a good 12 idea and I thought-it might even be a better idea for me to 13 make it than for him to make it because some of the things that (V) 14 I wanted to talk about come from my perspectives through 15 participation in the work that the BWR Owners Group has done 16 and with the NUMARC Containment Integrity Working Group.
17 when Bob originally brought thir issue forward some 18 twenty months ago, I was at that time Chairman of the BWR 19 Owners Group, and I had been chairman for about two weeks, and 20 I've often accused Jack Fulton o.0 talking to Bob Benero and 21 getting him to agree to wait these issues out until after he 22 was out of office so that he wouldn't have to address them, but 23 he denies it flatly.
24 From that involvement or during that involvement, the 25 early involvement, I had contact with a lot of my peers and Heritage Rep rtinc Corpdration (202) 62814888 en. , no
l 29 1, with a number of utility managers and executives trying to
( 2 figure out what we ought to do in response to the narrow 3 issues.
4 A number of the points have already been pointed out 5 by Bill Rasin in what some of the early discussions had to say, 6 and that was don't forget our plants are safe-enough, don't 7 forget the severe accidents that we're talking about, 8 challenging the containment, our low probability events, you ,
9 don't want to forget the context in which we're studying these 10 things. It is an important issue for us to go after. It is an 11' important issue to pay attention to, and there's going to be 12 new information that develops out of research, out of refining
[) 13 our analytical techniques, and that we need to pay attention 14 to.
15 But let's not get so caught up in the moment as we 16 look at is it going to melt through the liner and those types 17 of things as to forget that the probability of this happening 18 is sufficiently low that we don't have to go shut down our 19 plants and react to it in some other way.
20 Well, through that participation, and I won't rehash 21 a number of things that Bill said in terms of the low 22 probabilities and those things, I gathered a number of insights 23 of the'way that we ought to be viewing these things and just 24 really what we ought to do. What's the approach that we ought 25 to use when we get into looking at severe accidents.and how Heritage Reprting Corporation (202) 628-4888
30 1 they're going to challenge the containment.
( 2 Some of the gaps and things that I think we're trying, 3 to. fill now are what's the criteria for improvements that we're 4 evaluating, that we're going to weigh the information against
~
5 and decide whether or not we want to do it. Should it be a 6 cost benefit. Should it be risk. Is it an overall integrated 7 risk picture. Do we assume the core is on the f'.cor and then 8 zipping across the prob' ability of one in terms of core melt. .
9 How were we really going to approach it.
10 Also saw from talking to people that what we're 11 talking about is a very complex phenomenon and the way things 12 fit together, the way systems fit together, the way operators 13 utilize those systems, the way the containment is going to act
{a}
14 and there is a number of areas that we need to identify and 15 further analysis and research comes in.
16 But when we get all the answers, you can't just look i 17 at portions of it and say I got the answer. You've got to go 18 back and integrate it all together and look at how the whole 19 process goes once again.
20 The frequency of the challenge not just on an overall )
21 risk basis, but on some of the individual elements, we can say 22 station black-out is a dominant accident sequence, and I have 23 seen that in some reports, but even within station black-out 24 sequences, there is different types of sequences that you see l
25 through a station black-out event. It doesn't always progress IIeritage Reprt-61;g (202) Corporation
-4888 I
I l
1 31 '!
1 the same way and just saying that, you know, seventy percent of
() 2 the accidents are caused by station black-out, a very small 3 percentage of those are probably severe enough to really 4 challenge the containment.
5 The consequences of each of the challenges and how it 6 progresses is important and, then, finally, after all of that, ,
7 going back to the point of the integrated analysis being j 8 necessary. .
9 1 have developed a list of questions regarding 10 containment performance that as we sit here today and as we 11 have done in our past evaluations, that you sit back and ask 12 yourself. How likely are the conditions necessary to challenge 13 the containment. How likely are the potential' release
)
14 pathways. What's the severity and timing of those consequences 15 through each pathway. Are you going to get an early severe 1
16 release or are you going to get a late minor release.
17 Can the pathway be made less likely, and if we do 18 make it less likely, if you do go out and attach some of the 19 problems that we have seen, what becomes the next pathway. l l
l 20 After you've solved one problem, what problems have you created 21 for yourself next. What becomes the next dominant pathway for 22 release through the containment and what are the consequences 23 of those subsequent pathways. And, again, how does this all 24 fit in.
25 The bottom line question being, does the improvement Heritage Reprting Corporation (202) 628-4888
l 32-1 terminate the event, delay the event or at least reduce the O( ,/ 2 event. Those are the things that I think that we're looking to 3 do, to stop it, to either delay it or to reduce the overall 4 effect.
5 You need to consider competing risk. These are 6 complex phenomenological issues. We're using systems and we're 7 using systems that were designed to both prevent and mitigate.
8 When do you make a decision that a system that you're utilizing _ .
9 is going from the prevention mode to the mitigation mode and 10 are you detracting from something.further that you might have '
11 done trying to prevent it.
12 These are all things that you need to consider. It's
() 13 an iterative process. You need to keep going back. You need 14 to keep looking at l't . You can'.t decouple the analysis from 15 the boundary conditions which you enter into it with and their 16 likelihood.
17 From that, I have developed my own point of view in ,
18 terms of how we ought to step through this process and, again, 19 I guess I go back to -- I can't say this is an owners group 20 position, but this is a position which, in my interaction with 21 peers, executives, upper management, a number of people at the 22 NRC, sitting with many hours when this first thing came out 23 with Bob Benero and talking about how we ought -- a philosophy 24 of how we ought to take a look at these things, and I think 25 it's important to remember as we move through these today.
Heritago Re p rting Corporation (202) 628-4888
~r ,-
33 1 First off, let's not forget that the risk of these 2 challenges is low, and these challenges that we're looking at 3 all develop in different ways and are very complex. We need to 4 adopt an overall risk perspective, a perspective that takes us 5 starting with normal operation, walks us through the 6 challenges, those types of things, and gets us all the way out 7 to the severe accident.
8 Establish a regime of study in that context. I think ,
9 that we can make a big mistake by merely focusing our research 10 on what happened once the core is out of the vessel. There's a 11 number of things that you have to consider in terms of the 12 event, what's the mechanism for it ejecting out of the vessel,
() 13 what's the make-up of the core material, different things like 14 that.
15 You need to have the whole thing following through.
16 I think that, and this is a point that I guess our company 17 feels particularly strongly about, is I think that it's l l
18 essential that we coordinate the industry and the NRC research 19 efforts.
20 We have been spending a lot of money on severe l
21 accidents and research in severe accidents, and it seems as l
22 though we could do something, and I don't think that this is 23 closing us when I talk about this, but I think that it would 24 behoove us all to spend our money to get together prior to
'25 doing some of the experiments and developing some of the Heritage Rew rting Corporation (202) 628-4888 I
i l
34 l
. I analytical techniques and look at the conditions and 2 assumptions-that we're putting into this and ensure ourselves 3 that we're in agreement so that when we get results, we don't 4 have to come to a forum where we say it's X and you say it's Y 5 and we get into an argumentative situation.
6 I think that this kind of meeting is good. I think l 7 this is really a positive step in the right direction in terms 8 of talking about where we're going with severe' accident ,
9 phenomenon and sharing the infc rmation and developing a 10 stra,tegy to go from this point forward on what we ought to do 11 to close out the issue.
12 I don't think that we can ignore the plant specific 13 vulnerabilities. Once we have done research and developed 14 these things, we need to use analyses to understand the plant 15 specific vulnerabilities. The differences that we have in each 16 plant that wa've developed in terms of when we put our des'igns '
) 17 together for the containment, how we implement emergency i
18 operating procedures, and the whole gambit when you talk about 19 how a severe accident is going to progress.
20 And, then, fi a ly, the plant specific evaluations 21 once you've gotten to that point where you've identified the 22 analyses to understand the plant specific vulnerabilities, ;
i 23 you've come up with the inforrnation through research to put l
24 into those, and developed the whole program, and I think that's O 25 the point at which the industry has gotten to, which Bill Rasin Heritage Reporting Corporation (202) 628-4888
35 1 alluded to, not alluded to, stated that we feel that the IPEs 2 are very important.
3 We need to continue to learn from the research, and 4 then go forward with plant specific evaluations on that basis 5 because I don't know that this is a problem that we can hand'le 6 generically by regulation and that type of. thing.
7 So, those are some of the insights which I have 8 gained through participation in this process and since we ~ .
9 started this process back about twenty months ago, and that's 10 about it.
11 MR. HULMAN: Steve Sholly of MHB Associates.
12 STATEMENT OF STEVE SHOLLY, MHB ASSOCIATES 13' MR. SHOLLY: I'm here representing the Massachusetts 14 Attorney General's Office.
15 MR. LANE: We're going to check the mike system for ,
l ,
16 just a second.
l 17 MR. HULMAN: If you will hold it.
18 (Pause) a 19 MR. HULMAN: Steve Sholly.
, 20 MR. SHOLLY: Unlike the last two speakers, I will try 21 and give a little perspective here. I'm sure you realize that I 22 it will be a little bit different.
I 23 Following tt'e accident at TMI, now nine years ago, 24 there was a proposal that would have done something like an 25 IPE, it was called NREP or IREP, which would have been done in Heritage Reporting Corporation (202) 628-4888 i i l
e 36 1 1983. We're here about five years after that, just embarking on
() 2 a similar process. ;
3 From that* perspective, our clients at this point in 4 the Massachusetts Attorney General's Office thinks th.s is 5 about overdue.
. 6 The IPE process, we see this as an integral part of 7 what's going on here. The containment issue is sort of one 8 part of the problem. The other part is the accident sequence .
9 frequency issue. We think they both have to be addressed and a
10 concern about the IPE process as it's laid out in the IPE 11 methodology.
i 12 Now, I'm going to try and separate the methodology ,
4
() 13 from the_ concept. The concept is doing an IPE, looking for 14 plant specific vulnerabilities is a very good one. The IPE 15 methodology per se, though, we have some concerns about because 16 of its limited scope, limited depth, and also its cost, and 17 it's our understanding from talking with some folks in NRC and l
- 18 some in industry that the cost of the IPEs that have been done i
19 so far are beginning now to approach what it would take to do a 20 full-scope Level I PRA.
] 21 Now, in view of the fact that the methodology for 22 Level I PRA is relatively well defined, that the limitations of 23 them is rather well understood, and that the depth that you're
! 24 forced to go into in doing one of those is much greater than j
j 25 IPE, we would suggest to you that now is the time to consider Heritage Re orting Corporation (20 ) 628-4888
37 1 whether you want to stop with the IPE methodology or just go on 2 and do a full-scope PRA.
[)
3 ' For a variety of reasons, particularly in this 4 accident management context in which we're here now, we think 5 that the full scope PRA is the way to go. We.would like to 6 suggest to you that you consider that and not stop short of the 7 IPE methodology.
8 Turning to the Mark.I issues, I'll agree with the .
9 last speaker or two that the issues are by and large not new.
10 The conta'inment dry well melt through issue was identified in 11 1985, if not before. So, it's been around for a few years now, i 12 too.
13 Given that state of affairs, we hope that we can get O 14 past the point now where we quit arguing about NRC computer 15 codes or EPRI computer codes or IDCOR computer codes and get 16 down to the business of trying to decide what, if anything, 17 needs to be done.
18 Notwithstanding the Attorney General's desire for 19 something to get done rather soon, there is sort of a five-step i
20 approach that one could take, and I think this is moderately 21 consistent with what's going on now.
22 For instance, take a look and see what can be done in I 23 the short-term and some utilities are moving in this direction, i
24 Boston Edison is one of them, and they are in somewhat of a 25 unique position having been shut down for so long, I realize l Heritage Re w rting Corporation l (202) 628-4888
-s +
38 1 that, but there are others who are moving to implement what are
(} 2 really some common sense changes, making sure, for instance, 3 that you can rather easily get service water into the low l 4 pressure injection system to use it to save the core, if LPCI 5 fails.
6 Those sorts of things, you don't need to wait around 7 on NRC for. You don't have to wait around for another meeting 8 like this. You go on ahead and do those, and we'd encourage ,
9 you to do those. You don't need a detailed PRA to tell you 10 that that's a good idea. You do those things that you can 11 identify in the short-term.
12 The second step is to do the full-scope PRA, and when 1
() 13 I say full-scope, I'm including external events, detailed look 14 at operator procedures and possible mitigation steps that.the 15 operator can take, the whole ball of wax, essentially.
16 Third, once you've done your PRA, you identify those 17 plant specific things that you can do that weren't obvious 18 before and do them, and in so doing, you take account of your 19 PRA and revise it so that it now looks like your modified plant 20 looks.
21 The fifth step or the fourth step is to reach accord i 22 with NRC. Obviously, some of these things like containment l 23 venting is going to require agreement from the NRC before it !
24 can get implemented. I would hope once that the detailed work, 25 the PRA, the phenomenology studies and whatnot are done, that fleritage Rewrting Corporation (202) 628-4888
i 39 1 utilities and NRC can move forward rather quickly after that to ,
2 reach accord on the changes that are going to be made.
3 The fifth step, and I think this is an important one !
4 that often gets overlooked, and that is to then use your PRA, l 1
5 use your revised assessment of what your plant looks like. l 6 Integrate it into an overall tool that you can use for severe 1
7 accidents, for training, for emergency procedure modifications, 8 for dealing with new issues as they come up. Make it a living ,
9 document. Don't just do it and sit it on the shelf and wait 10 for something to happen and come back in five years and say, my i 11 God, this doesn't look like our plan anymore. You've got to 12 keep it up to date.
- 13 In closing, I would like to remind everybody that, 14 you know, whatever happens here with mitigation possibilities i 15 and the containment issues, one need remember that the core 16 melt frequency part of this is still an open issue in a lot of 17 plants. I think it's particularly true with Mark I's. There l 18 are not a lot of Mark I's that have been studied, at least in
) 19 publicly-released documents.
20 I am not as confident that it can be readily stated 21 and confidently stated that the frequency of accidents is low.
22 There have been enough PRAs done and enough surprises come out l 23 of those PRAs that one needs to remember that there is a
- 24 possibility out there that there are a few plants without I
25 liars. You've got to look for those.
l 1
Heritage Reporting Corporation !
(202) 628-4888 i
, I
40 1 That's all.
(I 2 STATEMENT OF JERRY HULMAN, CHIEF, SEVEFE ACCIDENT 3 ISSUES BRANCH, RES l- 4 MR. HULMAN: I'm aoing to try and give an overview of 5 the issues, basically a summary, and a little background for 6 eve rybody . j 7 Let's go to the first one. Then, I'll take care of
- 8 some housekeeping matters. The~ objectives of the. workshop. ,
l l
9 What we want to do principally is try and narrow and focus the
! 10 issues. Where there is disagreement, we want to identify what >
11 the disagreements are and what the ranges of disagreement. ,
12 On the technical issues, on the challenges, on
()
~
13 containment failure modes and on potential mitigation l 14 improvemen;s.
15 DR. SPEIS: Jerry, do you have the microphone? l 16 MR. HULMAN: Can you hear me in the back?
- l 17 DR. SPEIS: It was on before. I J
l 18 MR. HULKAN: Yes? No? Let me try it again. Is it I i
19 on now? John, can you hear me? Did I hear John? I'll speak l
)
20 up as loudly as I can. I'm not sure the microphone is working.
21 Now, it is. Okay.
) 22 There are issues that we'd like a little discussion
)
23 on related to risk levdis, prevention versus mitigation, and i
24 plant specific differences that we may not be able to capture i
- 25 with this discussion. So, let's go on to the next one.
Heritage Reprting Corporation
- (202) 628-4888 i
41 1 I want to present a couple of photographs courtesy of
( 2 Philadelphia Electric that give you an actual view of the dry 3 well of Peach Bottom.
4 I think i:'s important from a perspective point of 5 view for some of the discussions that you have an. idea of what 6 a dry well looks like, what a vent pipe looks like, what a vent 7 pipe cap looks like, and I'm going to try and illustrate that 8 by pointing at this photograph. .
9 This area down here is the floor of the dry well.
10 This is the vent pipe and this is the vent pipe cap. All of 11 this material, these spikes, all of these spikes and duct work 12 and pipes up here are in the dry well. It io very crowded.
() 13 Let's have the next one. This is_another photograph 14 of the same area. It's not quite as visible. This is the vent 15 pipe going down to the suppression pool. This is the cap to 16 the vent pipe. The floor of the dry well is right here.
17 Right here is the door, the door that goes into the 18 pedestal area underneath the reactor. The reactor sits up some 19 nineteen or twenty feet above that floor.
20 We've talked about probabilities. This is the j l
21 probability information that the staff has available to it for l 22 Mark I's. This gives you a little insight into the ranges of 23 core melt probabilities that have been estimated. All of the 24 PRAs that the staff has from industry plus those we've financed 25 ourselves are of variable quality and variable age, and if you Heritage Reprting Corporation (202) 628-4888 <
42 1 take a look at the core melt probabilities, they have quite a jlfh 2 range. From two times ten to the minus three to eight times 3 ten to the minus six.
4 What's interesting about the eight times ten to the 5 minus six that was published with draft NUREG 1150 is that 6 seven times ten to the minus six came from station black-out.
7 Next slide, please. I've tried to divide the 8 challenges to containment into several different groups and .
9 indicate their relative conditional probability. That is, 10 given the core melt accident that could result in vessel 11 failure, these are the probabilities, relative probabilities 12 and relative orders of importance of the various challenges.
l 13 I have also listed the categories of potential 14 improvement in mitigation that we want to discuss at this 15 meeting.
16 Next slide. Again, this is the slide that Dr. Speis 17 presented that shows the relative pressure levelu estimated for 18 containment failure. My understanding is that this is 19 principally based on Peach Bottom and there may be variations l
20 from one plant to another.
21 Some of you may not be familiar with some of the 22 numbers that go with a Mark I Containment. This gives y>u an 23 idea of some of the design parameters, some of the dimensions,
,s 24 some of the volumes associated with a Mark I reactor.
/ ;
\' 'l 25 For example, the estimateu free volume in the dry Heritage ReWrting Corporation (202) 628-4888
43 1 well is a 150,000 cubic feet. It's small by comparison to some fl
( ,/ 2 PWRs.
3 Can I have the next silde? There is some more 4 information on a Mark I and there is also what we might call a 5 cartoon that gives you a cross-sectional view of what a Mark I 6 reactor and containment looks like.
7 The containment is basically all of this area and 8 this is the suppression pool. This area is the dry well. This ,
9 is the reactor vessel. The little dots are reinforcea 10 concrete. l 11 Next slide. I've tried to identify or summarize the 12 issues associated with each one of the challenges. I'm trying
(')
V 13 to get a focus out of the meeting on things like the degree to 14 which the invert in-vessel vertical channel boxes could block, 15 could be blocked and influence the melt down process.
16 The fraction of Zirconium that's oxidized in the 17 vessel is very important, and the rest of these items that are 18 listed here are also important. I am hoping that out of this 19 neeting we can get additional information on these subjects.
20 Let's go to the next one. I don't want to go over 21 them in detail. I will have copies of these viewgraphs 22 available. You can look at them at your leisure.
23 Let's go on to early over-pressure. Same kind of 24 issues associated with in-vessel and ex-vessel pnenomena.
O 25 There are additional ones. There is concrete dogessing. There Heritage Rewrting Corporation (202) 628-4888
44 1 is heat flux to water to the shell and to concrete. There is 2 ahell pressure and temperature limits and location of failures.
3 There is questions relating to the benefits of venting and 4 spraying.
5 There is questions about the benefits one might gain 6 if we could use the fire protection system in the reactor 7 building in the event of a containment failure. Could you 8 attenuate the fission products and basically reduce the amount ,
l l
9 of fission products leaving the plant. l 10 Next slide, please. Again, with late over-11 temperature, the pressure challenges. I have listed the issues 12 containing the bypass challenges. If you remember from some of )
i I
<~g 13 the PRA studies, containment bypass, except for liner melt (J 14 through, was given a relativ'ely low probability.
15 Questions arise, do we continue to believe that.
16 Should we do anything to make sure that comes true.
17 Next slide, please. Hydrogen improvements is 18 interesting. When the issue of hydrogen control was raised a 19 couple of years ago, the emphasis was on the twenty-four-hour 20 period allowed during start-up and shut-down, when the 21 containment would be deinerted. ,
22 Boston Edison came in, speaking to that issue, and to j 23 two other issues on hydrogen control'. The adequacy of nitrogen !
24 for inerting for long-term station black-out and sneak sources 25 of oxygen that one might get from back-up compressed gas i-lioritage Reprting Corporation (202) 628-4888 1
. _1
45 1 systems used to operate valves.
f"/
1
(, ? Boston Edison thought they were important and 3 proposed some improvements. These are the kinds of questions 4 on that improvement we want to try and address at the meeting.
5 Let's go on to the next one. Containment opray 6 improvements. What do we get in the way of reduced fission 7 products. What do we cet in the way of either retarding-or .
I 8 preventing shell melt through, if anything, from the use of the l l
9 spray.
- 10. If the spray flow is reduced, there are analyses of i
11 record that are related to the design bases for every plant. l 12 The question is, could we be causing a problem for lesser g .
(j 13 accidents.
14 Next slide. Venting. Venting has become very 15 controversial. There are benefits potentially from venting and 16 there are negative benefits. There is a down side. We want to 17 look at both the up side and the down side. There are 18 questions relating to the adequacy of existing hardware as well 19 as some potential improvements.
20 Next slide. Core debris control. On the assumption 21 that you melt through the vessel, questions arise, can you do 22 anything to confine it to the dry well and the pedestal area,
~
23 or is there anything you can do in the Torus room that might 24 help, or in the reactor building. Those are the kinds of 25 questions we would like to. hear some comments on.
Heritage Rew rting Corporation (202) 628-4888
i l
46 f 1 Next slide. The ADS system, the automatic If, in a station black-out, it
) 2 depressurization system.
3 doesn't work, and you have a high pressure release from the 4 vessel, there are questions of early containment failure.
5 There are questions of direct containment heating, and there 6 are varied views on that subject.
7 It's also of interest that at least one utility has 8 already modified their automatic depressurization system to ,
l 9 increase the reliability during station black-out. I
. 10 understand the staff hasn't reviewed it yet, but there are 11 questions about whether there is the right thing to do or 12 necessary.
l f/-) 13 And, lastly, training and procedures. There are !
(_e' 1 14 questions about whether better procedures and better operator )
1 15 training are necessary. As both Dr. Beckjold and Dr. Spels
)
16 indicated, we have a program getting started on accident 17 management.
18 I'm not going to go into that much at this meeting, 19 other than if the opportunity arises to try-and seek some l
20 industry comment or some public comment on the adequacy of I 21 existing procedures.
22 John, next. Let me get to housekeeping before we 23 start the session.
24 We mailed an early copy of the agenda to everybody.
25 We have since revised it several times, including verbally at Heritage Rewrting Corporation (202) 628-4888
. . . _ ~ _ _
I l
i 47 1 the meeting. Our secretary, Ms. Kondulis, in the back has
() 2 copies of the agenda if you already haven't gotten them. If 3 you haven't signed in, ple~ase sign in. There's a sign-up' 4 sheet. I think it's either being circulated or Ms. Kondulis 5 has it in the back.
6 If you want copies of all the material that's 7 presented at the meeting, it will be made available about noon 8 on Friday. 'We're going' to have it all copied and copies will .
9 be available for everybody.
10 I'm asking all speakers, any of the material that you 11 used in your presentation, make sure that one copy gets to our 12 court reporter and one copy gets to Ms. Kondulis. The court 13 reporter will bind copies of that material into the record and (D) 14 we'll put that record in the Public Document Room. We want to l
15 have a verbatim transcript to make sure that we capture all of 16 the views as they're articulated.
)
17 If you have to leave for any reason before noon on 18 Friday, leave your name and mailing address with Ms. Kondulis 19 and we'll see that you get a copy of all the hand-out material.
20 If we get into an area where we think it may be 21 fruitful, we might ask several people to get together to see if 22 they can find a common ground to narrower focus the issues.
23 Friday, about noon, we hope to summarize and get out 24 of here well before 4:30, if possible. If not, we're prepared O 25 to stay as long as necessary. I realize it's the week-end, but Heritage Re w rting Corporation (202) 628-4888 .
48 1 we're going to try and get through the meeting.
'\
7/
(, 2 Now, any speakers that wish to have viewgraphs made 3 or additional material presented and would like some 4 secretarial help, Ms. Kondulis.is available in the back of the 5 room.
6 Rose Marie, will you please stand up and~1et 7 everybody see you? Thank you. So that if you need anything in 8 the way of typing or viewgraphs, see Ms. Kondulis. .
9 Last, as I indicated in the invitation, I'm going to 10 try to keep each speaker to,about fifteen minutes. I have no 11 hook to pull the speakers off stage, but I'm going to try and 12 keep it to fifteen minutes.
[s-13 With that, let me call on -- I've been reminded that 14 viewgraphs from Dr. Beckjord's presentation, Dr. Speis' 15 presentation, and mine are available at the back of the room.
16 You can pick those up at the break.
17 With that, let's take our first speaker. Our first 18 speaker is Mohsen Khatib-Rahbar from Brookhaven National 19 Laboratory.
20 Mohsen.
21 STATEMENT OF MOHSEN KHATIB-RAHBAR, BROOKHAVEN 22 NATIONAL LABORATORY 23 DR. KHATIB-RAHBAR: Thank you.
g-) 24 With Jerry's introduction, what I hope to do is to V
25 try to summarize for you the issues related to late containment Heritage Rep.rting Corporation (202) 628-4888
_u ..
49 1 challenges.
( ) 2 There was a report which I assume has been 3 distributed to most of you, if not already will be available on j l
4 Friday, in which we have summarized all of the containment 5 challenges that Dr. Hulman has summarized, and here I basically 6 want to give you the highlights of where we think what the 7 issues are and what are the concerns.
8 I'll first describe what are late containment i 1
9 challenges, then I'll get into the definition of the various 10 challenges, and I'll identify as we see the sources of ,
l 11 containment load and the uncertainties associated with those l l
12 and then discuss the containment failure modes aa perceived by 13 NRC as well as IDCOR and then I'll summarize. .
g 14 Containment challenges, Jerry has already covered 15 this. They are perceived as being early containment failure as l 16 well as late containment failures. Early containment failures 17 deal primarily with over-pressurization /over-temperature 18 failures occurring prior to or right after the actual reactor 19 pressure vessel failure.
20 The issue, such as direct containment heating, 21 containment bypass type failure modes, such as direct liner 22 attack on the shell, shell attack by the ccre debris or in LOCA 23 bypass frequencies. That particular failure mode is not going 24 to be addressed specifically as an agenda item, but I think it O 25 could be a question of discussion because that inter-system IIeritage Reporting Corporation (202) 628-4888
50 1 load pressure does not challenge containment or in different
( 2 ways because the containment is assumed to fail by definition, 3 and it's perceived by two reductions of frequency of that 4 particular event. One'can eliminate or reduce the challenges 5 associated with inter-system LOCA.
6 I'll be discussing base failure modes, specifically 7 over-pressure /over-temperature failure modes. Base failure 8 mode is considered to be much less likely and we're not going ,
9 to be discussing them, although they are addressed in the issue 10 characterization report.
11 What are delay containment challenges? As I 12 indicated, generally slow over-pressurization or heat-up of the
() 13 14 dry well atmosphere due to ste,am, if it's a flooded pedestal condition and/or non-condensable gasses as well as direct 15 energy transport from the melt into the containment.
I 16 The containment can fail with different mechanisms ,
i 17 depending on conditions in the containment. If a leakage in 18 the containment is sufficient to prevent further 19 pressurization, there will be primarily a small leak from the j 20 containment. If the leak inside is not sufficient to turn 21 around the pressurization, this could eventually turn into a 22 major rupture of the containment, and if the pressurization is 23 very rapid, rupture can occur.
24 The late containment challenges are primarily 25 resulting from containment conditions when they occur after the IIeritage Reprting Corporation (202) 628-4888
v -
51 1 actual pressure vessel failure. If the containment has not sa
( j/ 2 failed already due to other mechanisms, early failure at the 3 reactor pressure vessel failure can occur, the sources which 4 can contribute to containment loads are primarily steam. If 5 there.is water present to cool the core debris and the flooded 6 pedestal involvement, boil off of the over-lay water can 7 increase the steam partial pressure in the containment.
8 Beyond the actual water vapor from the concrete ,
9 decomposition can also be another major -- another source of 10 steam, not a major one, to the' containment.
11 The major source of loads are coming from non-12 condensable gas evolution resulting from interaction of the
() 13 molten core debris with the concrete.
14 There are three different types of concrete 15 aggregates. Linear aggregates, salacious aggregates and the 16 limestone common sand aggregates. The limestone concrete is 17 very gaseous. Typically, there is forty percent weight of the 10 concrete is gas. Therefore, compared to the other types of 19 concrete, it's perceived to cause the major challenge and 20 source of non-condensable gasses in the various containment 21 structures.
22 The other source of containment load is thermal 23 energy. Thermal energy can be transported to the containment 24 either directly as a result of high-temperature gasses, 25 ovolution of high-temperature gasses through the melt and/or Heritage Reporting Corporation (202) 628-4888 9 . --- ,
52 l' aerosols which are being generated during the core concrete 2 interaction which are quite hot and they can pose a heat source f'N%>)
3 to the containment environment.
4 But for direct thermal radiation and convection heat 5 transport to the containment and that is another major source 6 of thermal energy transport to the containment.
7 Combustion of combustible gasses is also a concern.
8 Although Mark I Containments are inerted, the inerted -
9 conditions can be perceived. For example, during start-up or 10 shut,down conditions at low power levels, normally twenty-four 11 hours time period is allowed for deinerting. Water containment 12 is deinerted.
'^T 13 Also, there could be other means of inducing air
/
14 and/or oxygen into the syster 4, ,
i the severe accident. For 15 example, there is an air-operated system instrument as well as 16 a valve which may be operated by air pressure.
17 The major uncertainties in the various loads as were 18 described in the previous transparency can be broken up 19 basically into two different sources of uncertainties. The 20 core melt progression uncertainty which basically provides the l 21 conditions for ex-vessel interaction. The quantity of the core 22 and its composition can play a major role on the time which is 23 .available to generate conditions which are conducive to failure 24 in the containment. I I
f')
~/ 25 Temperature of the core is another source of i
lieritage Rewrting Corporation I (202) 628-4888 !
53 1 uncertainty. The timing at which the reactor vessel fails is O
(/ 2 also a major uncertainty.
-3 The uncertainties related to the. core and concrete 4 interactions are primarily from concrete decomposition 5 temperatures, depending on the type of concrete. There is a 6 wide uncertainty in what is the actual decomposition 7 temperature of the concrete.
8 Core and water heat transfer. If there is water 9 present, whether-the core crunches or not and what is the mode ,
10 of heat transfer due to the over-laying water pool.
l 11 The heat transfer processes can also -- are also 12 major sources of uncertainty. In fact, that's one of the major
() 13 differences between NRC and IDCOR, computational predictions, 14 because of the split between heat transfer, upward heat 15 transfer, and heat transfer into the concrete is a major source l 16 of uncertainty.
17 The corium on the floor is something which is also a l 18 major source of uncertainty because it depends on the quantity 1
19 of the core which is available, its temperature and how it can j l
20 flow into the pedestal and spread over the floor.
21 Chemical reactions. How much of the decomposed 22 gasses can bypass the melt and what is the extent of that 23 uncertainty is also important.
24 The containment challenges have been looked at over 25 the last several years. As part of the containment load lieritage Reporting Corporation i (202) 628-4888 l
l 54 1 working group, I think, four or five years ago, there were a ir's/,l 2 number of parametric calculations performed to try to identify 3 what are the major challenges to containment.
4 At that time, it was concluded that there's normally j 1
5 a long time to containment failure for the basalty type of l
s 1 6 concrete aggregates because of the fact that there's less l l
7 gasses available to be released to the' containment, and there's l
8 normally any where between eight to ten hours to fifteen hours ,
9 of time available for containment failure, and this containment 10 failure was primarily thought to be caused by over-11 pressurization because temperatures did.not reach very high 12 levels at those time frames.
/)
V 13 _ However, for limestone type aggregates, because of 14 the large evolution of gasses, the time to failure is normally l
15 believed to be chorter because of much larger partial pressures 16 associated with larger quantities of gasses which are being 17 released to containment.
18 One of the major concerns is what is the containment 19 failure mode because that impacts the timing which is available 20 if the containment is going to fail due to a late failure 21 challenge.
22 The IDCOR analysis calculated a much higher upward 23 heat transfer from the core to the dry well atmosphere relative 73 24 to the heat transfer to the concrete and the side walls V 25 downward heat transfer versus what the NRC calculation IIeritage Re orting Corporation (20 ) 628-4888
-r e
t 55 i i cap 6bilitius show. ,
/') 2 As e resuit, IDCOR calculated that their containment (f l 3 failure primarily d.ue to an over-temperature failure mode while 4 the NPC calculations showed that the late containment failures 5 was due to over-pressuri?ition because the pressure is always 6 much higher relatively and closer to the failure limits of the 7 containment relative to the compartment temperatures.
8 As an example, I'm showing a typical calculation , j 9 which was performed in support of NUREG 1150, using the NRC 10 source term code package. Here, you'll see that the 11 containment is calculated to fail roughly about two and a half 12 to three hours after the actual pressure vessel failure, which
/' \ 13 was occurring at 780 or 790 minutes into the accident. This is O
14 a long-term station black-out where ADS was assumed to fail, 15 not to be operable, as well as there was six hours of battery 16 power available for running the turbine-driven pumps.
17 The containment was assumed to fall in this 18 particular case in a 130 PSI pressure and this is specifically 19 for the Peach Bottom Mark I Containment. If you look at the 20 comparable temperatures which are being calculated by the 21 source term code package, at the time of the assumed failure 22 conditions, the maximum temperature in the dry well is roughly 23 about 400 odd degrees.
24 Now, if you look at somewhat of a comfortable O 25 calculation, although not exactly for the same sequence, for Heritage .'te mrting Corporation (202) 628-4888 T - , , , . S w
53 1 the IDCOR analyses, one finds out that the time for failure is s
(_ 2 between the' actual pressure vessel failure and the containment 3 failure is much longer of the six or seven hours.
4 However, the containment still fails but not due to 5 an over-pressure condition. On the next slide, you'll find out i
6 that the temperature in the compartments have increased to much 7 higher levels than the NRC codes predict and, in fact, I think 8 the stated criteria was 1200 degrees F in this case. ,
9 This is an important consideration in the way that l'0 these type of accident sequences and challenges have to be 11 mitigated. Whether one will have to resort to sprays or 12 combinations of sprays and venting. It,'s important to sort of
/ 13 resolve the issue and try to see what is the most likely
\.)
14 condition in the containment.
15 To summarize, I think, in general, it's well known 16 that the significant time before containment failure and tnis 17 time depends on the logic sense and the magnitude of the ;
I 18 uncertainties resulting from core and concrete interaction 19 primarily.
20 There is large uncertainties with regards to failure 21 mechanisms and also whether the failure is going to be due to 22 over-temperature /over-pressure or a combination of the two. I 23 guess the presentation by Dr. Speis indicated that there is 24 both a pressure and temperature dependence on the containment 25 performance, and that's an impor't ant consideration here.
lieritage Reprting Corporation (202) 628-4888
l 57
- 1 The uncertainties in these type of pressures and ,
() 2 temperature loads are primarily due to uncertainties in 3 calculating non-condensable gas accelerations or the extent of 4 molten core concrete interactions, and also to the degree of 5 upward heat transfer versus cycle heat transfer has been 6 perceived.
7 That is al1 I have to say.
Are there any questions 7 8 I don't know if you want to take questions now, Jerry, or wait. ,
9 MR. HULMAN: Wait.
10 DR. KHATIB-RAHBAR: Okay. .
11 MR. HULMAN: Paul Hill from PP&L.
12 Paul.
() 13 After Paul, we will ask some questions of the 14 speakers or people in the audience on this particular issue.
15 STATEMENT OF PAUL HILL, PENNSYLVANIA POWER AND LIGHT 16 MR. HILL: My name is Paul Hill. I'm here
. I 17 representing Pennsylvania Power and Light Company. !
18 Our plant is a BWR 4 Mark II plant. However, the 1
19 issues that are to be discussed here today, we belle /e, will j 20 affect us every bit as much as they affect the Mark I and for 21 that reason, we're here to express our views on the subject 22 being discussed today.
23 My plan is to discuss the answers that we have -
24 provided to the NRC on questions 3, 6 and 5 on pages 39 ar.d 40 25 of the Mark I issue characterization paper which was Heritage Rew rting Corporation (202) 628-4888
l i
1 58 l 1 distributed.
() 2 On the first viewgraph, I'd like to point out that in 3 our view, it's extremely important to use an accurate-4 representation of operator error in assessing the dominant 5 accident sequences for a nuclear plant _and, in particular, we 6 think that it's very important to keep in mind that there is a 7 great distinction between operator errors that cause initiating 8 events or cause equipment unavailability in response to an .
l 9 initiating event from the type of error that operators may make ,
10 in response to an initiator accompanied by equipment failure.
11 We believe these two things are not at all. alike. We 12 think that it's important to give close attention to procedures
/'N 13 and training for emergency operating procedures that are U
14 intended to respond to a severe event in the plant.
15 We believe that if that's done, and if a careful risk 16 analysis of the plant has been done that identifies tne 17 dominant challenges to the plant, that the severity of damage 18 to the plant or the frequency of severe damage to the plant can 19 be reduced to what we believe is a truly negligible level.
20 Now, PP&L has identified over the past two or two and 21 a half years something on the order of a dozen operator actions
, 22 in response to a severe accident that are highly specific to 23 the Susquehanna plant, but these dozen actions, if you can 24 assure that the operators will take them, not only sharply O 25 reduce the frequency of severe damage to the plant, but also
. lieritage Reporting Corporation
'(202) 628-4888
59
, 1 has a radical impact on the nature of that damage and, in fact,
() 2 what you find is that it greatly reduces the severe risk 3 challenges. '
4 In doing our study of Susquehanna, one of the things 5 that we found that was quite interesting to us is that the 6 nature of the dominant challenges to the plant and the actions 7 that we can take to respond to them tend to be very plant
, 8 specific in nature and, in particular, they tend to be very .
5 much dependent on the design of the support systems in the 10 plant.
11 For that reason, we think that if the question of 12 conta'inment integrity or the more general question of risk from ,
13 nuclear plant operations is to be addressed, _that the first 14 step has to be a very careful risk assessment for each plant 15 and then a very careful examination of emergency reeponse 16 procedures and operator training to assure that the proper 17 actions will be taken to minimize the chance for the 18 conditional probability of a severe event.
19 20 21 22 23 0
Heritage Re wrting Corporation (202) 628-4888 .
l l
60 1
/~ 1 PP&L in order to support some of the assumptions that N~lT L 2' we've made with regard to operator actions and the reliability 3 of those actions. We have conducted an initial series of 4 simulator experiments in order to make measurements of operator 5 fidelity in following procedures. That is, to try to get a fix 6 on what the probability that a procedural step will not be 7 taken or will be taken incorrectly. And also to get a fix on 8 what the time required for the operator to take the action is.
9 What we found is that we believe that procedural 10 error for a well trained crew is essentially negligible insofar 11 as the contribution to severe events is concerned. And we 12 believe that we use the Haneman HCR method for determining 13 operator response time, and we find that that methodology gives
() 14 a rather conservative picture of operator response.
15 (Slide) 16 We found that the use of conventional risk assessment 17 models and assumptions, particularly those having to do with 18 ATWS and those having to do with station blackout, for example, 19 loss of HPCI from high suppression pool temperature, results in 20 a rather severe distortion of what we believe the nature of the 21 true dominant challenges to the plant are.
22 And as a result, we believe that if that picture is 23 adhered to in attempting to respond to such challenges, we 24 believe there's a real danger that the truly critical areas 25 will be overlooked.
O', Heritage Reporting Corporation j (202) 628-4888 .
l 61 fg 1 (Slide)
V 2 And of greater' concern, if this diversion of 3 attention due to conservatism in the models is allowed to 4 happen, we believe that it might become a self-fulfilling 5 prediction of the risk of severe damage end severe' consequences 6 from plant operation.
7 We believe that it's very important to be as 8 realistic as possible in the assessment of the equipment ,
9 unavailability in the plant, and the assessment of the 10 reliability operator involved in these procedures successfully.
11 (Slide) 12 The kind of improper profiles that we're concerned 13 about are those that lead to an inappropriate accounting for
/O
(_j/ 14 stopping core damage in the vessel prior'to reactor vessel 15 failure. For Susquehanna, we made a very serious attempt when 16 we carried out our risk assessment starting in mid-1985 through 17 the end of '85, to attempt to look at what the odds of success 18 in stopping the damage would be.
19 And what we found was that for all of the dominant 20 challenges to the plant, we were always in the process of 21 restoring some type of make-up flow to the vessel and the issue 22 that had to be addressed was whether or not we made it in time.
23 Using what we thought was a rather conservative model, we found 24 that we had about a 70 percent success rate, so that we reduced 25 the vessel failure frequency by a factor of three, roughly.
O lieritage Reporting Corporation (202) 628-4888 x
62 1 We t,hink, though, that that model was extremely 2 unrealistic. We are now using what we think is a more credible 3 model, namely that of the Oak Ridge Code BWRSAR and our 4 expectation from that model is that our conditional probability 5 of reactor vessel failure given core damage will be sharply 6 reduced, at least to the 90 percent range, and we expect to in 7 fact quite a bit better than 90 percent. ,
8 And then finally, a second issue related to that, ,
9 given that the vessel does fail, the modeling of the rate of 10 pour, the nature of the material in the pour and the energy of 11 the pour, if that's done in an overly conservative fashion, 12 then one concludes that containment will be severely 13 challenged. Once again, the BWRSAR models say that these pours
() 14 15 are much slower and of a much more benign nature than the '
conservative assumptions that typically are made.
16 We think that these what we call improper profiles of 17 the dominant challenges result in a greatly exaggerated view of 18 risk in plant operations.
19 (Slide) 20 I'd like now to go to question 6.
21 The PP&L view of the hierarchy of accident response 22 objectives is shown on this view graph. Namely, our first goal 23 is to prevent core damage. I think everybody's familiar with 24 these. 1 1
25 The next, however, is to avoid loss of the reactor l
(/)
s_ Heritage Reporting' Corporation (202) 628-4888 i
\
J
63 g~ 1 vessel, and we think this is extremely important and requires V} 2 very careful attention for the BWR plan.
3 And then finally, to avoid loss of containment 4 integrity caused by core melt and reactor vessel failure.
5 In the case of fuel clad damage or core damage, we 6 think that in the absence of some t'ruly catastrophic common 7 mode failure, such as station blackout accompanied by failure 8 in the BWR 4 case of both of the turbine make-up supply ,
9 systems, that fuel damage from loss of ability to get water to 1
10 the vessel is a negligible contributor. ;
11 There are a very large number of completely l 12 independent and very diverse ways of getting water into the l 13 vessel. It can easily be done in time except in a station
(). 14 15 blackout case. And for that reason, we believe that if the procedures are good and the operator training is. good, that the 16 likelihood of damage happening is extremely low, and it does l l
17 not represent a dominant kind of a challenge to the plant. I 18 (Slide) i 19 With regard to reactor vessel failure, we believe 20 that the core damage progression models in this code are as 21 credible as we were able to find and these core damage 22 progression models do result in giving us enough time, we 23 believe, to have a very high rate of success at saving the 24 vessel. At this time, we haven't completed our studies yet, 25 but we have seen preliminary calculations and done preliminary Heritage Reporting Corporation
-(202) 628-4888 e
l l
64 1 calculations which make us feel pretty confident about what the
',-)
\-'
s 2 outcome of our calculations will ultimately.ime.
3 And finally for containment failure,-we think it's 4 very important to use a realistic assessment ref what the nature 5 of the pour of debris onto the drywell floor will be and what 6 the state of the drywell floor will be. We tend to be 7 prejudiced in the direction of if we see that the vessel is 8 going to fail, that we would attempt to find some.way to get ,
9 water on the drywell floor. And in our particular plant, we i 10 have the ability to get a foot and a half of water on the floor l
11 if we can find a pumping source to get the water there. And in 12 most cases, we believe that that source would be available, 13 particularly if we take anticipatory action to prepare for the
) 14 event of reactor vessel failure.
15 And this issue of anticipatory action is something 16 that's worthy of a great deal of attention in the preparation 17 of emergency operating procedures.
18 (Slide) ;
i 19 One of the major problems that we've had is the l 20 availability of success criteria for saving the vessel and the 21 ability to quench floor debris on the drywell floor. We think 22 this is an area where there is'a need for further research 23 since we believe that full scale or even large scale tests to 24 look at these phenomena are really not practical and perhaps 25 not even desirable, at least initially.
() Heritage Reporting Corporation (202) 628-4888 l l
r_
65 1 We think it's a place where separate effects Q(~g 2 experiments should be considered. The reason for this separate 3 effects experiments has to do with the analytical models that 4 one develops to derive success criteria for stabilization of 5 either core damage in the vessel or the core debris on the 6 drywell floor.
7 (Slide) 8 The approach should be to do very thorough .
9 calculations of the dominant challenges to the plant for 10 whatever plant is in question. And it's very important that it 11 be for the dominant challenges to the plant, not some surrogate 12 set that's felt to be binding, and to apply the success 13 criteria that are derived to see what the success rate.can be.
() 14 Given that a reasonable level of success is found, it's then 15 necessary to defend the validity of the success criteria. And 16 we believe the way to do that is to do sensitivity studies on 17 the models to look f'or the sensitive parameters that are 18 questionable, be they models or be they basic fundamental input 19 data in the models, and then to attempt to devise separate 20 effects experiments to go test these out.
21 To date, we have seen evidence that some work is 22 being done in the area of stabilizing the core damage 23 progression before reactor vessel failure. On the other hand, 24 we have not seen a similar concerted attack on the question of 25 stabilizing core debris on the drywell floor.
Heritage Reporting Corporation l (202) 628-4888 1
. .n i
66 s 1 We think that experiments and analytical models and U 2 analytical investigations to explore the limits for success in 3 quenching this debris is very much a thing that should have a 4 high priority.
5 Thank you.
6 (INSERT) 7 MR. HULMAN: It's time for discussion.
8 Dr. Spies indicated to me that he did see necessarily 9 where the models used by the Staff and its contractors for late 10 over pressure failures were radically different than those .
11 being used by industry. That there is a great deal of 12 difference of opinion.
13 Does anybody care to comment on that, for example.
14 Is there a significant difference, for example,
{)
15 between the CORCON models and the models the Staff has been 16 using for late containment failure?
17 What's industry's perception?
18 Somebody want to speak for IDCOR or NUMARK or the 19 owner's group.
20 Please identify yourselves when you speak up for the 21 purpose of the Court Reporter.
22 MR. FULLER: I'm Ed Fuller from EPRI. I believe that 23 Mohsen captured the essence when he indicated the relative 24 effects of containment over pressure versus over temperature.
25 Having principally to do with the way the industry code map
() Heritage Reporting. Corporation (202) 628-4888
67 1 calculates heat transfer from the debris to the containment, 2 and consequently in the energy balance taking a lot of the 3 steam out of -- if you'll pardon the expression -- out of the 4 core debris concrete interaction process.
5 What this does, if you do not have water on the 6 debris which is a very worst case assumption, is that one 7 calculates with MAP that you get temperatures that would tend 8 to predict an over temperature failure of the drywell probably, 9 due to thermal expansion effects around penetrations.
'10 Chicago Bridge and Iron recently did a study for 11 IDCOR and IDCOR turned that into an assessment of what an over 12 temperature failure would be and it's somewhere in the range of 13 700 to 1,000 degrees F. If one looks at that range, and
() 14 compares it with the over pressure failure that was alluded to 15 by Mosan, you get containment failure times after vessel breach 16 and the same kind of time frame, namely, perhaps four to six 17 hours or thereabouts.
18 It's just a question of -- it comes back to how you 19 treat the energy balance from core debris on.the drywell floor.
20 MR. HULMAN: Any comment? Any questions from the 21 audience?
22 I'd like to ask one more question of Mr. Hill.
23 I didn't understand what you meant by success 24 criteria?
25 Would you please speak to that?
A
(_) Heritage Reporting Corporation (202) 628-4888 4
- . , . - . - - - - - - - - Q -- --- , , .
1 I
68 1 DR. HILL: Basically what I mean what are the 2 conditions for example if reactor vessel failure occurs for 3 success in quenching the debris on the floor. And that can be 4 any one of a number of things. That can be the success could -l 5 come about just simply from the nature of the damage 6 progression itself and the rate of pour, and the nature of the l l
7 materials involved in the pour and the timing of the pour. Or 8 it could come about by having pre-flooded the drywell floor in, anticipation of the possibility of vessel failure and a debris 1 9
10 pour.
11 So I think it can be any one of a number of things. )
1 12 We have not yet done, nor have we seen calculations that give 13 us any idea of whether or not there can be any success in this
() 14 process, but we believe it's very important to take a very hard i 15 look. And we dont see that that's been done at this point in 16 time. ,
1 17 MR. HULMAN: Any other questions?
l 18 Please come up and identify yourself for the Reporter 19 and the audience.
20 MR. CUSHMAN: My name is Bob Cushman. I'm from 21 Niagara Mohawk.
22 In your presentation, you showed a slide that was-23 titled, Core Damage Frequency Estimates, but you eferred to it 24 consistently as core melt frequency. Are they the same as far 25 as you're concerned? i l
Heritage Reporting Corporation (202) 628-4888 i
I
i i
i 69 l
- 1 MR. HULMAN: They are synonymous. As far as the
'~ 2 studies were concerned, there was no differentiation.
3 Any other?
4 Please come up.
5 MR. WARD: I'n e Ward of the ACRS.
6 Mr. Hill, on your first chart, the first bullet item 7 said the contribution of human error to the occurrence of 8 initiating events doesn't imply that operator failure in ,
9 response to a severe accident must be e significant 10 -ontributor.
11 I guess I didn't quite understand what you mean by I
12 that. Do you mean,'is there some reason to believe there's a 13 . differ, ant error rate expected from operators in controlling a 14 severe accident than there is in operation which might lead to
{)
15 a fault causing an accident?
J 16 MR. HULMAN: Mr. Hill, I think you went for a drink. )
l 17 Did you hear the question? !
18 MR. WARD: I'd hate to repeat that. l 19 MR. HULMAN: Would you please repeat it?
l 20 MR. WARD: What do you mean by this? Do you mean 21 that there's a difference in error rates in these two different l 4 22 sets of circumstances that can be expected?
23 DR. HILL: I'm reluctant to say that I think there's 24 a difference in error rate, but what I think is --
25 MR. HULMAN: Do you understand the question?
() Heritage Reporting (202) 628-4888 Corporation
l l
70 j 1
(~~ 1 DR. HILL: Yes, j V) MR. HULMAN: Why don't you rephrase the question the 2 ,
l 3 way you understand it and then answer it. ;
4 DR. HILL: I'll try.
5 I believe the question was what did I mean by the ,
l 6 fact that the failure rate for operator error or the operator 7 failure rate for emergency operating procedure response, was it !
8 different from the errors made in the mainten'ance of the plant -
i 9 or in the conduc t of normal operations.
10 I believe that's what the question was, what did I
]
11 mean by that statement.
12 What I meant by that statemant is not that the 13 failure rates are necessarily different but ratlar that the f' 3 vi,/ 14 fact that many initiating events and marty equipment failures 15 can be attributed to operator error should not casually be 16 interpreted to mean that the operators will commit significant 17 errors in response to an accident.
18 We believe that that is not necessarily true. In j 19 fact, at PP&L we're convinced that with good procedures and j l
20 good operator training, that the error rate in response in i 21 following the EOPs, will be negligibly low. The experience we l l
22 have from our simulator measurements basically are that cut of .
23 something like 1600 procedural steps and the measuroments we've 24 made to date, we have seen no cases of procedural error, that 25 is, failure to follow the procedura, with a few exceptions on O Heritage Reporting Corporation (202) 628-4888 i i
o . - J
71 1 the order of three or four_ cases. l
(~%)
(m-2 In those cases, what was involved was the procedure 3 was poorly worded, and in fact, testing the wording of our 4 procedures was part of the objective of. conducting.these 5 experiments, to see if in fact the procedures were clear.
6 The other case'-- in fact, in all of the cases seen, 7 we had a crew member-out of place. And the reason that we had 8 to do that is Susquehanna is a two unit plant and we have a two 9 unit crew in a control room with one shift supervisor. As a 10 result, -- and we can only test one crew at a time because we 11 have a single unit simulator -- and so as a result, in order to 12 test every crew, on half of them we have to make crew 13, substitutions. And in those cases we did, these were the cases A
is_,/ 14 where we detected what we would consider a potential, although 15 not a clear cut procedural error.
16 What I meant by that was that the crew did follow the 17 procedures if you interpreted the procedure the way they 18 interpreted it. The problem is they interpreted it 19 incorrectly. That's something that I believe we've since i
20 corrected.
]
21 In our risk assessment work and in our future -
22 simulator experiments, that's something that we'll be giving 23 very close attention to.
24 MR. HULMAN: Any other questions?
25 Please come up and identify yourself.
I J Heritage Reporting Corporation (202) 628-4888 i
i l
72 1 MR. BANKOFF: My name is George Bankoff and I'm with 2 ' Northwestern University.
3 I would like to speak first of all to the question of 4 the difference between the IDCOR and the NRC calculations of 5 the split between the heat going down.and the heat going up in 6 the absence of water on the drywell floor. And this.beems to 7 me to be a question basically of a technical nature which has 8 to be resolved by recourse to experimental data and to ,
9 analytical predictions and it cannot be done by general 10 statements.
11 Basically, we have a radiation transfer from the top, 12 we have natural convection, and we also have going to the 13 surface of the molten material, we have a turbulent thermal
() 14 dif fusivity and the only place where there can be very much in 15 the way of disagreement is in the magnitude of that turbulent 16 diffusivity. So it seems to me that one cannot resolve 17 questions of this sort without the detailed examination of I l
18 whatever experimental data there are, the German data, 19 whatever. And whatever single effects experiments are 20 available and whatever literature that is available for bubbly l 21 flow in this sort of system.
22 This has to be examined by. groups of experts and then l i
23 a decision made. It cannot be made on the basis of popularity 24 or votes or anything of this sort. And I have some concern i
25 that aside from this particular question, that we will have an Heritage Reporting Corporation (202) 628-4888 l
. l
73
- 1 effort-to resolve some of these issues by the nature of this 2 meeting without going through that sort of detailed. technical 3 examination.
4 -MR. HULMAN: Any other questions?
5 (No response.)
6 MR. HULMAN: Hearing none, let's take a fifteen-7 minute break and then we'll go to liner melt through.
8 (Whereupon, a brief recess was taken.) ,
9 MR. HULMAN: Ladies and gentlemen, could we please 10 take our seats.
11 Does anybody have the sign up sheet, the log in 12 sheet?
13 Will you see that it gets passed on?
() 14 15 Rosemary?
You've got it? -
16 The gentleman over here has it.
17 Would you please continue to move that so we know 18 who's at the meeting.
19 Perhaps the most controversial aspect of Mark 1 20 containment performance is the likelihood of shell melt 21 through.
22 The agenda indicates a rather large number of 23 speakers on that subject, so many in fact, that we had to make 24 two sessions. The first several speakers are researchers from 25 the National Laboratories to present a perspective.
() Heritage Reporting (202) 628-4888 Corporation q .
1 I
l 74 s 1 To be followed by some industry people and to be ;
l 2 followed by some consultants.
3 I'd like to start out by_first introducing Steve 4 Hodge from Oak Ridge who will be the lead off batter and in i 5 fact, it'll be Steve Hodge, Cliff Hyman and George Greene, Dana ,
6 Powers, Tom Kress, Raj Sehgal. And then we'll take questions J
7 and answers on those presentations.
i 8 Tonight after dinner, we'll have more speakers and , !
9 more questions and answers, and if we can't get through all the 10 questions and answers this evening, there is time tomorrow.
11 morning and we'll try and finish it, j 1
12 So let me start out by asking Steve if you'll come 13 up. Okay. And your platform. -
(
15 16 17 18 4
19 20 21 J
22 j 23 l l
24 i 1
25 l l
([) Heritage Reporting corporation l (202) 628-4888 1
l
- __ .)
- - ~ . .
75 g 1 STATEMENT OF DR. S.A. HODGE, OAK RIDGE NATICNAL LABORATORY 7.
' DR. HODGE:
2 Well, before I get into my presentation 3 this afternoon, let me point out that the first four or so 4 presentations in this session derived frem a task group that 5 was set up in October, 1987, by the Accident Evaluation Branch 6 in the Office of Research. And members of this task group 7 whose task is to examine this question of attack and possible 8 melt through of the drywell liner are Tom Walker is our group ,
9 leader, myself, George Greene from Brookhaven, Dana Powers and 10 Ken Bergeron from Sandia, and Jack Dallman from Idaho National 11 Engineering Laboratory.
12 Now, there is a task group report. The task group 13 report is expected to be placed in the public document room j ) 14 next week. However, that does not terminate the work of the 15 task group. It's going to continue and the Office of. Research 16 is sponsoring additional near term efforts to extend the work 17 to include improved shell heat transfer analyses capabilities, 18 improvements to the CORCON Code for core' debris concrete 19 interaction, and some separate effects experiments.
20 And there will be another report currently planned 21 for release in April and perhaps yet another in June.
22 I should also mention following these first four I 23 speakers who will talk about the specific task group work, Tom l 24 Kress will summarize some interim and related results of his on 25 the analysis of the drywell shell response. .
l
/" i (T) Heritage Reporting Corporation (202) 628-4888 i
4 _n n
I l
76 I) 1 with that'then, let me get into my part of the study.
,2 which is the calculation, trying to develop a methodology for 3 calculation of the debris pours that might be expected to 4 emanate from a BWR reactor vessel 1under severe accident 5 conditions -- I should say, under unmitigated severe accident 6 conditions.
7 We've developed a set of calculations, calculational 8 methodology to address this. As part of the methodology, we ,
9 assume that the core debris in the bottom head of the reactor 10 vessel can be considered to comprise three layers, three 11 somewhat different layers of debris.
12 There's an entire presentation that I could give to 13 say how we got to this point, but that's not my purpose today.
14 Let me start from here with a dryout'of the water in the bottom ;
15 head after relocation of debris into the bottom head. The '
16 bottom layer is comprised then of the first material to fall 17 into the bottom bead, which was control blade and channel box 18 material.
19 The middle layer is formed once the fuel stacks, 20 columns of fuel pellets collapse into the bottom head.
21 And the top layer is not formed until after bottom 22 head dryout at which time the control rod guide tubes which 23 support the remainder of the core are subsumed into the debris, l 24 lose any structural integrity and any remaining portions of the l 25 core fall into the bottom head, O Heritage Reporting Corporation (202) 628-4888 i
l l
i 77
() 1 2
(Slide)
Now, the fact that the control rod guide tubes join 3 the debris, are modeled to join the debris is important and it 4 has important consequences so I want to try to justify that.
5 On the left here, you see a sketch of the reactor 6 vessel, and in red is shown the height of debris, given that 7 the .ontrol rod guide tubes remain in place.. And you see that 8 there's enough debris to surround all of the control rod guide .
9 tubes, all up to just below the core plate.
10 So with that argument, and remembering that this is ,
11 after dryout and the debris is heating up under decay heat, it 12 seems reasonable to assume that the' control rod guide tubes are 13 ,
subsumed into the debris and the resulting debris .eight would
( 14 be as shown on th,e right.
15 (Slide) 16 This view graph shows the constituents of the debris 1
17 and the main point I want to make here is the large amount of 18 stainless steel. Now, you'll notice up at the top, it says the 19 debris bed that's formed in the bottom head would consist of 1
20 core constituents and they're detailed down here at the bottom. j 21 Then the top guide, which is normally predicted to melt some i 22 15,000 pounds of scainless steel, core plate, another 20,000 23 pounds of stainless steel, but mainly the structures in the 24 vessel bottom head, which are almost 200,000 pounds of I
25 stainless steel added to the debris in the bottom head. J
( Heritage Reporting Corporation j (202) 628-4888 ;
l 4
4 l
.l 78 1 And then down here of course, we have the zircalloy
}
2 from the cladding and canister, the fuel, and yet another
~
3 36,000 pounds of stainless steel coming in from the control 4 blades and other small components in the core region.
1 5 (Slide) 1 6 We divided, as I said earlier, up into three layers. ;
. 7 This shows a rough. schematic of how the three layers are I'm not going to -- I don't have time to go l'nto 8 formed. ,
9 detail as to how this works. But'let me just say that molten 10 material in, layer three, an upper layer, tries to run downward.
11 If there's void in the layers underneath, the molten material 12 is relocated downward. It can then either freeze in the layer 13 below, or if it's not frozen, it can recede into the layer yet 0) v 14' below that.
15 If there's molten material in a node and the layer 16 underneath it is full, all of its voided material has been 17 filled with frozen material, it can move to the right and down, 18 or indeed to the right again and down as necessary seeking its 19 way down under the impetus of gravity.
20 Once it reaches the wall, if the' penetrations have 21 failed in that layer at the wall, then it is removed from the 22 vessel and added to the debris on the drywell floor.
23 (Slide) 24 Penetrations as shown here exist in the bottom layer 25 and in' layer number two. So the pene'tration failure is limited
\ Heritage Reporting Corporation (202) 528-4888
l 79
() 1 2
to occur either in the bottom most layer, layer one, or in layer two.
3 Now, if you remember earlier where I showed.you what 1
4 is in each layer, you'll recall that there's very little UO2 in l 1
5 the bottom layer. Most of the decaying source is in layer 1 6 number two, and indeed that is the one that is predicted to l
7 heat up and become molten first. And we virtually always 8 predict the initial penetration failure to occur in layer ,
i 9 number two, and much later in layer number one. !
10 (Slide) 11 Here's a photograph that was provided by the 12 Philadelphia Electric Company. showing the stub tubes which of 13 course the control rod drive mechanisms passed tiirough and the
( 14 instrument tube penetrations in the bottom head. And you 15 notice that the penetrations, the instrument tube penetration 16 wells are at the bottom head. In other words, they're at the 17 lining of the bottom head, whereas the control rod drive 18 mechanism assemblies are some four inches above the bottom head 19 and stuck up into the debris.
20 However, for reasons that have to do with the catcher 21 assembly that should a control rod drive mechanism weld become 22 weakened, there's a catcher assembly that stops it from moving
?3 more than about an inch and a half out of the reactor vessel, 24 and therefore it wouldn't clear the eight and a half inch thick 25 bottom head. We believe that the main opening to pass debris Heritage, Reporting Corporation (202) 628-4888
80 1 'i nto the drywell is going to be the instrument tube
[}
2 penetrations which are some,two-inch diameter noles.
3 (Slide) 4 Now, with that as a background, I want to show you an 5 example. And this is merely an example calculation. This i 6 particular one is based on the short term station blackout-7 sequence for Peach Bottom which is initiated by a postulated 8 common mode failure of the DC batteries, so that not only have, 9 you lost your AC injection systems, you also-have lost HPCI 10 and RCIC since you have no batteries and you cannot operate the 11 ADS, again because you'have no batteries.
12 So you stay pressurized. And you see that here, this 13 is the vessel pressurs versus time and we stay pressurized
() 14 after core plate dryout. There's no steam source until core 15 plate failure and debris begins to be relocated into the bottom 16 head, at which time pressure is restored. And then after 17 bottom head dryout, there's no steam source to maintain 18 pressure in the vessel. The pressure slowly decays until 19 bottom head penetration failure at which time it rapidly 20 decays.
21 And here we.are assuming it's a strictly user input 22 but the whole equivalent to five instrument tube penetrations 23 being opened.
24 Let me say right now that the way we do the j
25 calculations, the hole size only effects the rate of blowdown.
Heritage Reporting Corporation (202) 628-4888
l 81 ,
j)
i 1 If we have molten material-in a node adjacent to the wall, we v
2' take it out. We do not'model it as passing through a hole, we i
3 don't go to that level of detail. ;
4 So if it's molten and it's in a cell next to the l
5 wall, it comes out. But the hole size does control the rate of !
l 6 blow down and is shown here it's equivalent to five instrument )
7 tube penetrations.
8 (Silde) ,
9 This shows the timing of events that are particular 1
I 10 to bringing debris out of the reactor vessel. I start here 11 with the bottom head dryout at time 179 minutes, then the 12 control rod guide tube structure is subsumed into the debris 4
13 and the romainder of the core is brought down some time 192 I
( 14 minutes. And then the penetration fails in layer two. This is 15 a creep rupture mechanism based on anconal stainless steel l
16 welds, creep rupture of anconal stainless steel welds, it 17 occurs at time 220 minutes.
18 And then you see I put down ten-minutes later when it 19 left the vessel. And in this particular calculation in that 20 ten minute period, we brought out 126,000 pounds of debris of 21 which 97.5 percent of it is metals.
l 22 Now what is happening here, as I will show you in a i 23 moment in some detail, what is happening is that you start 24 blowing the steam through the debris bed out the hole and'it 25 rapidly heats the debris and causes a lot more molten material
~.
Heritage Reporting Corporation (202) 628-4888
82
-() 1 to be formed and you get an initial very rapid pour from~the 2 vessel.
3 (Slide) 4 Now, this table which is in your handout, shows the 5 layer by layer composition when they're first formed, and then 6 over here the all layers total. And all I want to do here is 7 call your attention that it's in there if you want to look at 8 the individual numbers.
9 (Slide) 10 Very important and one thing I want to stress here is 11 that central to these considerations is what eutectics are 12 assumed to be formed. Central. Because clearly'in this 13 calculation you see we have a stainless steel' boron'zire 14 eutectic melting at 2100 F. If that weren't there, we clearly 15 would not melt as much metal initially and we would not have 16 nearly as much brought out in the initial surge.
17 So it's very important that we do experiments and 18 that we bank on all available information to try to determine 19 what eutectics would be formed and what is their melting 20 temperature in the bottom head.
21 (Slide) 22 And I think this stresses the point. For this 23 particular sample calculation using the eutectics that I showed
)
you', notice the very large initial spike which completely 24 i 25 dwarfs this is the debris flow rate from the vessel in pounds
/'T i f / 4
\#
Heritage Reporting Corporation (202) 628-4888
83
(~) 1 per minute versus time and at time 220 minutes over there, you w
2 see we opened the hole equivalent to five instrument tubes, we 3 get an initial spike of molten material. And I emphasize that 4 much of this molten material is being formed during the steam 5 blow down through the debris due to metal water reaction energy 6 release. And after that, the pour is much less.
7 Not only that, notice that this time scale goes out 8 to 1,000 minutes here and we started at 220. So we're talking .
9 about 15 hours1.736111e-4 days <br />0.00417 hours <br />2.480159e-5 weeks <br />5.7075e-6 months <br /> or so in which the debris is continuing to come 10 out. It comes out over a very long period of time.
11 (Slide) 12 This is the mass average temperature of the pour as a 13 function of time. Here I have a greatly expanded time scale at
(
~
14 the bottom. I'm only showing you a period of 30 minutes, 15 because I want to expand this initial pour. So we're looking 16 from time 220 minutes out here to 250 minutes, just a 30-minute 17 period, and you can see that there are little increases in 18 temperature as we go along. We shift from lowering melting j 19 eutectic to the next lowest melting to the next lowest melting 20 and so forth. And again, it's a mass average temperature of 1
21 what's being poured.
l 22 (Slide) 23 This is the total integrated debris mass expelled 24 from the vessel. And again, just to make the point that at the 25 end of this calculation, which I terminated at time 1,000 n\ ' Heritage Reporting Corporation (202) 628-4888
l 84 1
[)/ 1 minutes problem time, we have about 630,000 pounds out of the 2 vessel, but there's still another 200,000 pounds or so left.in ,
i 3 the vessel. This pour would continue over several more hours ,
1 4 if I had taken the calculation out that far.
l 5 This is the temperature, this is the large scale plot I 6 of the temperature as a function of time, and I merely want to 7 show this drop off over here at about time 740 minutes or so, !
l 8 is due to the penetration opening in layer number one. Layer - ;
\
9 number one which is primarily metals finally gets hot enough to 10 cause penetration failure late. And when this occurs of course j 11 we see some low melting eutoctics again starting to come out, 12 and the mass average temperature dips to acknowledge that.
13 (Slide)
- 14 Then this is in the handout and I merely call your 15 attention to it now. ~It shows at the termination of the 16 calculation how much is left in each layer and the total left ,
l 17 in the vessel and then how much is left in the vessel and its 18 passed to the drywell floor by individual metal and individual l 19 oxides and total.
20 And as I said, 624,000 pounds out, and some 280,000 21 pounds still in the vessel at time 1,000 minutes.
22 (Slide) 23 Now, this last view graph I put on here to show the 24 accident specific characteristics of these calculated pours.
25 This happens to be for the solid line is a short term station O Heritage Reporting Corporation (202) 628-4888
85
/"T
(,/
1 blackout case with depressurization. With depressurization.
2 The dotted line is the long term station blackout in 3 which of course there is no depressurization because you have 4 no batteries at the time. Notice that the -- this is the 5 tortoise and the hare situation here.
6 We've slid the time scale so that they both start 7 from time zero which is the time of initial penetration 8 failure. The dotted line,' which is the case without , l l 9 depressurization, shoots.up rapidly. That's the effect of 10 energy release in the debris due to steam blowing through from 11 the pressurized vessel.
I 12 However, after that initial pour, it levels off and 13 indeed falls below the solid line, eventually because of the s_ ,/ 14 greater decay heat in the short term station blackout case 15 during t'his period of time right after penetration failure I l
16, catches up with it, and the debris comes out.
17 But I merely want to use this one, again as I say, to 18 emphanize that first of all, what you're going to calculate 19 with this method very much depends on what eutectics you l
20 assume, and once you've' adopted a set of eutectics, then it 21 becomes accident sequence specific. I 22 I'll stop there.
l 23 (INSERT) j 1
24 MR. HULMAN: Let me make a housekeeping remark with i 25 regard to the task ' force report that Steve referred to. That i
Heritage Reporting Corporation (202) 628-4888
I l
86
('~% 1 report will also be available among the handouts at noon on i
\ jl 2._ Friday. .
l 3 The next speaker is Cliff Hyman from Oak Ridge, also.
~
q 4 Cliff? ]
^
Before Cliff starts, Themis? l 5 i 6 DR. SPCIS: I would like to ask the speakers when ,
l 7 they are through to spend about one minute summarizing what is !
. l 8 it th- they told us. What is the bottom line. Have they -
9 developed a model that they believe should be used, or depends 10 on research or what are the assumptions or whatever.'
11 And Steve, when this gentleman is through, I'd !
12 appreciate when your turn comes, to tell us what you think you ,
13 have done and is it believable or it depends on what. Just
( 14 give us about one minute bottom line. .That will help get some
, 15 perspective where we are viz a viz everybody else. ;
16 t
1 17 18 19 20 21 1
22 j 23 1
24 4 i 25 Heritage Reporting Corporation (202) 628-4888 l
- , 4 I
a 87 STATEMENT OF DR. C.R. HYMAN, OAK RIDGE NATIONAL LABORATORY
) 1 2 ,DR. HYMAN: Hello, everybody. My name is Cliff 3 Hyman. I work at Oak Ridge.
4 The title of my talk today is a Methodology for.
5 Calculating Debris Spreading on the Floor of BWR Mark 1. Steve 6 has told you how we calculate the debris pours and somewhat of 7 the mechanisms involved in the debris pours coming from the 8 failed reactor vessel. ,
9 I'm going to take this a step further and show how at 10 least we are calculating the thermohydraulic effects of the 11 debris concrete interactions on the floor.
12 (Slide) 13 The purpose, like I said, is to discuss a methodology
( 14- we developed in assessing the debris spreading in the BWR Mark 15 1 floor. Next, I want to show some sample results from a 16 calculation we've done for Brown's Ferry short term station 17 blackout.
18 I would like to caution everyone that this 19 calculation that I'm doing, that I'm showing results for is I 20 different than the one that Steve did, so you shouldn't be 21 confused that the times are not the same. It's for a different l
22 plant, also. l 23 Last, I'd like to say a little bit about some of the 24 deficiencies we've identified in using this methodology'and I'm 25 sure it will prompt some discussion.
Heritage Reporting Corporation (202) 628-488R
88 ,
1 (Slide)
[) ,
2 As Steve says, we bring out the debris from the i
3 failed reactor vessel at various temperatures. This is a plot 4 of the temperature of the molten debris as it leaves the 5 reactor vessel as a function _of time.
6 Notice the very very long time scale on the bottom 7 axis, somewhat over ten hoors, 12, 14 hours1.62037e-4 days <br />0.00389 hours <br />2.314815e-5 weeks <br />5.327e-6 months <br /> something like , I l
8 that. As Steve-said, we use different eutectic temperatures ,
9 and you can see this in the temperature of the pour as it 10 leaves the reactor vessel. You see the 2100 on the bottom, the 11 2400, the 3365 and the 4352. These are user supplied input 12 variables for the BWRSAR code, and they're subject to large 13 uncertainties.
4
( 14 (Slide) ,
15 These are some sample pours for three of the 16 materials we're looking at in the bottom half and on the floor 17 of the Mark 1 containment. You have top, iron; middle, 18 zirconium: and bottom, UO2, as a function-of time.
19 Noticed early we get metals pouring out of the bottom 20 head, and later we get U02 pouring out of the bottom head.
21 This has impact on the core concrete interaction on the floor.
22 (Slide) 23 This is a scaled schematic of what the drywell floor 24 looks like in relation to the pedestal and the drywell sumps.
25 I would like to point out that the drywell floor is very small.
Heritage Reporting Corporat' ion (202) 628-4888 1
l 89
/')
(_/
1 In the in-pedestal region, there's only about 30 square meters j 2 of floor area available for core debris to spread on. In the
)'
3 ex-pedestal region, there's about 102 square meters of floor 4 area for the debris to spread onto. l 5 Notice the small sumps. The sumps for most BWRs are 6 very small compared to PWRs. 'For Brown's Ferry, it's about 200 7 cubic feet, so it will only hold about ten percent of the core
> 1 1 8 debris as it leaves the reactor vessel. ~
l 9 So you have to ask yourself, what is the core debris l
10 going to do when it comes out of the reactor vessel. It's 11 going to fill the sumps, it's then going to fill out the area 12 inside the reactor pedestal, and if the conditions are such 13 that the molten debris remains molten, it can flow through the
( 14 doorway as shown in the upper right hand side there and flow 15 into the ex-pedestal area and then maybe contact the drywell ;
l 16 shell itself.
17 Notice the scale here. The distance between the
, 18 drywell shell and the outer ring of the pedestal is only about l l
l 19 ten feet, the same distance as it .. between the center line l
l 20 and the inner radius of the pedestal.
21 (Slide) l 22 Now, the code that I'm using to do my analysis with 23 is the CONTAIN code that was developed at Sandia National 24 Laboratory and it's a very very powerful tool. What I have l
25 here is a schematic of the drywell. Note, I have seven cells l Heritage Reporting Corporation <
l (202) 628-4888
! i l
l _
90 1 indicated here. CONTAIN will allow me to do an unlimited 2 number of cell simulations.
3 What I'm trying to do here with this seven cell model 4 is to capture the natural convection flow paths that develop 5 during the course of the calculation. CONTAIN also allows me 6 to do core concrete interaction analyses using the CORCON MOD-2 1 7 code. It a,1so has within it the Vanessa code and the Maeros 8 code to calculate the generation release of aerosols from-the ,
9 core debris and the transport of aerosols throughout the l l
10 atmosphere of the containment itself. '
11 (Slide) 1 12 Now CONTAIN and CORCON will not dynamically spread 13 core debris for me. So what I've done is I've segregated, I've 14 broken into time regimes what the core debris may look like 15 flowing across the drywell floor. Notice on the left hand side 16 I have the sump fill calculation. In the middle after the 17 sumps are full, I've spread my debris into just the in-pedestal 18 region. I did do the calculation for the amount of time, let 19 the debris build up, then it reliquefies, it melts and if it's l 20 high enough, I allow it to flow into the ex-pedestal region.
21 (Slide) l 22 Now, this is a sample result, purely a sample. What 1
23 I've done here is there's four independent sets of calculation i 24 results presented. I've merged the results of all four 25 calculations onto one graph. On the vertical axis is the in Heritage Reporting Corporation (202) 628-4888
91
/
b) 1 2
pedestal metal layer temperature.
time.
On the horizontal axis is 3 For the first time regime, I'm filling the sump and 4 you can see what the response of the sump fill calculation is.
5 We have a small volume of water that we assume to exist in the 6 sump at the beginning of the calculation. The temperature 7 therefore falls. As you dry the sump water up, the temperature 8 rises. It gets to the molten point of the metals which is the, 9 dominant material in the pour at this point.
10 And then I assume that the sumps overflow and flow 11 into the 30-square meter in-pedestal region. I then proceeded 12 in the calculation and the temperatures then fall because I've
,_ 13 increased the area some six times.
But sooner or later, I
(_/' 14 start pouring oxide debris, which has a very high melting 15 temperature, and the debris builds up and approaches the 16 melting temperature.once again.
17 At this point, notice the 1589. That's one of our 18 metallic eutectic temperatures. When the debris reaches 1589, 19 we assume the debris is also high enough so that you would have l 20 flow into the ex-pedestal region. And so with that, I'll just 21 stop and I'll come back and explain this a little bit later if 22 we have time.
23 But the other two calculations show the in-pedestal 24 response after I spread into the ex-pedestal region.
25 (Slide) f)
~
Heritage Reporting Corporation (202) 628-4888
92 :
1 Now, in order to do these spreading calculations, L
2 there are certain things I have to do. I'want to dynamically 3 calculate in-pedestal and ex-pedestal debris core concrete 4 interactions, and what I do in order to do this calculation is 5 that I spread the debris such that the heights are equal 6 between the in-pedestal and ex-pedestal areas. I also preserve 7 the volume of the debris as it ablates into the concrete.
8 Next, I spread the debris so that the debris ,
9 composition for the J coaestal and ex-pedestal interaction is 10 (he same.
11 Next to the last thing is I spread the debris such 12 that the temperature of the debris does not rapidly change, or 13 I try to make it so that it doesn't rapidly change. If the
( 14 debris temperature falls too fast in the ex-pedestal
~
15 calculation, then I've assumed an area that's too large. If !
16 the temperature of the debris rises, then I've assumed an area 17 too small. So that's really the only control that I use, the 18 only criteria that I use choosing an initial ex-pedestal area i
19 over which to spread my debris.
20 The last thing I'd like to point out is that I 21 continue to do the calculations with an assumed ex-pedestal 1
22 area until that temperature gets to the molten point. At that 23 point, since there's floor area in the ex-pedestal region 24 that's not covered, I stop the calculation and I reinitiate it 25 with an increased floor area.
Heritage Reporting Corporation (202) 628-4888 J
i 93
) 1 (Slide) 2 For this calculation, I assumed an initial floor area 3 for the ex-pedestal floor concrete interaction to be 200 square
~
4 feet or in metric terms, 18.5 square meters. Notice that the. l 5 temperature does not go up. This is a plot of the ex-pedestal 6 metal layer temperature on the vertical axis, and time on the-7 horizontal axis. Notice that I've started the calculation at 8 about 355 minutes. This is some two hours after the debris .
9 pours start coming from the failed reactor vessel.
l l
10 So it takes two hours to' fill the sumps and to build 11 up a layer a molten layer high enough for me.to justify flowing 12 the debris through the door in the pedestal and initiating an 13 ex-pedestal calculation. !
(} 14 But notice that the temperatures do not go up and
)
15 down initially very much. In fact, they actually cool off.
16 This signifies that the debris is slowly cooling off, is slowly 17 solidifying, and it tells me that I've done maybe not too bad a 18 choice in the initial floor area that I chose for the ex-19 pedestal interaction.
20 Notice that the vertical line that I've drawn on this 21 graph at about 566 minutes tells me the time at which the' l
22 debris has reached the molten temperature when I should spread i 23 again. What I did was I doubled the area. I went from 200 I l
24 square feet to 400 square feet. 1 25 (Slide) l O Heritage Reporting Corporation I (202) 628-4888 1
i 94
()
V 1 When I go from 200 square feet to 400 square feet, 2 look what the zirconium energy production does. It just 3 escalates. The temperature on the previous slide had begun to 4 escalate when it reached the molten temperature so when I 5 spread it to 400 square feet, you have more surface area for 6 which the debris could get contact with the concrete and the 7 zirconium reaction is just right for accelerating. It just 8 takes off astronomically. .
9 So that's what's driving the thermal response of the 10 debris. Notice that the temperature when I chose my new area 11 of 37 square meters continued to increase. That tells me that 1; my area's too small and that I should have chosen an even 13 larger area. In fact, I think I make that a~ point on a later
( 14 slide. -
15 (Slide) 16 Yes, that's the point on this slide is to show that I 17 should have instead of going from 18.5 square meters, I should 18 have not gone to 37 but I should have gone to some much bigger 19 number. And in fact, I've done later calculations which are 20 not in your handout that show that the temperature of the metal 21 laye'r will continue to rise had I spread from the in-pedestal 22 region all the way around the ex-pedestal region. If I had 23 done that calculation starting at 566 minutes, I would have 24 done an adequate calculation.
25 (Slide)
Heritage Reporting Corporation (202) 628-4888
.95 1 . Now, there are several notable deficiencies in what 2 I've done and what we're assuming. When I first tried to do i 3 these calculations with the debris pours coming out of the i
4 BWRSAR code, we had a problem. The problem was that if you 5 look at the calculated temperatures that CORCON comes up with 6 when you give it the debris es BWR says it exists when it comes 7 out of the failed reactor vessel, COhCON says it's frozen. BWR 8 says it's molten. So you do not conserve energy. ,
9 Now, what I did to try to address this problem was I 10 went into CORCON, I removed the solid liquid beh6vior, the ,
1 11 melting behavior that CORCON was based on originally and hard l l
12 wired BWR eutectic melting temperatures into CORCON, realizing i 13 that it was a gross over simplification of what we believed to O
V 14 be the actual'CORCON melting behavior of the debris 15 Now, we brought this to the attention of Dana Powers 16 and Dave Bracuey at Sandia, and they are actively nursuing a I
17 more sophisticated treatment of eutectic matal in the inside of i 18 CORCON. We hope to have this as soon as they can get it ready 19 -- several months probably, though.
20 The next deficiency that is inheront in my 21 calculations is due to one of the original assumptions that i 22 CORCOh was based on. As I understand it, CORCON was designed I 23 to study high temperature molten debris core interactions with 24 concrete. The BWRs don't have very very high temperatures. If 25 you believe what we're saying you have much lower temperatures.
O Heritage Reporting Corporation (202) 628-4888
96 4
1 So CORCON has within it a simplification which 2 neglects outgassing of concrete if the debris temperature is 3 below the ablation temperature. You do not have any 4 outgassing, and this is because CORCON's originally based on 5 ablation rates which were very similar to the thermal 6 propagation rates into the concrete, so you didn't have to 7 worry about outgassing before ablation.
8 Now, I must say that I don't want to sound too ,
9 negative about CORCON. CORCON's a very good code. It was 10 based on assumptions which at the time it was made were 11 acceptable. But now we're doing BWR calculations, and so we're 12 finding things that are limiting. And so they're being 13 addressed and I think that eventually we'll have a better tool
() 14 ' to do BWR analyses.
15 Finally, I'd like' to say that it takes a lot of hand 16 work to stop containment and then to reinitialize it every time 17 you want to do a spread. These things that I'm doing are noe-18 mechanistic. In other words, I do not solve any flow equo ions 19 to do my flowing across the floor. It's based solely on 20 thermal behavior of the debris. ~
21 This debris that we're pouring out of the bottom head 22 of the reactor vessel have very low superheats. And the time 23 frames involved are very long. And so it may be -- I know it's 24 an oversimplification to base it on thermal behavior, but this 25 is what I've done as a first guess.
O Heritage Reporting Corporation (202) 628-4888
?
97
}/~}
1 (Slide)
U 2 Now, one of the issues involved at this meeting is 3 the discussion of threats to d.rywell shells. You cannot base 4 your analysis just on CORCON results. There are limitations 5 inside of CORCON which do not make it amenable to doing shell 6 interaction analysis. First of all, CORCON does not know that 7 there is steel for the radial structure. It assumes it is l ;
8 concrete. Right there, it makes it inapplicable to do drywell ,
l 9 shell interaction calculations.
. 10 Secondly, there's no temperature profile in the 11 concrete structure that is in the radial structura of the ]
12 calculation. Finally, there's no thermal interactions with the l
13 radial structure if the debris temperature falls below the
( 14 ablation temperature of concrete assumes the boundary 1
15 conditions was 80 or better. l l
16 For these reasons, you have to do subsequent analyses '
17 to study the potential threat to drywell shell integrity.
l l
l 18 (Slide) !
l l 19 What I've tried to do with this view graph is to show l
l 20 as I understand it where I fit in, into a bigger picture. See 21 me, I'm right here in the middle, i do molten core concrete 22 interaction analyses used in CONTAIN.
I I 23 On the upper right hand side, I interact with the in-24 vessel analysis people at Oak Ridge, namely, Larry Ott and we 25 interact on a day to day basis.
O Heritage Reporting Corporation (202) 628-4888
98 1 We also try to blend in experimental results that i ( 2 George Parker is doing for us at Oak Ridge. He's done one ,
3 experiment. He has more to be done. As soon as his results 4 are complete, we'll try to factor this into the in-vessel )
5 codes.
6 On the left hand side, I also interact with people at 7 Sandia and at Brookhaven. I interact with Ken Bergeron and his 8 people on the CONTAIN code project development staff. Just to, 9 give you an example of how we interact, I just received a 10 preliminary version, a parametric model to do concrete 11 degassing. I received it last Thursday.
12 he also interact with Dana Powers, Dave Bradley on 13 CORCON, tell them what our problems are, and they try to help 14 us, give us suggestions and so forth. And George Greene, we
) ,
15 interact with him. He's providing model development for future I 16 CORCON incorporation. >
17 On the bottom, I pass my information, my results to 18 other people who are involved in doing the actual shell i l
19 analysis calculations. These people include Tom Kress at Oak j 20 Ridge, Ken Bergeron and George Greene.
i 21 Now, that's all I have to say, 22 (INSERT) 23 MR. HULMAN: Steve, would you give us a one-minute i 24 summary of the two presentations, if you could?
25 DR. HODGE: I think what Cliff and I were trying to i
) Heritage Reporting Corporation (202) 628-4888 l
-. .-. . ~ . - . - _.
99
,- 1 get across this morning is the fact that we've been charged I
(_/ 2 with trying to come up with a methodology to calculate the 3 scenario by which molten cora debris might leave a Mark I BWR 4 and how it might spread across the drywell floor with the goal 5 of ultimately providing the conditions at the debris shell 6 interface to analysts who are trying to determine the shell 7 response.
8 At this point in time, both of us have serious .
9 problems that we've got to overcome in order to do these 10 calculations properly. In n.y area, the in-vessel calculations, 11 bring the core debris out of the vessel, we've'got to determine 12 what eutectics would be formed. We've got to know that. It's 13 very very important and it has a lot of interplay back and 14 forth as how much molten metal would be swept out of the vessel J
15 during the blow down if the vessel were pressurized. In 16 particular, that's an important question. 1 17 Now, we're trying to solve this. We've got some 18 small scale experiments going on at Oak Ridge, and we're ,
1 19 talking about perhaps doing some larger scale experiments in l 20 which you mix up the components, call it witch's brew 21 experiment, if you will. You mix up the components that might 22 be expected to be in the BWR debris, heat it up and see what's 23 formed and determine their melting temperatures.
l 24 I think, however, we do know a couple of things, and 25 I tried to stress those when I spoke earlier. The things that
/~N
() Heritage Reporting Corporation (202) 628-4888
l 100 1 we know are, iirst of all, the BWR debris is going to be metals
? rich, very metals rich. And second, there's going to be an. ;
'1 3 awful lot of it, and third, we would expect that for all of the '
4 debris to come out of the vessel is going to be over as long as 5 15 hours1.736111e-4 days <br />0.00417 hours <br />2.480159e-5 weeks <br />5.7075e-6 months <br /> from the time of the initial pour.
6 I think we know that much now.
7 In Cliff's area, where he receives this debris coming 8 out of the reactor vessel and has to move it across the drywell 9 floor, he's explained to you the problems there. The CORCON 10 code.was not written for this. It does not understand low 11 melting temperature materials. If the BWRSAR code says this is 12 low melting material and it's. molten, CORCON takes one look at l 13 it and says it's frozen, and you've thrown awa'y the latent beat
() 14 of fusion out of the problem. j 15 This has got to be fixed. j 4
16 Another problem that we had is that there was no 17 degassing from the concrete unless the concrete were heated to 18 above its ablation temperature o that it was ablated, no gas 19 was predicted to some up out of the concrete and enter the 20 debris flowing over the floor. Consequently, the debris just 21 froze, whereas in reality, you would expect the steam released 22 from the concrete and the CO2 to cause metal water reactions in 23 the debris and actually increase its temperature rather than 24 have it freeze.
25 So this is where we stand. We're still working on O'# Heritage Reporting Corporation (202) 628-4888 m _
l l
l 101 l l
I this. In the meantime, of course, the analysts who are
('~))
u 2 considering the drywell shell responses conditions can assume
\
3 debris in contact with the shell and proceed'there. So we're ,
4 not holding them up. But we've got to someway or other get 5 across this breach where we can take the calculation all the ;
6 way through, and that is one purpose of the task group on shell l 7 response.
8 MR. HULMAN: That's fine. Thank you. ,
]
9 Next speaker is George Greene from Brookhaven _
10 National Laboratory.
-)
11 George, you ready? l I
DR. GREENE:
12 Sure.
13 (Continued on folle ing page.)
14 15 16 1 l
17 l
18 19 20 21 1 22 23 24 l 25 O Heritage Reporting Corporation (202) 628-4888 i
l l
)
1 102 :
1 DR. GREENE: I'm George Green, and I'd like to 2 discuss what I consider to be the technical basis for ;
3 evaluation of.the vulnerability of the Mark I steel shell to 4 melt through by core debris.
5 A little bit of background -- as part of the 6 containment load's working group, as long as possibly six years 7 ago, the activities were expanded to investigate the !
8 vulnerability of the Mark I shell to contact with core debris ,
9 as a method of achieving early containment failure.
10 And this was recognized very early in a containment 11 loads working group activities because of the small size of the 12 drywell floor and the confined pedestal with one opening.
13 As a result of that I was assigned to do some ss/ 14 parar #Lric mechanistic analyses; and the tools that I used with 15 the hybrid analysis with the CORCON Mod I code, and hand 16 calculations for shell response.
17 Before I go on I'd like to point out'that ironically, 18 for this problem, it turns out that CORCON Mod I might have 19 been better suited for doing this problem than CORCON Mod II in 20 its present configuration, as you may have detected from Steve '
21 Hodge's presentation.
22 The original analysis that I did assumed that the 23 failure criterion for the steel shell was complete steel shell 24 a b l a t '.o n . ;
25 The reasons why I assumed this analycos, other than Heritage Reporting Corporation (202) 628-4888
--n --
. l 1
l 103 1
1 the fact that I didn't know what else to chose, were to avoid
'( _/,_,] 2 the details of a detailed containment analysis, which at the 3 time, we didn't have the capabilities of.
~
4 To avoid detailed structure analysis; to avoid 5 complicated transcient analysis; and to avoid the appearance of 6 ambiguity and uncertainty in the results.
7 Unfortunately I failed dismally on all four of these 8 objectives, as you'll find out. .
9 However, the results of the calculations-were l
10 undeniable; and that is that most of the calculations I l
'll completely ablated the steel shell in minutes. )
12 The calculations I will show on the next slide -- the i
13 eight parametric calculations shown here in. tabular form -- six 14 of the calculations failed to steel shell in minutes; tue other
)
15 two did not fail to steel shell on complete melt-through.
16 However, if you notice in the right hand column, l
17 those two calculations that did not melt through suffered 18 partial ablation; and would have been considered to have failed 19 structurally due to loss of ultimate tensile strength at 20 temperature.
21 This concluded in the original analysis that the ,
i 22 steel shell's not only vulnerable to contact with core debris, j 23 but steel shell melt through is probable within minutes of 1
24 contact with core debris with the steel containment boundary; 25 and'should be considered, in my estimation, a dominant mode of A
(,) Heritage Reporting Corporation (202) 628-4888
" ^
~
.q l
l l
1 104
/~') 1 early containment failure in Mark I BWRs.
(s/ 1 2 After a considerable' period of time elapsed, a period 3 of inactivity, I should point out, the results of the 4 containment loads working group, particularly for the steel 5 shell analysis were factored into the NUREG 1150 activities.
6 In order to assess the impact of direct failure of 7 the steel containment shell on contact with core debris on
'8 risk. ,
9 A total of eight containment analysts were requested 10 to address the steel shell failure probability; and a' consensus 11 or an average, I should say, of the eight NUREG 1150 12 containment loads analysts agreed on a conditional probability 13 of failure of the steel.shell under all conditions of
.m 14 approximately 75 to 80 percent.
15 This was factored into the NUREG 1150 calculations; 16 and i.t should be no surprise that it resulted in a peak in 17 conditional early containment failure probability for the Mark 18 I steel shell of approximately 80 percent. I 19 This caused some great consternation. And two Mark I 20 meetings were held by the NRC late -- early last year --
21 approximately a year ago by Benero -- tried to resolve this 22 issue.
23 Without going into the details, the. meetings with the 24 NRC contractors and industry representatives were inconclusive, j i
25 No issue resolution'was achieved; and subsequent to this it
h l Heritage Reporting Corporation (202) 628-4888
105 became apparent that a new model had been developed by the L)
/ 1 2 industry to evaluate the conditional failure probability of the 3 Mark I steel shell by IDCOR.
4 Since I had been the, I guess, primary antagonist-in 5 this issue all along, I was requested by NRC to evaluate the 6 new industry submittal for IDCOR, for the IPE program on Mark I 7 steel shell failure.
8 A cursory examination of the BN1 analysis, this ,
9 analysic I did originally for NUREG 1079 -- NUREG 1079 is the 10 containment leads working for report. (ph]
11 And also the IDCOR model show the following. The 12 .IDCOR model resulted in unconditional survival of the steel 13 shell in contact with core debris.
5,_/ 14 - The BNL model resulted in quite frequent or highly 15 probable failure of the steel shell; and partial ablation of 16 those cases which did not fail by melt through. j i
17 Subsequently, a more thorough review of the IDCOR '
18 model in comparison to the BNL model was done. And it was 8 19 technical areas of dispute were identified between the IDCOR 20 model and the NRC model.
1 21 Just briefly, the reason why this is 'so important, l 22 some of you may not be particularly aware of this, is that in 23 factoring the steel shell analysis parametrically into the 24 NUREG 1150 calculations, just singularly the issue of liner 25 melt through on contact with core debris was sufficient to n
Heritage Reporting Corporation (202) 628-4888
106 f/') 1 increase the probability -- conditional probability of early j
(_/ .2 ' containment failure for Peach Bottom from approximately a range-I 3 of about 10 percent to a range approaching 80 to 85 percent.
4 So this was quite an eye-opener. And that's the l l
5 reason why we pursued this further. i 6 The next slide, please.
7 A detailed comparative assessment of both the IDCOR 8 and the BNL approaches to this problem were revealing. They I i
9 revealed dramatic differences in the problem -- differences l 10 that unless resolved eliminate any potential possibility for I
11 issue resolution in the state that the issue is currently 12 couched. )
13 The key features that we disagree on are shown here; C)
(,, 14 and they're not meant to be humorous. But they involve all the 15 underlying assumptions -- initial and boundary conditions; 16 physical properties of the debris, and key thermal hydraulic 17 phenomena that occur in the drywell.
18 I want to point out here and I want to point out very 19 strongly that there's a perception of a temporal perception in 20 this meeting of a growing level of technical uncertainty with 21 which to resolve these issues. It does not exist.
22 Data currently exists with which to resolve these 23 technical differences; and George Bancroft's point should be 24 taken well.
25 These are technical issues. The major physical Heritage Reporting Corporation (202) 628-4888
I 107 in
( \ 1 phenomena that form the basis for the dispute -- and I say
.Q
'2 dispute, not uncertainty. They form the basis for the dispute 3 between the industry position on this issue and my 4 position; and now the Mark I task force position on the 5 following.
6 That's the mode of boiling of water over melts. The 7 mode of solidification of cora melts attacking concrete.
8 Mixing of the phases -- the oxide metallic phases -- and the ,
9 transient phenomenology of spreading of melts in a dry well 10 configuration.
11 During a detailed examination of the technical basis 12 for the disagreement between the IDCOR and the BNL analysis, 13 the following eight major technical. positions were found to be
_/ 14 in disagreement.
15 First, the IDCOR position was based upon assuming 16 core debris in a very thin layer in the dry well. More 17 important than that, though, the IDCOR position-was based on 18 assuming that this debris was not only solidified, but 19 consisted only of low thermocon ductivity UO2.
20 The BNL analysis allowed for a mix of core oxides and 21 core metals in the dry well. In the IDCOR analysis, the heat 22 transferred from the debris to any structures including the 23 steel shell was by conduction only. l 24 The conduction, cssuming that this could be a melt to 25 structures was not allowed. The BNL analysis allowed for a O Heritage Reporting Corporation (202) 628-4888
,-. ,,e m ,-
3, l
{
108 l 1
('}
s_/
l molten core concrete interaction based upon prevailing i 2 thermohydraulic conditions.
1 3 And on freezing, allowed for a freezing mechanism by 1
4 slurry formation. j 1
5 Third point, heat was generated in,the IDCOR analysis I l
6 by decay heat. (ph) And this is okay if you agree with the 7 first assumption that there's only UO2.
8 There's no metal gas chemical reactions allowed. .
9 However, NRC analysis have shown that in fact it's the metal 1
10 gas chemical reactions that dominate the heat source in excess !
\
11 of BWR calculations; and they were allowed in the BNL analyses. l l
12 In the IDCOR position, the steel liner was convecting 13 .and radiating to a concrete shield wall at constant p,
(,,) 14 temperature.
15 The BNL position may be just about as bad. And that l 16 is that it assumed an 80 backslide boundary condition. [ph]
17 However, examination of_that assumption in the BNL 18 analysis reveals that not allowing for the backslide shield or l l
19 concrete to heat up enhances the artificial ability to radiate !
20 and add heat to that boundary, whereas allowing for the shield )
21 wall concrete to heat up will indicate to you, very quickly, 22 the temperature difference potential for thermoradiation 23 through that gap will shut-off, approximating a adiobadic (ph]
24 boundary condition.
25 If you want to pursue this further mechanistically, O Heritage Reporting Corporation (202) 628-4888
109. l
()
V
~
1 2
you can find,that after minutes that heat flux could be on the order of one percent of the total energy in the melt; and can I
3 be safely assumed to be an adiobadic boundary condition.
4 In the IDCOR position, water over the core debris was 5 assumed to boil at the critical heat flux. This is an old 6 argument, not soon to be resolved between IDCOR and us.
7 Film boiling was not a model. Clearly, though, at 8 the super heats achieved in all the analyses done for the Mark, l l
9 I steel shell to,date film boiling should be considered to be 10 the dominant mode. )
11 However, I should point out one other thing. In the' ,
J 12 original BNL analysis, there was no boiling capability in the
^
13 CORCON code.
gg .
(_/ 14 Later on in our containment loads working group 15 activities, this became more apparent when we were examining 16 conditions in the surry cavity.
17 And subsequently a cool and boiling model was placed l 18 into CORCON Mod II. However, if you compare a classical flat 19 plate film boiling, as modeled by, for instance, the Berenson 20 model, to radiation and "nvective heat transfer that one could 21 model from the surface of core debris, one would find they're 22 within 10s of percents of each other.
23 And assuming that there's a convective and radiative 24 surface for the core debris as opposed to potential for a 25 boiling surface is an adequate representation.
)
Heritage Reporting Corporation (202) 628-4888 vw i
l 1
110
; 1 The steel shell and the IDCOR analysis allows for I
( ) 1 2 heat transfer to an overlying pool by a' heat pipe' mechanism at 3 the nuclear boiling heat flux. !
4 In the BNL analyses, there was no overlying water 5 pool, and this was not pursued further. ;
i 6 The IDCOR analysis does not have any mechanisms i
7 whereby the core debris can reheat or heat up from its initial !
8 temperature. It is unconditionally a temperature drop. ,
)
9 Whereas the NRC models, which include metallic debris 10 and presence of gas phase chemical reactions -- metal gas phase 11 chemical reactions allowed to heat up -- this was in fact the.
12 case in the original BNL analyses we allowed to heat up. 1 J
1.3 And finally, in the IDCOR analysis, downward heat f
O 14 transfer from the debris to the concrete was assumed to be by
_,/
15 conduction. i l
16 This is not a criticism only-of the recent IDCOR 17 position on this. I have a similar criticism for other models.
18 Concrete was not allowed to ablate, decompose, or odor out gas 19 in the IDCOR analyses.
20 However in the original BNL analysis in NUREG 1079 a I 21 molten core concrete interaction was directly coupled to the 22 containment steel shell ablation analysis, thereby achieving a 23 consistent MCCI analysis with a structural analysis.
24 Presently, a comprehensive data base currently exists 25 with which to resolve the appearance -- and I say appearance -
O\- Heritage Reporting Corporation (202) 628-4888
i l
111 of technical uncertainty in the Mark I issue.
'}
~#
1 2 This is not only a criticism of some non-NRC people 3 doing these analyses. This also seems to be a problem some of 4 the NRC' contractors have.
5 It's a very extensive data base; and it should not be l
6 underestimated. The data does exist with which to resolve l 7 this.
l 8 In spite of the fact that there's a censiderable data 9 base for resolution of this issue, the NRC, .in particular :he ;
1 10 Accident Evaluation Branch under Mel Silberbarg, has redirected 11 several of their contractors -- myself included -- to address -
12 - to take their programs to address resolution of technical 13 issues with. respect to the mark 1 issue on a short term basis.
( 14 Short term basis meaning redirection later than the 15 formation of the Mark I task force which Steve Hodge said was i
16 last October. )
17 I want to emphasize that these recent research 18 efforts are not an expression of lingering technical i
19 uncertainty in the phenomenology.
20 On the contrary, all of the emerging technical 21 results from the redirection of.the accident evaluation 1
22 branch's research programs tend to support the original l 23 conclusions that the Mark I steel shell will fail early on 24 contact with core debris. !
25 I will discuss some of these recent experimental O' Heritage Reporting Corporation (202) 628-4888
~
{' ;
ss 1 results in what follows. j 2 At Brookhaven, I have redirected my experimental 1
3 program in short term within the last several months to address 4 the following technical issues. l 5 Trangient melts spreading across simulated Mark I 6 drywell floor; the effects of water on melts spreading across 7 . horizontal surfaces.
8 -
The trangient mechanisms of cooling, freezing and 9 crusting of bubbling metallic melt pools recalls Steve Hodge's 10 analysis indicates if nothing else that the initial melts 11 coming out of failed reactor -- Mark I reactor vessels should ;
12 be primarily heavily metallic. !
13 We have concentrated our efforts on looking at s- 14 metallic pools. In addition to that, ' fourthly, the I 15 thermoresponsive steel structures that contain bubbling j 16 metallic pools -- bubbling meaning the bubbling is to simulate 17 MCCI or molten core concrete interaction hydrodynamic activity.
I 18 We're also investigating liquid liquid boiling over 19 molter pools to resolve the boiling question. And also l 20 bubbling heat transfer to vertical boundaries.
21 This simulates the bubbling activity of the core 22 concrete interaction and contact with a not ablating solid 23 surface to simulate the Mark I steel shell.
24 I have an army of slides with which to go over the !
25 technical issues that I've just outlined in about as much O Heritage Reporting Corporation (202) 628-4888
+-p_ W I' c- -
e w-
113 e 1 detail as any of you would like.
f ]/
2 My better judgement has been appealed to, and I have 3 deferred that. So I will conclude that. However, I'd like to 4 let you know that I'm going to after the meeting compose a 5 document which will address the recent experimental results 6 from each of those six areas that I've just described, and 7 describe how they should be applied to the BWR Mark I steel 8 shell issue in resolution of the steel shell melt through. .
9 At this point, all I'll do is conclude.
10 Melt spreading appears to be hydrodynamically 11 ilmiting phenomenon. This is a recent result; Steve Hodge 12 expressed to you the difficulties they were having with a heat 13 transfer limited melt spreading.
() 14 15 All.of our evidence now, in recent experiments, with pouring of metallic melts at variable superheats, superheats 16 down to as much as 5 degrees Kelvin.
17 So the metallic melts are spread over horizontal i 18 surfaces as far as the hydrodynamics will drive them.
19 Simulated sumps, sir.ulating the pedestal sumps I l
20 indicate minimal retention to inertial effects. Since there 21 were no floor mounted obstacles in the Mark I drywell outside 22 the pedestal doorway, these experiments lead us to the l 23 conclusion that contact would melt with the steel shell is 24 likely within seconds of reactor pressure vessel failure.
25 GN Heritage Reporting Corporation (202) 628-4888 2 .
I
(
)
l 114 l
,_s 1 More recent discussions with the"BWR program at Oak l l ) ,
(_/ 2 Ridge is influencing them to change their melt spreading '
3 methodology.
4 We've also done experiments -- very limited 5 experiments with water on the floor to examine the impact of 6 water on spreading of metallic melts over horizontal surfaces.
7 Briefly, we find that the attenuation of core debris 8 flow toward the shell will be a secondary effect because water * .I l
9 in melts in this configuration interact into film boiling mode; 10 and that's not an efficient mechanism for retardation of these 11 flows.
12 Bubbling metallic melts with overlying cooling. This 13 is the melt, now, simulating the core debris. They cool in 14 three stages -- experimentally determined.
15 With superheat in.these metallic melts, they cool )
l 16 isothermally with no crust formation. When there's no i 17 superheat, they begin to freeze initially as a slurry; and the !
18 duration of tha't slurry formation with no crust formation is l 19 dependent upon the gas flux coming out of the concrete.
20 Finally, after a period of time, when approximately .
l 21 30 percent of laden (ph] heat has been removed from the melt, I 22 stable crusting can be sustained.
23 Water ingression into these freezing metallic melts 1
24 is not observed. Crusts are not expected to appear early in 25 core debris contact with the steel she,1f.
() Heritage Reporting Corporation (202) 628-4888
115
T 1 Steel containing boundaries that hold metallic
] 2 bubbling melts track the transcient temperature of the bubbling 3 melts. This is not a surprising result now, since I've already 4 presented information that crusts don't form thermodynamically, l
5 stable.
6 Initially there are no crusts; initially there is --
l 7 there is freezing by slurry formation -- finally crusting. The l 8 steel boundaries track the melt temperature identically.
9 Only the top centimeter of the steel shell in contact l 10 with the debris with an overlying pool boiling experiences any l 11 measurable temperature gradiant.
12 Pool boils over bubbling metallic melts, and film 13 boiling. Water ingression and nuclear boiling are not
() 14 observed, even under cases of intense gas injections simulating 15 concrete decomposition gas fluxes.
16 Crusting at the surface does not induce water ;
17 ingression or nuclear boiling. However the gas flux and the 18 sub-cooling effects enhance the film boiling significantly 19 above the flat plate film boiling.
20 This information has been submitted to the NUREG 1150 21 people for resolution of their issues in their steel shell 22 analyses, which were presented by Ken Bergeron in order to be 23 fair to the IDCOR position.
24 And they are also in the process of being compiled 25 and sent to Oak Ridge in order to be modified -- the cooling O Heritage Reporting Corporation (202) 628-4888 a
I 116 1 boiling mechanism with overlying water pools.
2 And finally, heat transferred from bubbling pools to l 3 verticle boundaries, simulating the steel shell -- is a 4 convective not a conduction process. l
)
S More recent experiments with Water pools to be i 6 repeated with oxidic pools and molten steel pool -- molten 7 metal pools -- indicate that at superficial gas velocities 8 substantially below one centimeter a second, which is minimum j 9 bubbling, heat transfer coef'ficients to the boundaries by this 10 bubbling mechanism, approached 10 to the 4 watts per square 11 meter degree kilogram. Very high heat fluxes, indeed.
12 This is what I call -- I've done hybrid calculations; 13 this is a hybrid slide.
14 All I'd like to do is conclude here. The original
}
15 NUREG 1079 analysis is old. And on the basis of its age -- six 16 to seven years old -- it's been criticized.
17 However, I will submit experimental evidence to 18 demonstrate that there are no compelling technical arguments 19 that are consistent with the data base which contradict the 20 original Mark I analysis that I did at Brookhaven -- the 21 underlying assumptions and modeling methodology are sound and 22 reliable.
I 23 MR. HULMAN: Before Dana Powers comes up here from l 24 Sandia National Laboratory, there are three messages outside --
25 one for John Fulton; one for John Kittinger; and one for Steve Heritage Reporting Corporation (202) 628-4888 l
i
k 117
/3 1 Floyd.
('#l 2 Please see Ms. *Kondulis for your messages.
3 Dana, you're next.
4 (Pause) 5 STATEMENT OF DANA POWERS, SANDIA NATIONAL LABORATORIES 6 DR. POWERS: The previous speakers have outlined 7 several phases of the NRC's research task force report on the 8 Mark I BWR liders. There was another area; and that.was the ,
9 question of how water effects the interactions of melts with 10 the liner.
11 The interest in this question arose because very 12 likely many accidents that there will be water in the drywell 13 of the Mark I.
( ,) 14 Further there may be interest in using water as a 15 deliberate mitigation effort. We were asked to examine the 16 available literature and experimental data to see if.there 17 would be significant effects of water on the interactions.
18 We were asked to do that examination in two areas.
19 First, would it effect the interactions of core _ debris with 20 concrete and the liner; and second, would it effect the 21 radionucleite [ph) source term in the event that the liner were 22 to fail.
23 I have to admit that in approaching this question of 24 does water prevent liner penetration we were heavily biased by 25 the results from TMI where the destruction of the core former Heritage Reporting Corporation (202) 628-4888
118 es 1 plate suggest there was considerable doubt about the efficacy 2 'of water at preventing liner failure.
3 One has to remember'that the water inventory in the 4 Mark I containment is going to be relatively small. It is 5 1.imited by the openings shown in the introduction to this 6 session -- by the downcomers, into the torus to depths of about 7 one to two feet.
8 If this is the only source, of course water of that ,
9 magnitude would quickly boil and disappear. If, on the other 10 hand,, there are sprays available, the water could be rapidly 11 replenished.
12 There've been a number.of contentions about what ,
13 water might do during the course of core debris interactions in !
(
,/ 14 the Mark I drywell region.
15 Some of these contentions are listed here. One of 16 the most important contentions, of course, is the mere presence 17 of water would quench the core debris.
18 The boiling would occur at a very rapid rate, 19 sometimes as high as the critical heat flux. In addition to 20 that conclusion, others have argeed that indeed quenching would 21 occur, but associated with this quenching would be the 22 production of a huge amount of hydrogen due to the metal water 23 reactions taking place simultaneously with quenching.
24 Others have argued that water on top of core debris 25 would simply boil. It would boil, and film boiling have a O Heritage Reporting Corporation !
(202) 628-4888 O , - a n
119
) 1 relatively modest effect on the course of interactions in the 2 Mark I drywell.
3 And finally, one of the long chestnuts of the reactor 4 safety community is the presence of water might induce steam 5 e.xplosions.
6 We looked at the available data base concerning these 7 contentions. And I have to be quite frank with you, we found 8 the data base far less extensive than a number of contention
- 1 9 that were made, i 10 The types of experiments we found seemed to be 11 directly applicable or listed here. The first of these is 12 really not a combined core debris-concrete-coolant interaction; 13 it is a steam-explosion experiment in which core debris was
( 14 created in a molten state on a magnesium liner, and water was 15 added to the top -- the so-called alternate contact mode.
16 Two experiments were done: one exploded and one did l I
l 17 not. I I
18 We concluded from this that in the area of steam )
19 explosions we were just not predictive; we couldn't predict 20 would they occur; and'if they did occur, we were not predicting 21 about what the consequences would be.
22 We proceeded on with the analysis as though steam 23 explosions did not occur. And indeed, no steam explosions have 24 ever been observed in the course of combined core debris 25 interactions with concrete with an overlying water pool.
O Heritage Reporting Corporation (202) 628-4888
120
,_ 1 Two test series directed to this type of' interaction
/T k_/! 2 were found. One called the Frag test series involved 3 fragmented core-debris that interacted with concrete for a-4 sufficient period of time.
5 The liquified concrete intruded into the core debris; l
6 and then water was added on the top.
7 What we observed from those experiments were there 8 were no violated (ph) interactions when water was admitted. On 9 the other hand, there was no hydrogen generation -- enhancement 10 of hydrogen generation.
11 Show you some results from those testa -- this is i
12 just a plot of the combustible gas generation -- that is 13 hydrogen and carbon monoxide as a function of time during the
% 14 test; and it shows you prior to wa'ter addition and after water 15 addition.
16 And I think you'll agree, there's no significant 17 difference in the amount of combustible gas generation in this ;
18 test associated with water addition.
19 On the other hand, you also see from that plot that 20 the interaction of core debris with the concrete was not 21 interrupted by the presence of water.
22 Rather, solidified crust formed; the water separated 23 from the core debris. Though this crust was impervious to 24 water, it was quite pervious to the gases being created.
25 Second set of test was so-called Swiss test; involved
) Heritage Reporting Corporation (202) 628-4888
,,,,y - .-- - - --
121'
,-w 1 liquified core debris -- molten core debris interacting with i
\) 2 concrete water added on top.
3 The conclusions were relatively similar. The 4 concrete interaction was not interrupted. The next slide will 5 show you a concrete erosion is a function of time.
6 During the test -- two tests; they were nearly 7 identical -- one with water and one without. And I think 8 you'll agree that there is almost no difference between the .
9 test.
10 We did identify other features of the combined 11 interaction, however. There was no added hydrogen. Boil ng of 12 the water pool at the top of the core debris was not at the 13 critical heat flux; it was at some lower heat flux.
() 14 15 This lower heat flux did suggest that there was some barbatoge. enhancement of film boiling.
16 We also identified a very important result and that l i
17 is that the overlying water pool, which in the cdse of the 18' Swiss test was about 60 centimeters deep -- that is about the i
19 depth of water pools that you could have in a Mark I BWR -- and 1 l
20 sub-cooled, did attenuate.substantially the aerosol generation, l
21 The conclusions we reached -- the first conclusion we l 22 reached was that the data base on combined core debris concrete 23 interactions is extremely limited.
24 There is a data base. But most of the experiments 25 can be criticized on.the basis of whether the debris was Heritage Reporting Corporation (202) 628-4888
- g _
,m 0 .-- . , . . _ _ . - -- .,
122 1 totally prototypic, and on basis of scale.
2 (Continued on next page.)
3 4 l 5 '
6 7
'8 .
9 10 l l
11 12 13 14 15 16 l
17' 18 l 19 20 21 22 23 24 25 O IIeri?. age Reporting Corporation (202) 628-4888
1 123
,/~ - 1 The preponderance of the information was that we
\"# 2 could not use any experimental information to prove that water 3 would significantly interfe in the attack of core debris with 4 concrete or the liner, or that it would in any way prevent the 5 attack of core debris on the liner. It appeared that George 6 Greene's conclusion that liner attack was possible in the core 7 debris at about the same rate as he had calculated without 8 water being present was being preserved by these analysos. .,.
9- We then turned to the question of source terms 10 recognizing of course that if liner failure were to occur 11 unless there are source term consequences of liner failure, it 12 really has no risk impact.
13 We recognized that water would be present in the
) 14 . cavity largely because of the action of sprays. The effects of 15 these sprays would be multi-fold. First of all, the sprays in 16 the drywell containment easily operate at flow rates 17 sufficiently high that they would create water pools overlying
^
18 the core debris, that the sprays themselves would act as 19 at+enuate aerosols and they would tend to cool the atmosphere.
20 (Silde) 21 The effects of pools on aerosol generation to remelt 22 concrete in*.aractions has been established experimentally.
23 Show you some data from the Swiss series tests. 'inis a plot of 24 the aerosol source rate as a function of time showing before 25 water addition and after water addition. You notice that prior !
l l
A Heritage Reporting Corporation (202) 628-4888
, , , -n ,
124
,s 1 to water addition, the source rate amounted to several 2 milligrams per second. After water addition, the source rate 3 dropped by over an order of magnitude. The predominant cause 4 of this drop in the aerosol source rate is the fact that the '
5 water pool in the Swiss test was kept subcooled. Any steam and 6 gas bubbles entering experienced condensation that trapped most j 7 of the aerosols.
i 8 (Slide) , l I
9 The r. rapping effect of the sprays themselves in quite 1 10 important. The calculations have been attempted for the action 11 of sprays in a Mark I type BWR confinement. I'll show you some l l
12 next results on che .> ext slide.
13 (Slide)
()
14 This is a plot of the suspended aerosol mass in a 15 Mark I BWR containment, as a function to time given the sprays 16 are actuated. Two limiting cases are looked at. These 17 limiting cases correspond to the characteristics of spray 18 nozzles in an exist:.ng Mark I BWR, with the exception of the 19 fact that the flow rate has been cut down by a factor of ten.
20 otherwise, the particle size of the spray droplets and the 21 initial velocity of spray droplets are as specified for the-22 specific plant sprays.
~
23 What you see that even in the worst case, the aerosol 24 mass has e folding time of about twenty minutes or so. There's 25 a sharp reduction in the aerosol concentration.
( Heritage Reporting Corporati~on (202) 628-4888 4
-Q- -
1 I
i 125 i l
1 Aerosol concentrations can be kept at a very low equilibrium 7-) l 2 level by the actions of these sprays.
3 (Slide) 4 The conclusions that we reached on this matter were 5 that drywell sprays, especially if they were augmented by 6 additional sprays in the torus rooms could attenuate aerosols 7 generated by melts after they had penetrated the drywell liner 8 could well attenuate radionuclide releases after liner failure.
9 (Slide) 10 We reached an additional speculation that these 11 sprays might also have another tenefit in that they could 12 significantly attenuate the so-called revaporization source 13 term, that is the vaporization of d,eposited radionuclides from
/i 14 the reactor coolant system late in the accident, because of U
15 their ability to keep the drywell cool.
16 This is the substance of the conclusions in the 17 Appendix on the effect of water, that is, that currently we 18 find the affects to be minimal as far as preventing drywell 19 liner penetration, but the affects of water could be indeed 20 substantial as far as reducing the source term.
21 I want to conclude the talk by noting that the NRC 22 has now sponsored some additional experiments on the affects of 23 core debris interacting with concrete in liners. The objective 24 of these experiments are to validate computer models that are 25
~
being used to reproduce the calculations that George Greene 1
(/ Heritage Reporting Corporation (202) 628-4888 '
i nw - - - - .
126 1 described as calculations done six years ago and not to try to
~)
2 simulate a SWR reactor accident.
3 The emphasis in these experiments is to describe heat 4 flux relationships which you will see in a variety of 5 calculations being presented today is a general schematic 6 portrait of core debris interacting with concrete shown on this 7 slide, that is, a liner with core debris adjacent' to it, 8 separated by a solidified crust. .
9 In order for this liner to be penetrated, the 10 convective heat flux from the core debris must first ablate the 11 crust and then ablate the liner. A major question that exists 12 in most of the models to date is what is that convective flax 13 .to the liner.
() 14 (Slide) 15 The first experiments that we are attempting to do 1
16 are integral validation experiments. They are relatively l 17 simple experiments in which core debris is poured into tubes, 18 steel tubes with concrete plugs. The temperatures on the back 19 side of the liner are then monitored to obtain data that could l
20 be directly used to the codes to validate those codes.
21 And in general, the tests are relatively small, can 22 be done quickly, a variety of tests and types of melts are l 23 being planned to be used in these experiments. The first of 24 these experiments has been done. I will show you some data 25 from temperatures of the back side liner as a function of time.
( Heritage Reporting Corporation (202) 628-4888
127 gw 1 (Slide)
O 2 You notice that the temperatures rise rapidly due to 3 the convective heat flux from the liquefied core debris to the 4 steel. The convective action of gasses evolve from s the S concrete into the core debris causing a rapid temperature rise 6 can be clearly seen in a second experiment. By comparing these 7 results to that of a second experiment in which instead of a 8 concrete plug, a steel plug was present. And you can see th'e .
9 temperature rises are much more languid.
10 (Slide) 11 Some initial comparisons of these data to at least -
12 one of the models being used for the analysis of liner 13 penetration have been made,. This is a plot of the axial
()
~
14 temperature profile in the liner at a particular function of 15 time. A comparison of the data and the calculation code shows 16 relatively good agreement between the codes and the 17 experimental data.
18 In this sense, I'd like to echo George Greene's 19 sentiment iu that it appears that matching these data is going 20 to be a relatively easy job with the codes that are available 21 to analyze liner penetration. It suggests to me that the 22 predictions of these codes may indeed be reliable.
23 (Slide) 24 From these kinds of predictions, the neat transfer 25 coefficient can be inferred and this is just a plot of heat
(, Heritage Reporting Corporation (202) 628-4888
-128
() I 2
transfer coefficient inferred from the data as a function of time. And you can see that the convective heat transfer 3 coef ficient amounts to something cn1 the order of 1500 watts per 4 square meter to degree Kelvin.
5 To get some more specific information on the 6 convective heat transfer caused by gasses sparging the core 7 debris, George Greene described to you experiments he's 8 undertaking using simulant mater'ials. We hope to be able to
)
9 extend and validate that correlation he develops by using more 10 prototypic materials in a series of tests called the witch 11 tests.
12 (Slide) 13 These tests are involved inserting into molten steel
( 14 melts sparged with gases at controlled rates, steel probes that
-15 are instrumented. The types of tests we plan to do are shown 16 here using melts of steel and stainless sceel and varying the 17 gas flux using inverse heat flux calculations to provide a 18 correlation of heat flux versus superficial gas velocity.
19 And with that, I'll conclude my presentation.
20 (INSERT) 21 MR. HULMAN: Dana, can you give us a one-mi,.ute 22 summary?
23 DR. POWERS: Certainly. I believe I summarized by 24 saying we don't find evidence that the presence of water in the 25 Mark I cavity will interfere significantly in the core debris O Heritage Reporting Corporation (202) 628-4888 l-
129
- 1 attack on the liner. The penetrations will be rather,similar
\/ 2 to those the previous speaker described.
3 Th'e water available in the cavity will have a 4 significant effect on the amount of radionuclide release that 5 can take place as a result of core debris concrete r
6 interactions. Finally, the NRC research department is 7 continuing to sponsor experiments to validate models of core 8 debris concrete liner attack, models being used for the ,
9 analysis of the Mark I situation.
10 VOICE: What is the barbot, age effect?
11 DR. POWERS: The question was what is the barbotage 12 effect enhancement of film boiling. Barbotage is a term 13 invented some time ago to describe non-condensable gases coming 14 through a porous plate and the heat flux effect on that. In
)
15 this case, it is non-condensable gases coming from the melt
- 16 concrete interaction enhancing the boiling heat flux.
17 MR. HULMAN: The next speaker is Tom Kress from Oak i
18 Ridge. j l
19 Tom?
20 21 22 23 24 25
' () Heritage Reporting Corporation (202) 628-4888
l l
130
) 1 STATEMENT OF DR. TOM KRESS, OAK RIDGE NATIONAL LABORATORY 2 DR. KRESS: Thank you.
3 The subject of my presentation is the thermal 4 response'modeling once the melt' arises at. .I'm not quite ready 5 for that slide but that's all right.
6 (Slide) 7 The reason we are making this analysis, one eason, i l
8 is that we happened to have available to use a computer code , l 9 called HEATING 6. Some of you'may be familiar with it. It's ;
10 been around many years. It's a very flexible three dimenslor-'.
11 heat transport code that allows phase changes, internal heat 12 generation, very flexible boundary conditions you can put in.
13 Time varying temperatures or heat fluxes or convection boundary.
14 conditions. '
15 (Slide) 16 The way we're using this code as illustrated by this 17 slide, we don't tell the code there is a crust there. We let 18 the code tell us whether or not a crust forms. It initially 19 determines the perfect contact temperature that one would get 20 with liquid against liquid when the melt' touches the wall. !
I 21 Theres a classical equation for that that's built into 'the l
22 code.
23 If that contact temperature is below the crusting 24 temperature of the melt, the code will immediately start 25 forming a crust which can grow -- it's a dynamic model n
V Heritage Reporting Corporation (202) 628-4888
131 1 depending on the heat going into and out of that crust, the 2 crust may grow to some equilibrium level or.it may grow to some 3 value and then begin'to shrink and entirely disappear as the 4 liner heats up and the heat losses become less.
5 The model uses as boundary conditions what the core 6 melts anu core concrete spreading equations tell'us exist when 7 the thing gets'to_the liner. These boundary conditions we need 8 for this problem are the heat transfer coefficient. This is a, 9 convective process. The melt temperature, the crusting 10 temperature which involves its composition, and in general, the 11 time varying values of-those and the internal heat generation 12 thet may be due to the K heat and or chemical energy.
13 It accounts for all of these. The crust is allowed 0)
(, 14 to have water on top of it that is boiling and we do model the 15 full flat plane horizontal pool boiling curve. We let the code 16 tell uc whether its film boiling or nuclear boiling depending 17 on the temperature difference. It allows heat to go downward 18 out of the crust into the concrete. The heat going into the 19 liner is conducted upward and downward. It also lost off the 20 buckside the natural convection and radiation to the concrete.
21 In other words, I think we model every important 22 phenomenon that could be occurrin: in this process.
23 (Slide) 24 The next slide shows an example calculation using 25 this code. This was for case rhort term rtation blackout. I t-
\ Heritage Reporting Corporation (202) 628-4888 m
i 132
. 1 does have a high superheat in this case. Ne use some numbers V 2 Cliff Hyman told us for the melt conditions when it arrived at 3 the liner, and so you should view this as an example 4 calculation, not an actual problem, please. Because in this 5 c,a s e , it does have a relatively high superheat and does have a 6 reactively high heat transfer coefficient as told to us by 7 CORCON.
8 But you can see what I've plotted is the liner ,
9 maximum temperature as a function of time at different 10 positions across the cross section of the liner at that maximum 11 temperature location. You can see it heats up and eventually 12 gets to what I would consider its melting point, and you can 13 see the melt progressing across the liner.
() 14 15 (Slide)
The reascn for the flat part of the curve -- there's 16 a number of reasons for that -- one was CORCON input to this is l '/ a time variable input, has a time varying heat transfer 18 coefficient, has a time varying melt temperature and a time 19 varying crusting temperature. Those are inputs. I 20 Not only that, the crust is beginning to form. There i
?
21 was a crust that started here and it grew giving some 22 protection to the liner, but then it melted away again and 23 disappeared. And so in this case the crust was only a l l
24 transient behavior. In general, it's difficult, it would be a 25 right thing to do to use a code like this a run a lot of cases
() Heritage Reporting Corporation (202) 628-4888 i
133 fS 1 to see what the sensitivities of this kind of result to the V 2 various boundary conditions and inputs and material properties.
3- I didn't have time to do that for this meeting so I 4- developed what I could call the correlational model. It's a 5 hand calculation put on a PC and basically is a correlation of 6 these HEATING 6 results.
7 (Slide)
R And the way it works is illustrated here. It's a ,
9 steady state problem. It doesn't calculate the transient 10 crust. It assumes a crust is there at some value that the 11 solution to the problem will tell you what the thickness of it 12 is. What it does is it starts out with a model equation that 13 is basically the Mitchell and Finn equation. Therefore it
() 14 15 would be a relatively exact equation when there's no crust.
As the crust gets thicker and thicker and you get 16 two-dimensional effects, the two major effects of the crust a:e 17 it allows heat losses off the top and bottom to the water and 18 concrete and it also redistributes the heat upwards towards tte 19 cooler parts of the liner.
20 A correlational model has to account for those heat 21 losses and that redistribution of the heat. The model does do 22 that and it does it quite accurately. But it has to be J
2 .1 benchmarked to the HEATING 6 results. And I did that by 24 running a number of Heating 6 results that I knew would be 25 under conditions that would not melt the liner.
~3 g,/ Heritage Reporting Corporation (202) 628-4088
134 -
1 (Slide)
{
N- 2 And an example of that is shown on the next slide. !
3 As a matter-of fact, I used by burlap bench test for boiling I
4 water rector liner analysis program. That's to let you know l l
5 it's not made of rteel. I used that to decide under what !
l 6 conditions the liner would not fail, and then I ran the Heating 7 6 code and this is the result I got. l 1
8 And I have built into this correlational model, ,
9 something I call an E factor, which allows me to adjust the i
10 heat going to the liner, and mine's a burlap liner, to account l l
11 for those heat losses and those redistribution of the energy. i 12 And it does it quite well.
13 This was a case where you had quite a thick crust
- 14 form. This is about an eight centimeter crust. The way you do i
)
15 that is by having a very low superheat. I did this for a-16 number of crust thicknesses and found out this E factor was 17 about .85. I thought it would be a variable as a function of 18 crust thickness and it turned out a constant value gave a good l l
19 correlation of the Heating 6 result under all cases.
20 I use material properties in this that are for an 21 oxide crust, by the way. This does not represent what might l I'
22 happen with a metallic crust. I then used this burlap model to 23 run a number of sensitivity studies.
24 (Slide) 25 These are shown on this slide. What I have plotted Heritage Reporting Corporation N-(202) 628-4888
~
l 135
~'T (J 1 2
~
here is the maximum liner temperature that would be achieved as a function of the various'cructing temperatures. .Three of them 3 are shown as a function of the melt temperature associated with 4 that crust temperature.
5 In essence, then, it's a superheat, the difference 6 between the melt temperature and the crust temperature as a 7 function of four different depths of the melt itself.
8 (Slide) .
9 This is for one heat transfer coefficient that's 10 really not shown over there, but it was for 10,000 watts per 11 meter squared per degree K. I did it for a number of those ;
12 downward to 5,000. I have not yet gone below 5,000. l 13 But in essence, you can see that by'the steepness of s
( ,) 14 these lines, that the maximum liner temperature is very very i
15 sensitive to the superheat. It is very sensitive to the depth l 1
16 of the melt you might have, and by the other curves which j 17 aren't shown here, sensitive to this heat transfer coefficient.
18 I cross plotted this, a curve like this I cross 19 plotted to show the maximum superheat that one could allow in 20 your melt before the maximum liner temperature would arrive at 21 1,000 degrees K. 1,000 degrees K is where the liner gets 22 pret.y solid. So this is the maximum superheat as a function 23 of the melt depth that you can allow in this melt before the 24 liner maximum temperature gets there as a function of two 25 different heat transfer coefficients, and as a function of two l
O'- Heritage Rep'orting Corporation (202) 628-4888 0
136-
/T 1 different crusting temperatures, one of them 1800 K and one of V
2 them 3,000 K.
3 (Slide) 4 You can see by this that for significant melt depths, 5 you have to have a fairly low superheat to keep that liner from 6 getting this hot. Bt t if the melt depth is not very high, you 7 can*have pretty sign!.ficant superheat. And'it's pretty 8 sensitive to-the heat transfer coefficient. Not only that, 9 it's interesting to note that the actual melt temperature is 10 not very important. The crust will protect that. It's the 11 superheat that matters, and I think that's an interesting 12 finding.
13 (Slide) *
() 14 So basically, I'd like to summarize. What we have is 15 a computer model that I think accounts for all the important 16 features in this. It's ready to go, verified validated model.
17 We can use it. You give me the boundary conditions, the 18 temperature of the melts, the crusting temperature.of th' heat 19 transfer coefficient, and I'll tell you what the liner does.
20 I'll give you its transient behavior.
21 Also have a correlational model that's pretty good.
22 It's very handy because I can run thousands of cases in a day's 23 time and I can plot them up and it gives you good insight. It 24 tells'you what the important features are.
25 With that model, I've already determined that the Heritage Reporting Corporation (202) 628-4888
137
'f~)
v 1 sensitive parameters are these: the melt superheat, the melt ;
2 depth, and the melt to crust convective heat transfer 3 coefficient. Those are the things I need to know to really 4 resolve this problem.
5 I've made an editorial comment, these are all 6 uncertain values. Those are the things that I get from the 7 core melt progression code and the core concrete codes as they 8 spread the thing out. I need those things before I can tell ,
9 you what the answer is.
10 (INSERT) 11 MR. HULMAN: The last speaker this afternoon before 12 questions and answers is Raj Sehgal from EPRI.
13 Raj?
14 15 16 i
17 l 18 l
19 20 21 22 23 24 25 O Heritage Reporting Corporation (202) 628-4888
. ~ .
138 1 STATEMENT OF RAJ SEHGAL, ELECTRIC POWER RESEARCH INSTITUTE 2 DR. SEHGAL: I'm going to make some comments on the 3 issue of the code to the attack on the containment liner, the 4 interchange of using the words shell and liner is the same 5 thing.
6 .(Slide) 7 First, I want to point out the differences that come 8 about in the various analyses. ,
9 (Slide) 10 Let me give the next slide which talks about the 11 parameters that are really important. Number one is whether 12 there's water in the drywell. That I think is a very very ,
l 13 important consideration. If there's water, lots of estimates 14 will change.
)
15 The second one is what is the melt whether it 16 contains alloy zirconium or it doesn't, which predictions weigh 17 on that, t o o', but the amount of melt that's available for 18 discharge.
19 Thirdly are the melt discharge characteristics which l 1
20 vary from one analysis to another, which I'll talk about. !
l 21 And lastly, the concrete type, there was discussion l 22 of various limestone, limestone or limestone sand, but it's 23 been escablished now that this is limestone stand concrete for 24 Peach Bottom.
.25 (Slide) ,
i Heritage Reporting Corporation (202) 628-4888 l
i 139 i i The water in the drywell at the Peach Bottom plant
{a"'g 2 can accommodate up to about 2.5 feet which is like .75 meters.
3 This may induce FCI interactions and there would be some debris 4 quenching and debris dispersal accompanying that. So th,s 5 initial water depth is important to consider if you want to 6 find out what is the depth of the debris which goes towards the 7 liner.
8 If there's a continuous spray oper, tion; there will ,
9 be water on top of the debris _and that will lead to heat 10 removal from top surface, and while the debris is traveling 11 towards the liner and secondly after it's reached the liner, if 12 at all,' the liner will act as a fan to remove the heat. '
13 (S2ide)
() 14 15 In the initial condition specifications, I think the injector modeling is the most important phase, as Steve Hodge 16 has been pointing out too. In the ORNL scenario, the melt is 17 quenched in vessel and then it's remelted, and metals segregate 18 from the oxidic melt. And this substantial amount of zirconium 19 present and this relatively low superheat. Its just melted 20 material which is attacking the vessel head and is exiting the l l
21 vessel. But there's no water in the vessel behind the melt. l 22 In the case of the MAP scenario, this melt is not 23 quenched in vessel and there's a lot of water available behind l 24 it. If the melt is relatively homogeneous, it's not segregated 25 into metal and oxidic melts. But'also has a low superheat. .
l
() Heritage Reporting corporation (202) 628-4888
. _. e !
i J
140 1 -(Slide) j b'
'- 2 The way the melts are exiting the vessel are, j 3 different too. In the ORNL scenario, about 50' tons of the melt 4 may exit in the first five to ten minutes, and then another.2$0 5 tons follow it and comes in a protracted fashion for the next 6 approximately four hours. Metals come out first and steam 7 follows the metal, ar.d then maybe some hydrogen generation
. .8 while steam is interacting with the melt in the in-pedestal. ,
9 The most, important point is the debris is not 10 quenched in the pedestal and it's available to flow ou,t towards 11 the 'iner l as soon as it comes out. It depends upon its 12 superheat. ;
13 In the MAP scenario, the melt is discharged in a 14 sudden fashion, about 80~ tons of it comes in the first one
) ,
15 minute or two minutes, then another 80 tons fo11cws in the next 16 100 minutes. It's a homogeneous melt, as I talked earlier, but 17 water is following that, and as soon as water comes out, it 16 quenches the melt in the pedestal and the driver.
1 19 (Slide) 20 The next picture shows you the rates of discharges of 21 the picture of comparing MAP discharge versus the ORNL and you l 22 see the differences as I listed in the last slide. The total 23 melt discharge in MAP cases is smaller but comes out a little 24 faster, quite a bit faster than tne ORNL case.
25 (Slide)
({) neritage Reporting (202) 628-4888 corporation I l
l
_ _ , % _. ~ __ g -. -
141
, 1 The next slide shows you in the MAP scenario, the
) ,
- s. / 2 quenching of the melt. This is the water following the melt i 3 into the pedestal and the temperatures drop for the podium, and 4 then reheating starts and it takes a long time before the melt 5 starts, is molten again and starts to move.
6 In the case of drywell area, the melt temperatures 7 don't reach steel melting temperatures for a long time.
8 (Slide) ,
9 Now, let's consider the melt spreading case. We 10 heard from George that it's a thermodynamic phenomenon and that 11 it should be treated mechanistically from Cliff Hyman. Cliff 12 Hyman computed on a thermodynamic basis we have done some 13 calculations to spread the melt in the hydrodynamic fashion,
("') 14 mechanistic calculations defined the controlling parameters of
(./
15 the melt discharge rate, the initial superheat, how much 16 zirconium is there, how much it oxidizes the water quench it.
17 without water in the drywell, the melt will spread to the liner 18 and the heat radiation rejection will be somewhat compensated 19 by the heat addition zirconium oxidation.
20 With water in the drywell, the superheat for the two 21 cases can be identified with the low superheat case, there'll 22 be some dispersal of SCI and melt quench is possible and maybe l l
23 the melt will not reach the liner.
24 In the case of high superheat, dispersal is helpful i
25 and maybe the melt would reach the liner, although we think h Heritage Reporting Corporation (202) 628-4888
1 i
i 142 1 that the data base really we need, we don't have it yet, and we 2 need to qualify these molds with some experiments. l l
3 This is a transient calculation of the movement.of l 1
-4 the. melt towards the liner. It's 54 tons of melt deposited in ;
. i 5 one minute. It's a homogeneous melt and it's exiting the j 6 pedestal doorway and as the transient calculation built up like i
7 a liner.and then it comes out in a 90 degree angle, This is .
8 the assumption, and hits the liner, then disperses in a planar 9 in the radial annulus. Also indicated is the equilibrium depth !
10 for uniform distribution of 54 tons of material.
11 (Slide) 12 The next slide gives you the superheat. Just'now, i 13 Tom ta'1ked about the superheat of the liner of the melt as it 14 comes to the liner. You see the' drop.in superheat. The 15 initial superheat was 157 K. In thi's case, it dropped to about 16 55 or 60 with the zircalloy oxidation present and without that 17 to less than 20. '
18 So the melt in this analysis looses a lot of !
19 superheat as it travels towards the liner.
]
20 (Slide) !
21 The next question we asked was whether the heat load 1 1
22 imposed on the liner can be removed by water if there's a water 23 layer on top. We looked at two cases. This analysis from 24 Professor Kazimin from MIT. Two cases, one in which there's no 25 water on top of the liner, it's only radiation heat loss and
]
0 Heritage Reporting Corporation j (202) 628-4888 I
l
t 143-1 the second one with a water layer, theres nucleate boiling in.-
2 -the water region. Simple analysis results.are shown'on this 3 slide that the melt'is contacting the liner. There's a crust 4 f ornied . The-heat flux into the liner is Q and this heat i 5 rejection from the tops of this from the top of the crust and 6 maybe too as radiation or as nucleate boiling to water.- !
7 (Slide) f 8 The case here plotted is for a depth of debris of ten 9 centimeters and there's the case on top which is radiation 10 only. There's no-water overlayer. And the second one is with !
11 the water overlayer. The point to make here is for a liner-I 12 temperature of thousands degrees centigrade, one for water :
13 overlayer case really you can' go to very high heat fluxes and !
14 not subject the liner to very high temperatures. l i
15 For the case of no water overlayer, the radiation ;
r 16 alone, temperatures of 1,000 C. can be reached in about eight !
17 to ten watts per centimeter squared heat flux.
18 (Slide) j 19 The next case is for 20 centimeters liner, the depth 20 of the debris is 20 centimeters at the liner, and here one can 21 tolerate heat fluxes of about 10 to 12 watts per centimeter 22 squared.
23 (Slide) 24 The question now is the heat fluxes in cases of i
25 molten melts meeting lower temperature slabs and this is a O Heritage Reporting Corporation ,
(202) 628-4888 j I
l l
144 1 calculation done in CORCON. This was also performed at MIT.
h 2 This is the drop in downward heat flux with the melt sitting on 3 top of concrete as the temperature is reached. These are 4 various cases here. One is steel metal melt, one is steel plus ,
1 5 z.irconium metal melt, and there are a couple of cases of coal 6 oxide and coal plus concrete oxide.
r 7 As this temperature is reached, the heat fluxes drop 8 by two orders of magnitude for the oxidic melt and by at least, 9 one order of magnitude for the metallic melt. So this question 10 really then revolves around whether the crust is stable. And 11 we have performed some experiments in MIT. This is similar to 12 material experiments. It's mixture of water and cyclohexane i 13 lightly layered representation of the layered melt in CORCON on 14 top of a porous plate which is removing heat and air is being 15 injected from the bottom. And the crustability was 16 experimented on.
17 It was found even with 1\26th centimeters per second 18 flow velocity, the crust was relatively stable. There were 19 little spots in the crust where the gas would come through. So j 20 we feel that with the crust stability being relatively shown by 21 this experiment and maybe some other experiments if necessary, l
22 it appears that the crust form will not al' low high heat fluxes I l
23 into the liner. l 24 (Slide) 25 I have the last slide which gives the conclusion.
Heritage Reporting Corporation (202) 628-4888 A
l I
l 145
'/] 1 The conclusion is that for debris depths less than 20 C
2 centimeters, liners may be coolable. Of cou.rse, the data base 3 is not really sufficient yet.
- 4. (INSERT) 5 MR. HULMAN: It's question time and discussion time.
6 I'd like to start off by asking Raj a question.
7 Do I understand correctly that this work at MIT l 8 indicates to you that the liner is coolable if you have water ,
9 on the floor? .
10 DR. SEHGAL: It appears to say that, yes.
11 MR. HULMAN: Is that a position indicated by 12 industry? Has industry generally accepted that work?
13 DR. SEHGAL: Yes.
O.
14 MR. HULMAN: Questions?
15 Please stand up end identify yourself for the i
16 audience and address.your questions to whomever you wish.
17 DR. HILL: Paul Hill, PP&L. i 18 I have a question for Cliff Hyman. He indicated 19 earlier --
20 MR. HULMAN: Hang on one minute. Why don't you come 21 up here. They can't hear in the back.
i 22 DR. HILL: This question is for Cliff Hyman. He 23 indicated early in his paper that the method of modeling the 24 spreading of the debris outside of the pedestal was to be sure l
25 that it didn't melt. )!
i O'
Heritage Reporting Corporation I (202) 628-4888 )
I i
146
, 1 That rather defies my physical intuition for that t 2 process. I would think that it would spread until it fully ;
^
3 occupied the available space or until it froze.
4 I'd like to hear an explanation of why that choice -
5 was made. -
6 DR. HYMAN: I would like to say that George does
~
7 agree with you. He's done experiments where the viscosity's 8 very low and he says the metals were spread almost ,
9 incefinitely.
10 But if you remember one of the slides that-I- ,
11 presented, it showed the in-pedestal debris temperature as a .
12 function of time. It dropped very very fast. It dropped below 13 the solidification temperature of metallic debris. And there 14 has to be a limit over which you cannot have an infinitesimally
)
15 thin debris layer all over the drywell floor. It takes time to 16 flow. So that's my argument.
17 I agree it is limited.
18 DR. HILL: Well, I believe that assumption should be ,
1 19 very carefully investigated. I think that it might have a very l 1
20 strong influence on a result. I 21 DR. HYMAN: Yes.
22 DR. THEOPANOUS: I'd like to agree with the statement 23 that Cliff made. I think that if you're not indefinitely )
1 24 spread things down to mini, mini sizes. And in this l 25 connection, I think I'd like to ask a question to George i
Heritage Reporting Corporation (202') 628-4888
147
_( 1 Greene, because I have some concern about this indefinite V) 2 spreading that I seem to be hearing.'
3 How well his experiment model the radiation heat flux 4 which in this case is going to dominate the heat losses and how ;
1 5 did he establish a relationship between the experiment and the J 6 reactor.
7 MR. HULMAN: George?
8 Please come up here to ask your questions,.and'those ,
9 people that are responding also please come up.
10 DR. SEHGAL: Well, they ca'n use that microphone there 11 if they speak loud.
12 DR. GREENE: As I began to say, our experiments do i
13 not adequately. represent the radiated heat losses on the
( 14 surface. That's because of the conditions under which the 15 experiments performed are with non-reactive materials 16 experiments intentionally done with lower superheats to 17 investigate whether undar those conditions the limitations to 18 the spreading is hydrodynamic or thermodynamic.
19 with the melts that we had used at superheats down to 20 5 degrees Kelvin, we have found -- and I never used the word, 21 indefinite -- that melts spread until the hydrodynamic 22 limitation. In these experiments, that hydrodynamic limitation 1
l 23 was a balance between buoyancy forces and surface tension l
24 forces. The melts spread and we take videos of them. j 25 And after the spreading stops, the surface is still O Heritage Reporting Corporation (202) 628-4888
l 148 1
) 1 outgassing and we can se~e bubbling activity in our spread out 2 melt, and we're investigating several of the geometries now and 3 several other impacts like water and sumps. But without sumps ,
4 and without water, we find-in our experiments with molten lead 5 -- and like I said, these are only very recent -- we only want 6 to find out whether the fundamental processes are thermal in 7 nature or whether they're hydrodynamic in nature. In our 8 initial experiments, I think we've performed 15 so far at .
9 variable superheats geemetries and masses.
10 So far, we've covered 4 kilograms to 20 kilograms i 11 superheats of 5 degrees to 150 degrees one geometry one height i 12 of fall. We found that in all of our cases, the limitation to !
13 the ultimate spreading for these experiments was hydrodynamic. I 14 We particularly rely on the Sandia Natibnal Laboratory to do
. 1
- 15 parallel experiments with real reactor materials in which they i
16 can use steel melts at those kinds of temperatures over i 17 concrete surfaces. We can't do that at Brookhaven. The safety i
18 people won't let us.
i 19 MR. HULMAN: Follow up question? l 20 DR. THEOFANOUS: Yes, just.a short one. I think that
! 21 I understand your experiments that's what happened. I only
! 22 word of caution because if you don't put the right heat fluxes,
- 23 then you distort your time scale and then we have to worry 24 about time scales and length scales which I'm sure you do the 25 experiment in small parameters.
Heritage Reporting Corporation 4
(202) 628-4888 1
i
149
) 1 .
So I don't disagree with -- and I don',t say that it 2 will not necessarily spread. All I'm saying is you have to 3 make sure from scaling point.of view that you're doing it right 4 because otherwise, you will tell all these people that in all 5 cases you have hydrodynamic and it may not be so accurate.
6 Just a caution.
7 DR. GREENE: I think there was a ques. tion there 8 somewhere. And the question that I perceived was that there ,
9 was a scale limitation. Those experiments I was referring to 10 were not limited in scale. They could spread in any direction g 11 they wanted. And your scaling considerations are well taken, 12 Theo. Thank you.
13 MR. HULMAN: There's another response to the 14 question.
15 Dana Powers?
16 DR. POWERS: Theo, you raised a good point on your 17 need to do tests for lead to get you some idea but really you
! 18 need to do things with high temperature melts. We've done a 19 test. We did a test kind of by accident because it was a 20 spill, an accident, but we did spill 186 kilograms of molten 21 stainless steel with a modest amount of superheat onto a l
l 22 concrete surface. It flowed out to a mean depth of 2.1 23 centimeters, a leading edge depth of 1.6 centimeter thick. It 24 appeared to behave just exactly like George's experiments did.
25 That is, it flowed out and then it stopped and it was still O Heritage Reporting Corporation (202) 628-4888 l -
150
~'
7 ; 1 bubbling and outgassing the concrete. You could still see it
~
2 was liquid. l 3 So it did seem that even in the higher temperature 4 regime, we were controlled by this hydrodynamic constraint and 5 not by a freezing front.
6 DR. GREENE: Can I respond also?
7 MR. HULMAN: He answered my question.
8 DR. GREENE: In addition to those experiments we did, 9 Theo, we're also doing some experiments in a channel geometry.
10 We're also doing some experiments in a channel geometry where 11 we confine a water pool. And we've done the experiments with 12 not water, with cold water and with saturated water. And we 13 find the effects of the heat flux which were probably modeled
_f 14 'through the effects of water or natural convection to be a 15 perturbation and we can model this based upon whether we have l 16 boiling and we calculate boiling and that's consistent. l 17 So if there is a heat flux that's domine.it, it should 18 show up in a prototypic test like the ones at Sandia. And any j 19 effects of water you should be very careful to how you presume 20 the effects of water on the flow of melts. I'll just refer you 21 to Channel 21 Nova program several weeks ago in which there was 22 show about the formation of a Hawaiian island under about 500 23 feet of sea water.
24 DR. THEOFANOUS: We don't want to monopolize the 25 discussion but I just want to make a short comment. That your 9 Heritage Reporting Corporation (202) 628-4888
151
,s 1 accidental experiment is well taken but even that one I think i
xs 2 needs to be scaled, needs to be seen how it really applies 3 because we need to know again what the length scales involved 4 are, what the temperatures are, what the heat flux is, how 5 steel makes a difference versus fuel. All I'm saying is just a 6 word of caution. We don't disagree. Just a word of caution. l l
7 DR. POWERS: Well, just recognize that steel is a '
8 major component of cora debris. , l 9 MR. HULMAN: Raj?
10 DR. SEHGAL: I just want to comment that the model I 11 talked about does consider both hydrodynamics and freezing l 12 phenomenon and takes care of the zircalloy oxidation while the 13 melt is spreading. But we need some data to check out that 14 model before we can say that this is how it happens. It does
')
15 through the hydrodynamics piling up of water against a dam and 16 then spreading around, but it needs some data to qualify it.
17 MR. HULMAN: Vince Boyer?
18 DR. BOYER: Vince Boyer. It appears that a lot of 19 discussion we heard this afternoon is based on Steve Hodges' 20 groups work on the way in which the core melts and the material 21 which is in the bottom of the vessel and how it comes out. And 22 the other people picked up from that point and went on and did j 23 their analysis.
24 Now, if I remember what Steve said, he said that 25 about 220 minutes after I guess the start of the accident, was Heritage Reporting Corporation (202) 628-4888
l l
152 ,
I when the five instrument tubes failed a'nd started to discharge
)
2 this core melt, which was sort of layered into three different' 3 layers. I have a physical ~ problem on trying to visualize that 4 material being in layers on the bottom of the reactor _ vessel 5 a.fter what has gone on and what is going on in there with-the l 3
6 melting process and the burping and disturbing and maybe some 7 water injection now and then. !
8 I believe Steve assumed that there was absolutely, ,
9 absolutely no water being injected to any of this process. If 10 there was some, I think it would be interesting to know how 11 much and at what time periods, and whether he has looked at i
12 well, suppose at various time periods some water in varying 13 amounts is put back into the vessel, what does this do to the
( 14 overall process. -Because as an operator, three or four hours -
15 without being able to get water in there is incredible.
16 MR. HULMAN: Steve?
17 DR. HODGE: Yes. Well, I certainly think that you've
- 18 touched a very important point here. Let me make it very clear 19 that this sequence of events as we're postulating it is based 20 upon station blackout. It's very important that this be a 21 station blackout accident sequence in which there is no water 22 injected whatsoever.
23 Now, let me digress just a moment and say that for 24 the purposes of the NUREG 1150 exercise, about three weeks ago, 25 we did a calculation at Oak Ridge in which we postulated that Heritage Reporting Corpora *. ion
] (202) 628-4888 4
, , , , - r
i 153 7- 1 the control rod drive hydraulic system was operating to the
/ )
's' 2 extent that it would inject about 100 gallons a minute ~
3 throughout this accident. It was the only means of injection, 4 and it of course is not enough to terminate the accident. It's 5 not enough to remove the decay heat. However, it is enough to 6 keep the core plate cool.
7 Under those circumstances, the core plate would-not 8 fail, and you would form a debris bed above the core plate up , i 9 in the core region and not in the bottom head after core plate 10 failure. So it's very very important that we're trying to 11 concentrate on station blackout which is the quote dominant 12 unquote accident sequence for the BWRs. Under these 13 circumstances, you will dry out the core plate. That's a very
() 14 essential ingredient to the sequence of events as we see them.
15 There are two cases here of course. If you do ADS, 16 you'll dry out the core plate immediately after ADS because i
17 you'll flash the water in the vessel and you'll have a dry core I 18 plate. If you don't ADS, then some of the debris has to fell 19 into the water above the core plate and boil it off, then 20 leaving you to a dry core plate.
21 once you have a dry core plate, you continue to rain 22 down control blade and channel box material plus some candled 23 cladding which raises the temperature of the core plate to the 24 point that it could no longer support the load of debris, 25 relocated debris. The core plate fails. Then the material Heritage Reporting Corporation (202) 628-4888
154 1 falls in the bottom head.
, 2 , The reason for the layers are the first thing down 3 there is control blade and channel box material followed by 4 fuel material later and so forth. So it's central to the f 5 approach that again I can't emphasize enough that it be station 6 blackout and that you have a situation in which you've dried 7 out and failed the core plate and relocated debris into the 8 bottom head while fuel was still standing. Okay? ,
9 MR. HULMAN: Question?
10 Dr. Canton, you want to come up? Ivan Canton from 11 UCLA.
12 DR. CANTON: I have a question for the people who 13 addressed the liner failure. We've heard somewhere between 14 what, 80 tons and 250 tons have to be dealt with?
't 0 15 What's kind of a lower bound, based on your analysis at which you won't 16 fail the liner.
17 And just a comment. I really find the story by 1
18 Mr. Hodge not very mechanistic. Is there anywhere that you 19 have a written description of the justification for the 20 conditions that you've postulated?
21 MR. HULMAN: I think that question, there's a two i
j 22 part question, Steve.
I 23 Can you come back up?
]
24 DR. SEHGAL: Tell him to stay here.
25 MR. HULMAN: And why don't you stay up here, please.
)
Heritage Reporting Corporation i
(202) 528-4888 4
e _ m a m .
155 7'~) 1 Two part question, the first part -- did you get V 2 both parts?
3 First part was why the difference between the amount 4 of material that comes out of the vessel between your numbers 5 and EPRI's numbers. Can you explain the difference.
6 And the second part was where is your mechanistic 7 treatment explained in the core melt process.
8 DR. HODGE: Yes, I don't know how EPRI comes up with .
9 their number. So that's the answer to that question.
10 I think I showed you in the handout there I listed 11 the individual weights assigned to each component-that we add 12 up to make the debris bed in the bottom head, so that's where 13 our numbers are coming from. We're eventually going to bring
() 14 the whole core out, metals first and the UO2 would be the last 15 to come out.'
16 I'm sorry, am I answering the wrong question?
17 DR. CANTON: Well, I don't think you are. How did it
'd get there? I mean, how do you rationalize that the whole core 19 becomes a big pile of rubble? How does it get there? Why 20 doesn't it freeze, remelt? Why doesn't a pool form and 21 gradually penetrate its way out, like most of us'have relied 22 upon in the past. If we're wrong, could you explain why we're 23 wrong?
24 DR. HODGE: Sure. I've had a feeling that we were 25 going to end up on this point. There's no question -- you
)
Heritage Reporting Corporation (202) 628-4888 l
n _. . . ..
1 156 j
^ 1 7' h 1 know, when I first started talking and I put up there and I l
(_)
2 said, these are the three layers in the bottom head, I said i
I 3 there's an awful lot of argument that needs to go into it to )
4 explain how we get to this point. As a matter of fact, we had j 5 a whole day meeting about three or four weeks ago at the NRC 6 where we talked about nothing but that. i l
7 So it's a rather long and involved process. But _ ll 8 try to summarize it.
9 In the boiling water reactor core, you have a 10 situation where you have a relatively low melting temperature 11 material such as the control bladec, and then the channel box 12 walls which are sendwiched inbetween the higher melting 13 temperature materials. Now, there was an experiment done at
( 14 Sandia called the DF-4 experiment, which is the only experiment 15 to my knowledge that has ever been done in true MBR geometry.
16 And in this experiment, it predicted that the cortrol 17 blades go down first followed very shortly thereafter by the 18 channel box walls and leave the fuel standing. So this is what 19 we see and this is what the code is cracking through.
20 You start with a unit cell such as shown up there at
'21 the top. As you uncover and you heat up, the control blade I 22 melts and runs away, leaving you this second configuration.
23 And then the channel box walls melt and run away, giving us, if 24 you can raise it up, this bottom.
25 DR. CANTON: There's no refreezing? )
Heritage Reporting Corporation '
(202) 628-4888 m
157 1 DR. HODGE: There's no refreezing of ccntrol blade or ,
. Q l
/ 2 channel box material until it gets all the way down. l 3 Now, having said that, then --
4 DR. CANTON: Is that an assumption? j 5 DR. HODGE: No. This is what happened in the 1
6 experiment. But what I have to say is that the experiment you j l
7 'see was only about half a meter lony. It wasn't the full 12.5 8 feet of rods. Now, we did the pretest, calculation,s for the .l l
9 experiment at Oak Ridge. In the pretest calculations, we had 10 predicted that the channel boxes, that the control blades and l'1 channel box walls would go early and leave the fuel pellets I 12 standing in zirc oxide sheets.
l 13 I should say there's also candling of cladding during 14 this period. But you're left with the zirc surrounding the
" 15 fuel pellets as one of two choices, two options. That they can l 1
16 either melt and run down and there is some freezing of the clad j 17 'on the way down and remelting and going on down. But the l
18 temperature profile in the channel boxes and the control blades l I
19 is almost vertical. Once it starts melting one place, it melts 20 all up and down and there's just no place for it to refreeze.
21 And so you just have a relocation of material down.
22 In the experiment, which again is limited by the fact that it 23 was not a Zull length test, by any means, but in that 24 experiment and there was a video tape, a very dramatic video 25 made of this experiment while it was in progress, and you can A
(' Heritage Reporting Corporation (202) 628-4888
. ~
..- -. .~. - .
. -. .. . .. = _. . - .-. . . . - _ _ _ - -
158
() 1 see this in fact happened.
'(/
~2 And so what is differ.ent from the BWR from the PWR in 3 what we're predicting is the effect of this movement of large 4 masses of material down onto that dry core plate before any of 5 the fuel has ever become molten. And it piles up on the core 6 plate and attacks the core plate and the core plate becomes so 7 hot, it looses all structural integrity. And of course, we all l
B know it was never designed to support anything anyway. It's ,
9 just there for lateral support of the control out guide tubes, 10 not vertical support.
11 And then this relocates the debris accumulated above 12 the core plate in the bottom head and that we believe is the 13 first thing down there. Then as the fuel' continues to heat up, r\ .
(_j/ 14 it approaches the temperature at which the zirc oxide sheets 15 lose their strength, and when they lose their strength, then 15 fuel pellet stacks tumble over and fall into the bottom head 17 and that makes this second layer I talked about. The fuel has 18 a lot of UO2 pellets in it.
19 It's formed from collapse of fuel pellet stacks when 20 the zirc oxide sheets can no longer support the fuel pellets.
21 And then the third layer occurs when the control rod 22 guide goes. You remember that the power profile of BWR falls 23 off dramatically at the edges, very much. In fact, the outer i
24 13 percent of the core has a power factor of 23.5. It doesn't I 25 even know an accident's going on. But once you've dried out i
i Heritage Reporting Corporation ]
(202) 628-4888 -
l i
g .
I 1
1 159 y' 1 the water in the bottom head and you have the-debris bed down
(_]/ - !
2 there, it st, arts reheating, the control rod guide tube's that l l
3 are holding up the remaining portions of the core lose their 4 structural integrity.and the remaining portions of the core.
S then fall down -- not because they'rn damaged up in the cor'e 6 region, but because their underpinninqs are cut out.
7 That is the model, that is the way we modeled this 8 accident sequence very much dependent on it being a station ,
j l
9 blackout and there be no cooling supplied to the core plate to 10 keep it intact during the accident. j 11 As I say, this is a subject that's fraught with 12 uncertainties. There's a lot of controversy about all this. I i i
13 think we need another experiment like the DF-4 with full length
() 14 -fuel rods. I'd be delighted if we could do such an experiment.
15 I'd like to have a core plate in there and mock up the debris 16 accumulating and so forth. I think it needs to be done.
17 MR. HULMAN: Another question? l 10 Bob Henry, you want to come up?
19 MR. HENRY: Bob Henry from Fauske & Associates.
20 I've got a question for Dana on experimental core that he 21 talked about this afternoon.
22 Dana, if I put together a couple of presentations 23 this afternoon, namely Steve Hodges' presentation that suggests 24 that debris might come out of the reactor vessel over many 25 minutes or tens of minutes, and Tom Kress' presentation that O Heritage Reporting Corporation (202) 628-4888
. 160 i l
1 said-the most important. thing in assessing the integrity is
)
2 'superheat and the debris depth, and also Theo's comment with-3 respect to radiation heat losses off the debris, I tend to 1 l
4 think.that we need to have some idea of how much energy 5 t.rans'fer might occur between the debris leaving the pedestal 6 and that's where the drywell walls. And said something- i 7 obviously that Tom said it was very essential to his model. ,
1 8 Are there plans in this experimental program that ,
9 you're talking about to represent the transported debris from
]
10 pedestal region to the wall? And are there some plans to do 11 that with and without water being available?
12 DR. POWERS: Bob, our initial experimental effort is l l
13 really focused on validating the convective-heat transfer l y% -
t, ,) 14 coefficient from the debris to any crust forms. Go we've not 1 1
15 considered experiments involving the transport from the '
16 pedestal region to the liner wall, though I believe George ;
1 17 Greene in his simulator program is. I 18 I'd like to offer a caution on the treatment of 19 radiation heat transfer in these melt concrete interaction 20 skits mentioned alosig. And the fact that there is a huge cloud 21 of aerosol forms over these materials when they are poured on 22 concrete. That's a pretty effective blocking agent for 23 radiation heat transfer from the top.
24 Your concern abou+ the heat loss is probably well 25 placed but recognize you are getting a melt concrete f'l N
Heritage Reporting Corporation (202) 628-4888
. _m, ,- y Q- - -- *y , - _ - , - - -
161 ,
f) 1 interaction that tends to replace any heat loss by chemical V. 2- heat between the metallic constituents of the core debris as a 3 result of metal water reactions or CO2-reactions.
4 MR. HULMAN: I'd like to suggest one more follow on 5 a.nd'then we'll break it up for dinner.
6 DR. POWERS: Again, I believe George Greene's 7 experiments do involve water being present'and not present in 8 his simulated experiments. We have not looked at doing any ,
9 transport experiments either with water or without.
10 MR. HULMAN: I'd like to suggest that we break for 11 dinner, and those people that need to check into their hotels. l l
12 Can we please reconvene at 7: 30.
13 (Whereupon, at 5:25'p.m., the hearing was recessed fj
(_
1 14 for dinner, to reconvene the same day, Wednesday, February 24, 1
15 1988, at 7: 30 p.m., in the same place.) ,
l 16 17 18' 19 20 1
21 22 23 24 25 O-- Heritage Reporting Corporation (202) 628-4888
1
- REPORTER' S CERTIFICN"E~
2 3 DOCKET NUMBER: N/A
' - 4 CASE TITLE: BWR Mark I Containment Information Exchange Workshop I
5 HEARING DATE: February 24, 1988 6 LOCATION: Baltimore, Maryland 7 I I hereby certify that the proceedings and evidence 8
are contained fully and accurately on the tapes and notes !
9 !
reported by me at the hearing in the above case before'the 10 U.S. Nuclear Regulatory Commission.
11 .
I
/
12 l
Date: February 24, 1988 i 14 i
15 i 16 11, 'ggg,' , f' fgy,g '
Official Reporter f g7 18 HERITAGE REPORTING CORPORA j 1220 L Street, N.W. ,
Washington, D.C. 20005 1 gg 20 J 21 22 p7 23 24 25 HERITAGE REPORTING CORPORATION * ,
, (202)628-4888
O
~
WR ME l CONTAltfENT PERFORMANCE WORKSIOP BALTIt0RE,f0
.FEB. 214-26,.1988
~
O -
BY MR. ERIC BECLCRD DIRECTOR, 0FFICE OF NUCLEAR EGULATORY RESEARCH USNRC I
l 1
O l S
- . , - n , -wwwe, , w e e-v e , w.----e---,-v,-,w---nw . - - - , , - ,,w -- ,,,-,--
. ,- .e,,,-n,--,e -- ,+ , ,,e,-. . .r,,,--,-- - - -.... . ,--, - - - --w+~..-----,-ve ----.~-re.---w,
O MARK I CONTAlWENT ltlJTIATIVE IWORTANCE OF PROBLEM IWORTANCE OF CONTAlttENT F:.RFORMANCE AWLY DEMONSTRATED BY TMI-2 AND CHERNOBYL NEED FOR DECISION 1
NEED TO CONSIDER PREVENTION / MITIGATION BALANCE NEED TO CLOSE SEVERE ACCIDENT ISSUES ON CONTAlttENT PERFORP#4CE O -
NEED BEITER INFORMATION WHAT ARE THE IWORTANT CONTAIN E CHALLENGES I
BENEFITS / DISADVANTAGES OF POTENTIAL IW ROVE M S 1
I 1
1 2
MODEL FOR. SEVERE ACCIDENT CLOSUE
'EBODOLOGY (IDENTIFYING CONTAlttENT GALLENGES AND POTENTIAL IPPROVEPDfTS) TOGETHER WITH RISK INSIGHTS FROM NUREG-1150 SERVES AS P0 DEL FOR OTHERS RESEARG AGENDA RESCLUTION FOR MARK l's EXPECTED BY SEPT' 88 RESOLUTION FOR OTHER CONTAINENTS BY SEPT' 89 SEARG FOR RISK OUTLIERS.(IPE)
ACCIDDIT MANAGEE NT-MAXIMlZE PREVENTION OR KEEP IT INVESSEL .
SOE PHEN 0ENA RESEARG ECESSARY FOR CLOSURE RISK INSIGHTS (NUREG-1150) TO BE FACTORED IN CONFlWL3RY RESEARG CONTINUES L0tGEPM O
3
9 N .
e W R MA R I-4 CONTAINMENT PERFORf#EE WRKSHOP BALTIt0PE,is FEB, 2I4-26,' 19P.E c
[ BY DR. WEMIS SPEls DEPUTY OFFICE F NUCLEAR O. DIRECTOR FOR GENERIC ISSUES REGULATCRY RESEARCH USNBC 1
1 j
l l 0 . .
4 5
-4,__ _ . _ _ . . _ . . . - _ _-._.,_._.._._.o____-_..,.-.__.- ..;._.___,.___ ___..,-..~a.....__
1 1
.O LICENSING /REGULATORYRECCIREFENTS/APPROA.Cli DEFENSE-IN-DEPTH (FULTIPLE, SUCCESSIVE BARRIEPS) l I
DESIGN FOR NORMAL OPERATION EPPHASIZES EQUIPPENT RELIABILITY, REDUNDANCY AND INSPECTABILITY)
DESIGN TO DETECT FAILUFE(S) AND SHUT PLANT D0kN i
DESIGN TO CONTROL THE CONSECUENCES OF F0FE DAMAGINC ACCIDENTS DESIGN BASIS EVENTS TRANSIENTS (A00s)
ACCIDBiTS Q
CONTAlfEhT IAND OTHER SAFETY SYSTEM (S)1 DESIGN
EXTERNAL EVBiTS TID-14844 FISSION PRODUCT SOURCE TEFtS i
)
l O
/.
i
- - N' b- - - - ,- . . - , , , , . , , _ _
LIC91 SING /PEGULATORY RECUIPEPENTS/ APPROACH (CONT'D.
p) t
'"'
- l TMI EXPERIENCE FEEDBACK
- MJLTIFAILUPE CONSIDERATIONS (PLATE E00lft9,7 SYSTDS AND OPERATORS)
SYPPTOM-0RIENTED EERGENCY OPERATING PROCEDURES PERSONNEL TRAINING DESIGN FOR HYDROGB1 RELEASE FROM CORE DEGRAEG ACCIDENTS FOR ' WEAKER' C0tJAltt1ENiS l CONTlf.1 JING OPERATING EXPERIENCE AND PRA It! SIGHTS FEEDEACK
- PEVISED AND/0R NEW "REGULATIONS" (E.G., AWS, STAT 10f4 BLACK 0LT GNE TIE GENERIC AND/CR PLANT SPECIFIC REQUIRDENTS
' ADDITIONAL /REVISCD) g , .
V
- SEVEPE ACCIDBiT CONSIDERAT10tlS (ACCIDENTS MDRE SEVERE THAM EAs)
PPAs INDICATE THE EULK OF PUBLIC RISKG ARE FRCM SEVERE ACCIEDSS POLICY STATB E R TASKS INDUSTRY AND STAFF Wim A SEAROi FCP ,
1 I
OUTLIERS POLICY STATENNT CALLS FOR STRIKitC A BALANCE LEihEEN j i
PREVER[lCtl & MITIGATICti !
IFE EFFORT IS EXPECTED TO BEGIN ltl APCOT 2 PONTHS CCt.TAlt0ERi PEPFOPPANCE PD/lEWS START WITH MARK l's O
3
~
O SEvt w ACCIDENT CONSIDeRATi0nS
- WASH-1400, OTHER PRAs, TMI-2 AND CHERNOBYL ACCIDENTS, ALL TELL US THAT SEVEE ACCIDENTS REPRESEhT THE MAJOR C0hTRIBlfi!Ch
.T0 RISK FROM C061ERCIAL tAJCLEAR PCSER PLAfGS A DEFENSE-IN-DEPE ISSUE (PREVEhTION VS, MITIGATION) 1 IPFORTANCE OF CCMAINFOR ITMI-2 VS, CHERNOBYLl; "PRACTICAL FIXES" FOR EXISTING, PLANTS, E.G.:
FUREER EFF0oTS IN PEValTION [Il0lVIDUAL PLAffT l EXAMINATIONS VIA A PRA)
EASONABLE ASSURANCE OF MITIGATION CAPABILIT( FOR
' DCMitWIT (MOE PROBABLEl BREATS TO CONTAlWENT [I'E(: f REASONABLE Ut0ERSTANDING OF THE POTSfilAL CChTAlfVdNT FAILUE MDDES AND EEIR ltTORTANCE! !
UTILIZATION CF EXISTING CONTAlWEhT PERFORPANCE PARGINS AND THE POTENTIAL NEED.FOR ADDITIOMAL ltPROVEFENTS (HARDWAPE/PROCEDUES) i I
1 i
O u
AI
~
COMTAltPENT BUILDItra O
- DESIGNEI) FOR: .
DBAs (E G. LOCA/SLB TEPPERATURES 8 PRESSUPES)
- EXTERNAL EVENTS (EARTHQUAES, FLOODS, TCRNADOES)
TID-14844 SEVERE ACCIDEhT FISSION PRODUCT SOURCE TERM (RADIATION: N_0 SEVERE ACCIDBR PRESSUPE OR TEVPERATUPE EFFECTS)
USE OF CONSERVATIVE CODES / STAT 0ARDS FARGINS (AVAILABLE) ABOVE DESIGN LEVELS:
- MARGINS ARE CONTAlteENT SPECIFIC (VOLUE, PATERIALS, .
C0tFIGURATIONS, ETC)
- IN GEERAL, STUDIES (EXPERIM/A!1ALYTICAL) HAVE INDICATED THAT CONTAltiENT SYSTBG CAN SURVIVE Lf(SSUPE CHALLENGES O
- OF 2.5 TO 3 TIES DESIGN LEVELS PESIDUAL RISK FRCP. SEVEPE ACCIDEhTS:
l FOR EACH CONTAlttENT TYPE THERE REPAlt! FAILURE ECHANISFE kHICH COULD LEAD TO CONTAltPBU FAILUFE' KEY OUESTIONS: (1) PEASONABLE UNDERSTAl0 LNG OF CHALLENGES TO C0tRAltfBUS (LOADS (P.T.), FARGINS AVAILABLE, FAILUPE MDDES (TIE, LCCATI0fD, (2) REASONABLE UNDER$TAl0 LNG OF PPEBABILTIES (E.G. , SOE FAILURES F0 DES, GIVEN A S. A. ,
APE MOPE PROBABLE THAN OTHERS), AND (3) ACCIDENT PPEVBU10N VS. MITIGATION (ELATIVE EFFECTIVB4ESS CR BALANCING CF THE M AFFR0 ACHES)
O
.h. .
- l l
CONSIDERATION OF THE COST-EFFECTIVENESS OF THE WO APPROACHES-SAFETY G0AL, DEFENSE-IN-DEPTH (EXTENDING DEFENSE-IN-DEPE lhTO BEY 0hD DBf ARENA) j l
NEED T0 IDENTIFY THE Iff0RTANT PLAMT SPECIFIC SECi.EhCES AND B EIR CONTRIBUT10f1 T() CORE DAFAGE AND ALSO CHALLENGE TO CONTAlWENT BEF6RE ONE IS AELE TO ARRIVE AT A MOPE BALANCED VIEW AND PERSPECTIVE OF THE RELATIVE EFFECTIVEtESS OF THE %0
~
APPROACHES (1.E.,PREVEOTIONVS, MITIGATION) l l
f O
O .
6
~ . , .-
MARK-1 STRilCTllRAL FAILURE MODES-ACCIDENT ESTIMATED CONDITIONS FAILURE MODE FAILURE PRESSURE RANGE PRESSURE / TEMPERATURE TORUS FAILURE 130 - 180 psIG-
~
CHALLENGE IN WETWELL (BEFORE VESSEL BREACH OR WITH C00LABLE DEBRIS)
LEAKAGE THROUGil 120 - 180 PSIG DRYWELL HEAD FLANGE SIGNIFICANT LEAKAGE THROUGH ,
125 - 150 PSIG (AT 800FO )
CORE-CONCRETE DRYWELL HEAD INTERACTION BUT FLANGE
~
NO SHELL MELT THROUGH 50 - 90 PSIG'(AT 1200FU )
0 PRESSURES REQUIRED TO BURST Tile CONTAINMENT SHELL ARE GREATER THAN FOR THESE MODES 0 WHETHER "LOW" TEMPERATURE FAILURE IS IN WETWELL OR DRYWELL DEPENDS ON. PLANT-SPECIFIC DETAILS OF HEAD BOLT DESIGNS AND PRELOADS USED.
0 MARK-1 OWNERS' GROUP IS CURRENlLY SPONSORING A STUDY OF PLANT-TO-PLANT VARIABILITY.
O "HIGH" TEMPERATURE SCENARIOS FORCE FAILURES INTO Tile DRYWELL.
v.
O -
SEVERE ACCIDENT CLOSURE INCLUDES CONSIDERATIONS OF:
IPE CONTAltfEhT PERFORfW4CE .
SAFETY GOAL IlfLEElHATION EXTERNAL EVENTS
-USIs/GSIs ACCIDENT MANAGEENT IPPROVED PLANT OPERATIONS RISK INSIGHTS (E.G , NUREG-1150)
SEVERE ACCIDENT / SOURCE TERM RESEARCH O -
O -
7
l l
0 FOR EXAMPLE, C0t! SIDER ACCIDENT PANAGEFEl#
UTILIZE ALL AVAILABLE WATER SOURES TO PREVEF.T OR DELAY CORE DAMAGE SAFE FOR VESSEL FAILURE SOE EM'PLES C0fRROL RCD DRIVES OPEN OR BYPASS MAIN STEAMLINE ISCLATION VALVES, USE STEAM DRIVEN REACICR CORE ISOLATION COOLING PUW 8 WATER IN TANKS O
i l
l O
8
..N-. --,-+*----we, .,v- - , e-, .w.----,w- cr-,,-m.--+r-- ---.-4 -m- ,m, , , ---,- -..--, - - ,-- ,--,-s -,,c w ,,-- - ,+ ---- ,-.- -- - - . . - - , -w y
sf'f O -
,.p.
c.)
-6 i
BWR MARK 1 CONTAINENT PERF0PPANCE
. WORKSHOP l
BALTIF0E, MD O F E. 24-26, 1988 _
f 1
i' BY 4 JERRY h'UlfAN CHIEF, SE'EE ACCIDENT ISSUES BRANCH OFFICE OF NUCLEAR REGULATORY ESEARGl USNRC 9
J r
O h
O .
0 OBJECTIVES OF WOPJ3 HOP NARROW & FOCUS TECHNICAL ISSUES IDENTIFY IPPORTANT CHALLENGES IDENTIFY RANGES OF ltPORTANT PAP /ETERS ASSCCIATED WITH OlALLENGES DISCUSS SPECIFIC MITIGATION SCHEWS & RElliTED PARNETERS CHARAGERIZE f%GNITUDE OF POTENTIAL EBEFITS PROVIDE A FORUM FOR EXPRESSION CF DIVERSE VIEh5 OH O -
RISx LEVELS i
PREVENTION VS, MITIGATICt! !
PLANT SPECIFIC DIFFERENCES OF ltFORTANCE O
2
l O.
TKO PHOT 0S OF PEACH BOTTOM DRWE.L (C0URTESY OF PHILADELPHIA ELECTRIC COPPAlfD l
0 SHcss. Bu1 n01 aEen00uCen Htet O .
3
~- -
I 0
COE DAMAGE FEQUENCY ESTIMATES
BROWKS'S FERRY (IEP) 2x10-4/RY (INT, EVEhTS ONLY)
COOPER (A-45) 3x10-4/RY (ItH, & EXT, EVEhTS)
MILLSTONE 1 (IPEP) 3x10-4/RY (INT EVE M S ONLY)
PEACH BOTTOM (RSS) 3x10-5/RY (thT, EVENTS CNLY)
PEACH BOTTOM (IDCOR) 4x10-6/RY (ThT. EVENTS CFLY)
PEACH BOTTCE (NUEG-1150 DRAFT) 8x104/RY (IE EVENTS ONLY) 7x10-6/RY-SB0 CUAD CITIES (A-45) 1x10-4/RY (IAT & EXT, EVENTS) a O, -
VALUES ESTIMATED AT DIFFERBxT TIES FOR DIFFEREhT PURPOSES, FCR EXAFPLE, THE TVA DRAFT STUDY WAS INTEhtED AS A CONSED/ATIVE BOUNDING ANALYSIS FOR bSE IN DIECTING NPE DETAILED ANALYSES OF THE FINAL PRA. SINCE THESE STljDIES bEFE PERFORPED OVER A SPAM OF 12 YEARS, SIGNIFICANT IFPROVEPENTS IN EFEFGENCY OPERATINC PROCEDURES AND IN ATWS PROTECTION IN RECENT YEARS ARE NOT REFLECTED IN SCFE STUDIES.
O 4
l l-
O rARx i CHAttENGES a e01ealAt treR0vBeRS l
CGRAlftDR 01ALLENGES PER DRAFT NUREG-1150 LINER E LTTHP0 UGH-HIGH CONDITIONAL PROBABILITY EARLY OVERPRESSURE /0VERTEPPERATURE FAILURE-MODERATE CONDIT PROBABILIU LATE OVEREWERATk.JRE/0VEPPRESSUPE FAILURE-F0DERAT '
PROBABILITY CONTAlt(bit EYPASS 'LCW CONDITIONAL PROBABILITY STEAM SPilES & MISSILES - LOW CONDITIONAL PR0EABILITY .
POTENTIAL IMPROVB O US l HYDROGB4 CONTROL CChTAlttB E SPRAYS VBRit0 .
CORE DEBRIS C0tRR0l.
ADS RELIABILITY TPf,1NltlG 8 PROCEDURES (ACCIDBIT.MANAGBER PROCPAM) l OTHER?
i l
l O
5
O O O :
MARK-1 STRUCTURAL FAILURE MODES ,
t ACCIDENT ESTIMATED CONDITIONS FAILURE MODE FAILURE PRESSURE RANGE .
PRESSURE / TEMPERATURE TORUS FAILURE 130 - 180 PSIG CHALLENGE IN WETWELL (BEFORE VESSEL BREACH OR WITH -
C00LABLE DEBRIS)
LEAKAGE THROUGH 120 - 180 PSIG DRYWELL HEAD FLANGE SIGNIFICANT LEAKAGE THROUGH 125 - 150 PSIG (AT 800FO)
CORE-CONCRETE DRYWELL HEAD INTERACTION BUT FLANGE
- NO SHELL MELT THROUGH S0 - 90 PSic (AT 1200FO )
0 PRESSURES REQUIRED TO BURST Tile CONTAINMENT SHELL ARE GREATER THAN FOR THESE MODES -
0 WHETHER "LOW" TEMPERATURE FAILURE IS IN WETWF.LL OR DRYWELL DEPENDS 0N PLANT-SPECIFIC DETAILS OF HEAD BOLT DESIGNS AND PRELOADS USED.
! O MARK-1 OWNERS' GROUP IS CURRENTLY SPONSORING A STUDY OF PLANT-T0-PLANT VARIABILITY.
O "HIGH" TEMPERATURE SCENARIOS FORCE FAILURES INTO THE DRYWELL.
6
r' PEACH BOTTOM PLANT DATA Os (BWR, Mark I Containment)
Nominal Power 3,293 MWt 1,124 x 100 Btu /hr Steam Pressure in Core 1020 psig (7.0 MPa)
Primary System Operating Temperature 547 F (206.1 C)
Primary System Coolant Inventory Reactor Vessel Subcooled liquid 7,847.4 ft3 (222.2 m 3)~
Saturated liquid 4,004.7 ft3 (113;4 m3 )
Steam 8,813.5 ft3 (249.6 m3 )
Piping Recirculation 1,227.7 ft3 (34.8 m3 )
3 Feedwater 815.9 ft3 (23.1 m )
Steam 3,125.6 ft3 (88.5 m3 )
Reactor Vessel Inside diameter 251 inches (6.375 m)
L Inside height 72.6 feet (22.1 m)
Design pressure 1250 psig (8.6 MPa)
Design temperature 575 F (301.7 C)
Thickness (with cl.ad) 6-5/16 inches (1.92 m)
Total Vessel Weight 1,501,000 lb (680,839 kg)
Total Weight of. Internals (Excluding fuel, control rods, feedwater spargers, vessel head cooling nozzles, in-core guide tubes, start-up sources, temporary control curtains) 462,000 lb (209,559 kg)
Drywell Free Volume 150,000 ft (4,503in ) l Lesign Temperature of Drywell 281 F (138.3 C)
Orywell Internal Design Pressure 56 psig (0.38 MPa) r, Drywell Vent Pipes Number 8
,_ internal diameter 6.75 ft (2.1 m)
YJ 7
m .
f Pressure Suppression Chamber j 3
.. . Free volume 119,000, ft3 (3,370 m )
Water volume 136.000 ft3 (3,851.5 m3 ) j j', Suppression Pool Downcomer Pipes l
Number 96 Internal diameter 2 ft (0.6 m) l
~
Submergence, nominal 4 ft (1.2 m)
- l 1
QNY[N///
- BLOWOUT BLOWOUT
,, PANELS PANELS RERJEUNG BAY 7 .
.'d ,
.a (O' e:t, -
- C . = .: .
i.
. .~-
', DRYWELL
' J ,'r :. (j '
' . -j ATCH.
l
. . :. ; . . f,:
J
=
V Q
.. .;ToauS
/' Y. SCHEMATIC OF CONTA!NMENT
.y. .. :: :
DESIGN FOR PEACH BOTTOM i
- r. emt __ . E.: . . .. . ..
. * ; . i. .,
J
, '. W.
~
. PLANT O
8 e
h CRWELL 91 ELL ELTTURCbGH ISSUES THE DEGREE TO WHICH INVESSEL VERTICAL CHANNEL BOX BLOCKAGE
?NFLUENCES ELTDOWN PROCESSES THE FRACTION OF ZlRCONIUM OXIDIZED INVESSEL RATE OF COE DEBRIS THAT LEAVES VESSEL /ENTEfd DRWELL (ALSO INVESSEL PRESSUPE)
ETAL TO CORlUM PASS RAT 10 DEERIS TEFFEPATURES
' iHE MANNER IN WHICH DEBRIS COULD INTEPACT WiTh & WITH &
WITHCUT WATER SPREAD ON DRWELL FLOOR HEAT FLUX TO WATER AND THE CONTAltPENT SHELL FOR SCENARIOS h -
SPRAYS NOT WORKING SPRAYS WOPr.ING WATER ON DRWELL FLtCR AT VESSEL EFEACH to WATER ON FLOOR PETEFITS CF VEhTif0 TO PEDUCE liVDPCGEN CHALLFJEE TO PEACTOR BUILDII;G i.EVEL OF FISSION PRUDUCT PELEASES TO ENVIRC&ENT
- GIVEN VESSEL FAILURE, THE LIEllHOOD & CONSEQUENCES OF DRWELL SHELL ELTTIEUGH O
t 9
W O EARLY 0VERPRESSURE/0VERTEPPERATURE 01ALLENGE ISSUES SAE INVESSEL 8 EXVESSEL ISSUES CONCRETE DEGASSING
- HEAT FLUX TO WATER, SHELL & CONCRETE SHELL PRESSURE /TEPPERATURE FAILURE LIMITS & LOCATIONS OF FAILURE EENEFITS OF VENTING & SPPAYS BENEFITS Ce REACTOR EUILDING FI'E MOTECTION SYSTBi O -
tEvEt Oe FISSION eR000CT REteASeS TO ENviR0it o:1 l
1 O -
10
i l
O LATE OVERTEPPERATUPE/0VERPRESSUE CHALL_EEE ISSUES l
SAE AS OLELINED FOR POTENTIAL EAPLY FAILURE
- IF LATE FAILURE cam 0T BE PECLUDED, AE THE RISKS LOW BECAUSE OF COMBINATION OF PROBABILITY, FISSION PRODUCTS THAT WOULD EE l ELEASED, AND ACCIDDE CCNSECL'ENCES? I O !
C0fRAlttER BYPASS CHALLDEE j ISSUES LIKEL! HOC:. W BYPASS SEQUENCES
- WHETHER EXISTIhG PPACTICES HARDWAPE, PROCEDURES & MitEENANCE)
APE ACEQUATE O
11
O POTENTIAL HYDROGEN ColGROL iPPROVBOTS THREE POTENTIAL AREAS OF CHALLENGE ACCIDEhT EXPOSURE DURING DEltERTED PERIOOS ADECUACY OF LONG-TEPM NITROGEN SUPPLY BACKUP COMPRESSED GAS SUPPLIES FOR VALVE OPERATION OF OXYGEN OUESTIONS RELATED TO IMPORTANCE OF PERIODS OF DEINERTING, NITROGEN SUPPLY ADEQUACY AbD BENEFITS OF POTEhTIAL IFPROVBO(TS COSTS OF POTENTIAL IMPROVBOTS 1
l i
O
, I 12 i 1
I
O POTENTIAL CONTAIN T NT SPRAY lW ROVEPENTS
- EXISTING SYSTEM PAY t0T WORK, OR WORK PROPERLY, FOR IMPORTANT SEQUENCES-
- POTENTIAL MITIGATION OF PRESSURE /TEFPERATURE, CORIUM CHALL$NGES T SHELL & FLOOR, FISSION PRODUCT ATTENUATION l
00EST10NS RELATED TO l
WATER SUPPLY DURit0 STAT 10tl ELACKOUT SEQUENCES
- HEAT FLUX CAPABILITIES ON CORIUM, IN CONTAINENT ATMDSPHEE, WATER
& ON THE SHELL FISSION PRODUCT ATTENUATION O - is SeRAv et0w iS REoUCEn, ARE 1 Heat NEcA11vE StrE1x le tC1S 30R NON-COFBELT ACCIDBsTS?
1 1
l0 -
13
O .
POTENTIAL VBRING IFPROVEE NTS kHAT AE POTUTTIAL BENEFITS OF VDHilE USING EXISTING HARDWAE?
HEAT IPPPOVEN NTS Call BE PADE? ,
hMAT AE POTB{ilAL BENEFITS OF VEhTING WITH IFFROVBelTS Ili HAPDWAE & PROCEDUES?
hhAT AE FOTENTIAL NEGATIVE SAFETY ltPACTS OF EXISTING - i HAPDWAPE & PROCEDUES, OR IWPOVED HAPDWAE & PROCEDUPES?
- CAN BENEFITS BE MAXIMIZED, AND NEGATIVES SUCCESSFULLY O MINIMlZED?
1 O
14 .
- ~ _
l l
\
POTENTIAL COPE DEBRIS C0hTROL IlPROVRENTS ;
- GIVEN A COPBELT ACCIDENT WITH VESSEL FAILURE, CAN CORIUM BE INTERDICED IN C0ffiAlttDTT WITHOUT SHELL lELTillP00GH?
WITHOUT EARLY CONTAlWENT FAILUPE?
- IF LINER ELTTHROUGH IS LIKELY, CAN FISSION PROCUCTS BE 1
- ATTENUATED EFFECTIVELY IN THE TOPUS ROOK UNDER THE SUPPEESS10ll l
PCOL?
COULD USE OF TlE EXISTitlG FIPE SUPPRESSION SYSTDE IN THE l REACTOR BUILDING EFFECTIVELY ATTENUATE FISSICt1 PRODUCTS FRCH A CONTAIMET BYPASS, CVEFFRESSURE/0VERTEMPERATURE FAILbFE, OR l
SHELL ELTTHRCtGl? l l
I 1
J l
I O '
j 15 i
- l l
t O PORYflAi. AUTOMATIC DEPRESSURIZATION SYSTEM (ADS)
. Pf.l.1 ABILIlY IWROVStNT CAN RELI T_1 VESSEL DEPRESSURIZATION DURING STATION ELACK0UT SEQUENCES SIGNIFICANTLY PEDUCE THE LIELlH000 & CCNSEQUENCES OF A COREELT ACCIDENT?
HOW MUCH CAN THE SAE IMPROVBENTS EE COUfffED C6 TO PREVEhT A COPHELT FROM HAPPENING, OR IN DELAYING CR PEVENTING VESSEL FAILUE?
O 1
/
l I
i6 i
e
POTENTIAL TRAINING & PROCEDURES IfMOVBE4TS IWROVEENTS IN TPAINING a PROCEDURES TO BE CONSIDEPED IN A SEPAPATE NRC ACCIDENT MANAGEE NT PROGRM)
ACCIDENT MANAGEENT PROGP#1 IS A COMPANION TO SEVEPAL OTHERS THE STAFF HAS UliDERWAY TO CLOSE ALL SEVERE ACCIDENT ISCOES
.. O -
O 17
) -
_-__-.....__4..
i IDJSEVEEPING O - REVISED AGENDA AVAILABLE FROM MS, KONDULIS l VERBARM TRANSCRIPT WITH CCPIES OF VIEWGRAPHS TO E$ PUT IN PUBLIC DOClfENT ROOM TRANSCRIPT BEING TAKEN TO MAKE SUPE WE CAPTURE THE SPECIFIC l
C0 m WTS OF ALL PARTICIPANTS COPIES OF PAPEPS 8 VIEWGPAPhS TO BE PADE AVAILABLE BY NOON FRIDAY. IF YOU HAVE TO LEAVE EARLY & WANT A COPY, LEAVE YCUR i
MAILifE ADDRESS WITH MS, KONDULIS
)
0 -
issue Paa meS 10 Be. mum 10 reeS issues. e mR1 BACK ON THURSDAY a FRIDAY i
FRIDAY ABOUT l00N WE PLAN TO SUFFARIZE ,
1 l
TURN IN YOUR WRIM PAPERS a VIEWCRAPHS T0 tE. VONDULIS ASAP l MS KONDULIS WILL TYPE SiMARIES a PAKE VIEWGRAPHS l
O !
16
- - __ 1- ._
AtEICIPATED PAJOR MILESTONES ITEM DATE 1, MARK I STAT 9ENT OF GEERIC PERF0PPANCE ISSUES E 1/22/88-CCFPLETE INVITATION TO RES/NRR WORKSHOP
- 2. INEL LETTER REPORT ON MARK I VENTING li19/88-CCPPLETE 3, ORNL REPORT COREELT PROGRESSION 2/148-C0FPLETE
- 6. ItCbSTRY/ESEARCHER/FUEllC WORKSHOP 2/24-2/2C/88
- 7. CCt111SS10N PAPER ON SEVEPE ACCIDENT ISSUE CLOSUPE 4/88
- 8. IPE C&EISSION PAPER ON GENERIC LETTER 4/88 -
A 9. PPELIMINARY STAFF EVALUATION OF PAPK 1 ISSUES 2/28-4/1/88 V
- 10. PRELIMINARY PESULTS ON PEACH E0TTOM FOR 1150 FINAL 4/88
- 11. CUtilSSION PAPER ON INTERIM STATUS OF STAFF F%RK I 4/29/88 PEPFOPFANCE EVALUAT10N
- 13. FINt.L RECatENDATIONS ON PATE I CONTAINENTS 9/88
- 14. WCRKSHOPS ON OTHER CONTAlttENT TYPES 10/88-2/89 15 CUP 11SS[0N PAPER ON OTHER CC6TAINENT TYPES 9/89
- 16. ACPS & CRGR TETit0S ON OTHER CONTAIIPENT TYPES 10/89 l
l O
19
. - _ ._--~--... - . . _ - --- - - - - - - -- -
l l
I l
1 I
I 1
. l l
l 1
i i
i l
I
, ADDITIONAL INFORMATION i
1 l
l l
l l
4 1
i 1 l
l l
1 O .
4 1
4
t NUREri/CR 4920
., cs BNL-NUREG-52070 0 )
O VOLUl1E 1 ASSESStiENT OF SEVERE ACCIDENT PREVENTION AND tilTIGATION FEATURES:
BWR, MARK I CONTAINiiENT DESIGN Prepared by W. T. Pratt, F. Eltawil a ,* K. R. Perkins, R. G. Fitzpatrick ,
W. J. Luckas, J. R. Lehner and P. Davis **
Date Published - March 1988 O Department of Nuclear Energy Brookhaven flational Laboratory Upton, riew York 11973
'U.S. leuclear Regulatory Comission, RES
- Inter-iountain Technologies Inc.
O 9
j' _
. . - - - . .- .. - . ~ =- . - , _ _
l I
NUREG/CR-4920 1 BNL-NUREG-52070 VOLUME 1 i
l l
l l
ASSESSMENT OF SEVERE ACCIDENT PREVENTION AND MITIGATION FEATURES:
BWR, MARE I CONTAlHtiENT DESIGN l
Manuscript Completed: February 1988 t
Date Published: Maren 1988 y*
Prepared by W. T. Pratt, F. El tawil a,* K. R. Perkins , R. G. Fitzpatrick, W. J. Luckas, J. R. Lehner and P. Davis ** ,
O .
Departnent of tbclear Energy Brookhaven National Laboratory l Upton, New York 11973 I
i "U.S. Nuclear Regulatory Conni ssion RES i **Intermountain Technologies Inc.
- 1 1
Prepared for -
4 Division of Reactor and Plant Systems Of fice nf Nuclear Regulatory Research U.S. Nuclear Regulatory Commission
, Washington, OC 20555 !
Cantract No. DE-AC02-76CH00016 !
i FIN A-3825 i
e n - ---,,-,
J ABSTRACT Pl ant features and operator actions, which have been found to be inpor-tant in either preventing or mitigating severe accidents in BWRs with Mark I containments (BWR liark I's) have been identified. These features and actions were developed from insights derived from reviews of in-depth risk assessments perfomed specifically for the Peach Bottom plant and fron assessment of other i relevant studies. Accident sequences that dominate the core-damage frequency I and those accident sequences that are of potentially high consequence were id enti fi ed. Vulnerabilities of the BWR Mark I to severe accident containment loads were al so identi fied. In addition, those features of a BWR ,* ark I, which are important for preventing core damage and are available for nitigat-ing fission-product release to the environnent were also identified. Thi s ;
report is issued to provide focus to an analyst examining an individual )
pl ant. This report, calls attention to plant features and operator actions and provides a list of deteministic attributes for assessing those features and actions found to be helpful in reducing the overall risk for Peacn Pottom and otner liark I plants. Thus, the guidance is offered as a resource in examining j tne subject pl ant to detemine if the same, or similar, plant features and l operator actions will be of value in reducing overall plant risk. This report i s intended to' serve sol ely as guidance. {
O I
e j
O iii
_ _ _ . . - _ - - .- - _ __ J
4
~
TABLE OF CONTENTS .
d) Page ABSTRACT................................................................ iii LIST OF TABLES.......................................................... viii AC K N OWL E D G ME N T S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x i NOMENCLATURE............................................................ xiii
- 1. EXECUTIVE
SUMMARY
................................................... 1 1.1 Core-Damage Profile.........<.................................. 2 1.2 Co n s eq u e n c e An a l y s i s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3 Plant Features, Operator Actions and Their Important Attributes..................................................... 3
, 1.3.1 Measures to Mitigate Fission-Product Releases........... 3 1.3.2 Measures to Control the Frequency of High-Consequence Sequences............................................... 4 1.3.3 Measures to Reduce High Core-Damage Frequency Sequences. 4 1.4 Using the Report............................................... 5 1.5 References for Section 1....................................... 6
- 2. INTRODUCTION........................................................ 9 2.1 Background..................................................... 9 2.2 Objectives..................................................... 9 2.2.1 Pl ant Featu res and Operator Acti ons . . . . . . . . . . . . . . . . . . . . . 10 O 2.2.2 D e t e rmi n i s t i c A t t r i b u t e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3 Organization of the Report.....................................
11 12 2.4 References for Section 2....................................... 12
- 3. OEFINITION OF GOALS AND RELEVANT BWR MARK ! FEATURES................ 15 3.1 Mitigate Fission. Product Releases.............................. 15 3.1.1 Pl a n t v u l ne ra b i l i t i es . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 3.1.2 Mitigating Features..................................... 17 3.1.3 Containnent Integrity via Wetwell Venting and Suppres si on Pool E f fecti venes s . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.2 Control the Frequency of High-Consequence Sequences. . . . . . . . . . . . 19 3.3 Re&;ce Hi gh Core-Damage Faequency Sequences . . . . . . . . . . . . . . . . . . . 19 3.3.1 Station Blackout via Alternate Reactor Pressure
, Ves s el Inj ect i on . . . . . . . . . . . . . . . . . . . . . . ................ 2) 3.3.2 Loss of Containment Heat Removal by Using Alternate Cooling............................................ .... 20 3.3.3 Reactor Pressure Vessel Depressuri zation Perf ormance. . . . 21 3.3.4 Support Sys t em Interdependenci es . . . . . . . . . . . . . . . . . . . . . . . . 21 3.3.5 Re a cto r Bui l di n g Fl oodi ng. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.4 References for Section 3....................................... 21 4 PLANT FEATURES. OPERATOR ACTIONS AND THEIR ATTRIBUTES............... 23 4.1 Mea sures to Mi ti gate Fi ssi on-Product Rel eases . . . . . . . . . . . . . . . . . . 24 4.1.1 Maintaining Containment Integrity via Wetwel.1 Venting and Suppressien Pool Effectiveness...................... 24 T
v
4.2 . Measures to Control the Frequency of High-Consequence Sequences...................................................... 24 4.2.1 Preventing and Mitigating Interf acing Systems LOCA. . . . . . 24 4.2.2 Preventing and Mitigating Anticipated Transients Without Scram........................................... 25 4.3 Measures to Reduce High Core-Damage Frequency Sequences. .. ... .. 25 4.3.1 Mitigating Station Blackout via Alternate Reactor Pressure Vessel Injection............................... 25 4.3.2 Preventing Loss of Containment Heat Removal by Using Alternate Cooling....................................... 26 4.3.3 Ensuring Reactor Pressure Vessel Depressurization Performance............................................. 26 4.?.4 Identi fyi ng Support System Interdependenci es. . . . . . . . . . . .
26 4.3.5 Preventing and Mitigating Reactor Building Flooding..... 27 4.4 Using the Report............................................... 28 4.5 References for Section 4....................................... 28 APPENDIX A - SEVERE ACCIDENT RISK INSIGHTS.............................. 43 A.1 C o r e - D ama g e P r o f i l e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 A.1.1 Peach Bottom Core-Damage Profiles - RSS, IDCOR, and ASEP/SARP.................................... 43 A.1.2 Peach Bottom Dominant Sequences: Differences in RSS, 10COR, and SARP Analysis........................ 44 A.1.2.1 TC (ATWS) Sequences............................ 45 A.1.2.2 TB (Station Blackout) Sequences................ 47 A.1.2.3 TW Sequences......................... . ......... 48 A.1.2.4 TOUV and TOUX Seque*ces........................ 49 A.1.3 Dominant Sequences Comparisos.3: Peach Bottom and Other BWR Analyses.................................. 49 A.1.3.1 TC Sequences................................... 50 A.1.3.2 "A Sequences................................... 50 A.1.3.3 TOUV and TOUX Sequences........................ S1 A.1.3.1 TB Sequences................................... 51 A.1.3.5 T P O I Se q u e n c e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 A.1.3.5 T PQE S eq u e n c e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 A.1.3.7 In t e rf aci ng Sys t ems L0CA. . . . . . . . . . . . . . . . . . . . . . . 53 A.2 Core-Meltdown Phenomena and Containment Response............... 53 A.2.1 In-Vessel Hydrogen Generation (NRC/IDCOR !ssue 5)....... 55 A.2.2 Core Slump. Core Collapse, and Reactor Vessel Failure (NRC/IDCOR Issue 6)............................. 55 A.2.3 Containment Failure 3ecause of In-Vessel Steam Explosions (Issue 7).................................... 57 A.2.4 Direct Heating of Containment (Issue 8)................. 57 A.2.5 Ex-Vessel Weat Transfer Model From Molten Core to Concrete (Issue 10).................................. 58 A.2.6 Sup,:ression Pool Bypass (Issue 13A)..................... 53 A.2.7 Containment Performance (!ssue 15)...................... 59 A.2.8 Seconda ry Contai nment Performance ( Is sue 16) . . . . . . . . . . . . 59 A.3 Fission Product Release................................... .... 59 A.3.1 Fission Product Release Before Vessel Failure (Issue 1)............................................... 60 vi
Page O '
k/ A.3.2 Fission-Product and Aerosol Retention in the Reactor Co ol a n t Sy s t em ( I s s u e 4 ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 A.3.3 Ex-Ve ssel Fi ssion-Product Rel e6se (Issue 9) . . . . . . . . . . . . . 60 t A.3.4 Revaporization of Fission Products From the Reacto r Cool an t System ( Is sue 11) . . . . . . . . . . . . . . . . . . . . . . . 60 A.3.5 Fission-Product Depositton t odel in Containment (Issue 12).............................................. 61 A.3.6 Second ary Contai nment Perfo rnance ( Issue 16) . . . . . . . . . . . . 61 A.4 O f f s i t e Co n s e qu e n c e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 A.5 S umm a ry a nd Ri s k In s i g h t s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 A.S.1 Co r e- Dama ge P ro f i l e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - 62 A.S.2 Co n s e qu e nc e An al y s i s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 A.6 R e f e r e n c e s . . . .' . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 APPENDIX B - PLANT FEATURES RESilLTING IN LOW PROBABILITIES FOR ACCIDENT SEQUENCES........................................ 79 B.1 References..................................................... 80
) -
1 i
l I
1 i
vii i
LIST OF TABLES Tabl e Pa ge 1.1 Plant Features and Operator Actions for Preventing and Hitigating Severe Accidents in a BWR with a liark I Containnent., 7 4.1 Important Attributes for SWR fiark 1 Containnents Relating to Plant Features and Operator Actions: Maintaining Containment In t e g ri ty v i a We twel l Ve n t i n g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 4.2 Important Attributes for BWR Mark 1 Containments Relating to Plant Features and Operator Actions: tiaintaining Suppression Po ol E f f ec t i v e n e s s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 4.3 Inportant Attributes for BUR Mark I Cor.tainnents Rel ating to Plant Features and Operator Actions: Preventing and Mitigating Interfacing Systems 1.0CA........................................ 34 4.4 Important Attributes for BWR Mark 1 Containments Relating to Plant Features and Operator Actions: o revanting and Ititi, gating Anti ci pated Tran si ents Wi thout Scran ( ATUS) . . . . . . . . . . . . . . . . . . . . . 35 4.5 Inportant Attributes for BWR Mark I Containnents Reitting to Plant Features and Operator Actions: Mitigating Station Blackout via Al ternate Reactor Pressure vessel (RPV) Inj ecti on . . 37 4.6 Inportant Attributes for BWR nark I Containnents Relating to Plant Features and Operator Actions: Dreventing Loss of Contai nment Heat Removal by Usi ng Al ternate Cool i ng. . . . . . . . . . . . . 38 4.7 Important Attributes for BWR Mark I Containments Relating to Plant Features and Operator Actions: Ensuring Reactor Pressure ve s sel De p res su ri za t i on Pe r fo rma nc e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 4.8 Important Attr1Dutes' for BWR Mark I Containments Relating to Plant Features and Operator Actions: Identi fying Support Sy s t en I n t e rd e p e nd e n c : e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 4.9 Inportant Attributes for BUR fiark I Containments Relating to Pl ant Features and Operator Actions: Preventing and Mitigating R e ac t o r Bui l d i n g Fl ood i n g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 A.1 BWR Conpari sons : Core-Danage Frequencies....................... 66 A.2 Sunnary of Charges Included in the IDCOR Connitted Core-Darage Profile *or Peach Botton................................. 67 A.3 Conditional Prcoabliities of Care nanage Given an ATWS in Pescn Bottom vs. Shoren8n.................................... 68 A.A T3 Sequences: Comparisons...................................... 6Q A.5 TW Sequences: Conparisons...................................... 70 A.6 TQuV and TOUX Sequences: Comparisons........................... 71 A.7 Compari son of the ;DCOR and SARP Con tai nment Ma trices. . . . . . . . . . . 72 A.R NRC/ICC07 Issues................................................ 73 A.9 Compari son of ICCOR and SARP Predictions of Fission Product Rel ease for an 3TUS Sequence Uith 'in Operator ac tions Taken..... 74 A.10 Caiparison of 10COR ind SARP Predictions of Fi ssion-Product delease for a Station Bl ackout Requence......................... 75 A.11 Conparison of IDCOR and SARP Predictions of rst Districotiun for 3 3tation R1dckout Sequenco (Fraction of Initial Core Inventory)...................................................... 76 1.1? Cornarison of ICCOP and 'tuPEG-1150 Consequence Resul ts
(?erson-Ren).................................................... 77 O
viii t
LIST OF TABLES (Cont'd)
Page 8.1 Typical Success Criteria . for ECCS in Response to BWR-4 LOCA..... 81 B.2 Typical LOCA Frequencies for a BWR From the IREP Browns Ferry Study........................................................... 82 O
e 1
l 4
l
- ix 1
ACKNOULEDGl1ENTS
(./ The _NRC manager for this program, F. D. Cof fnan, providc4 considerable input
~
and technical direction. In addition, the program has oenefited from the technical direction given by Ors. Z. R. Rosztoczy, B. Sheron and T. Spet s and from the technical review performed by R. J. Rarrett.
Tha- authors are grateful to Or.
N. Hanan for his significant input to the '
draft version of this report. in addition, B. J. Garrick, D. H. Johnson and D. R. Buttemer of Pickard, Lowe and Garrick, Inc. , J. W. Hicknan and A. L.
Camp of SNL, T. Kress and S. Hodge of ORNL and P. Cybulskis of BCL reviewed an early draf t of this document and provided many helpful suggestions. The re-port has benefited significantly from the detailed BNL equipment qualification and human reliability perspectives of B. E. Miller and Dr. C. ft. Spettell , re-s pecti v~ely , and the in-depth review and guidance given by Dr. R. A. Bari, Dr. G. A. Green.e and R. E. Hall of BNL. The authors also appreciate the sug-gestions and perspective provided by representatives of other organizations including J. F l'ayer of IEAL.
The authors are especially gr'a te f ul to S. Flippen for her considerable patience in producing numerous revisions to this report, ahe was acly assisted from time to time by C. Conrad. D. iiiesell, S. !!oore, and H. Nelson.
1 0 '
l 1
a 0
xi n .
~
a ac alternating current A large lost Jf coolant accident (LOCA)
ADS automatic depressurization systen ARC alternate room cooling ARI alternate rod insertion ASEP Accident Sequence Evaluation Program ATWS ' anticipated transients without scran BF Browns Ferry Nuclear Station BNL Brookhaven National Laboratory BWR boiling water reactor C failure of reactor protection system (RPS) '
C t mechanical failure to scram C
2 operator failure to actuate standby liquid control system (SLCS) or to control level with high pressure systen (HPS), or failure of SLCS CDEP failure of manual depressurization CDF core-damage frequency ;
CHR containment heat removal CLWG Containment loads Working Group C0tlT inadequate or no containment heat removal leading to loss of core cooling CPUG Containment Perfornance Working Group CRD control rod drive systen '
dc direct current OG diesel generator OGCli diesel generators conmon mode failure O DGREC OHR failure to rec.over diesel generators decay heat renoval E f ailure of coolant injection E
f ailure of injection after venting (used for ATUS sequences only)
ECC emergency core cooling ECCS emergency core cooling systems ,
EDG Energency Procedure Guidelines l Esu energency service water FW feedwater systen GG Grind Gul f Nuclear Station '
G1 generic issue HAOS failure to inhibit AOS HEP numan error probability HPCI nign-pressure coolant injection system HPi$ nign-pressure injection systens HPLC failure to control RPV water level wich HPCI (either due to operttor HPSU nign error and/orservice pressure hardware failure or malfunction) water I
f ailure of containment heat removal
!0COR In41ustry Cegraded Core Rulenaking Progran I i
!NJ fat'ure of injection with low pressure systens (LPS) af ter containment f ailure (CF) ,
!CRV inadvertent open relief valve i Id? [nterin Reliability Evaluation Progran IDE incividual plant examination ISL interfacing system LOCA J failure of the HPSu xiii
NCt1ENCLATURE (Cont'd)
LLRT local leak rate testing LOCA loss-of-coolant accident LOOP loss-of-offsite power (sometines denoted by LOSP)
LPCI low-pressure coolant injection LPIS low pressure injection systens LPLC failure to control RPV water level at low pressure (either due to operator error and/or hardware failure or mal function)
LWR ligh water reactor MCC notor control center MSIV main steam isolation valve NPSH net positive suction lead NRC U.S. Nuclear Regulatory Cennission NRC/RES U.S. Nuclear Regulatory Cornission, Of fice of Nuclear Regulatory Research P one or nore stuck open relief valves (SORV)
PCS power conversion systen PRA prooabilistic risk assessnent PWR pressuri zed water reactor 0 f ailure of feeawater system 0 PCS recovered early 0,1 PCS recovered late R$ reactor building RCIC reactor core isolation cooling systen RHR residual heat renoval systen RPS reactor protection systen EPV reactor pressure vessel RSS Reactor Safety Study RSSNAP Reactor Safety Study Metho1 ology Application Progran S t internediate LOCA 5 snall LOCA S$RP Severe Accident Researen progran SARRP Severe Accident Ri sk Reduction Progran 530 station blackout SI safety injection SLC f 3i! are at SLCS (due to f ailure of nanual initiation aralar 41ue *.c hardware n31 f uncti0n or f ail ure)
SLCS standby liquiet control systen SNL Sandia National Laboratories SNPS Snorehan Nuclear Power Station 509Y stuck open safety relief valse ITCP Source Tern Coce Pacxage Su service water
(?) unident1 fled contribution
(*) transient sequences are denoted oy T followed by letters denoting the relevant f ailure, e.g. , TC transients invol ving f ail ure af 9PS TOUV transients involiing f ailure at FJ, HPiS, and LPIS Atc.
T transient TAF top of active fuel T3 station alacrout sequence ( al so re ferred to as $30)
TC AiWS TCV IOSS Of CO' denser vn uun initiator l Xiv
NOMENCLATURE (Cont'd)
Tpw loss of feedwater initiator it isolation transients ,
T}25% isolation transients at power level greater than 25%
TM manual shutdown TNOPCS transients with PCS initially unavailable TpCS transients with PCS initially available TP01 transient sequence with SORV failure, feedwater failure and loss of CHR TQU transient sequence in which the feedwater and HPI systens fail TQUX transient sequence in w;11cn the feedwater and HPI systeas fail and depressurization does not occur Ti -
turbine trip initiator
'iT turbine trip transients at power level greater than 25%,
O f ailure nf hign-pressure injection system (HPIS)
USI unresolved safety issue V
failure of low-pressure injection systen (LPIS)
VC 3 operator systens failure to control level .and reactivity with low-pressure VENT containment (wetwell) venting failure W
X f ailure of containnent heat renoval (CHR) failure of reactor pressure vessel (RPV) depressurization O
O
. xv '
4
- 1. EXECUTIVE SUttfiARY The U.S. Nuclear Regulatory Conmission (NRC) has fomulated an approach for a systematic safety examination of existing plants to detemine whether particul ar severe accident vulnerabilities are present and wnat changes are desirable to ensure that there is no undut: risk to public health and safety.
This systenatic examination program is part of *.he overall severe accident policy outlined in the implenentation plan.1 As part of the foundation for the systematic plant examination progran, the Industry Degraded Core Rulemaking Progran (10COR) selected four reference plants for detailed analysis: Peach Botton, Grand Gul f, Sequoyah, and Zion.
The 10COR analyses perfomed for the reference plants have been documented together with tne nethodology used for the analyses and the technical basis supporting tne methodology.
Parallel witn tne 10COR work, the NRC under the tevere Accident Research Progran (SARP), perfomed risk assessments, audit calculations, sensitivity studies, and uncertainty analyses for five p1 ants. The five plants considered by SARP were Peach Botton, Grand Gul f, Sequoyah, Zion and Surry.
The purpose of thi s report is to identify plant features, operator actions and their important attributes to be used as a common starting point for the systematic safety examination of individual plants. This information is derived as a result of a review of all IDCOR and SARP analyses. It identi-fies plant features and operator actions tr'at studies showed were important for either preventing or mitigating severe accidents in eacn plant type.
Three basic goals for this severe accident progran apply equally to all plant types:
Goal 1: ititigate fission-product releases.
Goal 2: Control the frequency of high-consequence sequences.
Goal 3: Reduce hign core-damage frequency.
The aim was, therefore, to identify plant features and operator actions ano ineir important attributes that could be used to acnieve these goal s dur-ing the examination of individual plants.
In this report, those plant features and operator actions that were found to be irportant to eitner preventing or nitigating severe accidents in the reference plant studies are provi1ed for the utilities to use as part of e3cn individual pl ant examination (IPL). It is not ?ne intent of this report to specify a set of inprovements for either the reference plant or for any otner pl ant wnicn woula ce su'ficient to acnteve a certain level of safety. In-stead, the report inoicates potential improvements in various areas of pl ant design anc operation of unich eacn utility snnuld be aware unen concocting its IPE and maxing decisions on plant inprovements. The intent is to provide guidance to tne analyst performing an IPE, as to tne plant features and oper.
ator actions wnich were found to reduce overall risk. It is prulent.to check wnetner po ten t i .11 improvements identi fied in studies of otner sinit ar pl ants can ce of ne lp in baproving overall plant perfornance. The i n f o rm a t i o n ca n'.
tained in tnit report, theref ore , compl ements the iPEs not it snould not be construed to discourage the investigation of new ways to reduce over3ll risk.
1 s _ __-
In this report, the attributes whicn have bp5n identified as important to assess the performance of plant features and op;.rator actions that were snown to be beneficial in preventing or mitigating a severe accident are presented to provide de,.eministic (as opposed to .probabilistic) performance measures which are judged to be helpful . When a decision is made to alter plant fea- j tures or operator actions, the utility should address a set of questions re. j lating to the design, operation and availability of the needed equipment and !
the training of operators. For the example of wetwell-venting guidance, it is j important to assess the capacity of the venting system, the selection of set- l points to initiate venting, the availability of applicaale procedures and tne l accessibility of certain valves by operators. The section on wetstell venting ,
provides helpful infomation in assessing venting capability in each individ-ual pl ant .
Based on an extensive review of prior severe accident investigations, the authors have provided a list of detailed attributes related to each important plant feature or operator action which can be used to assess the capahility of individual boiling water reactor (BWR) tiark I plants to cope with severe acci-dents. Al tnougn much of the work is based on prooabilistic risk assessments (pRAs), the infomation presented in this report i s deterministi.c in nature ,
i.e., it describes specific features of key systens and operational procedures wnich have neen found nel p f"I in reducing the likelihood of severe accidents wi t hout regaro for the f ail. ' rate of tne pl ant feature or operator action.
The report takes into account oetailed severe accident experiments and analy-ses perfomed by the NRC/RES, the nuclear power industry and foreign govern-ments.
The following sections present the insignts gained f rori reviewing the PRAs. Sceci fically, the IDCOR Peach Botton Integrated Containment 1n31yses' and the SARP Peach Botton reports -s were reviewed in detail. These studies 3
were comparea wi th tne original Peach Botton risk assessment in the Reactor Sa fety Study (PSS) (WASH-la00 ) 6 and relevant BUR PRAs for other plants , nanely Browns Ferry,7 Limerick 3,3 and Shorenan,10.11 1,1 Core-Danace Profile P3As for E.n s 9 ave irdicated that acti cents initiated ny traastents ratner than l os s-of-cool in t accidents 'LOCAs) dominited tae tot al care .ia., age f requency (CDF) estimates. However, there appeared to be no consistent pat-l tern of rel ative cant 1ng of transient sequences among tne PRAs reviewed, it l
is also incortant to chserve that for a given accident secuence, contributors to di f ferences in quantitative resul ts between tne PR As incl uded sucjective modeling issumptions , plant di f ferences as well 3s data differences. For tne I four BWR ol ants considered in the si x PRAs tnat wera exanined, tne sane few l functional actident sequences figurec prominently in all of the respective CDF profiles, in the DSSi (unich used tre Peach Botton plaut) and tne Interin Pelianil.
7 ity Evaluation Program (IRED) study (wnicn usad ne t3rowns Ferry pl 3nt) loss of containnent neat m ov11 sequences were founn to ne important contrilutors to core nel? ' 300ut M ", ) . The more recent Acc id en t Sequence Evaluation Pro-gr!n (ASE?) ed ICCOR studies nave reduced the CDF , attrinutable to tnese se-quences , based on operating procedures that include Venting ano alternatin injeCClOn,
? /
For the Linerick P1A, Browns Ferry IREP and Shorehan PRA, accident se-O quences with failure of hign-pressure injection were inportant contributors to CDF. Most' of this contribution was because of a high failure rate for the automatic depressuri zation systen ( ADS).
Both SARP and IDCOR indicated that station blackout (SBO) and anticipated transients without scram ( ATWS) are the dominant core-melt sequences for the Peach Botton plant. Both studies calculated a total C0F approaching 10-5 per reactor year for these events.
1.2 Consecuence Analysis The assessnent of core-nel tdown phenonena and containment response in the PRAs indicated that the Mark I containnent is vulnerable to severe acci-dent containment loads. Unless mitigative actions are taken, a Park I con-tainment has the potential to fail shortly (in a few hours) af ter the core deDris melts through the reactor pressure vessel, if containment failure occurs in the drywell , any fission products in the drywell atmospnere could pass to the reactor building and ultinately to the environment witnout ne benefit of suppression pool scrubbing. Because of tnis vulnerability, the predicted offsite Consequences were relatively insensitive to tne definition of the accident sequence. In addi tion, di f ferences in the IOCOR and severe Accident Ri s k Reduction Progran (SARRP) assessnents of containment response and fi ssion-prod uc t release also did not resul t in major dif ferences in the predicted of f si te consequences. The only time that a major reducti,on in of f-site consequences was predicted by 10COR and SARRP was with successful wetwe'll venting and no suppression pool bypass.
SARRP estinated that the dominant accident sequences (nanel y, ATUS and S30) resul t in a significant probaoility of suppression pool bypass. Thus, tne bypass nechanisms identi fied in the SARRP analysis have to he addressed to ensure nitigating of fission products.
1.3 Plant Features, Ooerator Actions and Their [nuortant Attributes Table 1.1 crovides nelpf ul information to assess the performance of plant f ?atures and operator actions identi fied as inportant to prevent or nitigate severe accidents.
1.3.1 fieasures to ititigate Fission-Product Releases The assessment of core-meltnown phenomena and containment response indi-cates that the flark l Containment is vulnerable to severe accident Containment loads because of its relatively small v ol ume . Unless nitigative actions are taken, a lia rk I containnent has the potential to f ail shortly af ter the Core debris nelts tnrough tne reactor pressure vessel. For tnis reason two impor-tant categories of plant features and operator actions have been identi fled wnicn relate to nitigating fission-product releases under severe accident con-ditions.
! ten 1 - ftaintaining Containnent Integrity via Uetwell Venting fia rx 1 containnerts are very ef fective at condensing stean, but their O snail volone naxes then vulnerable to any connustibla and noncondensable gases 3
c_ _. __
that would be generated during a severe core-nel tdown accident. The inpact of hydrogen burning was not si gni fic ant for the (1 ark I containnent because the atnosphere is inerted. However, SARRP predicts that the acctriulation, of non-W condensable gases, released from core concrete interaction, will fail tne con-tainment because of overpressurization, 10COR, on the other hand, had pre.
dicted that the containnent would fall because of high temperatures before overpressurt:ation. 01 f ferences in the predicted drywell pressure and temper.
ature nistories will in fl uence the containnent perf ormance. The infornation
- presen ted in Table 1.1 on wetuell venting (relating to preventing overpressure l failure) and on containment spray (related to preventing failure because of l high temperature) night prove to be hel pf ul in naintaining containment integ.
) rity.
l
! Item ? - ttaintaining Suppression Pool Effectiveness '
The ability of the liark I suppression pool to trap aerosol fission prod-ucts is an important nitigative featuret since it l ead s to a d i rt*c t red uc ti o n in of f site consequences by a f actor of 10 or noro. Thus , any pathways that night open, which wo ul d all CW tile fission products to hypass the pool, are undestrable. Therefore, measures to prevent fluton products fron bypassing the suppression pool are provided i n Ta n l e 1.1.
1.3.2 t'easures to Control tne Frequency of Hiyh-Consequence Sequences f lten 3 - Preventing and flitigating Interf acing Systems LOCA 1
l
[n general, BWR lia rk l PRAs have found inter f acing systens' 'LdC A t0 he unitkely. However, the possibility of aigh relenes makes it inportant to ensure that the frequency of these t* Vents is kept very law at all "ara I plJnts.
Item 4 Prevoqting and fiitigating Anticipated Iransients 'Jithout $ cran An t i c i pa t ed transtents without scram ( ATUS) have been found ta Se inpor-t int contrihotors ta rist o r nany w a.. The NaC pronul .n ted ene Atus rule t3 r0'tuCe the f roquency of A',j$ oventg, Ihis i n f o rma t 10n pf*ewq t ed in this repqr*
t?'Th a s 1 ?c s 'Me it'p0 Fl a nt ** Of ( J r r PC t o'9 e r Qt'n t. y p r:.lC eiluri i ard Operat )r t
t r.l i n i ni) in retaverinQ ff'h1 an A I7) event,
- 1. 3. 3 lieasures to 4txtuce liigh Cord-Damage f requency Soquentes Iten 5 - titttgating Station Bl ac kou t vti Al t er n.i t e Reactor Pressore Vessel InjeCLion f o f' 1CCtdentS Involving the Iass df 0ffsi?e power antt onsito odergenty power, the NW C roc ormand s e T i,1) n 1 Og t he pr0po%e,) itjt1on nlatkout ($30) rule for applicahility, ihr inro rn a t i o n pros nted in tnis roport is intenae.i to et1pha s t te t he heel (J i e J er. it finr pl ant specific features ind pottntiil c Moo n $
C a%e f aillres whitn Could tigable s y s t em'. requirod to wo r k -luring an Q ),
f )r indt41jual p} ants ,v n 1 C h Ira f ounit to have ) vulnet abilit y ta iBd, tre repdr$ hldnlljnts 'ne i n p o r t, i n c e Of praptar v9erijef% y p rnt e*,t ur os a ni t ope r it a r tritning to recoverin.) fran an $dd event.
O
Item 6 - Preventing Loss of Containment Heat Removal by Using Alternate Cool- l ing l l
Accident sequences involving loss of containment heat removal (CHR) were i found to be quite important in the. earlier pRA studies reviewed. In WASH-1400 loss of CHR sequences accounted for 53% of the calculated CDF. In the Browns i Ferry IREP study, these sequences similarly accounted for 50% of the calcu-lated core-damage frequency. However, the most recent Peach Bottom studies, IDCOR and ASEP/SARRP, show a two-and-three-order-of-magnitude reduction, re- ,
spectively, in the quantification of these sequences. Therefore, in Item 6, i the mechanisms already effectively employed at Peach Bottom to reduce the fre-quency of loss of CHR sequences are highlighted.
Item 7 - Ensuring Reactor Pressure Vessel (RPV) Depressurization Performance One of the insights gained from the existing BWR PRAs is the importan,ce of the ADS in mitigating loss of high pressure injection sequences. Speci fic insights are presented to ensure that the likelihood of these accident se-quences occurring is low, item 8 - Identifying Support System Interdependencies Although the importance is difficult to quantify, one of the insights de-veloped in most risk assessment studies is the importance of support system interdependencies. For example, a draft of the SARRP Peach Bottom study in-dicated that loss of all service water was a dominant contributor 3
to core mel t . The final version of the accident sequence studies has reduced it to O- one percent of the overall core melt frequency. In order to ensure that sup-port system vulnerabilities do not cause unacceptably high core melt fre-quencies for other RWR Mark I plants, the importance of support system inter-dependencies is emphasized.
Iten 9 - Preventing and Mitigating Reactor Building Flooding Flooding of tne reactor building has been found to be a significant con-tributor to core damage in only one fiark Il plant. However, the concerns appear to te of general applicability to other designs. Thus, insignts are provided to assess the potential for flooding of safety-related eouipment for Mark I plants.
1.4 Usino the Reoort Numerous investigations, including PRAs, have been perforned for the ref-erence plants arid similar plants by both the NRC and the nuclear power indus-try. The insights gained from many of the studies have been used in develoo- )
ing the plant features, operator actions and their attributes contained in ;
this report. These insights are issued to provide guidance to the analyst l performing an IPE. This guidance is in the form of plant features, operator l actions and deterministic attributes for assessing those features and actions found to te 5elpful in reducing the overall ris< for Peach Bottom and other Mark i plants. Thus , the guidance is given to provide a resource in examining the subject olant to deternine if the sane, or similar, plant features and operator actions will be of value in reducing nyerall plant risk, s
l 1.5 References for Section i
- 1. SECY-86-76, "Inplementation Plan fnr the Severe Accident Policy Statement and the Regulatory use of New Source-Tern Infomation," NRC/E00, Feeruary '
28, 1986. .
- 2. "Peach Botton Atonic Power St a ti on-Inte grated Containnent An al yse s ,"
10COR Technical Report T23.1PB, l4 arch 1985.
- 3. A. ii. Kolac zkowski et al . , " Analysi s of Core Danage Frequency f ron Inter-nal Events: Peach Bottom, tinit 2," Sandia National Laboratories, NUREG/
CR-4550, Volume 4, October 1986.
4 C. N. Amos et al . , "Containment Event Anal ysi s for Postulated Severe Accid ent s : Peach Botton Atomic Pnwer Station, Unit 2 ," Sandia National Laboratories, NUREG/CR-4700, Vol ume 3, Oraft Report for Comment, liay 1987
- 5. C. N fcos et al ., "Evaluation of Severe Accident Risks and the Potential for Risk Reduction: Peach Bottom, Unit 2," Sandia National Laboratories, NUREG/CR 4551, Volume 3, Oraf t for Coment , April 19R 7 6 "Reactor Safety Study: An Assessment of accident Ri sks in U.S. Ccmer-cial Nuclear Power Pl ants ," U.S. Nuclear Regul a tory Connission, WASH-1400, NUREG/75-014, October 1975. l l
7 S. E. Mays et al ., "Interin Reliability Evaluation Program: Analysis of i tne Browns Ferry, Unit 1, Nucl ear Pl ant ," Idaho National Engineering Lao- I oratory, NUREG/CR-2302, July 1o42. l
- 3. "Procabilistic Risk Assessment - Limerick Generati ng Station ," h i l ad el -
pnia Eiactric Co., 1982.
- 9. I . A. Paca:oglou et al . , "A Review of the Linerick Generating Station Probabilistic Di sk Assessnent," Brookhaven National Laboratory, NUREG/CR-3023, February 1983,
- 10. "Procacili stic Risk Assessment - Shorenam Nuclear Power Station," Long Islard Li gnting Company , . lune 1%3.
- 11. D. Ilnery et al . , " A Review of the Shorenan Nuclaar Power Station Droo 3 bilistic Risk Assessment (Internal Events and Core Damage Frequency),"
l Brooxnaven National Laboratory, Nt! REG /CR 4050, June 1%5.
O 6
c;
- 1 Table 1.1 Plant Features and Goerator Actions for Preventing 'and rtitigating Severe Accidents in a BWR with a flark I Containnent l
Iten Plant Features or Operator A:tions Measures to liitigate Fission-Product Releases 1
Maintaining Containment Integrity via Wetwell Venting 2 tiaintaining Suppression Pool Effectiveness 2.A Preventing Suppression Pool Bypass 2.B Using Drywell Spray Measures to Control the Frequency of High-Consecuence Seouences 3 Preventing and ttitigating Interfacing Systems l'JCA 4 Preventing and fiitigating Anticipated Transients Without scran Measures to Reduce High Core-Damage Frecuency Seouence_s_
5 flitigating Station Blackout via Alternate Reactor Pressure Vessel Injection -
6 Preventing Loss of Containnent Heat Removal by Using Alternate Cooling 1
7 Ensuring Reactor Pressure vessel nepressurization Perfornance 3 Identi fying Support System Interdependencies l j
i 9 Preventing and tiltigating Reactnr Ruilding Flooding 1
1 1
7 i
g, maa2. .,A. A ama-a- a - -.. - A a h= -- - .+ w -- ---
e t
9 9
l I
(
0 3
1 u ~ -_-_ _____..-_______._____ ' ~~ ~~~ - - -- .- -. --- __ , , , _ _ _
. _ . - . _ _ ~ . . . _ . _ . _ _ _ _
l 2. INTR 00HCTION O
2.1 Rackground The U.S. Nuclear Regulatory Concission (NRC) has fomulated an approach
- for a systematic safety examination of existing plants to detemine whether particular severe accident vulnerabilities are present and what changes are desirable to ensure that there is no undue risk to ,public health and safety.
] As part of the foundation for the individual plant examination progran, l the Industry Degraded Core Rulemaking Program (IOCOR) selected four reference j plants for detailed analysis, namely: '
Peach Botton (a BilR with a Mark I containment)
Grand Gulf (a BWR with a liark Ill containment) !
. Zion (a PWR with a large-volume containment) !
, . Sequoyah (a ouR with an ice-condenser containment) .l The IDCOR analyses perfomed for the above reference plants have been ,
documented together with the nethodology used for the analyses and the techni. l cal basis supporting the methodology. )
parallel with the 10COR work, the NRC under the Severe Accident Research Progran (SARP), perfomed risk assessments, audit calculations, sensitivity studies, and uncertainty analyses for five plants. The five plants considered ,
4 include the above four IOCOR reference plants, and, in addition !
. Surry (a PUR with a subatnospheric containment) 3 The IDCOR and SARP analyses per fomeo for the reference plants, were reviewed in order to identify dif ferences in 'the analytical detnods and to understand the reasons for the di f f erences. The experience gained from these reviews was tnen used to identify plant features and operator actions found to be important for either preventing or mitigating severe accidents in each p1 ant *ype,
. in turn, this infomation should be helpful in the ind ivid ual pl ant exaninations (IPEs).
l The first pl ant reviewed was Peach Botton,t which ic a BWR-4 witn a liark I containment. The IOCOR Peacn Bottom analysis was documented in March 1995 and was supplemented by, additional sensitivity studies in July 1985 Tne SARP 2
Peacn Botton reports
- were reviewed in draft fom during 1986 These re-ports wer1 puolished early in 1987 and were summari zed in the "Reactor Risk Reference Document" (NupEG-1150),5 which was puolished for comment in February i 1987. The experience g'ained from the review of these Peach Botton studies al ong witn ntner EUR PRA studies (namely, l.inerick, Shorehan and Browns Ferry) was used to generate the plant features, operator actions and deteministic attributes wnicq are tne subject of this report.
! 2.2 Objectives i
- Tnroe dauic goal s for this severe accident progran apply equally to all
- plant types
j . co l 1: Mi ti gate fi ssion-produc t releases .
i e i _ _ _ . . .. _
l l
. Goal 2: Control the frequency of high-consequence sequences,
. Goal 3: Reduce high Core-danage frequency.
The aim was, therefore, to develop a list of plant features , operator actions, and deterministic attrioutes tnat could be used to achieve these goals during the IPEs.
2.2.1 Plant Features ana Operator Actions In this report, those plant features and operator actions that were found to be important to either preventing or mitigating severe accidents in tne reference plant studies are provided for the utilities to use as part of each IPE, It is not the intent of this report to specify a set of inprovenents for either the reference plant or for any other plant wnich would be suf ficient to achieve a certain level of safety. In s t e ad , the report indicates votential improvenents in various areas of plant design and operation of whicn each utility should be aware wnen conducting its IPE and making decisions on plant inprovenents. Tne intent is to provide guidance to the analyst perforning an IPE, as to tne plant features and nperator actions which were found to reduce overall ri s k. It is prudent to check wnether potential improvements identi-fi ed in studies of other sin 11ar plants can be of help in improving overall pl ant perfornance. The information contained in tni s report , therefore , pro-vides a connon starting point for tne IPEs, but it should not be construed to discourage the investigation of new ways to reduce overall risk.
The three goals were noted as applying equally to all plant types. Al-thougn the goal s are independent of plant type, the plant features and oper-ator actions tnat are needed to acnieve the goals are plant decendent, in general terns , Goal 1 inplies that there should ile effective means of nitigat-ing tne fi s s i o n-p rod uc t releases for the broad classes of accident sequences unicn dominate tne core-damage frequency. Therefore, tnese doninant 3CCident sequences have to De deternined and tnose plant features and operator actions nat are available to n1tigate tre release of fission products have to De i d en t i f i ed . Then, a list of deterninistic attrinutes can be .1d en ti fi e<1 to ensure that tnese coninant accicent sequences can be nitigated.
There may ne Accident sequences for snicn a spec 1 f'c pl ant will nave su;-
stantial fi sslan-product rel eases (e.g. , containnent cypiss sequences i. Thus, for sucn secuences Goal 1 nay be di f ficult to acnteve. Therefore, til reason-aole steps are Identified wnich could reduce tne frequency of these poten-tially hign-consequence sequences (nanely, Goal 2). Again, the accident se-quences have to De id en ti f i ed and pl ant vulnerabilities and/or operatar actions tnat leao to core canage fo r tnese sequences also have to be identi-f i ed , Detailed attrinutes of each plant feature or operator action can tnen ce identi fied inicn will aid in assessing an individual plant's capability to prevent tnese sequences fecn occurring, it is al so dest ricle to ensure that tne overall core-damage f requency ts l ow ( nanel j , rioal 3), again, the dominant accident sequences nava to be found so that 1 etae1 ittrinutes af eacn plant feature or operator action can he identi tlen ta reduce the f requency at tnese sequences, if necessary, In general , the follouing sc reeni n g process was used to determine the potentiallj inportant Nnctional sequences and to deternine pnetner or not to 10 1
i
. i I
identify a particular. plant feature or operator action to prevent or nitigate '
'that sequence:
. any accident sequence with a core-damage frequency g'reater than 10-6 per reactor year
. any sequence that contributed to more than 5% of the total core-danage fre-quency ,
. any event that caused a conditional probability of early containment fail-ure greater than 0.1 any sequence that resulted in containment bypass with a frequency greater than 10 7 per reactor year .
any sequence that was judged tn be uniquely important (example, very severe consequences)
This screening process led to identifying plaiit features and operator actions that can be used in the IPE of other BWRs with fiark I containments.
For example, venting of the wetwell was identi fied as an iten that woul'd help to achieve Goal 1 (namely, to nitigate fission-product releases) for tr, BWR t'a r k I reference plant. Therefore, in the IPE of other BURS with liark I con-tainments, the need for wetwell venting may need to be carefully assessec.
The identification of a particular plant feature or operator action for the BWR I43rk I reference plant does not imply that this plant or any of the O other plants in this, category need to confom to this' feature or action, it simply neans that analyses have indicated that this particular feature or action has the potential to significantly reduce risk. Thus, a. resource is provid ed for, examining the subject pl ant to detennine whetner the same or similar plant feature or operator action will be of value in recucino overall plant risk. Whetner or not tne plant feature or operator action is useful or needed in a particular BWR with a liark I containment deoends on plant-specific detail s and is oeyond the scope of this report and is therefore not accressed nere.
2.2.2 Deteministic Attributes In this report, tne attributes wnich have been identified as important to assess the cerfomance of plant features and operatcr actions that were snown to be beneficia', in preventing or nitigating the consequences of a severe ac-cident are presented. These attricutes provide deternir.istic (as opposed to probabilistic) perfornance measures which nay ne nelpful to the IPEs. When a decisinn is nade to provide a feature or an action identi fied in thi s report ,
the utility snould address a set of questions relating to tne design, cuera- '
tion and availacility of tne needed equipn nt and the training of operators, For tne e-anole of wetwell-venting guidance, it is important to assess the capacity of t.*e venting system, the selectinh of setpoints to initiate vent-ing, the availdoility of applicable procedures and the accessibility of cer-tain valves ny operators. The deterninistic attributes related to wetwell v,enting provide helpf ul i~n f o rma t i on in assessing venting capability in eacn individual pl ant.
11
The deterministic attricutes' address the general issues of (1) .surviva-bility of equipment (i .e., unenever credit is given for a systen or a conpo-nent to nitigate the accident, the ability of tne equipment to function under envi ronmen tal and fluid dynamic loads associated with severe-accident se-quences must be taken into account), (2) equi pment capabilities, capacities ,
and duration of operaoility, (3) accessibility of equipment , (4) availability of support systems, (5) identi fication of necessary components, (6) identi fi-cation of important operator actions, and (7) identification of parameters for initiation of mitigating systems and operator 3ctions, 2,3 Organization of the Reoort This report describes detailed plant features, operator actions and their attributes for preventing and nitigating severe accidents in SWRs that have a turk I containnent, it is the first of a series of five volumes that deal with assessnent of severe accident prevention and nitigation features for sev-eral di f ferent reactor and containment types. Other volumes in the series are:
. Vol ume 2: BWRs with Mark 11 Containments
. Volune 3: BURS with fbrk III Containments
. Volune 4: PWR5 with Large-Volume Containments
, vol ume 5: PWRs with Ice-Condenser Containnents.
Appendi4 A of this volune contains a review of the IDCOR and SARP anal y-ses for a BWR 4 with a Mark I containment alnng with otner pertinent studies.
The i nsi gnt % gained from these studies lead to the identi ficati on of the strengths and vulnerabilities of a BUR with, a Mark I containment. In Section 3, tne three nasic goal s of tne program are related to the relevant design features and operating cnaracteristics of a BWR 4 with a Mark I containment.
The plant features and o'perator actions that were identi fied to acnieve tne tnree goals are therefore initially presented in Section 3 In Section 4, the plant features and operator actions are restated and d et a il ed reasures to assess tnsir ef fectiveness are developed. Append i x B nignlignts the key sys-tems an3 features of 1 SWR Mark I pl ant wnich ensure that the seve*e accident ri sk i s low.
2,4 Refererces for Section 2
- 1. "Deach Bottom Atomic Power Station-Integrat+d Containment Analyses ," InCOR Tecnnical Report T23,lPB, March 1985, l
- 2. A. H. Kol ac zkows ki et al . , " Analysi s of Core T amage Frequency f rom Intar.
Peacn Botton, Unit 2," Sandia National Laboratories , NupEG/
nal Events:
CR a550, Vol une 4, Octcoer 19A6
- 3. C. N, 4-05 at II., "Containcent Fvent Anal yst s for oostulated Severe Acci-cents: Peach Buttom Atomic Power 3tation, Unit 2," Sand i a National Lanor-atories , NUREn/ Co a 700, Vol one 3, Dra f t Report for Connent , "ay IM 7, J. C. N. Anos e' al . , "Eval ua tion of severe Accident lisks and the ro tanti d for aisk 3 eduction: Peacn Botton, Unit 2," Sand 13 National Labor a tori es ,
- WPEG/CP aS51, Vni ve 3, nraf t f n r Comnen t , April IM7, 12
i
- 5. "Reactor Ri sk' Reference Obcument," U.S. 'Nucl ear Regulatory Comission, NUREG-il50, Oraf t for Coment, February 1987.
4 e
e l
O-s
! i I
I j
1 4
4 1
l l
l i
f G
s 13 i
1 ,
'*~~- *-,w-
- - _ _ _ _ _ , , , , _ _ , . _ """"e-
--- - - - _ . . _ _ , , _ _ "'"'WWvwmwmw,,, ,_ _
5a. 4_ , ..W.A B 7 4. E 1- J " - *- - - -- - - - --'-- - - - - '----- ' - - - - - - - - - - - - - - - - - - ' - - - - - - - - ' - - - - - - - - - "
l l
I l
l l
l
~
9
)
l l
I l
O
, - 14
1
- 3. OEFINIT10ft 0F GOALS Att0 RELEVAf4T BUR t1A2K I FEATURES l In this section, the three goals are related to the relevant design fea--
tures and operating characteristics of a BWR with a flark' I containment for the accident sequences and containment failure- modes that have been found to be ;
important. Both 'avorable and unfavorable severe accident attributes are !
discussed, The important accident sequences and containment failure nodes are presented in Appendix A.
Screening criteria have been used to identify those sequences that need to be addressed to deternine plant features and operator actions to accomplish ;
each goal . Specifically: '
For Goal 1 (liitigate fission-product rel eases) , all sequences have been exanir;ed which represent 5f. of the core-damage frequency or are estinated to hava a frequency higher than 10"6 per reactor year and which result in a 1 3 conditional probability of early containment failure greater than 0.1. l For Goal 2 (Control the frequency of high-consequence sequences), all se-quences have been examined which result in pool bypass and are estinated to !
have a frequency higher than 10-7 per reactor year.
Fo r Goal 3 (Reduce high core-damage frequency sequences), all sequences have been examined wnich "have the potential to occur" more frecuently than 10 - 6 per reactor year. Note that this screening criterion has been used to identify potential vulnerabilities from risk assessnent insights which do not necessarily apply to Peach Bottom itsel f, but may apply to other Mark I pl ant s .
This section provides the link between the goals (developed in Section 2) anc the pl ant features and operator actions (developed in Section 1) tnat may be used to assess the capability of specific plants to neet these goals. This sectiun i s organized into three subsections, which correspond to the three go al s .
1 3.1 Miticate Fission-Product Releases Goal 1 requires that there shall be ef fective neans of mitigating the i fissior.-product releases for the accident sequences that may lead to core dan. I age in a BWR with a !! ark I containment, in Appendix A, the most important contricutors to tne core-damage frequency (C0F) were found to be station blackout (530) and anticipated transients without scran ( AT'.lS) sequences.
Several stucies incicate that transients witn loss of injection into the reac-tor pressure vessel (RPV) are also potentially important contributors. Other 4
transients and l os s- o f-cool an t accidents (LOCAs) may also r.ontribute to the 00F. Two speci fic accident sequences for whicn mitigation by the fiarx I con-tainnent is iref fective are al so identi fied in Appendi x A. These speci fic se-quences (the interf acing systens LOCA and the ATUS with a power transient) are t
discussed further in Section 3.2, which attempts to deternine how the fre-quency of thete unmitigated sequences can be cuntrolled. This section concen-trates nn the troad classes of accident sequences for which existing plant i featuras proviue significant means of mitigating fis5 on-product release, in !
4
- ne following sactiuns not;n the favorable and un f avoracl e severe accident I attributes of *.he tiark I containment are identi fied. This discussion in turn l 15
4 leads to the identi fication in Section 3.1.3 of the plant features and cper.
ator actions that are related to Goal 1.
3.1.1 Plant Vulnerabilities The small volune of the Mark I Containment makes it vulnerable to pres-sure and/or temperature increases because of the nonconcensable gases ano neat released during a core-neltdown accident. There are differences cetween ene 10COR I and SARP2 analyses regarding the estinates of how long it will take to pressurize a Mark I containment to its ultimate capacity after the core deoris has failed the reactor vessel (and is interacting with concrete); but both studies concluded that the containnent may eventually ~ fail. Unless nitigative actions are taken, a fiark I containment can fail because of overpressure or overtemperature within a few hours of RPV f ail ure, If contairnent f ail ure occurs in the drywell, any fission products in the drywell atnospnere coula pass to the reactor building (and ultimately to the environment) witnout the benefit of suppression pool scrucoing.
An inspection of the Mark I containment configuration ( see Fi gure A.1) shows tnat tne pedestal below the RPV would tend to confine the core deoris a f ter a core-nel tdown accident. Extensive core-concrete interactions would be expected to occur. There 3re di f ferences between the [0COR and SARP analyses rel ated to now nigh the Core deoris tenperature ulll remain during these in-teractions and to the quantities of the less volatile fission product s that will be rel eased. However, at this time the possibility of the core debris remaining hot and releasing significant quantities of fission prooucts has not teen roleo out, Aftar the sunps directly underneath the RPV (pedestal region) fil l ed ui '.n core deorts, tnere would be suf ficient core naterials from a full core-neltdown to over fl ow onto tne drywell fl oor. If tne core debris remainea mol ten it could flow across tne drywell floor and reach the steel contairnent liner. Tnis steel liner would of fer little resistance to noiten cnce denris and it wa s pr ed i c t ed 3 to fail very rapidly if the core deoris reacneo tne l iner wall . 'Tnis was found to ne a mecnanisn for early loss of dryuell i n t e';-
ritj in the SA W analysis arc is thus anotner Park [ cont 31nnent vulnera:111ty rel a*i ve to sene ,taer vtal nnent desi gns i n uni:n tne geonetry ucu'4 tend .a prevent
- he Core deDr13 ' r o.1 reaCning the contalnnent wall. The $A4P analysis estinated a rel ativelj nign cona1tional pronability for liner f ailure for $30 sequences, in the sections that follow, suppression pool scruobing is noted as an ef fective nit 1 gative feature for tne fiark I containment pro v i d ed all of.tne fission products pass tnrougn the pool, it is, tnerefore, important to ensure that paths do not cuen unicn would al l ow the fi s si cn products to byoass the soppression 2001 The vacuun breakers between the wetuell anel d ry wel l woulq
- reate i patn tnat Dyoasses tne suppression pool, i f they f ail oprn, in aaq i-tion, tne vseious drywell genetrnlon seal s could De degraded at n1 9n tempera-tures and pressures. Fi11ure of tnese seal s would al so open up catns t,3t would Bypass tne s w ression pool, if the nain steam i sol ation v ilves (ns t ys; fall to close, anotner suooression pool nypass patn would .ni s t .
O 16 -
Al though' the Mark I containments aopear to be vulnerable to severe-O
(/
accident loads, they have seseral very important nitigative features, wnicn are described in the section that follows.
3.1.2 liitigating Features The suppression pool in a Mark I containment is a very effective nechan-ism for trapping any fi s sion-prod uct aerosols that might pass througn it.
Thus, to a large extent the ' suppression pool has the potential to compensate
- for the vulnerabilities identified above (in Section 3.1.1). For example, overpressure failure of the containment (and perhaps loss of drywell integ-rity) can be prevented by venting the wetwell. With controlled venting of the wetwell atmosphere, containment integrity is lost but the containnent function (retention of nost of the fission products in the pool) can be maintained.
High drywell temperatures and resul tant penetration seal degradation can be prevented by drywell spray. The potential for molten core debris to spread across the drywell floor and fail the containment liner may also be reduced by spray operation (refer to Section A.2.4). Drywell spray will also contribute to decontamination of the drywell atmosphere even for sequences with suestan-tial suppression pool bypass.
The atmosphere in a liar < I prinary containment is continuously inerted 4
(by introducing nitrogen and thereby lowering the oxygen concentration) during operation, wnich prevents hydrogen combustion. This is a very significant l nitigative feature, which is important to maintain during a severe accident. '
For example, wetwell venting and d rywell spray operation could resul t in a O vacuun in the containment, wnich could introduce addi tional oxygen and thus deinert the containment atnosphere.
An aren of si gni fic ant phenomenological uncertainty (refer to Section A.2) rel ates to core nef tdown with the reactor coolant systen at high pres-sure. If noiten core materials are ejected fron the RPV under pressure, .t i has Deen suggested in the SARP analysis that the materials form fine aerosols, whicn could ce dispersed into the containment atmosphere and directly heat it. This Could resul; in a large pressure pulse, wnich could tnre3 ten con.
tainnent integrity at tne tine of RPV failure. BURS have an auton3 tic <tepres-surization system ( ADS) wnicn can prevent high-pressure core-meltcown (on RPV low water level af ter a time delay or manually by tne operator). The ADS has a dual role as a care-mel t prevention system as well . For those sequences in wnich tne nign-pressure injection systems f ail (TOU), the ADS can depressurize tne reactor vessel ano allow the low-pressure systens to inject water into tne RPV.
Finally, the BWR tiark I primary containment is completely enclosed in a reactor Dui!ning. This building is, therefore, available as a seconcary con.
tain:ent to trao any fission products that night be released from tre prinary
- containnent during a severe accident. The anount of fission products tnat l night be trapped in the reactor building is uncertain, but it is a potentially !
1 inportant nitigating feature. '
- i i
s 17 I
3.1.3 Containment Integrity via Watwell Venting and sucorossion Pool :ffec-t1veness The above discussion has identi fied several pl ant features of tne BWR plant with a Mark I containment that have the potential to help 3chieve Goal 1, namely , mi tigati ng fi ssion-product releases. From the above discussion, plant features and operator actions that will aid in assessing whether soe.
cific plants meet Goal I have teen identi fied. The plant features and o p e r-ator actions address containment integrity, the effectiveness of the suppres-sion pool , and the various nechanisms for possible pool bypass. As long as the dominant release path is through the suppression pool, the Consequences of core melt accidents were shown (refer to Table A.12) to be reduced by at least a factor of 10 relative to sequences that bypassed the pool. One of the doni-nant suppression pool bypass mecnanisms is al so addressed ( n anel y , nel t.
througn of the steel containtnent) to ensure that tne release is subst3f.tially reduced or that tne frequency of such a pool bypass mechanism is kept low.
Both the ASEP and IDCOR studies id enti fi ed SB0 and ATUS sequences as being the nost si gni ficant contributors to core nel:. Al though the studies disagree on wnicn i s -lore they are in general agreenent on the total care-relt probability (aboutimportant, 10- s/ reactor year) Thus , i t is Selieved tnat tne ef fectiveness of the suppression pool must be maintained for bovi ATUS and 530 events.
Fo r sequences that threaten the containment by overpressure, wetwell venting has the potential to preserve the containnent function by relieving noncondensaole gases and/or saturated steam, thus preventing further pressure buildup while forcing fission products to be scrubbed by tne pool . Wouever, &
for tne two dominant sequences (SB0 and ATUS), existing venting aracedures W will be di f ficul t to aerform. For 5B0 sequences, power depenaencies nay pre-cl ude actuation of venting f rom the control roon, and hi gh radiation l evel i may hanper local manual valve actuation at the time when it i s n eed ed . For ATWS sequences, the large venting capacity requi rements , the small tina wi nd ow for operator action 3na possinle proniens with 1 sol ation systens nake success-ful venting un'1er sucn conditions operationally oif ficul t, retail eo neasares are identi fied in Sec; ion a.1.1 to ensure venting capacility for these dent-nant accident sequen;es, tous mintaining supprassion pool effectiveness.
The SaRRp analyses for Peach ?otton ind ic a t ed that core debris nel ting through the steel containment snell was a dominant suppression pool nypass necnanism. SA9RP estinated tnat the condi tional prnbability of containment snell mel t-througn was relativel y hign for SRO seluences. Although it could be argued that, even uith containment shell ,el t-tnrough , i f the d rywoll pres-sure is relieved Dy netsell venting, tne ariving torce to transport tne radio-nucl ide from ine dryuell to tne raactor building will t e r ed uc ed , the fric-tional f i s s i a n- p re1uc t release rWins uncertain. Therefore it appe3rs pru-dent to attemet to keeu th9 cantainment shril fron neiting. It nas hee, su u-gested that i f ceranic bric< curos are buil t to keep.the corium fron reaching the containment snell, containment mel t-tnrnogn could be precluJed, unwever, compl ete No ri s con fi nement would be di f fical t without impacting water resurn to the pool. !f notn overpressure f all ure an t cont 11nnent nelt-thr7ugn can Ne preven;+1 (5/ /enting and 3y Dullding coriun retainers) the 11 ul i noud af goal Dyuass c sn he r e uc +1 s;3stantially (Sy ahout i f ac tor 11 using sARP event tr0est.
13
1 i
I For sequences that still result in suppression pool bypass, the drywell sprays will tend to wash out aerosols from the containnent atmospnere and thus reduce the airborne fission product concentration during core-concrete inter-action. In some sequences such as 580, drywell and wetwell sprays would not !
be available because of ac power requirements. For the other dominant se.
quence class ( ATWS), these sprays may not be av'ailable because of suppression pool heatup and its effects on net positive suction head (NpSH) of tne spray p umps. The plant features, operator actions and their attributes identified in Section 4.1.1 address alternative power supplie: and suction sources to ensure that drywell sprays will be available for the two dominant sequences.
3.2 Control the Frecuency of High-Consecuence Seouences The plant features identified in Section 3.1 can effectively mitigate fission-product releases for the broad classes of accident sequences tnat were i i
found to dominate the core-damage frequency (C0F). However, two accident l sequences were identified in Appendix A for which substantial reducing of fission-product release for the BUR liark I plant cannot be ensured. Nei ther of these two sequences appear ' to meet the screening criteria (>10-7/ reactor year) for Peacn Botton, but their importance in other PRAs indicate that the {
l individual plant examination should ensure that specific plant vulnerabilities l
do not make the9 contributors to risk. l l
The first accident sequence that may defeat the plant containment fea- l tures identi fied in Sec ti.o n 3.1 is the interfacing systems 1.0CA sequence (event V). Although none c f the SWR Mark I PRAs reviewed in Appendix A indi- i cate that it is a significant contributor to core-nel t frequency, DQAs for other plant types (e.g., Shoreham) h. ave identi fied it as a significant con-O tributor to risk. It nas also been identified as a generic issue (91-105) ny the ?!RC. Thus , i t is important for other BUR liark I plants to assess the 1 i
potential contribution of interfacing systems LOCA to risk. The plant fea- i j tures and operator actions identi fied in Section 4.2.1 to keep the frequency
- of event V low should he considered appropriate pending reso'lution of GI-10,5 The second accident sequence that may defeat several of the plant fea-tures icenti fied in Section 3.1 is an ATUS with a power transient. n tni s sequence, tne operator fails to control the RPV injection at low pressure dur-ing an ATUS event. The rising water level in the reactor vessel produces a power transient that cannot be controlled by the normal containment heat re-moval systems. The containment will pressurize rapidly and may f ail with the resultant lots of coolant injection and eventual core nel t into a f ailed con-l tainment. The ability to contr)) this rapidly progressing sequence by wetwell l
venting is di f ficult, and thus mitigating this sequence appears to be un.
l i kel y. Therefore, tne risk of these ATUS sequences nust ne centrolled by en.
suring tnat its frequency is low. In Section 4.2.2, operator actions unich may be inportant during an ATUS event are identified.
3.3 Recute High Core-Damage Frequency Sequences In Section A.1 i t was found that only a f ew . acc id en t sequences figure proninently in the core damage profiles at all of the pD As raviewed. Tnts led to tne conclusion that if the frequency of t9is relatively small subset of at-cident sequences could be raduced, then the overall CDF could also he reciuced.
- 19. .
3.3.1 Station Blackout via Alternate Reactor pressure vessel Injection Most of the pRAs for BWRs (including the ASEP study of Peach Botton) in-dicated that the nost important contributor to the C0F is s;ation blackout (580). Therefore, plant features, operator actions and their attributes have been identified in Section 4 related to these accident sequences.
Station blackout refers to a loss of the of fsite power supply with con-current failure of the onsite emergency ac power divisions. Reducing the f re-quency of SB0 sequences is addressed by the proposed NRC SB0 rule. The infor-mation presented in this report emphasizes the need to search for plant spe-cific features and potential common cause f ailures which could disable systens required to work during an 580. For individual plants which are found to have a vulnerability to SB0, this report highlights the importance of proper emer-gency procedures and operator training in recovering from an SB0 event.
For the BWR 4 design, the two systens designed to operate in the presence of an SB0 are the high reessure coolant injection (HPCI) and the reactor core isolation cooling (RCIC) systens. By removing the long-term SB0 sequences related to dependent failure nodes of either system, the 5B0 core-damage f re-quency can be signi ficantly reduced. The long-tern dependent failure modes of HPCI and RCIC under 580 conditions are (1) battery depletion, (2) pump seal failure because of suppression pool heat-up, (3) loss of room cooling, and (4) high turbine exhaust pressure. Since the RSS did not address these f ailure modes, it is espected that their removal from the current design would lower the C0F because of SB0 back to essentially the RSS point estimate of 10 7 .
3.3.2 Loss of Containment Heat Renoval by tising Alternate Cooling Accident sequences involving loss of containment heat renoval (CHR)
(e.g., TW) were found to be inportant in the earlier PRA studies considered in Appendix A (see Table A.1). In the RSS.* the TW sequences accounted for 53; of the calculited CDF. In the Browns Ferry IREP5 study, the TW sequences sin-tiarly accounted for 507 of the calculated CDF. The nosi recent Peach Botton studies (IDCOR and SARP) show a two -a nd -t h ree -o r d e r -o f -ma gn i t ud e reduc *1cn, respectively, in W smences quanti fication. Therefore, the mechant sns al-ready ef f ect f uly employed at Peach Bottom to control the frequency of TW se.
quences (and other relateo loss of CHR sequences) have been identifled in Sec-tion 4 There are a number of f actors associated with this reduction in the fre-quency of W sequences. When the RSS study was performed, it was assumed that overheating of the suppression pool failed emergency core cooling (ECC) iy ec.
tion and the contairment failed with a conditional probability of unity. ECC injection f ailure care abnut either by f ailure to nalatain NPSH conditions for the ECC pumps because of the heated pool conditions or, surviving that. loss of their suction source by some overpressure failure of the cont ainnent it.
sel'. Since that early study, investigations into ECC punp survivab11 tty havo denonstrated tnat tne ounps have a substantial likel1 hood of successful opera-tion given a heated suppression pool. In additton, the containnent fa11gre concern was assured to be nitigated by' containment (wetaell) venting proce-dures at Poich Botten frefer to Table 4.1). Alternate sources of injection capability ha M also been established at Peach Botton to preclude reliance upon an oerheated suppression pool and are identifled in Table 4.6.
20
4 3.3.3 Reactor Pressure Vessel Oeoressurization Performance O
V As nentioned in Section 3.1.2, the ADS .is an important systen in nit'igat-ing loss of nign-pressure inj ection sequences. Al tnough neither SARP nor 10COR indicate that these are dominant sequences for Peach Botton, other stud- "
ies (Limerick and Shoreham) indicate that failure to nanually depressurize can be an important contributor to core melt (about 6x10-s/ reactor year). Pl ant features, operator actions and their attributes are identified ig Tacle 4.7 to i
ensure that other fiark I plants have a low CDF due to high-pressure injection failures. Mditionally, the ability to depressurize the reactor pressure
. vessel (RPV) is an important mitigative feature that hel ps maintain suppression pool ef fectiveness, 3.3.4 Suocort Systen Interdependencies t fiost PRAs have stressed the importance of interdependenc'ies having tne l potential to comprornise the perfomance of many critical safety systems. In many cases, risk assessment studies have identified such vulnerabilities very early in the study and "fixes" have been made which substantially reduced ri s k . Al though no such dependency-caused vulnerability has been id enti fi ed for Peach Rotton, it may be advisab.le to search for their existence in the IDE because such interdependencies are design-spect fic and may be important con-tributors to CCF, 3.3.5 Reactor Buildino Flooding One of the accident sequences , whose potential for contributing to the core damage frequency was specifically evaluated in the Shoreham fluclear Power '
Station (SNPS) PRA,6 is the release of excessive water into the reactor build- ,
ing. Botn the SNPS P9A and the Brookhaven National Laboratory (BNL) revieE i
l such an initiator I
of the SMPS contribute ORA revealed suostantially to tne CDF tnat accident (3.9x10-* and 2.0sequer1ces.
x10- induced hg/ react spect ivel y) .
To nelp ensure tnat ?tark I plants which may have a similar safety-related equi pnent flooding potential can be identi fied , the deterministic dttributes 11sted in Sect:an a.3 provide an aid unicn may be used to screen for sucn vul- l
, nerabilities curing tne IPE.
3.4 Refererces for Section 3
- 1. "Peach Botton Atonic Power Station-Integrated Containnent Analyses ," ICCOR l Technical 9eport T23.1PR, furch 1945, l
- 2. C. N. Anos et al ., "Containment Event Analysis fnr 00stul ited Severe Acci.
dents: Peaca ;otton Atonic Power Station, unit 2," Sandia National Labor- l atories , ','tFE9/CR-4700, Volone 3, Gra f t Peport for Connent , t'ay 1H 7
- 3. G. A. Greene, X. 4 Perkins anr1 S. A. Hodge "!npact of Core-Carc rete in-teractions in the I'a r y 1 Containment Crywell on Containment Integrity a nr1 Fsilure of the Drywell Liner." Paper IAEA-Sit.291/36, presented at the International Synposiun on Source Tern Evaluation for Accident Cunditions, Col umbus , r.h io , Nove"'ner 1 HS, l
O 21 l
4 "Reactor Sa'fety Study: An Aste%<nent of Accident Ri sks in U.S. Co: narcial Nuclear Power Pl ants ," U.S. Nucl e ar Regulatory Comi s sion , WAS4-1400, NUREG-75/014. October 1975.
- 5. S. E. Mays et al . , "Interin Peliability Eval uation Program: Analysis of the Browns Ferry, Unic 1, Nuclear Plant ," Idaho National Engineering Lao-oratory, NUREG/CR-2802, July 1982.
6 "Probabilistic Risk Assessnent - Shoreham Nu c l e,a r Povter Station," Long Island Lignting Company, June 1983.
7 O. Ilberg et al ., "A Review of the Shoreham Nuclear Power Station Proba-bilistic Risk As se s smen t ," Brocknavan National Laboratory, NUREG/CR 4050, June 1985.
9 e
22
I I
4 pl. ANT FEATURES, OPERATOR ACit0NS AND THEIR ATTR tRUTES t 1
g in Section 3, those accident sequences that doninate the core-damage 're- .
- quency (CDF) were identified as were those that are potentially of nign conse- l quence. Vulnerabilities of the itark I containment to severe-accident contain-nent loads were discussed and tnose features of a BUR with a tiark I contain-nent, which are important for . preventing core damage and are available for nitigating fission-product release to the environnent were identified. l Based on "insights" fron previous pRA studies and other severe accident research "plant features and operator actions which are considered important :
to ensuring acceptable risk for the reference plant"I have been iden ti fi ed , i Eacn of these features and/or actions are enunerated and related to tne severe accident progran goals discussed in Section 3. Along with tnese generalized features and actions, a detailed list of deteministic attributes nas been i developed and operating in order to providewnich enaracteristic "plant-specific are to be guidance exanined onby the the design fejtures utilities."
{
Nine pl ant features and operator actions were identi fied (Taol e 1.1)
- wnich reflect the potential importance of these features to plant risk. As discussed in Section 2.2.1 the infomation presented in this report provides a list of areas of potential inprovements for various areas of plant design and operation of unich utilities snould be aware wnen conducting assessnents, it is f urther noted t r.a t a nuncer of issues appear to overlap various generic issues as defined by the NRC. Final resolution and disposition of tnese generic issues may enconoass NRC-inposed requirements. However, the infoma-tion presented herein is intended only as guidance which may be nelpful in perfoming an individual plant exanination (IPE).
Tables 4.1 and 4.2 fission-product releasesaddress (Goal 1)nessures to ensure witn reference tne capability to naintaining to nitigate containnent integrity and naintaining suppression pool ef fectiveness.
i Taoles 4.3 and 4.a address measures for controlling the frequency of l niyn.consedence sequences (Goal 2) with reference to minimizing interfacing l systems 1.0C A freluency and nitigating anticipaten transients witnout scran l l a*.iS) sequences .,
Finally, Tanies 4.5 tnrough 4.9 address neasures for reducing hign cora.
danage frequency sequences (Coal 3) with reference to nitigating station bl ac kout (580) sequences, nitigating loss of containment neat renoval se-l quences, enhancing reactor pressure vessel (RPV) depressurization perfomance, examining ficoding.
support systen interdependencies, and nitigating reactor building Tne remainder of tnis settinn is organi:ed into three sunsections corre.
sonndirg to the t1ree Dasic goals. In each subsection, tne corre sponding plert features 3rd nDer3 tor dCtions are discussed f rni t whiCn d e t e 7ti n t s t ic attribut es are develope <l. Inese ittr1butes address pl ant features ano oper-atoa actions, urder severe 3Ccident Condi tions , wi th regard to the general issues of (11 survivattlity nf equipment. (i .e., wnenever credit i s q19 n f or a l i
sjstem or 3 component to nitigate the accident, tne an111ty of the equipment to t',nctiqn under the
.*ny t ronnen td l Condit1nns ano fluid dyndi'11c l oad s asso.
clar ed wi tn severe accident sequences should be taken into account),
23 I
l
(2) equipment capabilities, capacities, and duratinn of oper3eility, (1) accessibility of equipnent, (4) availaoility of support systens, ($) identifi- .
cation of necessary components, (6) identi ficati on of inportant operator actions , and (7) identification of parameters for initiation of nitigating Syste"1s and operator actions, 4,1 liessures to tiiticate Fission-Product oe leases For a B' R that nas a tiark I containment , the dominant core-canage se-quences were found to be station blackout (SBO) a nii anticipated transients wi t hout sc r am ( AT',lS ) , in order to mininize of fsite consequences for the se-quences, the containnent systens (both primary and secondary) should be able to retain a substantial fraction of fission products released aven under these severe accident conditions, a,1,1 tiaintaining Containment Integrity via Wetwell Ventina and Suocression Pool Effectiveness As aiscussed in Section 3, the nost im po r t ?,n t systems for ni t i gat i nij hign. consequence sequences are tne contain ient and its suppression pool, In 3ddi ti on to condensing the stean generated in an aC C i d en t , the suppression pool also acts to remove fission products f rom the containnent atnosphere, As 1ong as any release path is forced through the pool (e.g , during wetwelI ventin,;, the pool will act to reduce the environmental release fractions by a factor of 10 or more, Thus , the ni ti gati ve features deal with naintaining containnent integrity and ensuring the effectiveness of the pool as a fission-product niti gati on Sjsten.
Tables a,1 and i,2 gravide a list of deterministic attributes rel ated ta these general plant f eatures and operator actions wnich nay be used to evalu-ate each plant's capability to avoid breach of tne containnent and suppression pnol by? ass and possible suopression pool bypass neChanisnt that were identi-fled in 4ction 3, a.2 "P3surt's to Cuntrnl the FreduenCV Of High-Consequence Seouences lr $eC'100 . ' ,1, C l in ! #ejturas ind operjtcr actions wre toe nt'fj N tnt sneula hel a ens ure con tai n.,ent Inte.jrtty and nitigate tission product raleases flr t ne D r0 ad Clisses Of 3le id en t sequenc s tnat wer tound in Aepena,x a *. ,
be inportant ta the core-danage frequency, Ho wev er , two accident sequences were identifle1 in Aepoqi114 A for which the RWR tia rt [ containment has linitM neans of nitigating f i s s i o n-p rod uc t releases; n a.1e l y an i n t e r f .ic i n g 3 y s t e.n s LOCA and in AINS wita a power transient, in t*11s section, oper) tor ic
- 1 on s for controlling tne f requency of occurrence of inese po t en t i a l l y ni p-consequence se.;uances are 1denti fiel, a 2.1 drevent190 i m1 l'1 * ) jating Inter'icin g Svsteas ( OC A in gen e r a l , N "a r < l NAs nava found intortacing systris LDC3 I wont t"
- o 5e 3 ni f f unit te! f eve, (less tnan 10 7/re xtor year). H0,ever Enre, jee bne %R s (a,9,, S nu re'un ) for w h i t. h event 'l is rist signific1nt Secausa u the pten t1311 j ni ;n rei sases, Th y enjective of the intarma* tan pre s on t e nerdin is t0 "USJre t"at tha t rdQut"C y o f event '/ I S bept it 3n IC C e 91 D } y I ' M leve}, 3rQt)* naVen '. ) t 10 0 )I LJD0ratury ( Bill) 1 s presentl y per f ar"il ng 3. st f ad y v
. l 9
>o provide technical support to the f!RC for the resolutior of the generic issue related to the interf acing systems LOCA (GI-105). Therefore, tne infor-nation presented in Table 4.3 should be considered appropriate pending resolu-tion of the generic issue. .
Table 4.3 lists the nost important attributes of interfacing systens LOCA in a BWR which may prove helpful in controlling the frequency of such events.
4.2.2 Preventing and tiitigating Anticioated Transients Without Scran The important attributes of the anticipated transients without sc ran (ATWS) sequence with respect to' operator actions were found 3 to be tne likeli-hood of misleading instrumentation, the need to inhibit automatic safety sys-tems, the use of required mitigating actions which conflict with operator response to other accident conditions, and the need for coordinated actions and communications ainong control room staf f nenbers under hi ghl y stress f ul cond itio ns ,
i The infornation presented in Table 4.a is based upon the assunction that I each of the plants is (or will be) in compliance .with the NRC rule on "Reduc- l tion of Risk from Anticipated Transients Withuut t ran for Li ght-Matar-Cool d Nuclear Power Pl ants ."
4.3 tieasures to Reduce High Core-Damage Frecuency Seouences j
The najor contributprs to the core-damage frequency (CDF) are presented in Appendix A. The IDCOR and ASEP/SMP analyses indicate that the station bl ackout (530) and anticipated transients without scran (ATUS) sequences are i the dominant contributors to the C") F . The results of otner pRAs and pRA l reviews indicate that in addition to those two types of sequences other se- l wiences, nanely, loss of containment heat re'noval (CHR) sequences (TU, SI, and t Tar)! sequences) and sequences with failure to depressuri:e the reactor pres-s are vessal (DPV) for injection with 1ow-pressure e.ystems (TOUX sequences),
can also be major contributors to tne CCF.
4.3.1 Miticiat'nu Station Blackout via Al ternate Reictor pressare nssel in;ec un in nost DRAs for light-waters eactors (LWRs). station bl ackout (SR0) sequences have been major or prominent contributors to the CDF. As part of the effort to resnive tna un resol v ed safety issue ('JS! A-4A), the NRC is pro- l posing to anend its regulations "to provide further assurance that an 530 I (loss of notn of f site power and onsite energency ac power systee.si will not adversely af f ect the puolic health and safety."s For accident sequences , de-i vel oped by an ind1eidual pl an t exanination (IPEl, which invol ve the l oss-of- ;
offsite power and onsi te energency power , the propose (1 SB0 rule snould he ;
e nnined for appl ic ab il i ty. The deterministic attributes related to W are i n tend ed to epnasi ze tne new to search for plant-speci fic features and po-tantial connnn cause f ail ures which could disable systems requirH o work during an 530 For individual plants unich are found to have a vulneraDilitj to SAO, Tanle t.5 nt ynl ignt s tne inportance of prope e,ertjency proceduras and oper-ator traini g in rncovering f ran an 550 event.
25
4.3.2 Preventing loss of Containment Heat Removal by Using Alternate Cooling k
For some of the PRAs and the PRA reviews used in this study, sequences with sucesssful coolant injection but with subsequent loss of containment heat removal (CHR) (TW, 51, and TP01 sequences in Tables A.1 and A.9) can be impor-tant contributors to the CDF; in those PRAs it is assumed that containmert failure causes loss of emergency core cooling (ECC) injection. As discussed in Section 3.3.3, in the PRAs where those sequences are not important, tne main factor for the low contribution to CDF is because of credit given for wetwell venting and alternative sources of injection. Therefore, it appears to be important to have alternative injection sources available in addition to wetwell venting to ' provide adequate CHR during accident sequences with ini-tially successful ECC injection but with subsequent loss of CHR. Previous studies have shown that ECC injection is not lost in DJ sequences for 20 or more hours at wnich time the decay heat is about 0.5%.
Table 4.6 provides determi ni stic attributes for successful decay heat removal.
4.3.3 Ensuring Reactor Pressure vessel Deoressurization Performance In some of ths PRAs examined in this study, sequences with failure to de-pressurize the reactor pressure vessel (RPV) af ter failure of tne high-pres-sure injection systems (TOUX sequences) are important contributors to C0F. In all these PRAs, the automatic actuation of the automatic depressurization sys.
tem { ADS) only occurs on coincident signals of "high" drywell pressure and "low" reactor' vessel water level held for a time delay of 2 minutes. For a large number of transients with loss of high-pressure injection, these coinci-dent signals will not occur. Therefore, the contribution of these sequences to CDF is dependent Jpon the intervention by the operator to manually depres-surize the reactor. In Table A.6, 3it can be seen that these probabilities va ry f rom 1.Sx10 "/ demand to 6.0x10- /cemand. In Peach Bottom, the CDF was reduced by changing the logic of the ADS auto-actuation logic to eliminate the need for the "high' drywell pressure and changing the water level time-initia.
tion setpoint to 'i ow l ow" and the initiation time delay to 8 minutes.
Table 4.7 n19,11gnts those features that can ennance the performance of the aOS.
4.3.4 Identifying Supoort System Interdependencies One of the bene 'its of performing a detailed PRA is that the system in-terdependencies are modeled and are reflected in the results. However, not all PRA studies have performed rigorous interdependence analyses and, there-fore have not ferreted out all of tne possible subtle interdepenaencies. This may have prcfound ef f ects upon their results. An interdependency is aefined as tne f ail ure of one system leading directly or indirectly to the f ailure of another systen.
An in-deptn 3policat:On of basic PDA methodology aith re".pect ta interce-pendenCles y'elded significant findings on a previously heavily stuc1ed PhR.
To illustrate Inis point, reference is nace to tne ent study of syste, inter-actions (supsort syste9 Interdependencies) at Indian Point Unit 3.' The major finding of that study was that a spec 1fic single stati0n emergency battery lh 26
l l
could fail and, among otner things, negate the entire low-pressure injection
[)
(_s' function. The point to be emphani:ed nere is tnat none of the nunerous other studies and reviews of the Indian Point 3 design were able to detect this in-portant single failure nor did the BNL study until all tne support systens were explicitly modeled, linked together (the fault tree linking. approach 7) and solved using the SETS computer code.s NUREG-11509 has provided a thorough application of the latest PRA nethodJ to five reference plants-and the results point out nunerous insights into tne importance of specific design differences among the studied plants. Ho wev er ,
the NUREG-1150 authors empnasize the importance of support systen dif ferences and the difficulty of extrapolating the result from one plant to another, it is not suf ficient to make a single overall dependency table of t9e f ront-line and support systems for a given plant and simply compare that to the reference plant. No two plants will have the L e set of systen interde-pendencies. Support systems vary widely from p1 ant to pl ant even though the plants may be of a similar class and have the same set uf front-line systems.
Based upon the dominance of the SB0 sequences for Peacn Botton as well as for other BWR designs, it is important that as part of the IPE detailed inter-dependency tables be constructed for this sequence with all dependencies con-di tioned upon the existence of an SRO for various lengths c' time, sucn tables would also explicitly identify all of the expected failure necnanisns (e.g., identify whetner battery f ailure occurs because of loss of room cooling or charge depletion).
T Table 4.8 provides a list or attributes related to support system inter -
M dependencies.
4.3.5 Preventing and flitigating Reictor Ruildino Flooding Al tnungn medlun or snall leakages can be adequately niti gate 1 by t*e existing sunos or punpoack systems, large water leacages are of prinary con- I cern in reactor nuilding (3B) fl ood i n g. Potenti al water sources for excessive water releste into the lowest RB level include tne suppress 1on pool, ne cen, densate stor ye tank , tne reac tor cool ant systen, ne service water sy s ,em aM tne fire protection system stor3ge tank. Some of the major equipnent l oc a t e1 in the lowest CR level compartnent may include emergency core cooling (ECCi pumps and tneir electr1 cal control panels for the hign-pressure coolant injec tion (HPCI), reactor core i sol ati on cooling (RCIC), core spray and l o w-pressure coolant injection (LPCI) systems. l 1
Reactnr Duilding fl ooding can Se i n i ti a t ed by (1) a najor maintenance whicn raquires exposing 3 safety systP9 to the Rb atmospnere, ana (2) breus in the pressart :ed or tne non-cressuri zed part of piping or components. In 1:en 1, "majar maintenance" refers tn those actions which would requi re di s- ,
nantling uf system components tnus elininating a barr1.r between larga soortes ;
of utter nnd the D. W flooding can partij he preven tat an,1/ or a1 ti ga'ed througn proper training ano procedures. For example, nnce the Ra i s fl oo<1ed ,
tne operator ihould be abl e to follow the ins' ructions for responding to tne 31 d en to ident)ty the source of the fl 00d and i sol a ta 1; before the water
- l e <el n tne lowest conpartnent reacnes a critical level, The operator saould Ov sl ia enos aoout al territive devices or equipnent which can be utill
- ec to 27 i
i
- ~ .
provide coolant injection to tne RPV in case of emergency core cooling systems (ECCS) equipment f ailures in the flooded compartment.
The BNL study M of the Shoreham Nucl e ar Power Station (SNPS) revealed
- that although the SNPS Alarm Response Procedures give general guidelines for monitoring system parameters to detemine the leakage location and initiate the leakage isolation, spacific requirements for operators to systematically check the operation parameters of relevant systems are not included. BNL also identified that the random failure of an equipment protection electric circuit breaker coinciding with RB flooding may result in the propagation of f ailures to the upstream motor control center (MCC), other MCCs , and the associated load centers, It is important tnat this and/or similar potential common-mode failures be avoided.
Al t nough this type of 91nerability to flooding, was identified for a Park
!! plant, it is believe- the concerns are of general applicability to other designs.
Table 4.9 highlight ,se plant features and operator actinns that can prevent or mitigate reactcc building flooding, ,
4.4 Usino the Report Numerous investigations , including PP As, have s- perfomed for the ref-erence plants and similar plants by both the NRC p , t"> nuclear power indus-try. The insignts gained from many of the studies hc ;een used in d ev el o p-ing the plant features, operator actions and their attributes contained in this report. These insights are issued to provide- guidance t o ,,t h e analyst performing an IPE. This guidance is in the form of plant features, operater actions and deterministic attributes for assessing t90se features and actions found to be helpf ul in reducing the overall risk for Peach Botton and other Mark I p.3nts. Thus, the guidance is given to provide a resource in exantning the sucject n! ant to determine i f the same, or similar, plant features ano operator actions will be of va:ue in reducing overall plant risk. The in or-mation contaired in this report is intended to be used solely as guidance, but it may inclute / s a s e set) sor e requirements generated by tre NRC on generi:
3 issues.
4.5 :eferences for Section :
- 1. SECY-86-76, "Impl enenta tion Pl an for the Severe Accident Pol ic y Statement and the Regulatory Use of Ne w Source-Tern In f o rma t i en ," NRC/EDO, February 28, 1986.
- 1. R. Barrett, "Status of the Sev e r a Ac c id e n t Program for Operating peac-tors ," U.S. Nuclear Regulatory Ccmmission, of'1ce of Nuclear Peactor Regu-lation Staff Presentation to the ACPS Subcc*mmittee on Class 4 Accidents, February 24, 19 % .
- 3. W. J. Luc <is at al., "A Human Deliability Analysis for the ATUS Accident 9quence With MSIV Ciosure at the Peach Bottcm Ata n t c Ne r Station,"
Brnoanaven N3tton31 L3boratory. Tec hn t c al Report A-3272, April 1W.
O 23
,, 4 ATUS Final Rule - Code of Federal Regulations, Title 10 Section 50.62, "Requirements for Reduction of Risk frca Anticipated Transients Without V)
/
Scran Events for Light-Water Cooled Nuclear Power Plants," June 1984
- 5. NRC Station Bl ackout Proposed Rul e , Federal Register, Vol ume 51, No.
55/ March 21, 1986, pgs. 9829-9835,
- 6. R. Youngblood et al . , "Fault Tree Application to the Study of Systens Interactions at Indian Po i r.t 3 ," Brookhaven National Laboratory, NUREG/CR-4207, January 1986.
- 7. kierican Nuclear Society and Institute of El ectrical and El ectronics Engineers, " A PRA Procedures Guide," NUREG/CR-2300, January 1983.
- 8. R. B. Worrell and D. W. Stack, " A SETS User's Manual for the Fault Tree Anal yst ," Sandia National Laboratories, NUREG/CR-0465, S AN077-2051, November 1978.
- 9. "Reactor Risk Reference Document ," U.S. Nuclear Regulatory Commission, NUREG-1150, Oraft for Comment, February 1987.
10 K. Shiu et al . , " A Review of the Accident Sequences Following an Exces-si ve Release of Water at El evati.on 8' of the Reactor Building in the Shoreham Nuclear Power Station," Brookhaven National Laboratory, Craft Report, NUREG/CR 4049, BNL-NUREG-51835, March 1985.
f'"i
'd
- l l
l 4
O 29 4
Tabl e 4.1 Importont Attributes for BUR ffark I Containments Relating to Plant Features and Operator Actions:
Maintaining Containment Integrity via Wetwell Venting Concern: Uncontrolled breach of the containment boundary in the progression of a severe accident can lead to significant releases of radio-activity.
Basis: Severe accident analyses indicate that wetwe i i venting will si g-ni ficantly reduce the potential for loss of , antainnent integrity because of overpressurization events.
Caution: Wetwell venting should not be indiscriminately perforned. A clear understanding of the accident sequence in progress should have been attained before initiating venting. The ef fects of venting should have been assessed and nade known to the operators during the training program. The assessment should include the ef fects of wetuell venting on the operation of energency core cooling
. (ECC) injection systems and health consequences.
Attributes:
1.1. For accident sequences where wetwell venting has been assessed to be beneficial, previous studies have indicated that the issues important to consider when selecting the wetwell venting pressure setpoint are as foilows: ,
~
a, the ul tinate containment pressure capability would not be exceeded,
- b. the backpressure acting on the high-pressure cool ant i nj ec ti on (HPCI) and reactor core isolation cooling (RCIC) pump turbines as well as the sa f e ty- r el i e f val ve assenblies would not prevent then from perfonning their functions.
- c. the vent valve 2ssemblies would not be prevented from pe r fo rni n g their fonc tion.
During a station blackout, wetwell venting night he di f ficul t to accon-plish. However , dependi ng on the pl ant con fi guration , if venting Con-nences following the onset of the transient but before the depletion of the station batteries, the benefit of venting could be realized.
1.2. If the station batteries are not available and nanual ini tiation of wet-well venting i s deemed necessary, the tine requi red to perforn tni s function wnile following the energency procedures is important for nini-ni zing tne ex60sure nf personnel to the nazarduus environnent.
O 30
Tabl e 4.1 (Conti nued) 1.3. It is important that operator training and emergency procedures, speciI fying the plant parameters that prompt the operators to prepare, com-mence, and terminate venting, are consistent With the scope and tining of the required actions. An important ingredient of severe accident planning is that the training and procedures specify the flowpath(s) available for venting, specific components to be aligned, the required positions / states for thes'e components, and the course of action if it is not possible to tenninate venting.
1.4 Assessing the capacity of the vent lines and associated vent valves has been found to be important for determining whether venting has the capability to decrease containment pressure for a particular accident sequence.
1.5. A venting fl owpa th , that ensures that all nedia to be vented pass through the suppression pool thus reducing radiological releases by approximately an order af magnitude, would substantially reduce risk to the public.
1.6. Venting studies have demonstrated the importance of assessing equip-ment; desi gnated to support wetwell venting, for its ability to func-tion reliably for a sufficient period under the predicted environmental and fluid loads associated with the venting commencement pressure, in-cluding venting operability during the vaporization release phase of core-con. crete interaction.
1.7 Severe accident studies have highlighted the importance of assessing the ef fects of possible hydrogen burns, radiation, and/or stean on equipment important to the mitigation of accident sequences which is l oc ated in the reactor building outside of the primary containnent.
This would aid in the identificdtion of al ternate vent.ing paths.. judged not to be detrimental to important eq;ipnent. The effectiveness of the reactor building blowout panel s and fire sprays to acconnodate tne di s-Charge through the primary containment vents, thereby ensuring reactor building structaral integrity were al so identified as important issues for successful venting.
1.8. The ef fects of possible containment depressurization on the net posi-tive suction head (NPSH) of the ECC related pumps and identification of alternate injection sources which are unaf fected by venting were deter-mined to be important considerations for successful venting.
1.9. Th e capaoility to terminate venting and the conditions under which venting would ce terminated is an important consideration in the vent- l ing assessment as is the level of radioactivity in the wetwell air- l space. '
1.10. Operator training and emergency procedures wnich speci fy the possible actions to preclude deinerting the Containment by terminating venting before a negative pressure di f ferential is reached in the vent path woul d red uc e the potential for containment impicsion or hydrogen con- '
O bustion, V i
~
31 i
Table 4.2 Important Attributes for miR Mark I Contairinents Relating to Plant Features and Operator Actions:
Maintaining Suppression Pool Effectiveness Concern: Bypass of the suppression pool in the progression of a severe accident can lead to signi ficant releases of radioactivity that would otherwise not occur if tne fission products were retained in the containment and/or scrubbec by the suppression pool, use of the drywel*) sprays for containment heat removal , fission-product scrubbing, and debris cooling can al so impact suppression pool ef fectiveness.
2.A. Preventing Suopression Pool Bypass Basis: Implementation of the following attributes may reduce the poten-tial for bypassing the suppression pool .
Attributes:
2.A.1. An assessnent of the fraction of the core leaving the reactor pressu're vessel (Roy) that can be controll ed or confined to prevent contac*
with the steel containment shell is considered to be inoortant in view of the uncerta nty concerning penetration of the steel shell by nolten core debris, flote that it has been suggested that this mode of early containment failure can be precluded by erecting a concrete or magne-siun oxide curbing of suf ficient height to prevent or delay the debris f ron reaching the containment shell and thereby confining ,the debri s &
T in an area wnere i.t can be cooled.
2.A.2. Past PRAs have d emons tra t ed that aporopriate maintenance, surveil-lance, and energency procedures and training related to closure of the contai nment i sol at ion val ves and vacunn breakers are essential *o pre-vent suppression pool bypass.
2.8. 'Jsi ng nryweli se n Basis: Severe accident inalyses indicate that use of drywell soray wnuld aid in decontaniniting the drywell atmosphere of fi ssion products ,
would hel p Control the Containnent pressure rise because of the decay heat load , and may promote debris cooling.
2.8.1. If tne heat renoval provided by the drywell s pray-rel a ted components nave suf ficient c 30acity to re'1ove hea t load s antici pated d uring the d ominant accident sequences , containment pressure rise can de con-t rol l ed . These loads incl ude but are not l im i t ed to :1ecay neat and the cnenical energy released f ron netallic oxidation.
2.8.2. The cene fi t a f dryuel1 sprsy can be acnt eved i f 1 t connences unen tne c on t ii nment pressure exceed s a pre <1eternined v al ue cal cul atei in acco rd ance with the B'.id 61ergency Proced ure Guidelines (E?cas) or beforo *he iiryuell temper 1ture reaches the val ue at which the auto.
natic depressuri:3 tion syste1 (ADS) i s qual i f i e<l.
32 0-
l 1
Tabi'e 4.2 (Continued) j 2.B.3. An inportant aspect of drywell spray usage is spray ternination when the containment pressure decreases below a predetermined value calcu-lated in accordance with the EPGs to preclude implosion.
2.B.4 An essential consideration is the ability of equipment designated for d rywell spray to function in a reliable manner under the predicted contajnnent conditions associated with sequences. for which containnent
, spray operation is needed.
2.B.5. Operator training and energency procedures specifying the fl owpath s and components to be aligned and their required positions for initiat-ing drywell spray could be . beneficial . If a backup system and/or equipment is to be utilized, operator training and procedures identi-fying the flowpaths and specific actions required for any temporary
- systen cross-connections may provide additional benefit.
2.B.6. Operator training ano energency procedures specifying the plant paran-9ters that will prompt the operators to initiate and terminate the d rywell spray consistent with the time required to align the systen ann components may eliminate confusion during accident stressed condi-tions.
2.8.7 The capability of the drywell spray to reduce fission-product contani-nation of the drywell atmosnhere has been found to strongly depend on '
its ability to cover the entire drywell volume for an adequate time, Ci witn spray droplets of an appropriate size. Other inportant consider-h ations for assessing spray effectiveness were found to be the total amount of water available fo.r long-tecn spray operation, the pressure under which the water could be supplied to the spray headers by var 1-aus sources, the elevation of the spray in the drywell , the nozzle.
spray pattern, as well as large nbstructions in the drywell below the ,
spray neaders.
)
2.8.8. To ;;ronate debris cooling, flooding of the dryuell via sprays could be !
Deneficial if the sprays were initiated early enougn to finod the dry- )
well floor before RPV failure for the dominant accident sequences, and the spray termination point would not allow the debris to reheat af ter it has been quenched.
2.R.9. Alternate sources of drywell spray sucn as a diesel-driven fire puno witn capability to provide suf ficient flow and head for adaquate con-t a i rnent, neat renoval could prove beneficial.
l 33
- -. - s . e - -- -- -
Table 4.3 trourtant Attributes for BUR llark I Containnients Relating to Plant Features and Operator Actions: &
W Preventing and liitigating Interf acing Systens LOCA Concern: Al though the interfacing systens LOCA sequences are not considered to be laading contributors to core-damage frequency, they represent potenti all y hi yh release sequences and could contribute si gni fi-cantly to the overall risk for the plant under review.
Ba s'i s : Inplementation of the following. attributes may keep the frequency of an interf acing systems LOCA acceptably low. Note that the reso-lution of Generic issue 105 (GI-105), which deals with interf acing systens LOCAs for both BWRs and PWRs , may have an impact on plant features and operator actions used to ensure a low frequency for thi s event. However, the following attributes could be hel pf ul in keeping the frequency of interf acing systens LOCA sequences low Attributes:
3.1. If low-pressure lines that potentially could be overpressurized are pro-vided wi tn al arns to al ert the operator to the syrtptons of an overpres-sure event, the operator could respond in a timely fashion.
3.2. Periodic operability testing and local leak rate testing (LLRT) after each shutdown of tne equipment designated to provide isol ation and pre-vent overpressuri zation , such as the residual heat remov al (RHR) line i sol ati on val ves or the low-pressure injection system check v al ves ,
. could be beneficial.
3.3. The relief valves designated, to mitigate low-pressure system overpres-surization were not si zad with the possibility of an interfacing systems LOC A i n ni no . However, given that an interfacing systens LOCA occurs in a non-isol atant e pnrtion of a inw-pressure systen, there may he al terna-tives availaole to tne oper car such as taking advantage of additional relief val ve s . F3ctort ng such actions into :he appropriate emergency procedures could De hel pf ul .
3.4 Operator training and procedures speci f ying the actions to be taken to i solite the low-pres sure systems identi fied above or to depressuri ze tne reactor coolant system could be helpf ul in nitigating the consequences of the interfacing systens LOCA.
3a
,s Table a.4 Inportant Attributes for BUR fiark I Containments Reiating to P1 ant Features and Operator Actions:
(d) Preventing and liitigating Anticipated Transients
- Without Scram ( ATUS)
Concern: ATWS sequences have been shown to be one of the leading ciasses of severe-accident sequences both in terms of core-damage frequency and risk.
Basis: The information developed here is based on the assumption that each of the plants is (or will be) in compliance witn the ATWS final rule dated July 26, 1984 PRA studies have shown that the predicted core-damage frequency because of ATWS is signi ficantly lowered based upon 3di fications wnich comply with the ATWS rule.
The major thrust of the ATUS rul e is on the addition and/or up-
~
grading of scram rel ated systems and equipment to prevent an ATHS. Human reliability studies perforned in support of NUREG-1150 point to potential benefits for improved operator training and procedures to mitigate the effects of an ATUS and prevent core d amage from occurring. For any individual pl ant which may be l found to be vulnerable to ATUS, the following attributes reflect added measures that emphasize the operator's role and function in <
nitigating an ATUS initiator. !
Ouring an ATWS sequence, the operator is required to inject boron into the reactor pressure vessel (RPV),, to inhibit initiation of automatic safety systens and to attempt to nanually control and l v
s) mitigate the outcone uf the event. In contrast, nost other acci-dent sequences are prevented or mitigated by systems wnich allow l
the operator to monitor automatic system initiation and require intervention onl y wnen a system fail s to function ad equ ate l y.
Thus , an AT'.lS sequence requires operator responses that a r'e in opposition to the responses requi red for the recovery and ni tiga-tion of all other of f-nornal and accident events'. Therefore, operator training and emergency procedures which specifically pre-pare operators to perform the unique ATUS actions called for in i One BWR Energency Procedure Guid eli ne s (EPGs) were found to be i cene ficial ,
j Attributes:
4.1. Recent severe-accident reseirch has indicated the importance of operator l training and emergency procedures whicr speci fically prepare the opera-tors for an ATkS and nence reduce tne plant vulnerability to ATUS. The
- speci fic ATuS actions include those wnich
a) specify the pl ant parameters that are indicative of ATU3,and the actions to na taken to verify that the reactor recircul 3 ting pumps have Orlpped aut713tically (or the actions to be taken i f tne pumps au nut trip automatically).
n) specify tna pl ant parameters that indicate nanual standoy liquid control system ( SLCS) actuation and the actions to he t3 ken and veri fication to be made to ensure that the SLCS was actuated.
ON 35 O , ,
- Tanle 4.4 (Continued) c) ensure operator familiarity with RPV water level and flow control during ATWS including guidance for a situation where RPV water level indicators may be inaccurate and may provide conflicting indications of the water level, d) address the possible reluctance of operators to defeat a safety sys-ten, in particul ar, the need to inhibit the ADS inmediately af ter an SLCS initiation.
e) Specify the responsibilities of operating staff members and clarify how infomation will be exchanged among then. In pa rti cul ar , in-Strumentation readings may have to be relayed between the mencers operating the control boards and the senior staf f operator coordi-
. nating the staf f's response to the accident.
4.2. An important consideration is the capability of the systems and equip-ment requi r ed for mitigating an ATUS to function for an appropriate time of operation under pred icted environmental and fl uid loads associatec with severe-accident sequences.
4.3. Es tabl i shi ng the main condenser as a heat sink by reopening tne main steam i sol ation valves and turoine bypass valves is an al ternati ve strategy which may be useful for some plants.
O l x l
l
s Table 4.5 Important Attributes for BWR tiark I Containments 4 Relating to Plant Features and Operator Actions:
('s_/) ,
Hitigating Station Blackout via Alternate Reactor Pressure Vessel (RPV) Injection Concern: Station blackout sequences have been shown to be one of the lead- ,
ing classes of severe-accident sequences both in terms of core- l damage frequency and risk.
Basis: Signi ficant study and resea rch have preceded current work on severe accidents; in particular, reference is made to the rulemak-ing activity already under way on station blackout (580). It is assumed that seen the SB0 rule is finalized, some requirenents of the rule may be similar in form to the attributes listed b el ow. I Nevertheless, during an individual plant exanination (IPE), it is I important to hignlight those areas which previous PRAs found to be important contributors to the SB0 core-d ama ge frequency. For those specific plants which are found to be vulneraole to SB0 events, the deterministic attributes below will assi st in identi- !
. fying potential areas for plant improvements as well as identify- , l ing operator actions which are key to mitigating an 580 event. . l Attributes: j 5.1. FC I the BWR-4 design, the high-pressure coolant injection (HPCI) system and the reactor core isolation cooling (RCIC) system, are two systens
- intended for the purpose of RPV injection independent of ac power. How-( ever, it has been postulated that these systems cannot continue operat-
's ing in the presence of a prolonged blackout. Previous studies hi gn.
lisi ted the 'mportance of assessing HPCI and RCIC with respect to e x- ,
tending tneir capahility to function in the presence nf an S30.
l 5.2. Operator training and energency procedures specifying the plant pa r an-eters indicative of HPCI and RCIC initiation, as well as tne actions i requi red to pl ac e in operation and/or ensure continued oceration of i these systems under SB0 conditions are essential .
- 5. 3.. An important consideration is the capability of HPCI and RCIC syste,s to function for an appropriate tine of operation under predictett envi ron-nental conditions and fluid loads associated witn 530.
$.4 Previous studies have emphasized the importance of eliminating, to the extent oractical , comnon-cause fail ures f rom the design of L1e ac and .ic power systens and other energency support systems. ,
1 1
1 I
A t i V
37
Table 4.5 Important Attributes for BUR Mark I Containments.
Relating to Plant Features and Operator Actions:
Preventing Loss of Containnent Heat Renoval by Using Al ternate Cooling Concern: Failure to remove the decay heat buildup in the suppression pool (lots of containment heat removal) following a transient event nas been shown to create net positive suction head (NPSH) problems for the punps taking suction from the suppression pool and, therefore, can lead to injection failure, subsequent core damage and contain-nent failure. The Reactor Safety Study indicated that this was a leading class of core-damage sequences.
Basis: Recent severe-accident research has indicated that the frequency of core damage due to failure of emergency core cooling (ECC) in-jection because of loss of the containment heat renoval function (i .e., TW sequences) can be reduced by several operator actions.
Attributes:
6.1. For severe-accident conditions when suppression pool tenperature pre-cludes use of tne prinary ECC injection paths, operator training and energency procedures specifying netneds and actions for containnent neat renoval via alternate i nj ecti on path ( s), were shown to be beneficial ,
especially when used in conjunction wi th wetwell venting (see Tabl e 4.1).
6.2. The capability of alternate injection paths to provide suf ficient flow to preclude core damage is an important consideration in deternining their effectiveness.
6.3 If local operation of any equipment is required , an important consideri-tion is whetner the time required to pertorn these functions i s consi s-tent with tne tine availaole to help prevent core damage and whether personnel would be exposed to thc hazirdous envi ronment.
6.4 / n n *. re r impcr'. ant considerition is the capability of aquipment, use1 for alternate injec*. ion, to function for an appropriate tine under tne pre-dicten environnental and fluid loads associated with severe-accinent sequences.
O 38
Tabl e 4.7 Important Attributes for BWR Hark I Containnents
[f-~gJ Relating to Plant Features and Operator Actions:
s/ Ensuring Reactor Pressure Vessel Oepressurization Perfo rmance Concern: Sequences, in which there is a failure to depressurize the reactor pressure vessel (RPV) after failure _ of the high-pressure injection systems, have been shown to be leading contributors to the core-damage frequency. fiany of these sequences do not create the con-dition necessary to actuate the automatic depressurization systen (ADS),
Basis: Past studies have shown that the following attributes can signifi-cantly improve the response of the ADS to depressurize the RPV so tnat low-pressure systems may he used to provide eniergency core cooling (ECC) injection.
Attributes:
7.1.
- The threat of high-pressure sequer.ces may be reduced by improving ADS actuation capability for the dominant accident sequences. Note that a design cnange at Peach Bottom, wherein the high drywe,Il pressure signal is not requi red for automatic ADS actuation and tne time delay for actuation has been lengthened, decreased the assessed f ailure rate for depres suri za ti on. However, resul ts for Peach Rotton indicated a high likelihood of concurrent ac and dc power failure which also results in a g hi gn-pressure sequence. Such vulnerabilities may be elininated by a
(- ') .
dedicated backup dc supply to ensure depressurization capability under station blackout conditions.
7.2. An imp 3rtant consideration is whether the AOS is capable of initiation and operation under the environmental conditicns associated with the d omin ant accident sequences particularly tne maximun pressure antici-pated before venting (see Table 4.1).
- 7. 3. Operator training and energency procedures which ensure reliable manual control of MPV cepressur12ation appear to be bene'icial .
7.4. The capability of the systens required to provide RPV depressarization (e.g., batteries and air supplies) to function for an appropriate time under the predicted environmental conditions and fluid loads associated witn severe. ac:ident sequences is important.
n v
39 .
Tabl e 4.8 Important Attributes for BWR liark I Containnents Relating to Plant Features and Operator Actions:
Identifying Support System Interdependencies Concern: When conducting an individual pl ant examination (IPE) it is essen-tial that the support system interdependencies be fully developed, understood, and reflected in the final resul t s. Otherwi se there is no assurance that the dcminant core-damage / risk sequences have been identi fied.
Basis: It is important to ensure that the full set of support systen in-terdependencies have been id enti fied and have been reflected in the results of the IPE. The complex nature of a nuclear power plant makes it essential that this area of analysis be fully exam-ined on a plant-speci fic basis.
Attributes:
8.1. It is important to identi fy any systen that provides direct 50pport to either a frontline or support system along with its supported systen.
8.2. It i s important to condi tion each dependency as to wnat sequences or under what (if not all) circunstances it applies.
8.3. Linking dependencies together within the analysis will help discover the extent to which their influence reac,hes through the systens to a Conse-quence. Thi s will help identi fy seCondtry dependencies to e,nsure that no one f ailure in a support system has any unknown critical outcone on other support or front-line systens.
40
n Table 4.9 Inpnetant Attributes for BUR liark 1 Containments fj s./
Reiating to P1 ant Features and Operator Actions:
Preventing and tiitigating Reactor Building -
Floodi rg .
Concern: An excessive water release into a portion of the reactor building outside the primary containment which houses a concentration of safety-related equipment raises the possibility of. a common-mode .
event disabling all the equipment in the compartment. At least one plant with a Nark 11 containment has been identified in whicn the location of sa fety equi pment , including all emergency core cooling systen (ECCS) pumps, in the lowest reactor building level makes the flooding initiator a substantial contributor to core-damage frequency (CDF).
Basis: The following attributes are intended to help reduce the potential of a conmon-node f ailure of safety equipment because of reactor building flooding in Mark I containments where reactor building layout combines inportant safety equipnent in low-lying portions of the reactor building with exposure to possible inundation.
Attributes:
9.1. Operator training and procedures ensuring that the operator will diag- I nose and isolate any flooding of the reactor building that occurs could be benef.icial.
O O 9.2. Operator training 'and energency procedures ensuring that the operator is prepared to use al ternate injection sources still available if fl ood i n g causes a common-noce failure of ECCS equipment could also be helpful .
- 9. 3. A PRA study has indicated the importance of assess'ing the electrical s'yst ems for tne pnssibility of cascading failures because of flood-induced electrical snorts.
9.4 Providing alarmed water-tignt doors in the lower levels of ~ the reactor :
building, to onti ff the operator wnen tney have been left open, could Da I cene fici al .
l l
l 4
\
. 4 '.
A Aaa aa .n ..a. -s --- b-. ".4 .++ rana- aa ~m.A s12. A
, l t
42
APPEN0!X A 1
. SEVERE-ACCIDENT RISK INSIGHTS In this appendi x the IDCOR and SARP analyses for a BUR 4 witn a liark I containment are reviewed. Di f ferences between the studies are identi fied and the basis for these differences are explained. This information was then used in the development of plant-type specific insights for the reduction in the frequency df occurrence and nitigation of severe accidents (discussed in Sec-tion 4 of this report). In addition to the IDCOR and SARP analyses, other j studies relevant to a BWR-4 with a Mark I containment were al so evaluated. :
The information obtained from these studies has contributed to the conpilation i of the plant features, operator actions and deterministic attributes provided )
in the tables at the end of Section 4 1
Section A.1 describes the review of the various estinates of the core-d ana ge pro fil e. Core-meltdown pnenomena and containment response are ad-uressed in Section A.2. Di f ferences between IOCOR and SARP estinates of j fi ssi on-prod uct rel ease and offsite consequences are discussed in Section A.3. Section A.4 addresses the potential offsite consequences of the severe )
accidents, which were found to be important. Finally, Section A.5 indicates the insights gained from the review of these studies.
A.1 Core-Danage Profile i
The main objective of this section is to present, within the scope of t hi s s,t ud y , the Peach Botton core-damage profiles energing from the RSS,1 O IDCOR,' and ASEP/SARP I analyses. Also in this section, an attempt is made to clariff not only tne nost important dif ferences in the tnree 3each Botton ;
studies , out also the di fferences between the Peach Rotton studies and the I other POAs considered in this st6dy (Linerick DRA Review," Browns Ferry IRE? '
P .A,s and the Shorehan PRA Reviews),
A.l.1 Peach Botton Core-Canage Profiles - RSS. IDCOR, and ASEP/SARP In :ne 153,1 sequences ini tiated by a transient event and folicued n'y f ailure to scran the reactor ( ATUS sequences - 41" of tne C0F) or ny failure to provide long-ter, decay heat renoval (OHR) from tne containment (IW se-quences - W of C0F) were found to be the nost inportant contrioutors to tne core-d anage f requency. A total CDF of 3.2x10-s/yr was ootained in tne RSS and the breakdown by sequence-type is provided in Tabl e A. I. Note that the RS5 used nr1ian values for all system f ailures and mean values for the transient initia. ors.
)
The ICC0R study: presented two core-damage profiles for Pe acn Botton, n eely, tne IDCOR baseline and the IDCCR committed core-damage pro files. The d ' f fere 1ce in tnese tuo core-danage profiles was that the connitted pro fil e accounted for cnanges in plant configuration and operation conni tted to by tne utility but not jet incorporated. In thi s an,11 ysi s , nol y tne Ir;COR cyni ,ted core-o nnage profile will be discussed.
Ine IDC1P cennitted core-damage profile was obtained by updati,y tne RSS analjsis. This updating procass incorporated recent values for initiator fre-auencies, nore realistic success criteria, plant nodifications, and cnanges in 43 nO O A a
the emergency procedures; all these changes are shown in Table A.2. It i s in-portant to note that for ATWS, Tu and station blackout sequences new event trees were developed in the IDCOR study; these new event trees were necessary to better characterize the progression of the sequences and the operator ac-tions during the course of the accident.
The IDCOR committed core-damage profile is al so presented in Tabl e A.I.
From this table, it can be seen that the total CDF is equal to 8.1x10-6/yr and the nost important contributors were determined to be transients with failure to scram (TC sequences - 90".) and station blackout sequences (TB sequences -
6 ". ) .
The results of the ASEP/SARP analysis for Peach Rotton (Table A.1) found that station bl ackout contributes 86". to the calcul ated core-d ama ge fre-quency. The second leading contributor was ATuS with a 12". contribution. Al l other sequences combine to contribute to the remaining 2%.
In sunnary, the following can be stated about the relative contributions of the calculated accident sequences within eacn study:
(1) ATWS (TC sequences) were found to be important contributors in all three studies; they were the major contributors to tne core damage in the IDCOR study.
(2) Station blackout sequences were the major contributors in the ASEP/SARP study; but were less important in the IDCOR study and negligible in the RSS.
(31 The TW sequences were onl / dominant in the RSS. The nain reason for this dif ference comes f rom the inclusion of alternate methods for containrent neat renoval (containnent venting with alternative injection sources such as the control rod drive pumps) in the 10COR and ASEP/SARP analysis. The extended tire frane faoout 3n hours before core mel t) and 'issociated low decay neat l evel (acout 0. 5 ', o f full power) inply a nign 11 kel i hoc d of successful containment heat renoval for these sequences.
The naj v ci f fereces in the doninant accident sequences are di scussed in the next sucsection.
A.1.2 Peach Botten Onninant Sequences: Di f ferences in RSS. IDCOR. and SARD A alvsis This section discusses the main 11 t ferences present in the R$$, 10 CUR, and SARP analyses for tne f ollowing sequences:
. TC
. T8
. Tu
. ,TQu'l and TOUX 0
44
A.1.2.1 TC IATus) Secuences CJm} Before discussing the differences in the CDFs for the various studies, the similarities and differences in the basic assunptions used in the studies have to be clearly identified:
(1) Scram failure was defined as failure of all of .the 185 control - cods to insert in all studies.
(2) A nanually actuated standby . liquid control system (SLCS) with a capacity of 43 gpm was assumed in the RSS. The !D00R analysis 'and the ASEP/SARP analysis assumed an 86-gpm manually actuated SLCS, an alternate rod, in-sertion ( ARI), and RPV Level 1 ftSIV closure.
(3) The current RuR. Energency Procedure Guidelines- (EPGs) were used only in 10COR and SARP analyses.
The most important contributors to the core-damage f requency from TC se-quences are discussed bel ow for all of the three studies considered (RSS, 10COR and SARP).
- RSS: TC = 1.3x10 5/yr-In tne RSS, a very simple analysis was performed, and only one sequence was found to contribute, namely, any t,ransient event (T = 10/yr) with failure of the reactor protection systen (RPS) (1.3x104) and failure of the o'perators to actuate the SLCS or to manually insert the control rods (probability =
0.1).
10COR: TC = 7.3x10-6/yr The 10COR study presented a more detailed analysis of the TC sequences.
In thi s analysis, the current 8WR EDGs (e.g., maintain level at top of active fuel (' AF), control RPV injection with high- or low-pressure systens, avoid Automat c deoressurization), as well as more recent resul ts from ATWS deter-
'ninistic analysis were used. Also, in the 10COR analysis, i f injection with low-pressure systens was successful, c red i t was given for contaiment heat renoval by venting the containment. Even when the containment was assumed not !
to successfully vent (i.e., even if the containment fails), low-pressure in-jectign was assumed to continue with a probability equal to 0.5. l in the 10COR analysis, the transients were basically divided into two groups; nanely, transients with the power conversion system (PCSI available (turnine trips T)
T and transients with the PCS unavail aol e (isolation events - Tr). In addition, a distinction was made between transients at low power (<25() and at high power (>251,).
Isolation transients (T[; wnich i ncl ud e MSIV closure, inad vertencl y oven relief valve (IORV), loss of FW, loss of condenser vacunn, anr1 transfers from turbine trip events with early closure of ilSIV) contributed anout SM to tne ATWS CCF, and turbine trips (TT ) contributed the other 42'.. The nost inportant sequences were identi fied as:
O 4 45
l T{2%(1.03) *C1 (1.0 x10- s),C2 (0.515)*VC3 (0,5) 2. 65 x10 - 6 ( 36 ". )
T{25%(4.19)*C t (1.0 x10. 2 s) *C (0.115) 3 2.41(0.5)
- VC x10- i (33t)
T{2%(1,0 3) *Ct (1.0 x10- 5)*G(0.485)*U(0.23) *VC3 (0.5) 5.74x10 7 (8")
T<25%(0.222)*C t (1.0 x10-5)*C 2 (0.215)*VC3 (0.5) 2.39x10-7 (30 T{25%(1.03)*Ct (1.0 x10-s)*C2 (0.515) *7C3 (0,5) *7EIiT(0.0) *E(0,1) 2.39x10-7(3",)
A sensitivity analysis was performed by BNL to calculate t.ie ATWS C0F with no credit given for containment venting and assuming that i f contai nmen t f ail s , core damage would re s ul t . With those two assumptions the calculated CDF increased from 7.3 x10-6/yr to about 1. 2 x10 - 5/yr.
ASEP/SARP: TC = 1 x10- 6/yr The Peach Rotton ASEP/SARP study presented a more detailed quan ti fi cati on of the ATUS sequences (than either the RSS or tne IDCOR studies). The ASEP/
SARP study analyzed all the operator actions required during the progression of ATils sequences. The Peach Bottom Tran si ent Response Impl emen t a ti on Pl an procedures were used in the 'ASEP/SARP. quanti ficat. ion of the' f ailure probabili-ties of the operator actions. Five operator actions, nanely. SLCS actuation, innibit 30s, nigh-pressure RPV level control, controlled depressurization, and W
low-pressure RPV level control, were analyzed in detail .7 The ASED/SARP analysis obtained an ATWS core-danage frequency equal to 1.0 x10- 6/yr, and tne dominant sequences were found to be:
T [( 4. 2 ) *Ct ( 1.0 x10- 5 )
- m( .96 ) M( .36 ) *7( .99 ) *H PLC( 6. 7 x1,0- 2) .
C M 2.13x13-1) 4.a x10 7 i (t 4. 4 ) *C (t 1.0 x10-5 ) *SLC( .044) *H AOS( . !4)' CONT ( .2?d) 5.8x10-8 T [( 4. 2 ) r (t 1,0 x10-5) *SLC' .044) *HAOS( A.6 x10- 1) *CDE P( 2.13x10- 1) * ?( .51 1. 7 x10 - 7 i[( 4. 2 )'C (t 1.0 do-i) *9LC f .0a4 ) *HAOS( 9.6 x10- 1) *VE '!T(0.91' CONT ( .222) 3.1x10 7
_Surn a ry From the discussion of the most important sequences in the previous sec-tions, 4. can he seen tnat tna di f ferances batveen the 10 Cort and 9 Ait p an al js o s are principally in tre following areas:
O as
-q (1) Hunan Actions - In all the three studies, the CDF was mainly determined by the values used for the various ' probabilities of operator errors. A better perspective of where those values lie is presented in Table A.3, wnere the conditional probabilities of core damage given a turbine trip or an isolation ATUS are given for the two Peach Botton studies (in Table A 3 the results of the Shoreham PRA 8 and its reviews are also pre-sented). Fron Table A.3, it can be seen that:
For a turbine trip ATUS, the conditional probabilities of core damage vary f rom "negli gibl e" ( ASEP/ SARP) to 0.586 (Shorehan PRA review). 4 The 10COR value is 0.053.
For an isolation ATUS, the conditional probabilities of core damage vary fron 0.024 ( ASEP/SARP) to 0.957 (Shorehan PRA Review). The 10COR val ue is 0.251.
tiost of the dif ferences in these CDFs for botn ATUS initiators were because of differences in the success criteria which resul t in sub-stantial dif ferences in human error estinates.
The conditional probabilities of core damage calculated from the Shorehan PRA and its review are also presented in Table A.3 (they will be dis-cussed in Section A.I.3).
(2) Venting - The IDCOR and ASEP/SARP studies gave credit for containment venting and also assumed that even with a failed containment RPV i nj ec-tic'n wi th low-pressure systens was possible. If these two "assunptions" p are not nade, then the ATHS C0F in the 10COR and ASEP/SARP studies would Q increase to about 1.2x10-5/yr and 2.3x10~6/yr, respectively.
A.I.2.2 TR (Station Riackout) Seouonces The contribution to core-danage frequency from T3 sequences as presented in Tables A.1 and A. A is:
RSS: 1.0 x10 7/yr
!DCP: 4.5x101//r ASEP/ SARP: 7.0 x10 /jr '
Since tne RSS analysis assumed that RCIC/HPCI can operate for 27 hours3.125e-4 days <br />0.0075 hours <br />4.464286e-5 weeks <br />1.02735e-5 months <br /> given a blackout , the only contribution to C0F cones f rom a blackout for nore tnan 30 minutes with independent failure of RCIC and HPCI. Because of tnis assumption, the RSS analysis was very sinpli fied , and comparisons wito IDCOR and ASEP/ SAR D are not meaningful, i
l Fron Tanle A.4 it can ce seen that tne nain di f ferences betWeen ICCCR and !
ASEP/SARP cwe from:
l l
(1) The loss of orfsite power (1.00P ) initiator frequency used in the ASEP/ !
SARP study was larger than that used in IOCOR by a f actor of anout 2.6 l l
- 12) A con on-node f ailure of the dc power systen postulated in tne AsEP/SARD analjs1 s was responsible for about 7 0 '. of the station bl ickout , core-damage frequency. The IDCOR analysis did not postulate this f ailure.
' 47 .
~.
l l
(3) Ah nu t 151, c' the station blickout CDF in the ASEP study results f ron in-dependent failures of two diesel generators (Nuncers 2 and 3) because of failures of the diesel generators (DGs) themselves or failures of the energency service water (ESW). The IDCOR study did not account for these failures because it assumed failure of 3 out of a of tne DGs compared with the ASEP/SARP assunption of failure of DGs #2 and 43. Thi s di f fer-ence in success criteria in the two studies was related to tne ESW systen cependency on power from DGs 42 or 43.
(4) The IDCCR analysis used a probability equal to 1. 7 x10- 3/denand for connon-node failure of 3 out of 4 DGs , whereas the ASEP/ SAPP study used a {
probabt'.ity equal to 2.35 x10 "/denand for connon-node failures of two DGs (d2 and 43).
A.1.2.3 TW Seauences The contribution to core-damage frequency froni TV sequences as presented in Tables A.1 and A.5 i s :
RSS: 1.7x10-5/yr IDCOR: 1. 5 x10 - // y r ASEP/SARP: - 1.0 x10 - ajy c The main di f ferences in the three studies are:
(1) Containment Venting and Core Danage Given Containment Failure: The IOCOR and ASE P/ S ARP analyses , gave credi t for containnent venting. Venting procedures did not exist at the tine of the RSS and thus venting was not accounted for. Hunan error prnbabilities' (HEPs) equal to 1.0 x i n- 2 and 1.0 x10- 3 were used in InCOR and ASEP/ SARP analyses , respectively. The IDCOR analyses al so as soned that even if the containment failed there would only be a 107, chance tnat injection would also be iost and tne cora wo ul d be danagei. If ne RSS assumptions (i .e., no containnent ventino and core damage wi t h containment f a il ure) had been used in all three anal yses ,
- h9 C?F f on TV sequences would be ulven cy:
953: 1.7elq-sfyr IDCOR: ? . ') < 10 - 5/yr ASE?!SARP: -7.0x10-'/yr (2) PCS Recovery: The values used for PCS recovery were different in all tnree stuaies, is snown in Tao l e A.S. Tne vil ues used in tne IDCOR analysis for failure to recover PCS for DHR (Og.Q in Tabie A.5) were, in j general, h1gaer than . hose used in tho otner nio Peach Botten anal yses.
(3} Hun 5n Action: !DCOR was tne onl y analysi s that explici tl y considered the operator f a t i ure to recogn t :e t ne need for DHR in ,anout 20 nours; a cou-qitive qunan err 0r urCDIDility equal to 1.d t ll ~ ' Was Ised. Nte tnat this was the leadlog contrioution to CDF fren TW sequences using IDCOR a M unstions.
O w
A.1.2.4 TOUV and TOUX Secuences The contribution to core-damage frequency from TOUV and TOUX sequences as presented in Tables A.1 and A.6, is: .
RSS: 8. 4 x10- 7/yr IDCOR: 4.1 x10- 8/yr ASEP/ SARP: 6.8 x10- 8/yr The RSS resulted in the higher core-damage frequency because of the suc-cess criteria used for the low-pressure injection systems and Qecause at the time of the RSS the RPV required manual depressurization for all transients with failed hign-pressure injection (see item 2 below).
The nain differences in the three analyses (as given in Table A.S) are:
(1) Control Rod Drive (CRD) Systen Inje'ction: The SARP analysis was the only one in which credit was given for CRD systen pump injection.
(2) ADS: In the IDCOR and ASEP/SARP analysi s , the ADS was actuated on a "low-low" level signal alone for .8 ninutes. In the RSS, a nanual depres-surization was required because tne actuation of ADS. was assumed to requi re "I ow" level and "hign" dryuelI pressure signals for 2 n1nutes.
(3) Low-Pressure Inj ection: The ASEP/SARP analysis gave credit for inj ec-tion, af ter depressurization, with the high-pressure service water (HPSil) systen and the IDCOR analysis gave credit for injection witn the conden-sete system. - -
(a) Loss of oc Bus Initiator: The SARP analysis was the only study tnat treated this initiator, a
A.1.3 Doninant Secuences Comuarisons: Peach Botton and Other BUR Analvses in this section, an attempt to expl ain di f ferences in the dominant acci-dent sequences in several Peach Botton analyses and other BUR PRAs and PRA re-views i s made. The nnst compl ete FWR analyses that have cnaracteristics similar to Peach Botton were chnsen for this exercise. They are: Linericx PRA Review,4 Browns Ferry PRA (IREP),5 and Shorehan DR A Review. 6 The following sequences for these studies were:
. TC
. TW
. 70tlV and TCUX
. T3
. TPO!
. SI
. TVnE O
, _ , . . . , _ . ._,9,.- c , - . , _ . _ , . ._,,y. ,_.,.,_,,,-.,r r - - .
9 A.1.0.1 TC Sequences (1) Lirerick PRA Review (C0F = 3.7x10-8/yr)
The Limerick pl an t has ' an automatic SLCS with "6-gpn C3paci ty and an ARI. With these sucstantial improvements in the protection system, Lin-erick would he expected to have a louer TC frequency. But, in that study, the current SWR EPGs were not used.
(2) Browns Ferry - IREP (CDF = 5.5 x10- 5/yr)
This study does not have a detailed anlaysis oi: the TC sequences; only the failure of RPS was Considered. The dominant TC sequences is:
Ti (1.7)=RPS( 3.0 x10-5) = 5.1 x10- 5/yr A comparison with the IDCOR and SARP work for Peach Rotton i s also not warranted because of the sinplified treatment in tne RF-IREP.
(3) Shorenan PRA Review (C0F = 4.5x10-5/yr) .
The najor dif ference in the ATUS CDF between the two Peach Rotten analy-ses (IDCOR and ASED/SARP) and the Shorehan PRA review ( a nd also the Shorenan PR A) comes from the HEPs; both the Shorehan PRA review and tne Shorenan PRA usec higher HEPs tnan those used in the Peach Bott'en stud-ies.
Other iiapn r t a n t di f ferences were Caused by assumptions in the di f ferent analyses, i.e.:
Shorehan PRA and the $horenan PR A review did not give credit for con-t rol l ed low-pressura inj ec ti on; both the RARP analysis and the ICCOR analysis for Peach Botton did give credit for controlled low-pressure i nj ec t i on.
Shorean 02A and ine Snorenan PQ A Review did not gi ve crec i- fo r cer- i tainrent venting '
1 Tne effects of *nese di f ferences can be appreciated by referring to Table l A.3 wnere the conct -ional pronacilities of core damage given an ATus are '
presentea.
Another difference, or less inportance, is tnat the Shorenan ?R A rev'iew l ysed hi gher initi ator frequencies and did not account for the power levei at tne tine of tne transient.
A . I . 3. 2 'u Secuenu s The cont"out on to CDF fr:n Tu sequences in tne Peach Botton and 311 i
q t ne r P',,P in jijses ise l 'n this re"qr*. i s pr9sented in I.ib l e A.1, lod t n e n o s *,
inport.)nt Contributluns 1re jlven in I.1Dle A.6 The nost Inpor*)nt di f f erence was found to be becluse of C on t.11 rnen
- poting. Cnly the lhCC4 and SARP analyses for DeaCh R0 t
- 0n give C re11 t for 50
3 containment venting with failure probabilities equal to 1.0x10-2 and 1.0 x10- 3, respectively. As discussed in Section A.1.2.3, i f no credi t were given to V containment venting, all the results would be between 7.0x10-6 ( Asgpf sAap) and 1.3x10 " (Browns Ferry [BF]-IREP). There were two reasons for the high core damage frequency in the RF-IR'EP: (1) no credit was given to recovery of the PCS for the OHR function and (2) a very high unavailability of the residual 1 heat removal (RHR) systen was cal culated given no of fsite power (2.9x10-2),
A.1.3.3 TOUV and TOUX Seouences The contribution to C0F from these sequences is presented in Tables A.1 1 and A.6; in Table A 6, the most important sequences are presented. The main differences are:
(1) Limerick PRA Review (C0F = 6.0 x10-s/yr) . I 1
The most important differences come from the fact that at the time of the ,
l.inerick PRA review, nanual depressurization was necessary and a proba- )
bility of 6.0 x10- 3/d was used (compared to auto-depressurization in IOCOR !
and ASEP/SARP), and the unavailability of the high-pressure injection, 8.1x10-3, (compared to 2.0x10- 3 in IOCOR, and about 4.0x10 in SARP). l (2) Browns Ferry - IREP (C0F = 5.5x10-7/yr).
The nost important dif ference was the need for manual depressurization (vs. auto-depressuri zation in 10COR and ASEP/SARP analysi s).
s (3) -Shorehan PRA' Review (COF = 6.3x10- s/yr),
In the Shoreham PRA review, about 70". of the CDF for these sequences (1. 7 x10-s/yr) ' was because of transients not a nal y:ed in the otner PRAs ;
(see Tacle A.6). Sone of these sequences were because of particular ;
characteristics of the plant. '
For the otner 301, nf the CCF because of these set;uences, the nain di f fer-ences are:
. A nigner un avail ao il i ty of the nigh-pressura injection systen
( 1,0 x 10 - 2 vs. 2.0x10-3 in IDCOR and about 4.0x10 in SARPi .
A failure probability of 0.13 was assumed for the nigh-pressure injec-tion and depressurization functions i f of fsite pouer and OGs 41 and *2 were not rec overed between 4 and 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br />. The reasnn given for this ;
assunction was depletion of batteries.
l A.1.3.1 T3 Secuences The contribution to CDF from these sequences is presented in Tao!es A.1 and A.4 In aDie A.4 tne nos; inportant sequences ara shown.
The main di f ferences annng the various 09As are:
v 51 .
, ,-n -nc , .n - -- - - , , - -- ,e - - - - . -
1 (1) Li erick PoA Review (CCF = 3.1x10-5/yr).
The Linerick PRA review used' a LOOP initiator frequency and probabil-ity of failure to recover offsite power higher than those used in the IOCOR and ASEP/SARP analyses for Peach 90tton (see Table A.4).
The Limerick DR A review (and the Limerick PRA) assued that given a station blackout, HPCI/RCIC can only work for 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> (tine of battery dept etion) if al ternate room cooling ( ARCl was made available by oper-ator actions, and for 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> otherwise. In the IDCOR and SARP analy-ses, HPCI/RCIC are assumed to work (given a station blackout) for at least 6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> without any operator action (tino of battery deplation).
. The connon-node failure probability of 3 not of 4 OGs used in the Lin-erick PRA review was about the same as that used in the IPCOR analysis for Peacn Botton; however, it was much lower than the value used in the ASEP/SARO analysis for two DGs.
(2) Rrowns Ferry /IREP (CDF = 2.9x10-5/yr) .
The major di f f erence between RF-! HEP, IDCOR, and SARP analyses for Deach Botton, cones f ron the probabi.lity of f ailure of the DGs , which was equal t o 2. 9 x10- 2 in tne BF-[ REP. The IREP report did not speci fy how thi s value (2.9x10-2) was obtained.
(3) Shoreham PRA Review (CDF = 1.3x10-5/yr).
The Shorehan PR A review-(and the PRA itsel f) presented the nost Metailed analysis for tnese sequences. In the review, RCIC/HPCI were as suned to f ail at 10 nours af ter the onset of the transient because of battery de-pl eti o n . However, for the di f ferent time pnases used in the analysis M to 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />, 2 to 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br />, 4 to 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br />, and 10 to 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />) ne failure probability of QCIC/HPC[ increased nonotonicall y until 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br />, heyend whicn time tne probaD1}ity of f 3ilure is unity.
The nost important di f ference between thi s study and the Peach Botton analysi s sas the LCSP i ni ti a tor frequency tugethor wita :ne failure to recover offsite power.
A.1.3.5 TPOI Sequences The contrications to CDF from these sequences are presented in Taole A.1. The main dif ferences are:
(1) C on ta i nmen t ven ti n g: Th e SARo and 10COR anal yses for Peach kottom teor c r ed i t Nr co r.t a i ni en t venting, wnereas tne others did not; tm s was t"e reason ony tnis se.:yence tyce did not aupear in toe ID::Ca and uso ar. al y-ses for Peacn sotten.
', 2 ) Stock npen 33fet/ ?alie* Valve (50?V): The S A'IP analjsis used a pron 3 Oility of 5.440-2 for an SnRV foll owing 3 transient. Ry contrist, tne RF PDA ised i vilue of 5. 7 x10- ' for a LOOP initiator and 3 value af 3.1410-2 hr all tne otner transients. The Shorenan 3 U. review used a value of 2.0 410- 2 ( for two or more 50RVs).
52
e A.1.3.6 TPOE Secuences
( 3 V The contribution to CDF from these sequences is presented in Tabl e A. I .
The only studies that present sone contribution from these sequences are:
RSS: 6.0 x10- 8/yr (inferred from App. V, page V-42, of the RSS)
SARP: 1.Gx10-7/yr BF-IREP: 1.0 x10 7/yr The main reason for this difference came from the probabilities used fo'r SORV (see previous section iten 2).
A.1.3.7 Interfacing Systens LOCA In all the PRAs and the PRA reviews considered in thi s study, the core canage frequency because of interfacing systems LOCA varied from negligible to 2.0x10-7/yr. The hignest interfacing systems LOCA CDF, i.e., 2.0 x10 - 7/yr, wnich was cotained ir, the Shorehan PRA review, was nainly because of the f act that the automatic actuation of the ECCS af ter an interf acing systen LOCA would cause flooding of the reactor building. This in turn had the potential to flood the ECCS pumps in a very short ' tine (in some cases less than 10 ninutes). Therefore, in tne Shorehan PRA and its review, credit was only given to injection with the condensate pumps with nakeup to the hotwell. Ho w-9 ever, ,other generic studies .10 have obtained CDFs that were much hi gher than tnose obtained in the PRAs analyzed in this report. Reference 12 has developed the following reconnendations, which are intended to "signi ficantly reduce the reactor accident risks" associated with interf acing systens LOCA:
O' (1) Disable toe non-sa fety-rei ated air operator of the testable i sol ati on check valve on the injection line in the safety systens involved *,
(2) Consider the need f o'r leakage testing of the testable isolation cnecx valve bef arp plant startup af ter each refueling outage nr af ter nainte-nance , repair, or repl ace,ent work on the valve, as an al ternative to reconnendation 1.
(31 Inprove wnan reli ability in maintenance and surveillance testing ties to !
reduce errors.
(4) Study reducing the f requency of surveillance testing of the i sol a tion carriers of the energency core cooling systems during pnwer operat in.
It snould he noted tnat RNL is currently performing a study on inter-f acing systens LOCA to provide technical support to the NRC, Reactor and Plant l Safety Issues Branch, far the ne 3n i n gf ul resolutinn of this generic issue 4 (GI 104) . !
A.2 Core-Pel'.down Phenonena and Containnent Resuonse In the previous section, inportant core-nel tdown accident sequences were identi f j ed in terms of tne overall core-nel t frequnocy, in tni s section , a review of tne core-neltdown pnononena and containnent response appropriate tn tnese accident ssquences is presen ted. In ad o '. t i n n , iccident sequences are vanined wnien, altnougn tney do not appear to be important to the overall 53 n . - - -* - -
core-nelt frequency, may ;:ose a unique. or very severe threat to Cont 3i nnent integrity. The review will again rely heavily on the [0COR and SARP analyses whicn were speci fically carried out for the Peach Aotton plant. The review also will take into account other studies pertinent to a B'R-4 witn a Dark I containment and, in particul ar, the resul ts of the Contairnent Loads Working Group (CLWG) (NURtG-1079)ll and the Containment Performance Working Group (CPWG) (NLREG-1037),12 A typical Mark I containment building is shown in Figere A.I. .The Dark I containment volumes are relatively small and relj on water to condense any stean that nignt he' released from the reactor cool ant systen during an acci-dent. Containments of this desigr are called pressure-suppression contain-ments. Mark I containments are very ef fective at condensing stean, but their small volune nakes then vulnerable to any conbustible and noncondensable gases that would be generated during a severe core-nel tdown accident. In fact, the NRC considers Mark I containments to be so vulnerable to combustion that the NRC requires all Ma rk I containnent atmospheres to be continuously inerted witn nitrogen aur1ng plant operation.
The aim of this,section is to identi fy severe-accident containnent loads (pressure / temperature hi stories) 3ssociated wi th the accident sequences iden-ti fi ed in Section A.l. These loads are then used to deternine the most proba-ble nodes of containnent f a il u re . Th e se , in turn, Identi fy the potential release paths for fission products to reach the envi ronment. This section ,
therefore, provides the link between the identi fication of core-nelt10wn acci-dent sequences and the determination of fi ssi on-prod uc t release paths. The results of an assessment of core-neltdown phenonena and cnntainment response is usually expressed in terns of a containment natrix. A containment natr1x provi tes the conditional probabilities of a particular acc id ent seuuence re-sulting in a variety of contatnnent f ailure nodes (or fi s si o n-prod uc t release paths) .
The IDCOR and SARP :entaim,ent matrices are ulven in Table A 7 From an inspec'.1on o f Tao' e A. 7, i t is clear that the SARP approacn includec 3 nigner potential for earl y containment f ailure than the 10C0R approacn. ICCnR pre.
dicted earl y containment f ail ure only for sequences with loss of uM. For tnese sequences, cantn'ecent tallare resolten in l os s-o f -c no l an t injection and, necce , core damage. ;or secuences witn coro ,neltdown in an intact con-tainment, IOCPR pr ed i c ted ritner no containnent f ail ure or f ailure nany nours af ter reactor vessel failure, in contrast, the SARP analysis yielded a rel a-ti vely h1 ;n l i k el ihood of early containnent f a il ure for nost se qunoc es , in addi tion , tne SARP analysi s indic r.ed a higher potential for suppression pool bypass after containment f ailure and gave 1955 crMit fo r viet we l l venting than the IOCOR analysis. Di f ferences in the pronabilities in Table A.7 are because of differences in model ing assumptions f or core mel tdcun and containment ra-sponse in t9e !CCCR ano 5ARp stjdjes, Tnese di f f erences aro discuss M in ae-t311 In tnp 'nllouing sec*.1 Ins.
The rw/1ew of tw ICCN and 9?? in al yse s o f Cr Al'9 I !down ph on un.fr i in 1 containnent response tidi 'jre j tl y as 51 s ted by t he [DCOR/ NRC 9eetings that were neld spec i f ic al l y to utenti f y di f f erences hetween tho approiccas viootet by the two grTip s anv to deveiVp 3 May of re s ol v 1 n ij these differences, hesA mootings 1 d e:n t i f ' el '. d oroad *N C/ I CCIM issuos that hi ghl i gn t e l s i .jn 1 f i can t differences Detue"q the aporoJChes of the two groups. Ihese issues are listed i
i
I I
(q
\_/
j in Table A 8, but they d'o not all apply to a BUR and sone are not related to core-meltdown phenomena and containment response. Out of the 13 issues, 8 I
have been id enti fiec that are. appropriate to the subj ec t of this section.
Each issue is discussed, in turn, in the following sections. 01 f f erences between IDCOR and SARP will be identified and their significance will be indi-cated.
A.2.1 In-Vessel Hydrogen Generation (NRC/IDCOR !ssue 5)
There were (and still are) significant differences between the IDCOR and SARP predictions of hydrogen generation during in-vessel core melting. During the early stages of core heatup and degradation.(while the fuel rods are still in place in the core region), both IDCOR and SARP predicted sinilar hydrogen generation. However, after the fuel rods and cladding began to melt and relo-cate into the boti;om of the reactor vessel , the SAP.P analysi s indicated more hydrogen generation than the IDCOR analysis.
Hydrogen (4 2
) is important to containment loading because it is a combus-tibl e and nonconcensable gas. The flark I containments are inerted with nitro-gen during plant operation and, consequently, hydrogen conoustion (. Issue 17) is not a threat to the tiark I containments' unless their integrity is lost and oxygen is introduced , thus deinerting the atmosphere. However, fiark I con-tainments are snail and, therefore, any signi ficant buildeo of noncendensable gases (such as H 2 ) could threaten containment integrity by overpressure. The larger amount of H 2 generated in-vessel in the SARP analysis resulted in a higher predicted containment pressure before vessel failure than in tne 10COR (g
v j
analysis. BNL staf f performed an extensive assessnent l3 of in-vessel 42 98"-
eration , particularly with regard to accidents that resul ted in core canage )
but which were terninaten by subsequent cool an t i nj ecti on. Tne resul ts of tnese calculations indicated the potential for nore H generation than pre-2 ditted by IOCOR. However, both studies al1ocated a very 1ou probaDility for overpressure f ailure of the !1 ark ( containnent from in-vessel H generation.
2 Tne authors concur that there is a low probability of Containnent ,f a i l u re Decause of H 2 accumulation befo/e RDV f ailure and , therefore, tne issue does not appear to signi ficantly af fect the potential for large fi s sion-p rod uct releaSa in ?tark I contairnents.
A.2.2 Core Slono. Core Collaose, and Reactor Vessel Failure (MRC/ TCM lssue t; '
This i s another area in wnich there were si gni ficant di f ferences between the 10COR and SARP analyses. The importance of these differences to overall risk again depended on pl 3nt specifics. Sec ti on A.2.1 i ndicated tnat the pre-dicted nycrogen generation during core slung was quite di fferent in tne IDCOR and SARP analyses but that tha impact was not great for the I! ark I cantlinnent because tne atnospnere was inert.
The inportance of the core slunp and reactor vessel f ailure nodel s is on new ney in *19ence tne ini ti al conditions for ex-vessel intar1cticas of tha core debri s w.itn wa ter or concre te. The IOCOR cora slunp noel as M ed that af ter 7m of the core nad nel ted , it rel oc a t ed into the hottom of $9a 7 actor vessel , whttn , in turn , rapidly failed necause of local penetratfor '3ilure.
Tnus , only a rel ativel y small fraction of the core, was ini tially rul esed frcn b
V ine reactor vessel . The remainder of tne core melted down over a noen longer 55
tiue pe ri od . A similar philosophy has been 3dopted in the draft MC staff issue paper on direc*, heatino, b This work states that the BWR core support desi gn (which provides indi 'iC' support for each group of four fuel bundl es hW fron the vessel bottom head, ~
, judged to minimize the probability of hign-pressure ejection of core debris into the containnent. Slunping of rel atively small quantities cf core decris (because of localized failure of the supports) i s anticipated to result in depressurization of the vessel (because of local melt-through) before large quantities of noiten core material have coll ected in the bottom head.
On the other hand, the SARP analysis (with the Source Tern Code Package) assumed total collapse of the core into the bottom of the reactor vessel af ter 7 5 f, of the core was predicted to nelt. Thus , all of the core debris sas available to be released when the vessel was predictea to fail in the SARo mod el . The nuch larger quantity of core materials released from the vessel at the time of vessel failure in the SARP nodel has important implications for the Mark I containment. If the reactor cool ant systen was at high pressure du.-ing core mei tdown , then the mol ten core naterial s would be ej ected und er pressure from the reactor vessel when it f a i l ed . In Section A.2. A, tne pne-nonenon that could occur when molten core debris is ejecte1 from the reactor vessel under cressure is discussed. Since more core debris was predicted to be ejecteo (SARP model), the resulting pressure / temperature loads in contain-nent were correspondingly hi gher , in support of NUREG-llM , SARD nas al so perforned an uncertainty study which exanines the range of possible core slunp benavior and attaches a low likelihood to the high core-nel t fraction Slung model .
If the reactor cool ant system was depressuri ud during core mel tdown ,
tnen the core deoris would f all under gravity into tne region below the reac-tor vessel after it fai'ad. Coviously, i f nore core decris was predicted to fall into tre pedestal regi on , then the resul ting noiten pool would 5e deecer and there would oe a greater potential for the core noterial s to flow across the d rywell fl oo r and reacn and fail the steel Containment liner. This was iden ti fied as a mechanism for e s rl y loss of dryuell integrity snortly after vessel f a il ure. Tnis potential f ail ure mode was allowed for in tne SARP con.
tainnent event trees out nnt in tne 10COR analysis (refer to Taale A..
I CN nas recer tl y sunni t tM an anal ys i s l5 uni c n indicates trit tne steel liner can survive for several ninutes aven witnout d rywel l sprays. The (ej questions are whetner tre debris is coolaole and unat is the ul timate deptn of deDris wnicn contacts tne liner.
From the above di scussion , i t is t,' .ar that di f f er #Ces between the IDCOR and SARP nodels for core si ung and vessel f ail ure are significant and do nave an important influence on :ne potential for early containment f a il u.'e. These di f ferences etntrioute to tne di f ferencas in pronacilities of early contain-nent f a il ure 111 oc a ted in tne ILCOR and SARP containment event tr-as (refer to Ta bl e A.7 ) .
Tni s is in area of chenomenological uncerta t atj wi tn very li t tl e meri-rental e tdence to support eitner the IDCOR or SARP nodels. Botn ,onel s are crH mle and sean tn 'ndicate the range of possible outcom e s nf a core-nel tclown aCC 1Rn t .
O 56
d l
A.2.3 Co'ntainment Failure Because of In-Vessel Stean Exol'osions (Issua 74 v The potential for an .in-vessel steam explosion to occur and generate a missile capable of failing containment was investigated by a group of experts '
and the results were publisned in NUREG-1116.16 The conclusion of tnis expert group was that the occurrence of such an event ha's a relatively low probabil-ity. These results are reflected in the SARP containnent event trees. The 10COR event trees allocate zero probability to. an in-vessel steam explosion resulting ir containment failure.
A.2.4 Direct Heating of Containment (Issue 8)
This was identi fied in the SARP analysis as an area of significant phe-nomenological uncertainty related specifically to core meltdown with the reac-tor cool ant systen at hign pressure. If molten core naterials, were to be ej ected from the reactor vessel under pressure, e xpe riments !' at Sandia l Nati onal Laboratory (SNL) have indicated that tney could form fine aerosol s ,
which mignt be dispersed into the containment atmosphere and directly heat it. An additional concern was the nxidation of the metallic content of the core debris. These reactio'ns ar e exothermic and would add an additional heat load to the containment. Exothermic chenical reactions were less of a concern in Hark I containments because tney are inerted. However, the zirconium-steam reaction could still contribute to containnent loading. l 1
The pressure rise in containment because of direct heating is directly proportional to the quantity of core debris dispersed from the reactor vessel . l Section A.2.2 noted that the SARP analysis predicted sigoificantly nore debris '
release at vessel tailure than tha IOCOR analyses predicted. Thus, the poten-tial for early containment failure because of direct heating was considered in the SARP analysis, but it was not considered credible in tne 10COR analysis. l The assumption tnat all tne core debris is released at vessel failure (SARp 1 an al ysi s ) is clearly conservative (SARP assumption) with respect to contain-ment loading. In addition to the pressure loads imposed by the dispersed core materials, tnere is the concern that the hot core debris could contact tne steel containnent snell and rapidly. f ail i t. 1 Therefore, gi ven tne present state of phononenological uncertainty asso-ciated witn this issue and the existence of relatively simple mitigative solu.
tions, the following points are of fered. There are two ways of potentially nitigating the ef fects of a high-pressure meltdown. The first is to convert a I nign-pressure sequence into a low-pressure sequence hy ensuring tnat ne auto- !
natic depressuriza . ion systen ( ADS) i s activated. Note that venting may also be required for sona stu ences (e.g. , station blackout) to ensure that the containment pressure doe., not increase beyond the rel ie f valve C3paDilitj, The second way to nitigate a hign-pressure nel tdown is ny drywell spriys.
Crywell spray operation nay reduce the pressure pulse associated with direct heating, an 1 flooding the drywell floor may impede tne dispersal of the core deoris and reduce the potential for the core to attack tne containment wall.
In addi tion to reducing pressure and c nol in g the core denris , the d rywell sprays will aio in decontaninating the drywell atnosphere and may substan-t1al l y reduco the released f,i ssion products even for cases witn drywell fail-
. ure, j
V 57
A.2.5 F v-Vessel Weat Transf er Medel F rem Molten Core to Concrete (Issue in)
This issue was of concern to "ark I containntents because heat transfer from the top of.noiten core materials (on the drywell floor) directly heats the drywell atmosphere. Thus, di f ferences in tne assumptions for heat trans-fer from the top of the core debris resulted in significant dif ferences in the pred ic ted drywell atmospheric pressures and tenperatures. The IDCOR nod el transferreo more heat f rom the top of the core deoris than the SARP nodel .
Thus, IDCOR predicted much hi gher drywell temperatures than tne SARP analy-ses. However, because 10COR predicted high heat transfer from the top of the core debris, the concrete erosion velocities were much lower than the SARP predictions. Lower concrete erosion results in less gases and aerosol s re-l eased from core-concrete attack and thus lower pressures in containment.
Therefore 10COR predicted containment failure because of high drywell tempera-tures, whereas SARf precicted containment failure because of hign pressure.
In addition,10COR predicted nuch longer times to f ail the containment because of overtemperature than the times predicted by SARP to fail the containment because of overpressure.
Di f ferences in the predicted drywell pressure / temperature histories in-fl uenced the potsntial for suppression pool bypass (Issue 13A) and containnent perfornance (Issue 15).
A.2.6 Sucoretsion Pool Byoass (!ssue 13Al If the fission products pass through the suppression pool, botn IDCOR and SARp pred ic ted significant retention of fission product aerosols in the water.
The anount of retention oepended on seve'ral factors such as submergence, water temperature, aerosol particle size, carrier gas composition, and otners. Tne hW ability of the fiarx ! suppression pool to trap aerosol fission products .was found to be an important mi ti gat iv e feature. Tnus, any pathways tnat night open , wnicn would allow ne fission products to bypass the pool are very un-des.r3ble. The following are possible ways in which the suppression pool naj be hypassed:
. lass of 1rywell isolation f ail ure 3r s acw nee ws netu+en tne drywell arn wetuell f ail ure ut drywell penetrations hecluse of hign teoperature structural failare of One drywell DeCause of high oressure f ailure of the drywell wall as a result of contact witn noiten core nater-ials Because of tne inportance of tne suppression pool as a ni ti gati vo fea-ture, the <ulneracili ty of a ftarx ! containment to any of the above nypass patnuays was caref ully assessed. The degradation of thP d rjuel l penetrations oecause of n'r;1 tenveratares in the SARP analysis was based on the sort of tN Cpmi, anien had significant R' L inuut. Fa il ur9 of tne containment m el wal' as a resul t or contact wit, mol ten core naterial s nas been identitled as 3 po t.en t 1 il fitlare nude hj "reeno, Perkins, and %t ge ld and the revi sed sm event tree s 'jave it 11000 i W 11(elinood for station bl1Ckout wequenCdS.
Gi /on inA lack of rel e van t e xperinont s , the authors were unabl ? to rulo not liner nelt-through as a potential cause af early containment failure. Thus.
5 action a.1 cevoleps a wt af deterni ni st ic att riout es to reduce tv likelt-nood of liner nel t-tnrough as well .is other suppression pool bypiss nodes, sd
i I
. l A.2.7 Containnent Performance (Issue 151 V The response of ' a liark I containment to severe-accident loads is uncer.
tain. In Section A.2.5, it was noted that IOCOR predicted very high drywell !
temperatures and thus predicted containment failure because of overtenpera-ture.
10COR assumed that a relatively small opening would occur wnica would l allow gradual leakage nf the drywell atmosphere to the reactor buildino. By '
compari son , the SARP analysis also allowed for the possibility of primary ,
containment f ail ure because of overpressure and assumed an opening large {
enough to rapidly depressurize the primary containment. In addition, the SARP '
analysis allowed for degradation of drywell seals because of hign tenperatures (but lower than the tenperature used by IDCOR as its failure criterion). Se al degradation was assumed to result in a gradual leakage from the drywell in the SARP analysis. This was again based on the work of the CPWG.12 Dif ferences in containmentI performance can influence the tining and quan- L tities of fission products released to the reactor building (refer to Section A.3). However, tnese di f ferences do not lead to major di f ferences in t*e pre-dicted overall risk as discussed in Section A.4 A.2.3 Secondary Contairnen't Perfnrnance (Issue 16)
The secondary containment (reactor building) surrounds the prinary con-tainment and has the ' potential to trap fission product s relea sed during a severe accident. There were di f ferences between the 10COR and SARD analyses with regard to nydrogen combustion in the secondary containnent. These di1-ferences resul ted in significant dif ferences in tha amount of fission penducts predicted to be retained in the secondary containnent. Tni s issue is di s-cussed in more detail in Section A.3.6 t
A.1 Fission Product Release Section A.? id en ti fi ed potential containnant f ail ure modes or fi s si o n-product release catns appropriate to tne inportant co re-mel tdown accident .
- sequences identi fied in section A.I. The purpose of tnis section is to pro- l vice in formation on
- ne timing and anount of fission products released frco tne danagc1 f;el is .,el l as tne subsequent deposition and ratention of these fission products along the release patns identi fied in Section A.2. Tne IOCOR and SARP analyses for tne Peach P,ottom plant were used as the nasi s for tnese cal c ul a ti ons .
In order to review differences in approach, tne IOCOR and NRC contractor anal jses (perfomed for SARP) are ca, pared in Tabl es A.4 and A.10 tor ATHS and station blackout sequences respectively. The (DCOR netnods predicte1 sligntly h i,;ne r relsases of the more vol a ti l e f i ss i on- p rod uc t groups (1edine and cestun) wnereas the SARP methods predicted much nigher releases of the less-vol atile tission-procuct groups f st rontiun, l anthanun, etc.) . The masons for the iti f ferent uredictions in Taoles A.9 and A.10 are complex but were clearly identi fied during tne nonerous 10C0iUNRC neetings and they ire incl un eo in tne list of 18 MC/IDCOR i ssues in Table A. A. Out of the 13 issues 6 are per-
- tinent to f i s s i on-p rod uc t release and t r a n spo r t . Hnw v e r , nnt all of tne6 are najur ccotributors to tne dif terences in Tables A 9 and A.10 F. a c h Jf
- ne 6 1ssues is d); cussed in the ful \0 wing subsec t ions, 0 !
A,3,1 Fissice Dreduct Release Sefore Vessel Failure IIssue 11 This is one issue tnat does not contribute significantly to the di f fer-ences between tne IDCOR and SARP analyses in Tables A,9 and A,10 Botn stud-ies predicted similar releases of the nore volatile fission products during in-vessel core degradation with the exception of tellurium (Te). However, a recent report by 10COR assessed the inpact of Te treatment and nodeled similar i n-ves sel Te releases to the SA'1P analyses. 01 f f erences in tne predicted environmental raleases of Te in Tables A,9 and A,in are, therefore , not he-cause of dif ferences in the in-vessel Te release and retention models, but are due to dif ferences in the amount of retention predicted to occur in the sec-ondary containment (also refer to Section A,3,6),
A,3,2 Fission-Product and Aerosol Retention in the Reactor Coolant Systen (Issue 41 Di f ferences in the ini tial reactor coolant systen retention predicted 5y 10COR and SARP were again not too significant and dif fer by less than a f 3ctor of t.vo, Tne inportant di f f erence between the 10COR and SAPP nodels was tnat in tne SARP analysis, fission products retained in the reactor coolant system at the point of vessel failure were pernanently retained, unereas in the !0COR analysis , revaport :3 tion of these fission prod uc t s ifter vessel f ailure was mod el ed , This is discussed in more detail in Section A,3,4, A,3,3 E x-Vessel Fission-Product Release (Issue 4)
There were si gni fic ant di f ferences between the IDCOR and 'SARP an al ys es for f i s s i o n-p rod uc t release as a result of core-concrete interactions. The nigner releasas of tne s t r un ti um , lanthanun, and cestun groups in Tanl os A,4 and A,10 in tne SARP analyses were b ec ause of the nooeling of e x-ves sel fi s si on- p rod uc t release, The potential for f i s s i on- p roo uc t release and inert aerosol generitton dur!ng core concrete interactions was not nodeled in :ne 10COR analysis of De3cn aa tton,19 [DCOR argued that by noceling tne aerosol generation ducirg core.corcrete interactions , the increased aerosol densitj in containment wo oli1 in:cet;o terosol aqqln,eratian and settling, tnus rad uc ' nij tre pred1:!9d environ,ent:1 release fractions rel a tive to those p r ed i c. t s; nitncut tni s adq i tional tee nsal saurce.
The laC@ g r ed i c ted core debris tenperaturas during core concrete inter-ac; tons were high Ind , based on experinental evidence , one would espect the release of some of t"e refractory f1 s s i on- p roiiuc t groups at these tenov e ).
tures. Tne SADP ansl /sti nod el et tne release of the refractory tissian proc.
ucts and tne inert aerosol s and the envirunnental rel ea se tractions wre not c al c ul a t ed to Se low (rerer to Tables A,9 and A,10i, A , 3,3 9eva00r12'atiqn if F1ssiqn DFO'l uc t s Frnn the Reactor CColant Syste
'(ssue i!$
sec ti on A,1,? i nl i c a t eq Nt revaporization was in aroa or myr di# far-enc e, %3 tueeq tne { dCiN inal $ARP .)nalySes, DeCall (nat $ARP does 30t model re-sapor 121!1on <)f fisstun pr )d 7C t s f rnal the reac tor c.ool in t s ys ton a t *or ro 1C ta r "
/essel f a ll aro , W%4 r o 15 ' 3C 0'l loes model (ni5 effeCt, Io obser/e the i n f l u-o n e. s nf t"9 '1*fer nt e 1pp rodc nt+ s , the distrthution of c e s l a'1- l od ' d o ICs!} 15
- racked at tarious stages of a sta*1on blackout soi;uence in Table A,ll, na i
1 l
i i
At the point of vessel f ailure, both IDCOR and the NRC contractors' anal- !
[
tg'] yses performed for SARP indicated significant reactor coolant systen retention i' of Cs! (100% retantion for IDCOR and W, retention for SARP). Inned i s tel y before containment failure, tne NRC contractor analysis has 85% of tne Csl permanently retained in tne reactor coolant systen, whereas the InCOR nodel l has revaporized 24*, of the Cs! from the reactor coolant systen. At the end of the calculation, the IDCOR nodel has revapori zed al*, of the Cs t.
The IDCOR revaporization nodel means that significently nore of the vola- l tile fission products were predicted to be released to the reactor building '
than in the NRC contractor approach in which revaporization was not nodeled.
For the sequence under consideration in Table A.11, the environnental releases were simil ar despite the major dif ferences discussed above. The reason for I the similar environnental releases was because of tne much greater retention of fission products in the reactor building predicted in the IDCOR nodel vs.
the SARP model, This is discussed in more detail in Section A,3,6.
A,3,5 Fission-Product Deoosition Model in Containment (Issue 121 This was another issue that does not contribute significantly to the dif-ferences between the IDCOR and SARP analyses in Tacles A,9 and A 10 Issues 9, 11, and 16 really drive the dif ferences in these two analyses, Kowever, tnis' issue may be of nore importance to other containment designs.
A,3,6 Seconnary Containrent Perfornance (Issue 16)
Section A,3,4 i nd i c.a ted tnat s'econdary containment (reactor building) performance was an area of major dif ference between tne IDCOR and SARP analy-( ,/ ses, The extent of the dif ference can be seen f ro'n Tabl e A.11, wh1cq shows that IDCOR precicts tnat only 6; of ne Csl entering the reactor building would ne released to the environnent compa red wi t h 20 ". rel eased in the NRC contractor analysis.
The abcVe d1 f ferences were signi ficant and were found to be bec3use of s ev +ral factors, For example, the gas flowrate through the secnndary contain-ment was h1gner in tne SARP analysis, wnich allows less tiae for aerosol set-t!ing. ln aM i ti a r , ICCN nodel s .1a tural ci rcul 3 tion patns in tne second ary containment, nnich further increased the residence times. Such patns were not mod el ed in tne 9RP analysis, Finally, SARP calculated several hynrogen buras I to occur in the seconcary containnent, wnich rapidly blew out the i tmo s ,:he re of t *1e huild ing, ICCCR Calculated nore gradual Comb u s t i o n phenonena, wniCh did nc t result in rapid blowout of the secondary containment atmosphere, 1 Inere was uncertainty with regard to how nuch retention of aerosol fi s. l l
s1;n pruducts would oc c ur in the secondary containment, Even in the SA1P analysis, inere was the potential for signi ficant retention.
A,4 Offsite Cansequences In tni s sec tion , the potential of fsite consequences of the severe acc1-dents <1escriaea in tre' previous sections are exanined. There is ano %C/[nCntt l 1ssue rel a tej to of f site consequences , wnich concerned di f f erences, in tne issu.%1 ev acua ti on nodal s. Di f ferences in the evacuatino modal influence tne pr eli c ted early heal th effects. The issue was largely resolved and was v
61 e
rel ated to tne froction of the population asWned not to participate in tne evacuation.
Table A.12 gives the person-ren calculations perforned by 10COR 20 and in NUREG-1150 h for several accident sequences and f ail ure modes. Inis taol e indicates that if the containnent is predicted to fail (either early or late) and the suppression pool is bypassed, then the offsite person-rec 1 predictions are similar (within the range 0.1 to 3x10 ) for 7 the accidents considerea. The only time that a significant reduction (to 4x10 s) in person-ren was calcu-lated , was with successful wetwell venting and no pool hypass. These resul ts clearly show that wetwell venting and the prevention of pool hypass are impor-tant in nitigating the fission-product releases for a liark I containment.
A.S Sunnary and Risk Insights A.S.1 Core-Danage profile As has been observed by others for BWRs,1 3 transients rather than LOCAS dominated the core-damage ri sk profile for the studies exanined in Section A.I. Otherwise, there was no consistent pattern of relative ranging of tran-sient sequences among the studies, it is also inportant to otserve that for a given accident sequence in Taole A.1, the major contributor to dif ferences in quantitative results cetween the studies was hecause of suojective nodeling assumptions ratner than plant di f ferences or data dif ferences.
For tne four BUR plants considered in the si x PR As , the same f ew acci-dent sequences fi gured prominen tl y in all of the respective core-damage fre-quency pro fil e's. This suggests tnat if tne probability of tnis relatively snali subset af acctdent sequences can he mininized, tnen tnere is a reason-abl e expectation that the overall c o r e-d ana ge frequency will oe ninint zed.
This principle is used to develop pl an t features , operator actions and asso-ciated attributes to redace tne overall core-damage f requency (Goal 31.
For the PSS Peach dotton stucy and the Grouns Ferry IRE 3 study, loss of cont 31n,ent ,ea t r.-nov al sequences (e.g. , Tu and TDQ1) w re inportant contrin-utors to core mal t lanout 50"!. Al t nou gh the nore recent .B E P snd 10CJR stuc-ies nave requce1 ese secence frequencies nasea on opera t i ng pr xequres < 3 r venting and il terna .1ve inj ac tion , de ter"1t ni s tic a tt ributes *1 ave teen identi-f i ed in Sec ti on 4.3.2 to ensure tnat these sequences are not d on n an t for other BUR itark I pl ants.
For the Linerick review, Rrowns Ferry IREP, and Shorenan review, Tauv/
TQhx sequences were inportant contributors to core nat t, f *0 s t of Enis con-tribution was because at a hign f ailure rate for reactor cool 3nt s ys t en d e-pressurization. Improved A05 reliaoility is addrassed in Section a.~i 3 It i s oportan . to reengnize that the quali tati ve iccident sequanca o-sc-1ptors ire rather general and Droad and tnat different nardware anaior nyerational ' 311 o res la the v a r i o u s R,.6 4, "tr( l plants could l a id ta the sdne generli 3 C C I'le n t s e q u e r'C d . [n Order to identi f y Ene )IJnt-s Wc1fli. (and ottwn un i co ) ml r'er m 1I i t1"s (na t contributa to a 91 von genaard ! 5equenC" rie-sc ri ptor Ie.,,, st3tton nl3ckuut) in 3 given pidnt, a pl dnt-speci f ic e uni na.
tlun Isuch 15 a ? ] t l Jre "ude ind effects analys1% C oupl t With a ' Tul t tree /
event tree analysis or an equivalent netnod) would he needed, u
l A.S.2 Consecuence Analysis The assessment of core-neltdown pnenomena .and containment response indi-cated that the Mark I containment is vulnerable to severe-accident containment loads. Unless mitigative actions are taken, a Mark I containment.has tne po-tential to fail in a short time (a few hours) after the reactor vessel fails.
If containment failure occurs in the drywell, any fission products in the dry-well atmosphere could pass to the reactor building without the benefit of sup-p' ression pool scruobing. Because of this vulnerability, tn'e predicted offsite consequences were relatively insensitive to the accident sequence definition, in . addition, differences in tne 10COR and SARP assessments of containment re-sponse and fission-product release also did not result in najor differences in the predicted of fsite consequences. The only time that a major reduction in offsite consequences was predicted was with successful wetwell venting and no pool bypass. For station blackout sequences, the dominant f ailure nodes were failure of the drywell shell via" contact with core debris or overpressure because of the builduo of noncondensable gases. Thus, both of these failure mecnanisms must be reduced substantially to ensure mitigation of fission prod-ucts (Goal 1).
A.6 References
- 1. "Reactor Safety Study: An Assessment of accident Risks in U.S. Conner-Cial Nuclear Power Pl ants ," U. S. Nucl e ar Regul atory Comni ssi on , WASH-1400, HUREG/75-014, October 1975,
- 2. "Ri sk Reduction Potential ," IOCOR Technical Report 21.1, June 19A;.
l j 3. A. N. Kolaczkowski et al . , " Analysis of Core Danage Frequency Fron Inter-nal Events: Peacn Botton, Unit 2," Sandia National Laboratories, NUREG/
- CR 4550, Vol ume 4, Octooer 1946, 2 I. A. 03pazoglou et al., "A Review of tne Limerick Generating Station 9rcoacilistic Risk Assessnent," Brookhaven Naticnal Laboratory, NUREG/CR-
! 3023, Feoruary 1933, i
) 5. S. E nijs et al . , "interin Reliacility Evaluation Progran: Analysis of tn.e Browns Ferry, Unit 1, Nuclear Plant," Idaho National Engineering Lab-orato ry , NUREG/CR-2?.02, July 1982.
l
- 6. D. Ilterg et al ., "A Review of tne Shorehan Nuclear Power Station Proca-oilistic Risk Assessment ( In tern al Events and Ccre Damage Frequency)," )
l Brooknaven National Laboratory, NUREG/CR 4050, June 1435.
7 W. J. Luc kas et al . , " A Human Reliability Analysis for the ATUS Accident Sequence Witn MSIV Closure of the Peach Botton Atonic Pouea Station,"
3rnoknaven Na tional Lanoratory, Tecnnical Repnet a-3772, Acril 19R6 '
.2 2
- d. "9 onab11istic Ris( Assessrent -
Shoren an Nucl ear Douer Station," Long isl and Lighting Cumoany June 1041
- 1 3 i
Lan, "herpressurization of Emergency Core Cooling Systens in Railing.
ua ter Peactors ," 2E00 (NoC) , February 1985 5
63
10 J. D. Harris and J. W. fiinarick, " An Evaluation of Q.UR Overpressure !nci-dents in Low Pressure Systens ," Oak Ridge National Laboratory, ORNL Pre-lininary Report, fiay 1085,
- 11. "Estinates of Early Containment Fail ure Fron Core-i'el Accidents ," Jon-tainnent Loads Wo'<1ng Group, NUREG-1079, Oraft Report for Connent, December 1985,
- 12. "Containnent Perfornance Working Group Report " Containment Pe r f o rmanc e Working Group, NUREG'-1037, Draf t Report for'Connent. May 1985.
- 13. J. W. Yang and W. T. Pratt, " Analysi s of Hyd rogen Production During a BWR-6 Core Heatup Transient," Brookhaven National Laboratory, Tecnnical Report A-3808, September 1985.
14 "Reactor Risk Reference Document," Volume 3, "NRC Staff Posi t ion on Di rect Containment Heating," Appendix J5, U.S. t:ucl ear Regul a tory Conmission, NUREG-ll50, Draft Report for Comment, February 1937,
- 15. " Approximate Source Ter . t'ethodology for Roiling Water Reactors ," Fauske
& Associates Inc. , FAI/36-l Decenner 1086.
16 "A Review of Current Understanding of the Potential for Containment Fail-ure fron In-Vessel Stean Explosions," Stean Exposion Review Group, NUREG- I 1116, June 1085.
17 W. Ta rbel l et al . , "Pressuri zed fiel t Ej ec tion into 59aled Reactor Cavi. !
ties ," Sandia National Laboratories, NUREG/CR A512, Octooer 1986 !
- 18. G. A. Greene , K. R. Perkins , and S. A. Hodge, " 'np ac t of Core-Concrete Interactions in tne t' ark I Containment Drywell on Containment Integrity ano Failure of tne 'lrywell Liner ." Paper IAEA-St'-231/36, Presented at tne l I n t e r n a *.1 o n al Symposi a on Source Tern Evaluation for Ac c id en t Conni- I tions , Col unous , On10, Novencer 1985, i
- 19. "Deacn Ro t t cr' Atemic Power Station-Integrated Containnent Mal yses ,"
[JCOR Te rnic d hp]rt 723.1?3, March 13 5,
- 23. "Tecnnical Suoport *or Is s ie Resol ution ," IDCOR Tec nn i c al Repart TAS.?,
July 1986 O
l l
[ _
l i
s J
i l
Aesc::r evilcing (sec ncary C0rit34Mrnent)
Crywell 2 closure cas
-+. , n -
L
^ i '
l
= _
7,1._ _
\...... / ; .
g, , , {
snieio wari . * :f.
- 0. .-
I g.,
Ny y .Reacter e,...
w*
/*/ p /. : .,.,,,,
j y Pecestal p......' - -
70 h ;
"" " * " 9 s l
.,2) l ,
y:IN. interau g
=== neg . exer
/, O
.; ,V C ew a ,,. j /I/ onne:r ers
,,. b ), ) 74Nt suceressico
.I .._.1 I r e.,, - ,
I
'N t =. u u - e,
- .2
= y- j j 3, 0 I
-ls....:u,e i
u::ression 20C1 l
1 Figure A.1 BWR Mark I containment configuration.
65
.i
. _ _ ._ . _ _ . .. . . . _ . . _ _ . o _ -._ .
Table A.1 DWR Comparisons: Core-% mage Freq0encies 10COR RSS ASEP Task 21.1 Limerick Browns Shoreham Sequence Peach Peach Peach PRA Ferry PRA Type Bottom Bottom Bottom Review IREP Review TW 1.7x10-5 1.0x10-a 1.5x10-7 3. 2 x10- 6 1.0x10-" 9.0x10-6 j (1.1x10-5)
TC 1.3x10-5 1.0x10-6 7. 3 x10- 6 3.7x10-6 5.5x10-" 4. 5 x10- 5 TQUV 8. 4 x10- 7 6.8x10-8 4.1 x10- 8 6.0 x10- 5 5. 5 x10- 7 5.0x10-5 and (6.8x10-5)
TQUX TB 1.0x10-7 7.0x10-6 4. 5 x10- 7 3.1 x10- 5 2.9 x10- 5 1. 3 x10- 5 TPQI --- --- ---
9.0x10-7 1.1x10-5 1.7x10 7 TPQE 6. 0 x10- 3 --- --- ---
-1.0x10-7 ---
AE 1. 5 x10- 7 3.2x10-s 1.6x10-8 2. 4 x10- 9 ---
3.0x10-7 At --- --- ---
SE g 2.0 x10- 7 7.5x10-8 1.2x10-7 --- ---
2.5x10-7 5E 2
- 5. 2 x10- 8 ---
- 4. 2 x10- 9 --- --- ---
SI 2 l 2X10" --- ~~~ --- --- ---
5J 2 1.1 x10- 7 --- --- --- --- ---
Total 3. 2 x10- 5 1. 2 x10- 6 d .1 x10- 6 9.9x10-5 2.0410-" 1.4<10-*
O 66
1 i
Table A.2 Sunnari nf Changes Included in the IDCOR Coanitted Core-Damage Profile for Peacn Botton Plant Modi fications
. Symptom-based energency procedures including venting, alternate fnjec .
tion sources, ATWS procedures, and using drywell sprays in certain cir. ,
cumstances
. Alternate cort insertion ( ARI) -;
86 gpn equivalent manual SLCS
. Level 1 ttsty closure setpoint
. Elinination of "high" drywell pressure permissive for ADS actuation Systen Analysis
. Fiew initiating event frequencies - l l
. Increase in operator error contribution to achieve subcriticality
. Lower scran systen reliability
., . Cognitive operator error considered
. Itore realistic RSS Case !! success criteria *
. RHR-HPSW intertie, condensate, and CR0 punps included in some sequences
. Revised treatment of PCS availability and alternative nethods of heat renoval
. Containment failure nnt assumed to always resolt in core damage
. ::ew val aes for recovery of of fsite power .ind diesel generatur relianil-ity anc recovery
. Battery depletion considered in station blackout sequence
- The RSS gave two sets of success criteria for coolant injection. The more stringent Casa I criteria were used in the RSS calculations. In the 10 CUR aralysis tne less stringent Case !! criteria were considered nore re'alistic and used for 'octoining the core-damage profile.
67 *
~.
Table A.3 Conditional orobabilities of Core Canage Given an ATliS in Peach Bottom vs. Shoreham Analysis Turbine Trip ATWS 3 Isolation ATWS D 10COR Committed -
. High Power 0.067 0.317
. Low Power 0.015 0.160
. Conbined 0.053 0.261 ASEP/SARP Negl i gibl e 0.032 Shoreham PRA
. High Power 0.248 0.893
. Low Power 0.038 0.620
. Combined 0.121 0.831 Shorenam PRA Review
. All Power Levels 0.Sd6 0.957 aTrantients in which PCS remains available, b includes transfers from turbine trip with early closure of fiSIVs.
t l
O 68
_a
- . . - - .-~ . - ,
i feele A.4 79 Seeveacoes Comeerisene -
' seco,,e Recov er Mign Pressere RCtC/ W 1 Of t site Po.or leou sat e LOCp Otisite Po or i nj ection in uenust ' (Givea u 1 Core Ceae;*
S tue v 7 0 0 0004 CGttC Control Q F r eq u enc e [
RSS CJ 2.01 2.03 1.03 0.0 4
.*eech Botton (l/2 nr1 791.07
- ICCD' O.44 l.73 0 .72 0.$ 3.7 2 2.7 7
.**erS Botton () nc) (6 nes) (9 hest r3 4.4 7 2.1 2 0.44 1.73 0 .72 swcces s(0.33 1.96 2 1.47 (12 mest 0.44 2.03 1.75 4.0 4 alte/5*** 7.5 5 54 6 7986.7 6 0.4 (04 et de power systus) l t *1 hel Success 2.3 4 0.5 0.1 3.4 7 io, of 2 Desi i6 to n,s> i. ,o . a,si 7.32 0.4 Svec ess 3.15 0.1 8.44
($u Slagle failure >
0.4 Soccess 1.34 0.6 0.4 2.37
( I nd eo cad ea t (6 to 4 nr.1 Failure of 2 OGs1 0.4 Svetest 2.34 0.6 0.1 4.47 tt OG Failure ami Otner in assinton.
ence) 0.4 Succ es s. 3.3 4 0.6/0.5 0.1 5.5 7 (Co.elastion et Failures in tae (lej ems 0G Fea tures t I
u .e i e. - > w .0. i... 3 0.4,ia. ,s> 3.i. ...a., 2 .i . , ;
- '8 ') .t ? 0.66(2. arts AmCe 0.11 1 .86 5 4.6 4(2.ari 4.J 6
'4* ! .i . t 0.54 0.13 1.86 5 0.9 5(1/2.a r i t .3 4 l l
See*** 'erev 1.32 Svec ess 2.9 2* 3.0 2t6 'o a nest 2.65 i e(se r .1 5 5 .J.2 4.2 2(DCIC) 2.92* 3.02 1.16 I***aa** *** 7.74 3.6 5 2.1 6 () 0.* ri , ,$.:11J.*ri
' s.14 e m i e. ceCIC/ e l (2.owt ce.5p
?8at .3 5 act t et t es et 7.19 3.3ft1/2.ari 10 ar bu t essweed t e l l ed after 10 nel I.12 2.3 3 tall 3 OCs> i.46
etivre oe atteraere soo. ha l ag . -
' b elast4oas of 3 steseis.
. 69 1
f est e 4.9 fe See we eta Came eri neas as e e,... C i nj ec tion c e, %et ee. Felle se ter Cecer met nev e t t Cea' e l m Cea' e t a**** 1eessat e Cegal fl e o eCl secoveres M3 eece'reres Care Caseqe teri, Late 8Cl f or chi e el (Cagi t e en, s ea t i ng 'ei sere f r e s i e** . ween treer
" I P#*****'
l'we r 7, se Og 02 Od2 7.0.) 4 J.4 I .t.1 est i. t .a s o
.eesen ansees i e i .t.) 2 JH2 !J4 S J*I f *0J L
S J+4 IW f *0,111 Cs
.aoes e tot +an l.01 3.34 to.i .l 7 f 0 .32 7 1.04 2.b4 f,.o J e g 1 J.4
', *4 .1 4 i.04 2.02 t J-2 9J 2 2 J.4 I.02 I .J.4 I .l .4 f = 0 .9 31 i .0 6.01 4.44 2.34 f .0 2 t .0.e e.3.i t f,,*0.32 ? 1.e.6 1.04 9.b3 t .6 2 ! J-4 t .0 2 f .J.f D J.4 0
'e=0 J e t
% *8'
.s oece %,ean t see .J.4 0.19 1 J .4 9.44 I J.4 4 8 *** ' 4 8 ** 4 I,*' dl 0 .44 so s ee
? ,0 2 3 ,s . f
'*1J4 7
- 1.i t 7.02 4 .S.)
s 9,g. t t .* .4 f . ' t .i F ttets se sei i ,0 g.7 5 g ,9.g
- * , *
- t .?
l' '* .e ' ** ' s i .sg ei
' es t , ).4
,.o.= i g ., g g ., 4, .g ; , ,,,
N oa m *e a
- e. i ee
'**** ' a 1.. % 3.02 3 f.2 1.14 i,..,
6 J .i i J .a i ,i .4 .g.,
e.n W i C o- .67 t .3 2 1.t . ) 4 ,4 . g i,..,
','*.>
- J .J
.p
'g') .e 8 e .)-4 3.02 1J2 e .4.)
.. e ei ..
w'* I ;.e hMI% c. SoIS *I ef e9 !4 M g O
70
Table A.6 TQUV and TCUI Sequences: Ceparisons
\
/ Migh Pressure Dooressurite- Loe Pressure $eo'wenc e g transients Fw/PC5 Inj ec tion tion injection Core Cesege i o u x
. 5 tway I v Freawency 45$
[ T gelo I.02 2.0-3 1.S-3
- 3.0-7
-peecn Bottaa i 1.0-2 2.0-3 1.3-3 , 3.0-?
7 =0.2 0.2 2.0-3 I .5-3
- iJ 7 g
2.0-3 1.5-3 I .2-7 10 Cat 4 t 0.537 1.0 2.4 5 3.7-6
.s e ,cn gotton
- C0F=6.S-4 ([TQl *0X T,,*0.32 7 1.4-l 2.0-3 T*=0.245 e 3 .0=1 2.5-6 4.4-9 7 e6.18 2.0-2 t [fQ) .VV 7
45(8/ $ AA8
.seecn Botton
- 0F e2.4 7 '
Limerics pe4 f a t .2 3 0 .61 8.1 -3 6.0-3
- eevie.
' 3.7-5 COFe6.0*S T *0.17 4.1 -3 6.0.) 6.3-6 7 6,17* 2.02 g ,1 3 6.0-3 8.0-6 s
i Brosas Ferry 7 et .7 t18(P) 1.0 t.8-3 I.44* 1.$.1 ZFe$.S.7 Swene NA f 7*4.0 S.22 1.02 4.44 S.$.6 wevie.
00F=6.8-5 f *0 .5 t .0 ) .0 2 4.44 4.26 f,'O.t! 4.51 1.02 9.4 4 2.16 3'4.3 3.0-2 1.5-2 8.4 4 i.g.6 (CCP/u q t e; L ov el 1 y ,g. g i Floodinq 8 1.35
- 09. (*)/0e Cool 184] 1 J.;
Loss et fB 1m S g.g Loss of DC 24 6
%e %es 2eoressur41stion.
5
%e I sa % i490
'travestiollte, t .e,= .4 5mi aco .
1av y 0.or.gg,ragetlen.
- a 4se si sat %)a00 (2.8 48 am con eesete laj ection (1.0-2 3.
' t e41 t Pr) cutsets 49 provi4*f.
71
Table A.7 Cceparison of the 10COR and SARP Containment itatrices Station Blackout ATUS Containnent Failure fiode 10COR SARP 10COR SARP No Containr.ent Failure .72 .06 -
.00 Wetwell Venting or Overpressure -
.17 .78 .38 Late Temperature-Induced Leak in Drywell .28 .02 - .00 Drywell Breach f rom tielt-Structure Attack - 40 -
.03 DW Overpressure Failure -
.20 .01 44 Late Pressure-Induced Drywell Failure with Core Quenched - -
.12 -
Containment Venting with Loss of Drywell Integrity -
.15 .09 .15
. O 72
l Table A.8 NRC/10COR Issues
(-
issue Subj ect 1 Fission-Product Release Before Vessel Failure )
2 Recirculation of Coolant in' Reactor Vessel !
3 Release Model of Control Rod Materials I 4 Fission-Product and Aerosol Retention in the Reactor Coolant System ;
5 In-Vessel H2 Generation l 6 Core Slump, Core Collapse, and Reactor Vessel Failure 7 Containment Failure Because of In-Vessel Steam Explosions 8 Direct Heating of Containment -
9 Ex-Vessel Fissior,-Product Release 10 Ex-Vessel Heat Transfer Model Froni Molten Core to Concrete 11 Revaporization of Fission Products. From the Reactor Coolant System 12 Fi.ssion-Product Deposition !!odel in Containment 13A Suppression Pool Bypa'ss (Pool Scrubbing) .
13B Retention of Fission Products in Ice Beds 14 Moceling of Emergency Response 15 Containment Perfomance 16 Secondary Containment Performance 17 Hydrogen Ignition and Burning 13 Essential Equipment Performance (o) i l
l l
I l
G l 73
-. ]
Table A.9 Comparison of 10COR and SARP Predictions of Fission-Product Release for an ATWS Sequence Witn No Operator Actions Taken Event 10COR* NRC Contractors **
Containment failure (hr) 1.4 1.4 Start of Core fielt (hr) 3.0 2.2 Vessel Failure (nr) 3.9 3.8 Fission-Product Release Fractions *** :
Xe-Kr 1.0 1.0 1-Br 0.1 0.03 Cs-Rb 0.1 0.03 Te-Sb 0.1 0.26 Sr 0.0004 0.49 Ru-Mo 0.001 Ne9 La -- 0.01 Ce -- 0.02 Ba --
0.39,
- IDCOR Technical Report 23.1PB, Maren 1985.
- NUREG/CR-4624, Vol. 1.
- Fraction of Initial Core Inventory.
O
! 74 1
l l
l l . - m__
Table A.10 . Comparison of 10CORa nd SARP Predictions of Fission-Product Release for an SB0 Sequence i
Event 10COR* NRC Contractors"
, Loss of Injection (hr) 6.0 6.0.
3 Start of Core fielt (hr) 11.4 10.7 1 i Vessel Failure (nr) 12.0 12.2 Containment Failure (hr) 18.0 15.2 Fission-Product Release Fractions ***: L xe-Kr 1.0 1.0 I-Br 0.05 0.012 Cs-Rb 0.05 0.014 Te-Sb 0.06 0.22 Sr Ne g. 0.37 Ru-lio 0.0001 f49 La --
0.03
. Ce --
0.05 - '
Ba --
0.28
- IDCOR Technical Report 23.198, liarch 1985. '
- NUREG/CR 4624, Vol. 1.
"* Fraction of Initial Core Inventory.
O ;
i i
l l
1 i
1 l
O 76 i
l
._ _,. - - - . .a
Table A.11 Comparison of IDCOR and SARP Predictions of Cal Distribution for an SB0 Sequence (Fraction of Initial Core Inventory)
Events Containnent Enc of Vessel Failure Failure Calculation 10COR* NRC** 10COR f4RC 10COR i4RC Location (12 hr) (12 hr) (18 hr) (15 hr) (60 hr) (22 hr)
Reactor Pressure Vessel 1.0 0.74 0.76 0.74 0.09 0.74 Drywell -- 0.12 0.20. '" -- 0.09 liel t Suppression Pool -- ***
0.04 ***
0.04 0.14 Reactor Building -- -- -- --
0.32 0.009 En vi ronment -- -- -- --
0.05 0.014
- IOCOR Tecnnical Report 23.1PB, tiarch 1985.
"NUREGCR-4624, Vol . 1.
- "The location of the remaining f raction of Cs! was not reported in the Battelle Columous Lacoratory's report. -
O 76
. - l
Table A.12 Comparison of IDCOR and NUREG-1150 Consequence O Accident Results (Person-Rem)
Sequence Containment Failure Mode 10COR NUREG-1150 ATWS Wetwell venting no pool bypass 3. 5 x10 5 ..
ATWS Wetwell venting with late pool bypass 1.0 x107 -- - I Station Containment failure at RPV failure --
0.2 to 3.3x10 7 ;
B1ackout i t
7 Station Containment failure after a few hours 1. 3 x10 0.1 to 1.2x10 7 Bl ackout l
O f
I l
O 77 e ^
, , _ - , _ .was.. am ---44aa._h - __ E .. -4M.hM--4--Amm 6_a_ .--m C .w-- ama- ...he-s-. h_a44-h_.A .22m m -- - - J.. e_m..ea_Whe_ m a. m mm., ma am4_.
I I
l I
)
i l
i 9
i
.i I
l i
O 73'
. . . _ - - - -..- - , . - . - ~- - ..-
Appendix B PLANT FEATURES RESULTING IN L0u PROBABILITIES FOR ACCInENT SE0VENCES This appendix examines those BWR-4 plant features which result in insig- j nificant probabilities for many severe-accident sequences. The purpose of i i this examination is to identify the plant features that produce these low l orobabilities, and attempt tn provide some means , i f necessary, to screen other RWR-4 plants to ensure that these same features exist.
In a typical probabilistic risk - assessment for a WR, several thousand '
~ severe-accident sequences are examined. Typically, only a few (on the order j of 5 to 10) of these sequences are found to contribute significantly (nore :
than a few percent) to the total probability of core damage or risk. The !
reasons that all other sequences are generally found to oe of such low prona- ;
,bility are: (1) the frequency of the event that initiates the accident is ;
very low or (2) the protection provided by existing plant systems is hignly r eli abl e. In many cases, the low sequence probacilities are the result of !
coth of these reasons acting sinultaneously. !
l l All accident sequences exanined in a PRA are initiated ny one of two gen-i eral types of events, transient or loss-of-coolan't accidents. These two ini- i l tiatnes are generally suodivided into several speci fic classes. It snould oe -j l
noted that there is a special class of transients wnien can lead to a LOCA by virtue of a stuck-open safety or relief valve. However, tnis sequence is l still initiated by a transient, and the stuck-open valve represents a systen (component) failure rather than an initiating avent.
As indicated in Append'ix A of this report, ili ORAs exanined consistently showed tnat transient-initiated, sequences dominate core . melt pronacility.
Althougn tne T3 and TC sequences are important cont.-ioutors for tne Peacn Button and Browns Ferry studies, various utner transients 3150 snow up as sig-nificant contributors in some of tnese studies. ;
From the preceding discussion, it can h( conci d ed inat AWR a plants as {
represen t ed by Browns Ferry and Peacn Cotton acorar t, nave reliaele '
urotectiQn ig)1nst loss-of-Cool an t initi dten acci te"t s. In ill cases LCCA sequences contribute less than h to the tots! GF . ina the naxinxi ococa-ciiity for any LOCA-initiated core-damage seawca fm t*e four ins pendent studies for Browns Ferry and Peacn dotton is 1.ix10*' (3 E sequence for the ASFP Peach Rotton study).
The BWR-4 design includes nunerous systems for cravicing snergency core-cooling during loss-of-coolant accidents. These syst~,s 3re shown in Taole B.1 (taken from Ref. 1) wnich illustr3tes *.ne sjste,$ capaele nf geoviding ideyyste cooling as a function of the 3ssu%u creu 11 :3. Ine taole also snous (last column) the decay heat ranoval tacass c r' ter'a , wnt ch are inde-pea nent of assumed break si ze or location. as t edicat+1 u the second calunn, tre WR-4 design includes at laast tnree saaarate systems Te conninations of sjstens caoable of providing adequate coolant injection fol'.0 wing a LOCA for al1 nreat configbratinns wi tn tne exception of the 1 ir ;e !iauid Itre neeak
( ..nicn , as will ne noteri later, nas a lower estimated prcoacility tnan other n re n sizes). Thus , the RuR a design inc!11es cansider1Sie r ef und 3ncy and diversity (resulting in hign functional rellacility) in orotection against 79
J LOCA initiators. This design feature has co .e about as a result of NRC rego.
lations and criteria emphasizing tna need for adequate energency core cooling following LOCAs.
With respect to decay heat renoval, Table B.1 indicates that either one or two of four pungs, depending on cooling node, are required and no diverse systen exists to provide this function. However, the systen is not required for some time af ter the LOCA, and failure does not lead innediately to core damage. Thus, opportunity e xi st s for repair to the sy s t e.n , and thi s is accounted for in sone of tne PRA studies. For these reasons, loss of decay heat renoval following LOCA initiators has et been found to be a significant C0F contributor.
Table B.2 provides an estimate of reactor coolant systen LOCA break size frequencies as used in the Browns Ferry PRA.: Sinil ar values were used in the other BWR assessments. As can be seen from Table 8.2, the larger sizes have lower estimated f requencies. The relatively low LOCA frequencies coupled with ECCS and decay heat renov al reliability produce insignificant co r e-d ama ge f requencies from LOCA initiators compared to transient initiated sequences.
The general ECC systen arrangement depicted in Tabl e R.1 f ar Browns Ferry 15 expected to exist for all BWR-4 plants since this design is in response to prescriptive fGC requirements for energency core cooling capability, nocumen-l tation by the General Electric Company confims the design similarity anong BWR-4 pl an t s . Thus , i t is not expected tnat any BUR-4 plant exists wnich would nave ECCS design features that would res ul t in a significant estinated core-damage frequency froc LOCA initiators. It i s, therefore, not considered necessiry to require screening to ensure tnat BUR-4 pl ants have adequate ECCS capab111ty for ;.ireventing core damage f rom LOCA initiators.
B.1 References
- 1. " Ad d i t i o n al Information Required for NRC St3ff Generic ?eport on Bo il i n g wat(r Reactors," General El ec tric Co. , NEDO-24N A, August 19 N .
- 2. "Interin Rel i ac 111 *.y Evaluatien Progran: An al ysi s o' the S rwn s Ferry
' n 1 *.
. [, % clear P1 ant ," ER.G 11 a no , In c . , W E ro" 2- ? m 2, Ju l j '.q.
O
a t
Table B.1 Typical Success Criteria for ECCS in Wponse to bur 4 LOCA 1 Break Si ze, Type Energency Cool ant !
& Location Injection 4
Cecay Heat Renoval l Large. Liquid Two core spray loops and Two of four RHR pumps with i (0.3. to. 4.3 f t2 two of four LPCI pumps associateo heat exchangers pump suction) or in torus cooling node Eur,of four LPCI pumps or or he of four RHR pumps with Une of two core spray associated heat exchangers loops and two of four 'in shutdown cooling mode LPCI pumps (one LPCI pump per injection loop) '
Large, liquid - Two core spray loops Two of four' RHR pumps with (0.3 to 4.3 ft2 or associated heat exchangers punp discharge) E e of two core spray in torus cooling node i
loops and one of two.LPCI or i
pumps on unaf fected side Une of four RHR pumps with i associated heat exchangers in shutdown enoling mode large, Stean Two core spray loops Two of four RHR pumps with (1.4 to 4.1 f;2) or i
associated heat exchangers
' E ur of four LPCI pumps in torus cooling node or or We of two core spray We of four RHR pumps with -
' locos and one of four
- associated heat exchangers LPCI pumps in shutdown cooling noce In t erned ia te , One of one HPCI punp Li quid Two of four RHR pumps with or '
associated heat excnangers I' (n.12 to 0.1 ft ) 2 Eur of six ADS relief i'n torus cooling node valves or or Ee of four RHR pumps witn Ee of four LPCI pumps associated heat exchangers or ;
in snutdown cooling node Ue of two core spray t loops Int ermed i a te , One of one HPCI pump i Stean Two of four RHR punps with !
or (puno discnarge) assnciated heat axchangers '
E e of four LPCI punps in torus cooling node or or Ee of two core spray Ee of four RHR pumps with loops I i associated heat excnangers i in snutdown cooling mode Saall , Li quiet One of one HPCI pump Two of four PHR punps wi th ar Stean or
! fup to 0.12 ft2) associated heat exchangers Eur' of si x ADS relie f in tnrus cooling node valves and one of four !
or :
LPCI punus one of four SFR punps with 1
t or l
- as soci,itet 'hea t e < changers Tour of si x ADS relier in shutitown cooling node 4
valves and one of two {
j spriy Icops :
l l 81 I
I l '
L
Tahl e B.2 Typical LOCA Frequencies for a bur Fron the IREP Browns Ferry Study 2 Frequency Type Si ze Location (Per Reactor Yesr)
Li quid large Suction side 9.9 x 10- 6 Di scharge side 3.9 x 10- 5 Stean large --
- 5. 2 x 105 Li quid Intemed iate -- 4.0 x 10- 5 Stean In temed i a te -- 2.1 x 10 "
Liquid or stean Small -- 1.0 x 10- 3 42
i 1
J, . <
i s
1 i
4 4
$ BWR MARK I PRELIMItiARY ISSUE CHARACTERIZAT10ri i
i j
i i
1 l
!O 4
$ JA!!UARY 21, 1955 i
'l J
l 4
1 1
i l
d a
i i .
i i O i
}
- - . . - - - - - - . - - , - - - . _ -?- - - - - . -.-- --
TABLE OF CONTENTS
(;ge 4
List of Tables 5
List of Figures 6
- 1. INTRODUCTION 7
- 2. CONTAINMENT CHALLEftGE3 AtiD FAILL'RE MODES 2.1 Early Overpressure or Tercerature Failures 7 2.1.1 Definition of Challenge 7 2.1.2 Potential Failure Modes 16 2.1.3 Assessment 18 2.2 Core tetris Attack on Steel Containment Shell 18 2.2.1 Definition of Challenge 18 2.2.2 Potential failure Mcces 20 2.3 Late Overpressure Or Temperature Fai'ures 22 2.2.1 Definiticr. of Challenge 22 2.3.2 Pctential Failure Modes 27 2.4 Ccr.tainment Eypass 28 2.4.1 Definition of Challenge 28 2.a.2 Potential Failare !'edes 32 2.5 Rapid Stea Pressure ard Missiles 34 2.5.1 Cefinition of Challence 34 2.5.2 Potential Failure Modes 37 2 . f. Sur-ary of Irportant Centain.sent Che lenge 3C
!ssues 2.7 References 41 1
3 s
Page
- 3. POTENTIAL IMPROVEMENTS 46 3.1 Hydrogen Control 45 3.2 Containment Sprays 47 3.3 Venting 49 3.4 Core Debris Control 53 3.5 Automatic Depressurization 55 3.5 Procedures and Training 56 3.7 References 57
() Appendix: Questions Asked of Research Contractor Personnel and Industry at February,1987 and March, 1987 Meetings. 71
. b I
9 I
e
4 l
a .
LIST OF TABLES Pace Table Typical MARK I Containment Response Prior to 58 2.1.1 Core Degradation Ranges of MARK 1 Containment Leading During 58 2.1.2 Core Degradation Pressure-dependent Leak Area Estimate for a 59 2.1.3 BWR MARK I Containment Potential Failure Modes of the Peach Botton 59 2.1.4 Containment as Suggested by D. Clauss (Subject to Quasi-static Overpressuri:ation and Temperature (800 F)
Typical Debris Heights as a Function of Floor 60 2.2.1 Area 2.2.2 Summary of BWR Mark I Local Failure Calculation 60 Results 2.3.1 Comoarison of Potential BWR MARK I Failure Mode 66 Times 2.4.1 Contribution of Containment Leakage to Offsite 69 Severe Accident Risk (Perion-rem / reactor year) 2.4.2 Likelihood of Undetected Braaches of Containment 69 Integrity
~2.6.1 Timing cf Key Events for Peach Bottom 70
5 O LIST OF FIGURES Figure Page l 2.2.1 Conditional Prebability of Co.ntainment Failure 61 for Peach Bottom (Oraft NUP.EG-1150) 2.2.2 Risk Significance of Various Systems, Containment 62 and Source Term Issues as Calculated for Peach Bottom (Draft NUREG-1150) '
2.2.3 Schematic of MAPX I Drywell and Pedestal Regions 63 2.2.4 Thermal Response of Drywell Liner as Calculated 64 by IDCOR Model 2.2.5 Impact of Debris Depths on Drywell Liner 65 .
Temperature as Calculated by IOCOR Model 2.3.1 BWR MARX I Orywell Response Without Leakage 67 2.3.2 Break Area vs. Containment Pressure 68 O.
1 l
i l
l O
%F ,,
6
- 1. INTR 00UCTI0t' In 1986 Nuclear Regulatory Commission (NRC) staff proposals for potential improvements to BWR MARK I plants to enhance centainment mitigatien in the event of a severe accident were made to the owners of those plants. These proposals had the common objective of preserving the suppression pool water inventcry to filter important fission product releases from a breached reactor vessel. The bases for the proposals included concern over the large uncertainties in f. ARK I containment performance given reactor vessel meltthough, as reflected in draf t NUREG-1150. The staff met with a group of NRC research contractor personnel (Feb. 1987), and industry representatives (Mar. 1987) to discuss the proposals. At these meetings 15 ouestions were asked and reproduced in an appendix for reference purposes. It became obvious that a consensus on the efficacy of several potential changes to improve accident mitigation did not exist, and that further technical evaluations were necessary.
To provide an assessment of potential improvements, the NRC staff is re-evaluating important MARK I severe accident phenomenological and potential improvement issues in parallel with the assessment of the MARK I plant in fina; NUREG-1150. The phencmenological issues are those associated with containment challenges. The improvement issues are primarily associated with either mitigating cantainrent challenges or recucing radioruclide releases to the ensironment. Depending en the outccme of the evaluation, a regulatory initiative may be considered to recuire improvements in EWR containment performance. The staff exoects to narrow and focus the important phenomenological and improvement issues with the help of a conference with researchers, industry representatives, interested members of the public, and the NRC staf f in late February 1989.
The folicw ng material describes MARK I containment phenomenolcgical arc i
improvement issues that the staff believes merits further discussion. The material has been gathered f rcm a number of sources. It does not, hcwever, recessarily reflect NRC staff opinions. The cbjective of this material is to I
_ . ~
provide a preliminary statement of the issues that can be used in the staff's
- focusing and evaluation process.
The present effort is directed towards the mitigation of severe accident
- consequences at BWR plants using MARK 1 containments for some low probability severe accident sequences for which the integrity of the containment function can be seriously challenged. The effort considers both means of either preventing or delaying containment failure and means of minimizing releases of radioactivity to the environment from a degraded containment. Except where means of accident prevention and mitigat. ion cannot be separated (as discussed in Section 2.5), reduction of the likelihood of accident initiation is not included.
- 2. C0ftTAlllMEtlT CHAU.ENGES A!i0 FAlt.URE MODES The phenomenological issues are those associated with containment challenges resulting from a variety of core melt scenarios. The estimated overall likelihood of core melt accidents has varied from 2 X 10-3/RY to 4 X 10 /RY in eight MARK I PRAs of varying completeness that the staff has reviewed.
The following sections sumarize the types of BWR MARK I severe accident containment challenges that have been postulated in PRAs. The sumaries include estimates of important parameters associated with the challenges. The uncertainties associated with these estimates may be important in evaluating public risk. .However, they are believed to be adequately stated in the descriptions.
2.1 Early.0veroressure er Temoerature Failures 2.1.1 Defini. ion of Challenge.
Although in MARX I containment designs the suppression pool is provideo to ccndense steam released from the reactor coolant system, the relatively smali containment volume (compared to large dry PWR centainments) makes tner more vulr.erable to pressure and/or temperature increases that night be associated Y with a severe accident. An early overpressure or temperature containment
l 8
failure is defined to include any failure that may occur prior to or very soon (within a few hours af ter fuel damage) af ter breach of the reactor vessel.
Should such failures cause the resulting mixture of steam, non-condensiole gases and radioactive fission products to bypass the suppression pool, the releases to the environment could be quite large.
Sources of containment loading resulting from a severe accident that may.
lead to early containment failure depend on the sequence, the progression of the accident, and include phenomena such as,;
(a) Containment tempera'ture and pressure loadings prior to core damage;
-(b) Hydrogen generation and pressure rise during core degradation; (c) Pressure rise due to vessel blowdown; (d) Pressure rise due to ex-vessel debris quenching; (e) Pressure rise due to high pressure melt ejection; and
- h' s
(f) Attack of the steel shell with high temperature corium that could lead to shell failure. (Corium is a molten mixture of core fuel and vessel internal and core structural supports.)
Containment leads prior to vessel breach do not usually lead te containment failure themselves, but are important to define the conditions subsequent to vessel breach which could be a challenge to containrent intagri ty. The aim of this section is to identify the ranges of severe accident leads (pressure /temoerature histories) appropriate to the accicent sequences that are risk significant. The quantification of ranges of contairrent loacs will rely heavily on the IDCOR and NPC analyses performed to date. This includes early results of the Containment Loads Working Group (NUPEG-1079)[1].
W
. l 9
Containment Loadino Drior to Core Degradation To assess the containment pressure and temperature rise at the onset of core damage due to steam plus heating effects, the results of Source Tenn Code Package (STCP) and MARCON calculations (2 4], together.with other calculations performed in support of draft HUREG-1150 [5], were reviewed. The containment load depends on the progression of an accident until core damage begins.
Typical MAPK I containment response prior to core degradation for accidents important to risk is presented in Table 2.1.1.
For a station vlackout with failure of the batteries, a potentially high risk sequence (5], there would not be water injection to the reactor coolant system and the centainment sprays and pressure suppression cool cooling would also not be available. The pressure rise in containment is that due to the boiloff of the inventory of water in the reactor pressure vessel (RPV), the pressure rise calculated in support of draft NUREG-1150 [5] for Peach Bottom is 0
n 8 psi (4]. This corresponds to a suppression pool temperature of 156 F (fren 0
l'~/ 900 F initial temperature) and a drywell temperature rise frcm 135 F (operating temperature) to 350c F (approximate result from loss of drywell coolers). Since the ratio of vessel inventory of water to the mass of suppression pool is similar between the BWRs with MARK I containments, these loads would be typical fo- all MARK I containn.cnts.
For a station blackout in which the batteries supply power fcr 6 to 8 hcurs, injection to the RPV would occur from HPCI or RCIC systems. Mcwever, the drywell ccelers and RHR are unavailable. The drywell temperature is elevated due to heat transfer from the RPV and from the uninsulated discharge tailpipes on the safety relief valves (SRVs). Heating-of the dryweil coulo j I
result in degassing of unliced concrete. Both the temperature increase and the liberation of water vapor from concrete increase containment pressure. In addition, the ruppression pool is heated to approximately 225 F. Uncer these l
circumstances a 19 psi partial pressure of steam is possible in the torus. A Source Term Ccce Package (STCP) calculation for this scenario for Peach Bottcm (2] predicts a pressure of 27 psia in the containment at the onset of core damage. Ccncrete degassing and heat transfer from the PPV to the drywell is not modeled in the STCP. Results of a MARCON calculation (4], which model these phencre ,
predicts a higher pressure rise of 38 psia.
10 For anticipated transient without scram (ATWS) events in which'injecticn is g insufficient to maintain adequate core cooling, a,nother potentially high risk sequence, even though the ECC systems are operable (failure-to-depressuri:e er sequences in which the level is not maintained), the pressure rise in the containment is not sufficient to challenge ,its integrity. An STCP calculation for this scenario (TC2) for Peach Bottom predicts a pressure rise of 30 psia
[3].
For an ATWS sequence with adequate core cooling provided by the low-pressure systems and with the lack of successful venting the containment pressurizes rapidly. As the containment continues to pressuri:e, there is' a possibility that the SRVs will reclose due to elevated containment pressure early enough to prevent pressurization of containment to the point of failure.
In Reference (4) the (.ontainment pressurization for Peach Bottom has been addressed by a single hand calculation assuming 470,000 pounds of water to be boiled off to the pool after SRV reclosure prior to core damage. Assuming injection lost with.centainment pressure at 110 psia that calculation demonstrates that the containment pressure is raised to 213 psia. Hcwever, this calculation assumes the vessel water level to be maintained at the normal level. This is not the expected case. Rather, one expects that the level would oscillate between the normal level and top of active fuel (TAF). Thus, the containment. pressure rise is. uncertain and depends on the water inventory of the vessel at the time the vessel pressure reaches the cut-off pressure for icw pressure injection. For transient-initiated secuences with less of containment heat removal and with high pressure systems failure on high containment pressure or high pool temperatures, the RPV will repressurire eliminating low pressure injection. Control rod drive (CRO) flow is also not available so there is no injection. Unlike the core-vulnerable 70 sequencesi at the time injection fails, the RPV water level will be in the norral range.
As indicated by the calculations described for that case, the containn.ent pressurizes rapidly to the point of failure. Hcwever, it should be noted that the probabilities of these sequences are estimated to be very low.
For scenarios in which the initiator is a large LOCA and there is no subsequent water make-up to the vessel, the containment pressure could rise 35
. psia if the containment sprays do not operate. This value is taken frce g
11 .
' BMI-2104 (Volumi II, Table 6.4) [6] and corresponds to the calculated containment pressure for a MARK I AE sequence following the blowdown of the RPV. Steam condensation in the pool and on structural heat sinks lowers _the pressure rise from the peak valve of 52 psia (Volume II, Table 6.5, (6]) which occurs during the first 30 seconds of the primary system blowdown.
For accidents initiated by small breaks ,followed by initial high pressure injection and continued operation of the RHR system, injection is lost 2 to 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> into the event _due to reduction in RPV pressure. Given the long period of RHR operation, an equilibrium condition will have been established in the containment and the pressure will be near atmospheric (pool temperature approaches service water temperature). However, if the RHR system dces not operate initially, the containment pressure rise is bounded by the results for a large break sequence discussed previously plus 5 psi (which is the steam partial pressure increase due to adding 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> of decay heat to the pool assuming equilibrium with the containment is achiev,ed).
Containment Loading During Core Degradation During core degradation the containment pressure rises due to hydrogen generation, steam production and temperature ircrease. Hydrogen generation is important to containment loading because it is a combustible and noncendensable gas. The MARK I containments are inerted with nitrogen during plant operation and, consequer.tly, hydrogen combustion is not a challenge to the MARK !
containments unless their integrity is lost, or oxygen is introcuced. However, MARK I contair: rents are small and, therefore, any significant builcup of noncondensable gases (such as2H ) could challenge containment integrity by overpressure.
To assess containment loading during core degradation the results of STCP calculations, MARCON calculations ard available MAAP [7] calculations were
~
reviewed. There are significant differences between the industry degraded core rulemaking gecup (10COR) code (MAAP), STCP, and MARCON predictions of hyorogen generation during the in-vessel phase. IDCOR's HAAP Code predicts oxidation of a to 10 percent of the available zirconium. MARCH predicts oxidation of 30 to 60 percent of the active fuel cladding. MARCON predicts 40
i 12 .
l to 80 percent oxidation of the available zirconium. The resulting range of containment loadings are sumarized in Table 2.1.2 and are discussed in the following paragraphs according to the potential progression of accidents during core degradation process.
For transients with the vessel at high pressure an STCP calculation for the Peach Bottom TB1 (2] sequence (station blackout) predicts 25% zirconium oxidation and a pressure rise of 24 psi. A slightly higher pressure rise is expected for short term station blackout scenari s. Based en MARCON results, anestimatedpressureriseof72psiisreportedinReference(4].
For transients in which the RPV is depressurized during core damage, blowdown of the RPV during core damage may increase hydrogen production. The pressure rise value for this case has been estimated based en an SICP Peach Bottom TB1 result (4]. A correcticn has been made to adjust for the fact that the RPV was at low pressure by adding the paitial pressure which would be exerted or the containment by the hydrogen which was in the high pressure RPV imediateiy before vessel breach. The gas temperature assumed in the partial pressure calculation was that calculated for the drywell in the TS1 sequence imediately orfor to RPV meltt'hrcugh. The resulting pressure rise was estimated to be 38 psi. However, assuming an in-vessel zircaloy oxidation in the rarge of 5.2% (MAAP calculation of TOUV at LaSalle (7]) to 51". 'or the liARC0!; calculation reported in Reference (4]), a pressure rise in tne range of 22-98 psi was estimated.
In a large or intermeciate-break LOCA the range of containment pressure riseduringcoredamagewasestiratedbasedonaBMI-2104(6]calculationofa Peach Bottom AC secuence. 'n that calculation, the pressure was predicted to be 106 psi with 36'; of the zircaloy reacted. A pressura spike of 63 psi is associated with core slumping. The lower bound for the pressure rise was estimased to be 16 psi by assuming 5*, Zr oxidation (typical of a MAAP code 1
1 -r- ..-= ._
13 calculation), gradual core melt, and no steam spike associated with core slumping.
In a small-break LOCA, the containment pressure rise during core degradation depends on the extent of flow from the RPV which passes directly to the drywell. The range of containment pressure rise is similar to scenarios in which the RPV is at high pressure. However, a correction was made to account for the elevated drpell temperature which is caused by direct leakage from the RPV.
For accidents in which injection is restored during core degradation, depending on the timing of injection being restored, significantly larger amounts of hydrogen may be produced. However, there is a very low probability of containment failure due to hydrogen accumulation before RPV failure.
Containment Leading Due to Vessel Blowdcwn The failure of the RPV would be followed by a increas'e ir the containment pressure as the internal energy of the steam and gases retained within the reactor coolant system (RCS) is releaseo to the containment along with the cciten corium.
The increase would be substantial if the vessel were not depressurized. In accident sequences in which the vessel is depressuri7ed at the ' time of vessel breach, pressure rise in the containment as a result of the equilibration of the pressure inside and cutside the vessel is not considered to be significant.
An approximate quantification of the pressure rise due to bicwdcwn may be accomplished with a simple hand calculation, but there are significant uncertainties in the in-vessel bulk-gas temperature, the hydrogen content, anc the final temperature in the containment. The impact of possible cirect heating in containment pressurization, a phenon.ena that may be associated with vessel failure at high pressurc, is not considered here. The issue of direct containrent heating is discussed separately.
For high reactor pressure in the vessel during blowdcwn, an STCP k calculation for the Peach Bottcm TB1 scenario predicts a pressure rise of 70 psi [2]. In this calculation stean passing to the pool is assumed to be
- , . --- n , , . - -. - -- --
14 condensed and hydrogen is cooled to the pool temperature. A bounding g calculation assuming isothermal expansion of the in-vessel gas into containment results in a pressure rise of 94 psi [4].
Containment Leadino Oue to Ex-Vessel Debris Ouenching ,
Quenching of the molten core debris (corium) as they are released into a flooded reactor pedestal region may cause rapid steam generation (so-called steam spike). The containment pressure rise due to a steam spike depends en the quantity of molten corium available for release and' the amount of water on the drywell floor (or sumps). Assuming sufficient water is available for quenching, a range of 5 to 30 psi was used in the, containment event tree analysis of Peach Bottom (4]. However, it is not clear that sufficient water would be available for quenching and, therefore, the estimated pressure rise may be considered a bounding value.
Containment Loadino Oue to High Pressure Melt Ejection In certain reactor accidents, the degradation of the reactor core can take place while the reactor coolant system remains pressuri:ed. In these accidents, molten core oebris will tlump and collect in the bottom of the reactor vessel. The core materials will start attacking the bottom head of the reactor. When the bottom head of the reactor vessel is breached, the core relt will be ejected under pressure. It has been postulated that the ejected materials may be dispersed out of the reactor pedestal'into the surrcunding drywell as fine droplets, quickly transferring thermal energy to the atmosphere, in addition, the metal components of the ejected core debris, mostly zirconium and steel, can react with oxygen (if presen'.) and stean in the atmosphere to gererate a large quantity of chemical energy, heating and pressurizing the containmer.t further. The term "Direct Containment Heatir;"
(CCH) is used in the cresent discussion to describe this complicated physical / chemical process.
6
I l
1 i
15 l -
1 t
Exothermic chemical reacticns are of less concern in MARK I contairments
- because the containments are inerted (low oxygen content). However, the zirconium-steam reaction and the thermal energy transfer could still contribute to containment loading. If even a small fracticn of the core debris (on the ,
order of 10 to 25% of the core debris initially suspended in the drywell atmosphere) participates in directly heating the containment atmosphere the pressure rise is expected to be suff'icient to challenge containment integrity.
The core debris transport and dispersal out of the reactor pedestal region bears profound consequences on what follows in the containment response and ultimately on the potential for DCH. Much remains to be learned from ongoing experimental programs concerning the potential influences of the transport path, including the effect of geometry and structures, on DCH. Results of Sandia tests seem to suggest that entrainment of melt particles in high-speed gas flow may be the dominant mechanism for debris dispersal. However, efficient ejection frem the vessel does not guarantee that the debris will have p an uninhibited opportunity to interact with the large volumes of gas assumed in
'I the simple bounding calculatien. OCH will be affected by the following MARK I design features:
- 1. The EWP core support cesign (which provides individual support for each group of fcur fuel bundles from the vessel bottom head) could minimize the quantity of core debris ejet..ad into the containment at vessel f ailure.
TMs is because localized f ailure of the support could result in slumping of relatively small quantities of core debris. Therefore, depressurization of the vessel (due to local meltthrough) could occur before large quantities of molten core material have collected in 'the bottom head (5).
- 2. One view is that if the melt progressicn results in large quantities of molten core debris release at vessel failure, the confined geometry of the vessel support pedestal, and the restricted opening to the drywell (2 meters by 1 meter for Peach Bottem) will effect dispersal of debris into the crywell airspace. Also, the pipes and associated whip restraints, as well as pumps and equipment in the drywell, may also tend to har.pe' debris oispersal [4].
5 -ouJ 16
- 3. The operation of BWR Automatic Depressurization System (ADS) cculd f virtually eliminate the potential for high pressure melt ejection.that could lead to direct containment heating. Credit for ADS operation was not given in draft NUREG-1150 for Peach Bottom for some station blackout s,equences.
The A05 has the additional benefits of (1) improved fission
~
product scrubbing by directing the RCS inventory through the suppression pool, and (2) reducing the core melt frequency for sequences in which the high presture injection systems fail, by depressurizing the RCS to allow water injection into the RPV via the low pressure injection systems (see also section 3.5).
2.1.2 Potential Failure Modes Analyses indicate that containment performance plays a dominant role in the assessment of risk associated with severe accidents. A key insight emerging from research on severe accidents is that the mode of containment failure strongly influences the offsite consequences. That is , early containment failures would produce higher source terms than late failure sequences, as would suppression pool bypass sequences. One primary concern for containment performance is how we'll the containment can witnstand the pressure and temperature loads associated with severe core damage accidents.
Ancther primary issue is the likelihcod of liner meltthrough. For scenarios in which containment integrity is maintained, fission product release will be small. For those scencrios leading to containnent failure, fissien product release cepends on the timing as well as the size and location of the break in containment. The mode of containment failure, (i.e., gross failure versus leakages thrcugh seals or the f ailure of penetrations) influences the amount of radioactive materials inside the containment that would be released to the reactor building.
Two basic mooels can be used in severe accident risk estimation to characterize the loss of containment integrity from overtemperature and overpressure challenges; the "threshold" model and the "leakaoc-before-f ailut c" )
model. A threshold pressure is defined as that pressure at which the containment l I
is predicted te fail. Failure is assumed to result in rapid derressuri:ation and release of the containment atmosphere (which may contain a large amcunt of fission products). If the containment pressure loading is calculated to be
17 below the threshold pressure, the containment is considered to be intact and fissicn product release is via containment leakage. Some recent analyses have pointed out that gross containment failure is not the only pathway to substantial fission product release and that significant leakage on the order of 100 volume percent per day or more may result (if it occurs sufficiently early in the accident sequence) in substantial offsite releases of radioactive fission products. The leakage-before-failure model provides a means of accounting for this condition when performing risk assessment analyses.
In order to assess the possibility that leakage paths not explicitly censidered in "threshold" estimates might lead to significant leakage before
~
the calculated capability pressure, a study was undertaken for six plants with different containment designs, including Peach Bottom. The approach taken was to estimate the increase in leakage as a function of increased pressure.
Assumptions about leakage through penetrations were made. Results were reported in f:UREG-1037 [8]. Table 2.1.3 provides a sumary of the resul'ts (for peachBottem).
Q .
3 .
In conjunction with the ongoing QUASAR research program at Ef;L, an j assessment of the Peach Bottom containment performance was also made. Eight l pntential failure modes (shown in Table 2.1.4) were suggested by the QUASAR l Expert Review Group (9]. Several of these failure mechanisms are apolicable to early containment failure as indicated in Table 2.1.4 These are leakage thrcugh unseating of operable penetrations, drywell rupture and melttnrough.
The leakages cue to the pressure unseating the ecuipment hatch and control rod crive (CRO) remeval hatch are insignificant compared with leakage through the drywell head.
The 10COR program considered the dominant containment f ailure mcde to be leak befcre failure, which would cccur as a result of a large strain of the containment boundary. For BWR containments, IDCOR indicatcc that, if venting is initiated, containmert failure due to strain o' the containr.ent boundary is of seconcary importance. For !! ARK I sequences in which containment venting is not initicted, the f ailure mode w/s assumed by 10COR to be due to elevated
[ temperature in the dryweli a long tire af ter vessel f ailure.
,-.w-- ~,--nn- , - .- - , - - - - - - -
18 .
A study by Mokhtarian et. a1 [44] of Peach Bottom has indicated a f l
potentially significant leakage path through the containment head bolt seal and a strain induced tear failure of the containnent shell in the wetwell airspace at about 160 psig. This study, as well as the others, made a number of assumptions of accident pressures and temperatures, and structural responses, that may be indicative of early or later overpressure /overtemperature severe accident containment challenges and responses for MARK I steel containment shells.
2.1.3 Assessment Early overpressuri:stien is a significant containment challenge that can lead to loss of containment integrity. These pressure loads arise primarily frem direct containment heating (DCH), steam spikes, and vessel blowdewn frcm high RCS pressure sequences. These are reascnably well understood with the exception of direct containment heating. Containment loading due to.DCH must be assessed through detailed parametric calculations using CONTAIN or any other lumped parameter code. These parametric calculations should include a spectrva of initial, boundary, and phenomenological conditions relevant to BUR MARK I severe accident conditions and containment configuration.
2.2 Core Cebris Attack cn Steel Centainment Shell 2.2.1 Cefinition of Challenge Studies performed as part of the Containment loads Working Group (CLWG)
(NUREG-1079)[1] and tne Centainment Performance Working Group (CPWG) (8], and others [8,44] have incicated that FARK I containments can contain pressures several times the design cressure even under elevated temperatures. Scre of >
these studies nave also identified potentially important challenges to MARK '
containments not previously considered ir. earlier risk studies.
One such important challenge involves molten core debris attack on the ,
steel containrent shell Icading to tbc potential for shell meltthrough. Should 7 {
this failure occur, a flow path to the reactor building (secondary containment) $ i
{
1 1
4,---- M 1
l 19
/*
' \)
bypassing the suppression pool would be available. This failure would, l
therefore, lead to rapid blowdown of the radioactive drywell atmosphere to the 1 reactor building with subsequent releases to the environment.
This failure mode was initially analyzed by Greene, et al [10], who performed a series of calculations assuming various quantities of molten core debris contacting the steel shell of the drywell. The effect of different corium temperatures was aisc examined. These calculations indicate that this contact could result in melting of the steel, inducing significant containment leakage. In calculations performed at ORhl (Ref. 6, Chapter 3), the metallic portion (Fe, Zr, Cr, hi) of the corium is postulated to be a liquid at 2280 K I ar.d released f rom the vessel to the drywell floor. The oxide debris (U02 ' l Zr02 , Fe0), with a higher melting temperature, on the other hand, is postulated j to be released frem the vessel to the drywell floor later. IDCOR calculations indicate a substantially lower temperature for the core debris which reaches I the drywell shell, but the assumed terperature of the melt was very low compared to estimates frem STCP, MARCON and MELCOR calculatiens performed by
'G NRC contractors. All three analyses assured no water pool in the drywell, nor '
the presence of spray cooling of corium or the shell.
At RPV failure, core debris which may be at or near the solidus temperature has been postulated to be poured (or ejected) onto the ::ecestal I ficer and into the drywell sumps. The core melting is expected to continue, and the molten debris is envisioned to flow out of the failed vessel cver several hours. Because of this ficw period, the rate of concrete erosion would be influenced by the rate that core debris is added to the volume that has already lef t the vessel. Under such conditions, the time at which the melt reaches the steel shell of the drywell could be important, also important is whether a significant fraction cf the core-concrete releases have occurred before shell failure. This is because a late failure (after the bu h cf the fission product release frem the core-concrete interact'.cris) has far less risk significance than a drywell failure shortly af ter vessel breach.
In the recently published draft Reactor Risk Reference Cocument (5] the issue of direct core cebris, attack en the drywell steel shell was considered.
Figure 2.2.1 shows the calculated relative likelihood that core damage
20 accidents would result in early, late, or no failure of containment. Note that p two ranges of the conditional probability of early failure are shown. The range on the lef t (solid box) includes drywell failure due to direct ccre debris attack on the shell; while th range on the right excludes this failure mode.
As indicated in Figure 2.2.1, the likelihood of early containment failure, given a core damage accident, has been esticated to exceed 80". when the shell meltthrough is included, and up to 40", for the no-drywell-shell meltthrough assumption, a factor of 2 reducticn in early containment f ailure probability. .
This relatively high probability of early containment failure results from the ,
dominar.ce of static- blackout sequences in the distribution of core darage frequency which in draf t NCTsEG-1150 leads to one of two early containment failure scenarios; (1) early overpressure and (2) direct attack of dryweli wall by molten core dcork .
The uncerta,inty in shell failure by direct debris attack has been shown to have a significant impact on the risk at Peach Bottem as shown in Figure 2.2.2 where a significant fracticn of the variance for early and latent f atalitics is calculated to be due to challenges associated with this failure mode.
2.2.2 Potential failure Modes Figure 2.2.3 shows a schematic of a typical MARK 1 drywell ficcr cross section. It is seen that the drywell floor area is relatively small and the drywell shell is within few meters of the pedestal doorway. Therefore, following RPV failure, the molten core debris released would spreac into the pedestal region and fill any existing sumps and, provided that a sufficient quantity of core debris has been released from the vessel, it,can then ficu across the remainder of the drywell floor with pctential for attacking the steel shell. Table 2.2.1 shows the height of the whole-core debris above the drywell floor as calculateo by Hodge [11] for the Browns Ferry reactor assuming debris port:i*ies of zero and 0.40. Also given is the volure of the whole-core debris (at :ero porosity) trd the total volumes of the drywell sumps.
q -- ,- -
~ ~
4 l
i 21 l 5
The physical properties of corium, particularly the viscosity of the molten debris, are conducive to rapid flow and spreading, provideo the heat losses from corium to the baser.at concrete, the containment atmosphere or the overlying water pool (if present) are not sufficient to freeze the debris Furthermore, as shown in Figure 2.2.3, it is clear that molten material will be directed out of the in* pedestal region through the doorway towards the drywell shell. As long as the corium remains at temperatures greater than the steel melting temperature (1750'K), there exists a potential for local drywell sheli meltthrcugh, and a flow path directly to the torus room of the reacter building ,
[10]. Some of the gap between the drywell shell and the concrete is packed l with fiberglass and polyester foam. Although, some aerosol retention potential may exist in the gap, it is believed to be small. Therefore, the radiological consequences wculd be e'xpected to be significant.
Parametric calculations performed by Greene, et a1 [10), on direct core debris attack are surrarized in Table 2.2.2. The table lists the concrete type, corium temperature, percent of core participating in the interaction., estimateo
- time to failure, total downward erosion at the end'of calculation and thickness _
of the.Shell iblated. These parametric calculations are based on several assumptiens, incl.uding no spray cooling of corium and the shell.
Analyses have also been performed by Fauske and Asscciates (12] for IOCOR using a finite difference model representing the two-dimensional conduction within the steel shell 6nd the supportive concrete inside and outside the shell. Mcjor assumptions of this model include: (1) solidified debris consisting of primarily UO2 only (n substantial metals present), (2) debris heat source is due to decay heat only (chemical energy source due to metal gas phase cxidation ignored), (3) efficient heat removal by boiling at the Critical Heat Flux (CHF) limit of the overlying water pool (if water present), (4) heat losses on the out0ide bouncary of the shell by convect.icn and radiaticn, anc (5) nc spray cooling of the shcIl.
Using this model, the transient thermai resperse of the steel shell was evaluated for initial conditiens of 2100*K debris temperature, 0.06 to 0.12 r' debris depths, with and without an overlying water pool. Figure 2.2.4 shcws the IDCOR estimate of the therral response of the hcttest steel node. As
9 illustrated, the temperature increases over approxinately E00 seconds to abcut 80C'K. Also shewn is the transient response when an overlying water pool is assumed, showing rapid quenching of the core debris leading to a dramatic postulated reduction in shell temperature. This quenching process is shown to be more efficient at higher containment pressures. Figure 2.2.4 illustrates the thermal history of the hottest shell nede corresponding to different debris depths up to 0.12m. As shown, the debris depth changes the shell temperature at 200 seconds by 150 K when quenching is not considered, in sunnary,10COR analyses do not indicate shell failure [13], but researcheranalysesindicatethelikelihoodishigh(5,10,11,13].
Further, the analyses to date reflect various views en initial conditions, transport process, heat transfer processes, and the influences of water as a spray and in a pool on the drywell floor.
2.3 Late Overoressure or Temoerature Failures 2.3.1 Definition of Challenge In thosa accident scenarios in which the reactor vessel is postulated to fail, high temperature core debris are likely to fall into the reactor cavity and crywell where they will interact with structural concrete and any water that n:ay be present. The consequences of these thermal and chemical core-concrete interactions may significantly impact containment loadir.g and tne mode of centainment failure. This section is concerned with the nature and magnitude of centainment loading during core-concrete interaction.
0 At high temceratures (approxicately 1,300-1,500 0), concrete decomposes; the ablation products cor=cnly include water vapor and carbon dioxide, as well as condensec pnsse oxides such as Ca0 and SiO 2
. At lower terperatures that may result from liquid metal / concrete interaction, concrete degassing but not ablation may be expected. The liquefied oxid1c components of the concrete mix with the uranium oxide fuel and metallic oxides of the debris. The core debrit could initially be all or partially molten and gases released at the debris. concrete interface would bubble through the debris pool.
i l
l 23
~
f} The fraction of cor. . rete mass that is converted into gas by thermal ablation or liquid metal / concrete interaction depends on the cceposition of the concrete. For concr'etes made primarily of limestone aggregates, the gaseous l decemposition products may carry away over 40 percent of the total initial l mass. Highly siliceous concretes containing mostly SiO2 produce much less gas. l (10% or less of the initial mass) upon decomposition. The gas content of limestone / common-sand concretes is somewhere between these two extreme values.
For example, in the case of limestone /coninon-sand concrete, approximatelv ,
l one-fourth of the original mass wiH beceme gaseous reaction preducts, primarily 1 water vapor and carben dioxide. As H 2O and CO 2 bubble through the decris pool they may react with liquid metals to form H2 and CO. The mass of unreacted water vapor will enter the containment atmosphere and may or may not condense, l depencing on prevailing thermal-hydraulic conditions.
Hydregen combustion events inside containment have generally been considered to be of low likelibced in the MARK 1 containment despite the potentie.lly larce amount of hydrogen that may result from core melt and core-concrete h interactions. This is due to the fact that MARK I containments.are operated in the 'Iinerted" condition most of the time at oxygen concentrations of 4 percent or less. A potential containment challenge due to hydrogen combustion may accrue, however, in the event that the containment is cperated without being inerted during an accident. Current plant technical specifications allow for deinerted operation during start-up and shutdown (i.e. low reacter power levels) on the orcer of 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> to facilitate maintenance and inspections. i.
likelihood exists that a severe accident initiated during these times mignt causa hydrogen ccmbustion inside containment. Other means also exist to cause oxygen to enter the containment. After onset of a severe accident, for instance, the operatiun of. f ront-line or back-up instrument service "air" systems and/or SRV accumulators may introduce significant quantities of air / oxygen into containment. These additional quantities cf oxygen may, in the Icng term, add enough oxygen to eliminate the initial ir.ert conta1cment environment and lead to ccmbustion. Additional challenges may also exist in that the centairrent challenges inerting status shculd be maintained in accidert secuences such as station blackout to 'eliminete combustion events 'rce takingplacinglateintheaccidentprogression[46].
A V .
l 24 are ncncondensible and contribute directly to The H'2, C0, and CO2 centainment pressure loading. The partial pressure of these gases can be ccmputed directly from the gas laws, the atmosphere temperature, and the containment f ree vclume. Considerable data concerning the decomposition products of concrete exist. However, the prediction of the relative abundarce and rates of generation of the various species of gases leaving'the debris pool 1 depend in large measure on the ability to mcdel the chemical phenomena that are resconsible for modifying the decomposition gases after they leave the concrete. The chemical oxidation processes are particularly important for SWRs because of the high stainless steel and zirecnium inventory in the core.
Liquid :irconium is extremely active and its oxidation highly exothemic so that, during the reaction, the debris temperature may be significantly elevatec resulting in accentuated fission product release.
Energy leaves the debris pool in the form of high-temperature gases, hot aercsois (a highly efficient mechanism for heating the atmosphere), and thermal
~
ra di a t.i on. Heat transfer wili be established between the pool, the atmosphere and the available heat sinks. Steady temperature increases would result..
Structurcs in tne line of sight of the pool are heated by radiation anc ;
centribute incirectly to elevating the containment atmosphere temperature. The rise in the temperature of the atmosphere is accompanied by a corresponding increase in pressure. The magnitude of these effects depends on the power of j the heat source, atmostneric compcsition, nature of available heat sinks, anc the free volume of the centainment.
The CORCCR code [14] has been developed by Sandia. f:ational Laboratories l for the NRC as a best-estimate computatienal tool to calculate the physical arc therredynamic variables needed to characterize the progression of high-temperaturc core debris as it erodes concrete in the reactor cavity. Frem an assun.ed initial axisymmetric "crucible", the change of cavity shape is tracked as the ablation front advances in time. As the decomposition products of the concrete enter the cebris pcol, chemical interacticrs are taken into account. Mass arc energy are conterved ano the temperat0re profile of the system is folicweo.
The composition and release rates of gases leaving the pool are ccrputed.
Radiative and convection heat fluxes from the surf ace of the debris pcol to
25 these surroundings or to an overlying water pool are calculated. These outcut variables computed by CORC0t! provide the data needed to characterize containment iceding. CORC0ft MOD 2 [14], which includes improved interfacial heat transfer models, the effects of crust formation, and the influence of an overlying water pool, has been incorporated in the NRC Source Term Code Package.
Earlier containment analyses for the MARK I concluded that the most likely failure would be from overpressurization of the drywell. The possibility of seal failure resulting from high drywell temperature was also recognized (leak before failure). The latter failure mode concern provided the impetus for the CPWG [8] to analyze the respcnse of seals in the drywell to the high temperatures resulting from core /cencrete interactions [2]. The CL'aC developed a MARK I standard problem specifically designed to address temperature predictions in the drywell, m
The predicted pressure /terperature histories during core /concre:e interactions are sensitive to model assumptions and several parameters such as time af ter scram, temperature anc composition of the core materials, concrete type, core / concrete interf acial area, and upward heat transfer from the core materials. Scme of these parameters are accident secuence depencent (e.g. ,
time af ter scram and initial temperature and composition of the care materials) and others are plant specific (e.g., concrete type and corc/cencrete interf acial area). Sensitivity studies 'aere perforfrec by the CLWG covering a'i of the above paran;eters using different model assumptions [1]. The CLWG predicted pressure /temperaturc histories, using various model assumptions, as provided in Figure 2.3.1 [10]. Clearly, the dryweil temperature can vary widely depending en mcdel assumptions, particularly the initial moiten core material (coriun.) temperature, the interf acial area, and the concrete type.
Refer to Reference [1] for discussiens of the details of the sensitivity studies.
The severe accident analyses carried out by the ICCOR program were der.e with the 1r.tegrated systems analysis code FAAP [7] (Mocular Accident Analysis Prcgrans). The core / concrete interactions aspects of the analysis used the DECCMP[4]mocel. In the MAAP analysis, convective and radiative neat release frem tne debris pool surface are coupled to natural circulation processes in the containment.
26 Also, unlike the C0RCON code, the CECOMP model in MAAP is one-dimensional; it treats concrete ablation of'the sidewalls of the cavity the same as hori:: ental surfaces. These two differences in mathematical approach preclude the expectation that computational results could ever be expected to be identical. The other significant differences in the 10COR approach are:
- 1. It is assumed that at the time of vessel failure, only a fracticn of the original core mass exits the vessel; debris are assumed to continue to flow into the reactor cavity for a period of hours.
- 2. In virtually all cases, a crust is assumec to exist on the surface of the dabris.
- 3. In all cases where water exists above the core debris, the debris pool is assumed to'be quenched.
For all the above conside. rations, it is evident that direct comparison o' ICCOR results with the NRC predictions cannot readily be made.
The ex-vesse' heat transfer model from molten core to concrete is an issue of concern f:r PARK I containments because heat transfer from the tcp of molten core materialt tun the crywell floor) directly heats the drywell atmospnere.
Thus, atiferences in heat transfer from the top of the core debris result in significant differences in the predicted drywell atmospheric pressures and temperatures. The 'CCCR model transferred more heat from the top of the core debris than the NPC model. Thus, 10COR predicted much higher drywell terrperatures tnan the HEC analyses. However, because IDCOR predicted high hea t transfer from the tcp of the core cebris, the concrete erosion velocities were much lower than the 'IRC predictions. Lcwer concrete erosion results in less gases anc aeroscis released frcm core-concrete attack and thus lower pressurcs in containment. Therefore, IDCOR predicted centainment f6ilure because of higr drywell temperatures; whereas NRC analyses predicted containment fcilure because of high pressure. In addition, IDCCR predicted much longer tir.es to cor.tainment f ailure due to overtenperature than those predicted by the NRC analyses due to overpressure, ,
27 2.3.2 Potential Failure Modes Earlier. certainment analyses of the BWR MARK I concluded that the mest likely failure would be from overpressurization of the drywell. The uhimate failure pressure of the Peach Bottom MARK 1 containment was estimated ;o be predominantly in the range of 117 psig to 138 psig by the draf t flVP.EG 1150
- expert review group. A recent study by Chicago Bridge & Iron (CBI'; has indicated a containment ultimate failure capability of 159 psig based on their
~
assumed strain failure criteria. In addition, the CBI study indicated centainment leakage through the upper head region of the drywell at pressures lower than the ultimate, failure pressure [44]. The possibility of seal failure resulting from high drywell temperature was also recognized. The sensitivity of these two types of failure modes has been examined and was reported in Reference [1] with respect to concrete composition and debris temperature. In g Table 2.3.1 frcm Reference [1], some of the calculational results are shown.
U The input data were based on the assumption of a TCUV accident scenario (loss of all coolant injection at scram and f ailure of the automatic depressurization system) at the Browas Ferry Nuclear Power Station. The calculated times te crywell failure fron overpressurization and high temperature are ccmpared in Table 2.3.1. These results were derived thrcugh application of the CORCON-FCC1 cece in conjunction with the MAC.E subroutine in MARCH 1.18. The importance of concrete cor.pesition is cemonstrated; the higher pressures associated with the two limestene cases reflect the higher gas generation associated with limestore deccmposition.
As noted earlier in Section 2.1.2, an assessrent of Peach Bottom containment performance was made in conjunction to the ongoing QUASAR research program in 5'il. The break area /containter.t pressure relations suggested by the OUASAR experts review group are reproduced in Figure 2.3.2.
As shcwn in Figure 2.3.2, in the first part of co,ntainment pressure / break area curve, the break area increases gradually as a function of pressure, which represents leakage from the drywell head. The sudden step increase in the break area occurs at the estimated f ailure pressure of the containment drywell shell (cylindrical part), which would most likely rupture, resulting in a very large break area. Two groups of curves corresponding to two different
l l
l 28 h\
containment temperatures were provided by Clauss (9]. Each group contains upper beund, median, and lower bound estimates for the break area. These correspond to different assumptions on the resiliency retained by the seal material and on the yield stress of the containment shell material. The median ,
estimate for the silicone seals used in the Peach Bottom plant is 10%
resiliency, with an upper bound of 25% and a lower bound of ?%. The major uncertainties of these estimations were due to the variability in the preload of bolts (50 kips per bolt useo in estimation), depending.on the ,
lubricant and the actual torque applied to the bolts as well as on relaxation effects. The genert.1 yield pressure establishes a icwer bound for failure due to shell rupture; the median estimate corresperds to a pressure 1.2 times tne general yield' stress (which is the ratio of the rupture pressure to general yield stress obtained in the Sandia 1:8-scale steel model test), and the upper bound is based en an estimate of the additional strength imparted to the shell-after contact with the reactor shield building.
The containrent break area / pressure relation suggested by Ahl [9 ) fcr varicus "gasket springback" levels is also shown it. Figure 2.3.2. In this estimatien, a bolt preioad value of 107.3 kips per bolt level was usec anc it was also assured that there is a 50 F temperature dif f erential between the containment shell and the bolts. Ahl also pointed out that the estimated brea, area is very sensitive to tne bolt preload.
At this time this issue is not considered risk significant for MARK I containments.
2.4 Contairment Bycass ,
2.4.1 Definition of Chal'enge The most important systems for ritigating hign-consequence sequences in a EVR are the centainrent ard its suppression pool. As long as any release path e
l l
l l
I l
l O 29 during an accident is forced through the pool (e.g. , during wetwell venting),
the pool will act to reduce the environmental releases. Containment bypass occurs when a path of sufficient size to allow relief of gas and fission .
products is opened between the drywell and the reactor building. This can lead to significant releases of radicactivity.
Containment bypass is thus a potentially important challenge to MARK I containments; and includes the following:
(1) Suppression pool bypass (2) Interf acing systems loss-of-coolant accident (3) Failure to isolate containment on demand ,
I l
(1) Suppression pool bypass If the fission products pass through the suppression pool, both 10COR and l NRC analyses predicted significant retention of fission product aerosols in the water. The amount of retention depended on several factors such as submergence, water temperature, aerosol particle size, and carrier gas composition. The ability of the PARK I suppression pool to trap aerosol fission products was found to be an important mitigative feature. Thus, any pathways that might open, which would allow the fission products to bypass the pool are very i undesirable.
If containment failure occurs in the drywell, any fission products in the crywell atresphere could pass to the reactor buildirg and frcm there to the environment witnout the benefit of suppression pool scrubbing. This can result in substantial environmental releases of radionuclides. Suppression pocl bypass cculd also occur due to a failed (stuck open) SR" tailpipe vacuum breaker, or f t.iied drywell-to-wetwell vacuum breakers during severe accident conditions. Because of the MARL I vulnerability to pool bypass, the predictec offs 1te consequences were relatively insensitive to the definitien of the O
V accident sequence. In addition, differences in the 10COR (l'7] anc NRC [5]
assessments of containment response and fission-product release aise did not result in major differences in the predicted offsite consequences. The only
30 g timethatamajorreductioninoff-siteconse'quenceswaspredictedbyIDbORand NRC was with successful wetwell venting and no suppression pool bypass. The IDCOR analysis indicated a lower potential for suppression pool bypass af ter containment failure than the NRC analysis.
As an examplo, the dominant accident sequences, namely Anticipate 0 Transient Without Scram (ATWS) and station blackout, result in a significant prcbability of suppression pool bypass.
(2) Interf acing systen.s loss-of-coolant accident The interfacing systems LOCA (the so-called V sequence) is caused, for example, by the failure of two valves in series that isolate the Icw pressure injection system from the high pressure reactor coolant systen, (PCS). If this were to occur during cperation, the high pressure in the RCS would be imposed on the low pressure injection system. Overpressuri:stion of low pressure system may result in rupture of low pressure piping This, if combined with failures in the emergency core cooling systems (ECCS) and other systers that ray be used to crovide makeup to the reactor coolant systen, wculo result in a core melt accident. Any radioactivity released into the RCS can escape out o' the containmert. This accident sequerce, therefore, may defeat the plant ccntainnent features designed to mitigate accident consequences, arc can leac to very large releases cf radioactivity to the environment.
In general, E'r:R MARK I PRAs have found the interfacing system LOCA to be a highly unlikely ever.t (les5 than l'i/ reactor year). The ccre damage frequency (CDF) due to interf acing systems LOCA varied f rom regligible to 2.0x10"'/yr.
This hignest ;3F, ie., 2.0 t 10" /yr, which was obtainea in the Shoreham PPA review, was ma rly because cf the fact that the automatic actuation of the ECC?
i after ur inte" oc'ng system LCCA would cause flooding of the reactor building.
This in turn had the potential to flood the ECCS pumps in a very short time (ir l some cases less than 10 minutes). Therefore, in the Shoreham PRA and its l review. creoit was only given to injection with the ccrdensate punps with mak eup to the hotwell (18]. However, other generic stucies (19.20) have obtained CDFs that were much higher.
l l
O 31 An interfacing systems LOCA could initiate a core melt leading to fission product release into the secondary- containment, bypassing the primary containment boundaries. Therefore, the challenge is one of accident prevention. Mitigation is to reduce potential offsite consequences. ,
NRC requires establishment of system reliability, periodic inspections, implementation of post-THI-2 requirements, and reviews of such interf aces in implementing fire protection requirements so as to reduce the likelihood of such events.
Cu'rrent estimetes are that interfacing LOCA's contribute little to the ,
overall risk at MARK I BWRs. Should future estimates indicate an increased risk contribution, further att'ention would be required due to the high pctential consequences of these sequences, b
V (3) Failure to isolate containment on demand .
Failure to isolate containment on demand was categorized in WASH-14CC
[24). It refers to excessive leakage rather than the total loss of containment integrity, and would leac to the loss of noncondensibles f rom the containmen .
Any accident will proceed almost independently of the holes in the containment, except that the pressure build-up will be reduced de;ending on the hole si:e.
There are many possible causes of containment isolation failure, mestiy related to open valves or failed seals in centainment penetratiens. 'those which have 2
occurred involve small openings belcw 0.01 m and usually below 6.5 x 10-5m;'
2 (0.1 in ). Only a small fraction c'f the fission products have been postulated tc escape through openings in this size range. However, it is not inconceivable that a large equipment hatch could be left open during shutcown or refueling cperaticns.
A study performed at Ook Ridge National Laboratory [21] to determine the contribution to pcpulation risk as a function of leakage rate incicates that for leakage rates of less .ian about 10 percent per day, the risk (person-rem / reactor year) cue to leakage during a severe accident is smeli ;
i compared with the risk from c7ntainment failure and that the risk due to leakage varies roughly linearly with the leakage rate. The Oak Ridge analysis l l
32 g used the "Siting Source Terms" based on Reactor Safety Study methodology, anc contained in NUREG-0773 [22], to calculate that contairment leakage at a rate of 1 percent per day would contribute about 0.2 percent of the total expected risk in person-rem due to all reactor accidentt. The results for other leakage rates were linearly extrapolated as shown in Table 2.5.1.
Infomation from operating experience indicates that there is a possibility that reactor containment cculd leak at rates exceeding the allewable leakage rate. Probabilities of various leak si:es anc rates have been estimated fron infomation centained in Licensee Event Reports from 1965 thrcugh 1985 ar.d Containment Integrated Leak Rate Test (CILRT) reports that have been recently compiled [23]. Caution should be exercised in the use of these data, however, because the range of uncertainty has not been analy:ed. The likelihood of varicus leakage rates (or areas), estimated on the basis that single valv.es in a train are open, is sumarized in Table 2.5.2 based on data averaged over all plants [5]. As might be ex;ected, the larger the leak , .he smaller the occurrence probability, e.g., the prooability of simultaneously opened airiocks is estimated to be about. 5 X 10-5 . For BWRs, the probability of an existing leak with an area of 0.6 square inch or less has been estimated to vary from 4 percent to 16 percent for various hole sizes. Overall, there appears to be about a 30 percent prcbability that cantainment leak rates would exceed the alicwable rates by factors of cre to ten ar.c up to a 1 percent probability of leak rates in excess of 100 percer.: per day. (Cecause of the ccmplexity of the des'.gn of containment cenetrations and the lack of standardized designs, estimates will vary from plant to plant).
This information can be useo to detemine the need for additicral assurince that containment integrity has not severely deterioratec curing the period between Type A* tests, which are nnw conducted abcut once every 3 years.
2.4.2 Potential Fcilure Modes O
- Type A tests involve pressurizing the entire containment and measuring the leakage rate.
O 33
.(1) Suppression Pool Bypass Suppression pool scrubbing is an effective mitigative feature for the MARK I containment, provided all of the fission products pass through the pool. It is, therefore, important to ensure that paths do not open which would allow the fission products to bypass the suppression pool.
The following are possible ways in which the suppression pool may be bypassed:
- loss of drywell isolation
- f ailure of vacuum breakers between the drywell and wetwell
- failure of orywell penetration seals because of high temperature
- failure of main steam line isolation valves (MSIVs) to close
- structural failure of the drywell because of high pressure
,- failure of the drywell wall as a result of contact with molten core materials (see Section 2.2). l (2) Interfacing Systens LCCA 9 The interfacing systems LOCA (a sc-called "V" sequence) is caused, for exarrple, by the failure of two valves in series that isolate the low pressure injection system from the high pressure reactor coolant system (RCS). If this were to cccur curing operation, the high pressure in the RCS woulc be imcosed I en the low pressure injection system. Overpressurization of the low pressure systen may result in rupture of low pressure piping. This, if combined with failures in the emergency core cooling systems (ECCS) and other systems that j may be used to provide makeup to the reactor coolant system, may result in a !
core melt accident. Radioactivity released into the RCS can escape cut of the l containment. This type of accident secuence, therefore, may defeat the plent contair. ment features designed to mitigate accident consequences, anc can lead tc very large releases of radioactivity to the environment.
O in general, BWR MARK I pRAs have found the interfacing system LOCA to be a highly unlikely event (less than 10 / reactor year). This value may be indicative of a class of non-dominant severe accident sequences. .
~
34 (3) Failure to isolate Containment The location as well as the size of containment isolation failures can have a bearing on the overall accident consequer.ces. If the containment isolation failure takes place in the drywell, fission prcducts released f rom the core may leak directly to the secondary containment. In the event of secondary containment f ailure this may result in a larga release of radioactivity to the environment, even if gross f ailure of the primary containrent is avoiced. For isolation failures in the suppression chamber gas space, on the other hand, the leakage of fission products will be through the cool and then out to the secondary containment. The water in the pool would be expected to retain most of the particulates and a large fraction of the iodinc.
Thus even in the event of secondary containment failure there would be a significant reduction in the fission-prcduct release for these cases.
2.5 Racid Stean pressure and Missiles 9
2.5.1 Cefinition of Challenge Rapid steam pressure rises anc missiles resulting from hot fuel-coolant interactions are a potential challenge to the BWR MARK I containment. Stear or vapor explosiens cccur when two liquids at different terperatures ccme into physical conuct in such a w3y that the internal energy, transferrec from tre hot liquid to the cold liquid, causes rapid boiling. Explosive rapid boiling can result. It is theorized that the most violent type of volcanic eruptions i are stear explosions (28). Explosions have also occurred in the tretal crocessing ir.dustry. Extensive damage in steel and aluminun foundries results l f rom accider. tai spills cf molten metal into water [29]. Similar steau spikes and explosi0ns rave been cbserved in charging electric melt furuces with scrap metal thet haocens to contain rain water. Every year, about 1 percent of the pulp recovery boilers in Ncrth America are destroyed when water is accicertally intrcduced onto the smelt [203 Cryogenic-water explosions cccur *n'en a cryogenic liquic such as liquefied natural gas is poured ento water [3: 2 If an accident at a nuclear plant led to severe core damage, continuing to onset of core melting, debris from the dar. aged core would at some point begin
35 to fall into the lower plenum of the reactor vessel. If an appreciable amount of water remained in the lower plenum, moiten core material falling into it could potentially cause a steam spike and if severe enough, an explosion.
Indeed, there is evidence of a steam spike during the Three Mile Island accident when coolant pumps were turned on several hours into the event. This would be associated with sudden disintegration of part of the melted core material into fine particles. The rapid increase in '.,urface area could generate large amounts of steam through a sudden increase of heat transfer, before pressure relief through expansion could take place.
The pressure surge in the pool of water could possibly cause some water and some of the core and structures above the pool of water to be accelerated upward as a slug. If this were to impact the vessel head with enough energy, it could break the bolts holding the vessel head in place, and accelerate the head upward as a missile, threatening the containment structure. This mode of O eerix conte 4mmeet feiture es treetee 4n wiss-14co c243, ene wes celiee en "alpha-mode failure". The mechanism of a steam-explosion-induced slug acceleration anc impact was demonstrated in the SL-1 accident, where the impact of the water slug severely strained but did not fail the vessel upper head
[32].
For a large, energetic steam explosion that challenges vessel integrity to be possible, a number of factors must occur simultaneously. The progression ano extent of r.elting in the core would be important because, for example, an upper limit to the amount of heat that can generate steam quickly would be set by the fraction of core falling into the water pool in a molten state.
Furthermore, the explosion, if it were to occur, would be very brief and would only involve that part of the melt falling into the pool in a relatively "coherent" manner. There is a consensus that a certain amount of breakup of I the molten mass into smaller sized pieces dispersed in the water is necessary.
This f ragmentation is called "premixing" . premixing can take place thrcugh the ef fects of turbulence and viscosity as the melt f alit through the water.
O Detonation started in part of the melt can propagate through the water to otner car'.s of the melt, triggering a Fore global enercy release. bnder
, certain circu.mstances even th.e trapping of a small amount ot 5:ater between a
36 wall and a mass of molten core impinging on it can generate a steam burst that initiates a steam explosion.
The kinetic energy of any slug of material accelerated upward or downward .
by the detonation is found from experiments to be at most a small fraction of the sensible heat energy in the molten mass falling into the water. Most measurements indicate that this conversion to mechanical work is a very small fraction of the total heat energy in the melt. Some experiments in the d.X.
have shown that only part of the molten mass falling into the water participates in a steam explosion. During UK experiments (about 25 kg of melt was used), under certain, conditions of geometry and pressure, up to 80% of the available melt participated in the steam explosion. However, analysis is leading to increasing confidence that this fraction would be much lower for equivalent conditions at the reactor scale. Premixing is very difficult (or impossible)atlargescales[16].
The probability of a steam explosion causing early containment failure, @
assuming an accident leading to a severely damaged core, was assumed in WASH-1400 to be between 0.1 and 0.01. A review by the NRC-sponsored Steam Explosion, Review Group (SERG) [33] led to a nearly unanin,ous opinion that this prcbability should be at most about 0.01. This was followed by a team review using senior analysts. Their conclusions [34] were that the contribution to risk frcm this class of events can be neglected, and uncertainties in the failure probabilit" are not a dominant consideration. Theofanous [35] has oeen evaluating the fa e.ure probability using the best information available on the statistical distr,cution functions for tre variables describing the phenomenon.
He uses these distribution functions in a probabilistic analysis that explores the different aays a steam explosien scenario could progress, and the relative likelihoods cf these patnways. This is an advanced version of analysis previcusly done by Corradini, Bermn, and others [37,3S], using simpler models.
He firds the probability c.f containrent failure to be smaller still than the SEPG limit by at least two orders of magnitude, thcuch with considerablo nurarical uncer'.airty in the result [16].
Exper' rents on stccu explcslcns have boor. perforn ed J*. a number s-locatiens, and raw been underway for scc.e jears [4. Cf these, ;he tests of
O 37 Koopmans, et al, [41] where LNG was poureo on water have a scale similar to that of a reactor accident.
Other experiments using materials more directly relevant are from Sandia National Laboratory (SNL) [38,42-44]. They report induction (delay) times of 15-300 ms between initiation and explosion in their experiments and conversien of thermal to mechanical erergy of up to 1.6% [16]. These experiments have provided a phenomenological dev.ription of steam explosior.s that has been developed by many investigators into a probabilistic de'scription of the progression of events. ,
SERG was convened to provide estimates of the likelihood of alpha-mode failures. The review group consisted of experts from universities, national laboratories, and the nuclear industry and included both domestic and foreign i representatives. The findings of the SERG are documented in NUREG-1116 [33),
O ene were er4e<>> ineicete4 ia sectica 2.8.1 or th4s report. Ta4s was roilowee byareviewteamheadedbySNL[34].
Steam explosions are well enough understood in a descriptive sense for use of a statistical treatment of the sequential steps in the process. The most sophisticated treatment of this kind is due to Theofanous [35]. A nuccer of
~
analytical stucies have been made of premixing, based on turbulence and the presence of Rayleigh-Taylor instabilities [16]. !
2.5.2 Potential Failure Modes l l
A potential mode of early containment failure, caused by a steem explosion-inducea failure of the vessel head and its acceleration upward as a ;
missile, has been called ' alpha-mode failure". This is also a potential !
l failure for the BUR MARK I containment. j i
l Though it is not simply the total kitietic energy of the upwaro slug that micht t,reak tne head bolts ar.d generate a missile, the phenomenon is usually O discusseo in terns of this energy. Calculations indicate that mcre than 1500 cego;oules of energy woulc tie required to sever the t,ults. Some calculations
38 g
have indicated that because of the heating of the upper head bolts through natural circulation in the vessel containing the damaged core, the head bolts might fail from impact by a slug with a kinetic energy some 500 megajoules l e s s .- On the other hand, the lower head of the reactor vessel could be broken free by a detonation that propelled the slug upwards with only about 500 megajoules of kinetic energy, because of the reactive forces associa'ted with upward propulsion. Failure of the lower head would greatly reduce the energy of the upward slug (16].
Two other mitigating points deserve mention. As the pressure is increased, a more energetic trigger is required to induce a steam explosion.
Above a pressure rise of about 10 bars (150 psi), an explosion would require an energetic external trigger [39,40]. Second, the efficiency of the detonation is seen to be reduced substantially by intervening structures and by unmelted fractions of the material dropping into the water pool.
Besides alpha-mode containment f ailure, there can be other generally less sericus but still possibly significant consequences of steam explosions that may occur during the progression of severe accidents. They include the potential for direct snock failure of the vessel icwer head and its consequences, the effect of debris dispersal anc nonexplosive steam generation uccn in-veste; core .relt orcgression, and the effect of steam explosions cr.
additional hycrcgen generation and fission product release to the containment atmosphere. Furtr.erciore, the possibility does exist of an ex-vessel steam explosion or a steam spike resulting frcm rapid interaction between core debris and water in flooceo crywells or in the pressure-suppression system. For the latter, the quantity of molten debris has to be suf ficient to flow out of the reactor cavity and, via the vent-pipes, drop into the suppression pool.
However, studies of steam explosions have concluded that events with significant energetits in-vessel are uniikely, and the energetics in e.v-vessel interac*icns aould be even lower. Therefore, this issue is not consicered as r19 signif icent ' o r "A ' con ta t or. e n t s .
2.5 tunnary of Dmor* :r e Cont.iiwent Challsro Issues i
.- ~ _- ._ _ - . - . . . . ..
O 39 The importance of phenomenological issues is determined by the oegree to which they possess the following attributes, namely, (a) early containment failure time, and (b) driving force for potentially large radiological release into the environment.
Containment challenges for which sufficient time is available to devise accident management strategies are of less significance, provided systems already exist to deal with containment challenges involved. Table 2.7.1 lists {
the timing of key events during some of the important Peach Bottom scenarios as j calculated by the Source Term Code. Package [2,45],
Based on the discussions given in this chapter, the most significant containment challenges are:
l (1) core debris attack on the steel shell, .
! (2) early overpressurization, and ,
I (3) isolation and bypass modes.
It is recognized that measures to mitigate significant early containment l challenges could also reduce the challenges associated with late containment l
! failures.
OUESTI0ftS i l
- 1) 'Are the discussions of phenomenological issues adequate? Are there other important mechanisms which can challenge MARK I containmente involving phencrrena not enccmpassed by the identified issues?
- 2) Are the magnituces of the parameters representing potentia' containment challenges all included " the ranges identifie(, or can i credible sequences be postulated siitt. ,alues which exceed those included in the issues as discussed?
i j O 3) The manliestar.ioni of the phenutena can be af fected by hurcan acticns 4
t,et.h prior *.o the accident and through errors in, accident diagnostics i ano a:enagement. Dc the identified phencrenological issues irrply
40 g undue reliance upon favorable maintenance, operation or mitigative behavior. Conversely, are there~ reasons to assume that any of the phenomenological issues can be further narrowed by reliably predictable human behavior?
- 4) Can any of the phenomena be either precluded or limited in magnitude of challenge to the containment by a feasible addition to or removal from any region of the containment of materials, or by alteration of current maintenance or operation procedures? Are there other phenomena which diminish the challenges?
- 5) Are there separate effects experiments which have the potential for demonstrating low likelihood or limited effect of any of the phenomena?
- 6) Are any of the phenomenological' issues related to one another or cctbina tions? If so, how are they related?
- 7) Is the containment response (failure modes, locations of leakage paths and tear locations) to overtemperature/ overpressure challenges evaluated for Puch Bottom [44] likely to be similar for all MARK I cc r.ta i nmen ts ?
l l
l O
- . i l
l l
l 41 2.7 References
- 1. USNRC, "Estimate of Early Containment Loads from Core Melt Accidents", A Report by Containment Loads Working Group, NUREG-1079 Draft, (Dec. 1985).
- 2. R.S. Denning et al "Report on Radionuclide Release Calculation for Selected Accident Scenarios", BWR, MARK 1 Design, NUREG/CR-4624, BMI-2139, Vol 1, (July 1986).
- 3. M. Khatib-Rahbar, et al. , "Independent Verification of Radionuclide Release Calculations for Severe Accident Scenarios", NUREG/CR-4629 BNL/NUREG-51998, (July 1986).
4 C.N. Amos, et al., "Containrent Event Analysis for Postulated Severe V.ci!ents: Peach Bottom Atomic Power Station, Unit 2", ilVREG/CR-4700, O. S = 86 nas. vol. 3. (nex 1987).
- 5. "Reactor Risk Reference Document", fiUREG-1150, Draf t (February 1987).
- 6. J.A. Gieske, et al., "Radionuclide Release Under Specific LWR Accident Conditions" Vol 2, BWR MARK I Design, BMI-2104, (July, 1984).
i
- 7. "l'AAP Analysis of the LaSalle County Station for Postulated Severe Accidents, Fauske & Associates, Inc., FAI/85-41, (October 1985).
- 8. USNRC "Containment Perfomance Working Group Report" NUREG-1037, Oraf t Report for Coment, (May 1985).
- 9. S.D. Unwin et al., "The Formulation of Probability Distributions For The CUASAR Program" Brookhaven flational Laboratory, Technical Report A-3286,
) (September 1987).
, 10. G.A. Groene, K.R. Perkins and S.A. Hooge, "!1ARV ! Containrent Drywell:
Impact of Core / Concrete Interactions on Containment Integrity and Failure
) ,
of Dryweli Liner", Proceedings of An 'ntrenational Symposiun on Source Term Ew:luation f or Accident Ccnditions, ; AEA neeung; Coluxous, Ohio (19E6).
42 g
- 11. S. A. Hodge, Enclosure to Memorandum Entitled "MARK I Containment Meeting (February 3,1987) Summary-Draf t" . From L.G. Hulman to Distribution , U.S.
Nuclear Regulatory Conmission; February 6,1987.
- 12. "Approximate Source Tem Methodology for Boiling Water Reactors" Fauske &
Associates, Inc. FAI/86-1 (December 1986).
- 13. L. G. Hulman; "Sum.ary of March 27, 1987 Meeting with BWR Owners Group /IOCCR on MARK I Containments"; USNRC; May 6,1987; and "Sumary of MARK 1 Containment Meeting" with NRC Contractor Personnel; USNRC; Feb. 24, 1987.
14 R.K. Cole et al. , "CORCCfi/ Mod 2, A Computer Program for Analysis of Molten Core-Concrete Interaction, tiUREG/CR-2920 ( August 1984).
- 15. R.E. Davis, M. Lee, E. Cazzoli and 11. Khatib-Rahoar "QUASAR Screening Sensitivity Analysis: Applicaticn to the C0RCON and VANESA Codes", BNL Technical Report A-3236, 6-22-87 (1987).
- 16. H. Kouts, "Review of Research on Uncertainties in Estimates of Source Tems f reu Severe Accidents in fiuclear Power Plants", NUREG/CR-4883, Bt;L-NUREG-52061 (1987).
- 17. "Technical Support f or Issue Resolution", 10COR Technical Report T35.2, (July 1986).
- 18. W.T. Pratt, et al, "Assessment of Severe Accident Prevention and Mitigation Features: 3WP,l'. ARK I. Containment Design",
NUREG/CR 4920, Vol '. (December 1987).
- 19. P. Lam, "Overpressurization of En:ergency Core Cooling Systems in Boiling Water Peactors, "AE00(NPC). (February 1986).
O 1
c 43
- 20. J.D. Harris and J.W. Minarick, "An Evaluation of BWR Overpressure Incidents in Low Pressure Systems", ORNL Preliminary Report, (Fiay 1985).
- 21. 0.W. Hemann and T.J. Burns, "Impact of Containment Building Leakage on LWR Accident Risk, "Oak Ridge National Laboratory, NUREG/CR-3539, ORNL/TM-8964, (April 1984).
- 22. R. Clond et al., "The Development of Severe Reactor Accident Source Tems 1957-1981",NUREG-0773,(November 1982).
- 23. P.J. Pelto, K.R. Ames, and R.H. Gallucci, "Reliability Analysis of Containment Isolation Systems, "NUREG/CR-4220, PNL-5432, (June 1985).
24 U.S. Nuclear Regulatory Comission (USNRC), "Reactor Safety Study--An Assessment of Accident Risks in U.S. Commercial Nuclear Power Plants".
O
~
W^ss-1400 (nuREG-75/014). (October 197s).
1
- 25. Letter to all LWR licensees on "LWR Primary Co'olant System Pressure Isolation valves", frem Darrel G. Eisenhut, Acting Director, Division of Operating Reactors, Office of Nuclear Reactor Regulation, USNRC, (February 23,1980).
- 26. "Order for Modification of License Concerning Primary Coolant System Pressure Isolation Valves", Letter f rom Darrel G. Eisenhut, Director, Division of Licensing, Office of NRR, USNRC, to Eugene R. Mathews, Vice President, Power Supply and Engineering, Wisconsin Public Service Corporation,(April 20,1981).
- 27. L. Chou, Brookhaven National Laboratory, Personal Ccmunication, (Dececber 1987),
- 23. K.H. Wchlet: and R.G. McQueen, "Experipental Studies of Hydrcragnetic Volcantin", Studies in Geophysics, National Academy Press, p. 158. (1984).
O .
j
. l 44 g
- 29. P.O. Hess et al., "Molten Aluminum / Water Explosions", Proceedinr,s of
. Technical Sessions Light Metals 1980, American Institute of Mechnical Engineers, p. 837,(1980).
- 30. T.M. Grace R.R. Robinson, and J. Hopenfeld, "Steam Explosions: Energy Conversion Efficiencies of Steam Explosions from Two Major Accidents in the Pulp and Paper Industry", Proceedings of the USNRC Thirteenth Water Reactor Safety Research Information Meeting (Gaithersburg, MD),
NUREG/CP-0072, Vol . 4, p. 443, (February 1986).
- 31. T.G. McRae, "Large-Scale Rapid Phase-Transition Explosion", Lawrence Liverrore National Laboratory, UCRL-88688, (May 1983).
- 32. J.L. Kin:e et al. , "Fuel Report on the SL-1 Recovery Operation",
100-19311,(1962).
- 33. Steam Explosion Review Group, "A Review of the Current Understancing cf the Pctential for Containment Failure from '-Vessel Steam Explosions",
NUREG-1116, (June 1985).
3a. V.L. Behr et al. , "Containment Event Analysis for Postulated Severe Accicents at tne Secuoyih Nuclear Power Plant". Sandia Nauvrcl Laboratories, SAN 026-1013, (March 1986).
- 35. T.G. Theofanous et al., "An Assessment of Steam-Explosion-Inducce Containmtnt Failure", Parts I, !!, III, and IV, Nuclear Science and Engineering, Vol. 97, No. 4, pp. 259-326 (Cecember 1987).
- 36. ft.L. Corradini and 0.V. Swenson, "Prcbability of Containmcrt Failure Cue to Steam Exolosions During a Pcstulated Core Meltdcs.n Accident in 1 LWR",
Sandia "ational Laboratormt , SAND 80-2132 (1981).
- 37. P, Perman, et c'.. "r. ore Felt / Coolant Interactions: Mucel l : m' . 'andi.
National Laboratories, SA"C 33-1852C., (1983). g
a i
45
- 38. M.L. Corradini, "Analysis and Modelling of Steam Explosion Experiments",
~
Sandia National Laboratories, NUREG/CR-2072, (1981).
- 39. R.E. Henry and H.K. Fauske, "Nucleation Processes in large Scale Vapor e Explosions", Journal of heat Transfer, Vol. 101, pp280-287, (1979).
- 40. R. Wilson, et al., "Report to the American Physical Society of the Study Group on Radionuclide Release from Severe Accidents at Nuclear Power
- Plants", Reviews of Modern Physics, Vol. 57 No. 3 & Part 11 pp. $1-5154, ,
(July 198S).
- 41. L.D. buxton and W.B. Benedick, "Steam Explosion Efficiency Studies" .
Sandia National Laboratory, SAN 079-1399; NUREG CR/0947, (1979).
- 42. L.D. Buxton, W.B. Benedick, and M.L. Corradini, "Steam Explosion '
O Eu4ciencx Studies", eerk u, Saedia Nemeet tebmtory, SAND 80-u24, NUP.EG/CR-1746,(1980). -
- 43. D. Mitchell, M. Corradini, and W. Tarbell, "Intermediate Scale Steam i Explosion Phenomena: Experimen a and Analysis", Sandia flational !
Laboratories, Sand 81-0124, NUREG/CR-2145,(1981). ;
I l
Aa. Mokhtarian et al., "MARK 1 Containment Severe Accident Analysis, CBI, April 1987.
- 45. M. Lee, Brookhaven National Laboratory, Personal Comunication, January 1988. j i j
- 46. R. G. Bird, Letter to S. A. Varga, NRC, "Information Regarding Pilgrim ;
Station Scfety Enhancement Program"; July 8, 1987.
l I
O
46 g
3 POTEllTIAL IMPROVEMENTS 3.1 Hydrogen Control Post-accident containment hydrogen control at MARK I plants is provided primarily through inerting (using nitrogen to reduce the oxygen concentration below 4%), recombiners or active purge repressuri:stion systems, and the elimication of the major sources of oxygen inleakage. I 'e demand for nitrogen will be influenced by the venting strategy that results from long-term accident management. For example, venting may cause air to be introduced into the containment resulting from vacuum breaker operation following venting. If so, long-term reinerting will be necessary. Two areas of potential improvement for severe accidents have been identified as follows:
- 1) Existing Technical Specifications (terms of the reactor cperating license) allow periods of up to 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> of power operation while the containment atmosphere is above 4% oxygen. Prior to a scheduled shutdown air may be pumped in to displace the inerted atmosphere for up to 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />, and af ter startup 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> are permitted for nitrogen addition.
This ta hour allowance could be reduced or eliminated to further tiinimize the likelinced af sufficient oxygen being present during an accident to surrort hydrogen cc-bustion.
- 2) Assuring that oxygen sources are eliminated for Icng-term hydregen control would be achieved by an adequate nitrogen supply for ler.g-term make-up. This improvement was identified by Boston Edison for Pilgrim.
It consists of the addition of a liquid nitrogen /vapori:er trailer, assorinec pioing and valves. A n:odificatien to a containment iso'ation valve was also to be made to acccrurodate actions during a station blackout. Primary systen air sources f or instrument and valve operatier j in certainment were to be elininated under in:plementation at W W 50.43 in Gener'c Le.ter r/-09. Bos'.on Edison bas suaqes tml that bac',-up ,
wt:p l i e s .t g a l t o tm irrrrrtant oxygen sources during seveie xcidents. l O l
- ___ _ _ - -- -- _ J
47 CUESTIONS
- 1) Do the periods of deinerted operation allowed by technical specification (24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> before shutdown and 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> upon startup) present a sufficient vulnerability to necessitate a reduction?
- 2) Are the nitrogen supplies provided at MARK I plants sufficient to maintain an inerted atmosphere for long tem station blackout scenarios?
- 3) Are the risks of oxygen addition from compressed air supplies used for primary and back-up operation of valves and instrument lines too high? If so, should nitrogen back-up supplies be sutistituted for oxygen supplies?
3.2 Containment Sorays i
O all cwas with naax t coateinmeats except Ox<ter Creex ead nine sile eoi#t 1, are equipped with a dual header drywell spray system. The two spray headers are rings located well up in the cylindrical part of the drywell With branches holding spray nozzles pointing down at an angle. The headers are fed through each division of the RHR system with spray operation'as an alternate mode of RHR operation. Due to the characteristically large size of RHR pumps (2000 -
10,000 gpm) the drywell spray has a very high flow rate. Precautions are usually included in operating procedures to avoid excessive use of this powerful spray system. Oyster Creek and Nine liile Point 1 have separate dedicated spray systems, Most plants have other systems already connected to the spray header feed
- lines outside of containment. They include such systems as RHR Service Water, Condensate, and in some cases, bolted blind flanges which are removec to install l'ines for periodic containment integrated leak rate tests.
Thus it is
- pcssible for a plant to provide one or more backup supplies for the drywell
- spray, even in tha event of a stdtion blackout, because of the availability of' l fire suppressicn systems with independer.t pumping capability. But the
( - nailable backups are all smaller systems, on the order of 10'J the size of the PFR. If they were used they would prebably nut be able to develop sufficient header pressure for even spray flew distribution in the drywell, it has been
48 g
suggested that by blocking most of the spray nozzles, a more even spray flow distribution could be achieved using the smaller systems.
Assuring a drywell spray even ir the event of a station blackout may decrease the severe accident risk based on the following considerations [1):
- 1) Reduction in likelihood of containment failure due to:
a) steam overpressurization and overtemperature-induced failures, b) direct drywell shell attack and f ailure by potentially retarding corium ficw towards the shell and by shell cooling.
- 2) Use of fire suppression systems with independent pumping capability can result in a reduction in 'the likelihood of core melt due to the availability of additional supply of water during station blackoui;. This h is because the spray equipment interfaces directly with the core cooling systems and, thus, the adcitional water supply would be available for core cooling (assuming ACE actuatien).
- 3) Recuction in tevere accident consec,uences due to:
a) Scrubbing of airborne fission products by sprays, b) Delaying time te containment failure cue to overpressure, osertempernure and core debris attack. (The magnituce of such a delay cepends upon the use of other potential improvements discussed in Sections 3.3 and 3.4).
The neg;tive implications of the containn:ent spray enhancement include
[1].
'. ) Actuation of drp e:1 sprays ucy decrade elettrical equirrent sw r <.,
mattet 'or containc ent isola tion valves e,r cantaitrent cooling 'ans.
ci the'.o systens for recovery ray be hampered.
1;s e g
i
-i
, l 1
I 49 l 2) The likeliheed of ex-vessel steam explosions.may be increased due to greater availability of water on the drywell floor.
, 3) The addition of backup equipment for the spray system may result in new interfaces between safety grade and non-safety grade systems.. These new interfaces could lead to system interactions which degrade the reliability of the safety systems.
- 4) By plugging spray nozzles to accomodate the reduced flow, the maximum flow for which the spray system was designed would no longer be available.
Drywell cooling under design basis accident, conditions would be reduced.
(However, the full RHR flow is still available to cool the core in the vessel).
Some adverse impacts of centainment spray modifications have also been O suggestee m. These iecluee reeecee sprey for coping w4th other potent 4el accidents that have been previously considered to result in core damage, interface issues between safety and non-safety systems, etc.
I OUESTIOMS :
- 1) What are the potential negative impacts of blocking most of the containment soray nozzles (for example en RHR system operation), and providing a crosstie to diesel fire pumps? How can these impacts be ameliorated?
- 2) Fission produce scrubbing is a known advantage of spray use. What quantifiable advantages also could be realized with respect to shell and debris cooling (separately, and in combination with venting, or in ccmbination with venting, core debris control and use of fire sprays in the rer.ctor building)?
3.3 'lenting l
'lenting rt reactor containmente, has been prnposed as a means to prevent s
ccntaircent caerpressure and preserve its ,ictegrity during a severe accident.
50 g The act of venting has been considered a last ditch effort and may occur when significant amounts of radionuclides are present in the primary containment.
In addition to substantial mitigation benefits, venting can also reduce the likelihood of core melting in many scenarios. To reduce their release, external filtered vent systems have been designed for both BWRs and PWRs. Such systems have been and are in the process of being installed in Sweden, France, the Federal Republic of Germany, and some U.S. Dept. of Energy reactors.
All U.S. BWR MARK I containments utilize pressure suppression pools. The subcooled water in the suppression pools would provide an excellent scrubbing or filtering medium if radionuclides were vented through it. The BWR Owners' Grcup Emergency Procedure Guidelines (EPGs) contain provisions for both drywell and wetwell venting and these have been impleniented for sorre BWRs. It is not known, however, if the venting procedures have been widely accepted at all the plants, or if procedures and equipnient have been evaluated for a wide range of potential accident conditions. Wetwell venting was credited in draft itUREG-1150 with substantial reduction in the probability of core melt and containment failure due to transients with loss of long-term decay heat removal at Peach Bottcm. However, an !!iEL study (tiUREG/CR a696) on venting using hardware and procedures in place at Peach Botton (at the time of the study) concluded that they would be inadecuate to cope with station blac'outk and anticipated transients withcut scram (ATWS) sequences. Significant prcblems identified involved the survivability of haroware, delays in executing procedures, and unir. habitable environments for the operator.
l l
Possible hardware and operational modifications have been discussed that
]
would presumably enhance the effectiveness of venting. Piping capable ot !
withstarding severe accident blowdown loads could replace lightweight ductino.
Irrproved piping in cenbinaticn with heavy duty valves would increase the likelihood that releases wculd be through the desireo location and the vent l could he reclosed. Valves could be made to be AC-independent and remotely cuntrolW to 31 tnw cperattun ct. ring statiun blackout sequences. IN;, tu re d i- --
c.tuld be inciuded in the vent path as a pssive restraint to he;u prevent inadver*ent ope ings i
i
, J
51 Operational modifications that have been suggested include the implementation of clear, "walked-through" procedures and operator training.
Administrative control of venting with consideration for potenially affecting other systems should be well defined. Shifting pump suction to the condensate storage tank (as suggested by recent EPGs) could eliminate the loss of key systems during an ATWS venting scenario.
Any discussion of the use of containment venting in the context of severe accident management must include consideration of any potential negative impacts. These possible undesired effects primarily stem from the possibility of inadvertent or unnecessary venting. In certain specific scenarios, venting ;
may actually result in a release of radioactive material which would not otherdise cccur. In other situations, venting may actually induce core damage.
These sequences are best illustrated through examples. Suppose a station blackout event has occurred and all core cooling has been lost. Core damage '
O bes4as. end the pressere 4n conteirment r4ses to the posat where 4ts intesrity is threatened. The decision to vent is made and carried out. However, a very short time later AC power is restored thereby allowing the containment sprays to operate. In hindsight the venting and its resultant release may have been unnecessary. A second situation may occur during an ATMS event where both SCRMi and SLCS have f ailed. However, core cooling is maintained, first by the nign pressure systems then subsequent to RPV cepressurization, via the SRVs, by 3
the low pressure systems. Eventually, the low-pressure systems are realigned to take suctiun frcm the suppression pool. If venting were to occur at this j point all the operating low pressure pumps woulo fail, thereby failing core l cooling and inducing a core. melt which may otherwise have been averted, in those sequences, wetwell venting could result in the loss of net positive j suction head (NpSH) of pumps drawing water from the torus, thereby causing the l loss of vital safety systems. In those cases, venting would be more likely to I cause or to haster the onset of core damage. These scenarios are presented as simple exarrples of some of the possible negative impacts of inadvertent containmt;rt venting. l.cditionally, the actual risk significance of these igt. cts has not been determined. !
52 In some instances, it may be likely (possible) for the vent to the suppression pool (and isolation between the drywell and the wetwell steam vol'.me) to fail (i.e. , the bellows could fail or the vacuum breakers could stickopen). This would allow a direct vent from the drywell to the environment, bypassing the wetwell . A second issue that should be considered is the effects of a failure to reclose the vent path after opening. Thirdly, venting of the containment would impose an additional requirement on the nitrogen supplies that should be available for re-inerting of containment for hydrogen control during long-term accident management strategies. The effects on existing regulations and technical specifications, vis-a-vis the relationship with existing safety systems (e.g. , containment isolation) have not been assessed.
Another concern relates to the use of existing equipment in conducting venting and the resultant effect on habitability of some plant areas. Some venting ficwpaths may result in high radiation fields for areas whi'h reautre access by plant personnel.
The above discussion attempts to describe the obvious negative aspects associated with containment venting as a severe accident management strategy.
There may be others which a more thorough review would reveal. It appears that the parameters necessary to determine the need for venting, (i.e. , such things as contairment pressure, the rate of containment pressuri:ation anc the probability of recoverir.g AC power), require an integrated kncwlecce o' severe accident possibilities. As such, a venting strategy may not be easily incorpcrated into the existing EPG framework.
CUESTIONS
- 1. A ke' j cuestion is whether or not there is a net benefit to plant safety derived f rcm venting. Can specific procedures and hardware n.odifications be ceveloped to demonstrate that a ccmplex severe accident at a MARK I plant is understooo encugh such that an adequcte ventinn e.trategy may bo developed?
2 Car *ne pntertial negative atrec t ct venting te miniminc? Ucw? g
O 53
- 3. Another issue of major cencern centers around the pre-accident decision of who has the authority to cause venting to take place? Should the operator be vestec with that authority, the utility management, or should
~
a passive system be designed which will provide absolute, invariate action?.
- 4. From a design perspective, the question arises as to the need for a safety-grade system. Can a modified vent system that includes non-safety. -
grave components be relied upon just as a safety grade system could?
3.4 Core Oebris Control It has been suggested that a curb (say of refractory or heat absorbing material) coulo te built around the drywell floor of MARK 1 containments and, alone or in conjunction with water cooling, be used to prevent or inhibit O contect eetween cere eebr4s ane the steei coetaiement s8eii. oeiay or prevention of, cuntact with the shell could substantially reduce the likelihood of early shell meltthrough.- A curb could also delay, limit or prevent debris frem entering any of the vents leading to the suppression pool. Here debris to enter a vent it could perforate the vent pipe or suppression pool shell and permit material to enter the torus room of the reactor building without first passing thrcugh the suppression pool.
The efficacy of a curb and water to control debris depends upon the amount and composition of the debris, heat transfer capabilities and the strength and dimensions of the curb. Although the fuel volume of a MARK 1 core is about 11 3 3 m , another 25 m of other debris, including cladding, control rods ano drives, steel and vessel internals might also be present. For comparison, the drywell 2
floor areas are about 30 m within the. pedestal and 100 m in the annular area outside. Debris could exit the vessel under high or low pressure deoending on whether or nut the automatic depressuri:ation system ( ADS) , System was initiated and worNd. Uncer high pressure, it n.ay be unlikely that a coherent rms of debris would attack the shell. Pather, it appears more likely that debris wnutd te swept out of the pedestal area arm vould splatter on the shell in a tcra coolable geometry (cssuming the sprays work) than as a large mass. A curb f or a low pressure meltthrough wculd have to bolo back a very hot anc. der,se
l 54 g mass without permitting penetration or undermining of the curb. Given the necessity of not interfering with other drywell features and operations (such as access to the vents by escaping steam and maintenance of control rod drives) and the potential for debris expansion by frothing, curb design becomes difficult. In combination with drywell flooding, shell cooling and the use of sacrificial materials to create a coolable geometry, a refractory curb might offer reliable shell protection for low pressure melt sequences. While such curbs might delay and potentially prevent shell or vent meltthrough, sufficient information does not exist to fully establish the technical or economic efficacy of such schemes in the limited drywell floor space.
Curbs or weir walls within the torus room of MARK l's to help provioe a coolable gecmetry for core debris and to scrub fission products in the event of a vent pipe, drywell shell meltthrougn, or suppression pool shell failure, have also been suggested. That is, even if the vent pipe is perforated by debris, water from spray operation wculd be expt.cted to provide water for fission product scrubbing and potentially establishing a coolable debris bed in the torus room.
CUESTIO!!S 1, Are drywell curos technically feasible?
- 2. Are torus room curts with or sithout drywell curbs feasible?
- 3. Would curbs in either location interfere significantly with reactor operatinns? In addition, would curbs in the drywell alter the outccme of accidents within the original design envelop with respect to such tooics as blockage of vents and the drywell beat attenuation capability?
- 4. Several ne:rs of protecting dryuell concrete f r om debris at tack , fer exwole by covering the ficar with a water-purcus pebble ted of thoria at d alumina, bee been suggested. What would be the ett1CaC:. 3rd practicality o# luch reasureO Is wawr in addition to a curb i.ecessary ,
i to protect the curb' l l
I 55 i
j 3.5 Automatic Deoressurization
.An area of significant phenomenological uncertainty relates to core meltdown with the primary system at high pressure. If molten c' ore materials i
are ejected from the RPV at high pressure, it has been postulated that' the core debris can be dispersed into the containment atmosphere and thus directly heat it. This could result in a large pressure and temperature increase which could challenge contairment integrity at the time of RPV failure, leading to substantial release of radienuclides to the environment (see discussion: in Chapter 2). This moce of containrent failure can be essentially eliminated by :
er.hancement of the Autocatic Depressurization System (ADS). The ADS enhancement has the additional benefit of reducing the core melt frequency.
For those sequences in which the high-pressure injection systems fail (TQU)..
the ADS can cepressuri:e the RCS and allow the low-pressure systems to inject r water into the RPV. ,
The A05 norrally serves as a backup to the high pressure coolant injectier
' system. It performs the function of rapic vessel depressurization when high pressure injection systets are inoperable or are unable to maintain adecuate I unter inver tery. The A05 control operates' solenoid-actuated air (or nitregen) j valves that allow the gas to open safety / relief valves anf rapidly der *essuri:e ti.e eacter vessei. The ACS receives electrical power fece the piant direct current system and is autcmatically or manually actuated, f
^
l Actuatien of the A05 is dependent on an adequate air or nitrogen supply and centrol pcwer to the solenoids which control the air or nitrogen supply.
NUREC-C737 1 tem !I.K.3.28 addressed the adequacy of the air supply and l l
ccroiTance with II.K.3.28 has been considered sufficient to assure a relicie !
air or nitrogen supply. In a protracted loss of alternating current power, the l i . i 1 l
!, O ;
I
E6 batteries providing power to the solenoids may be depleted rendering A05 inoperable. Alternative dedicated backup power (batteries or small generator driven pcwer supply and appropriate connections) for the ADS could substantially increase the availability of the ADS for station blackout events and avoid containment failure due to high pressure melt ejection-induced direct containment heating. Also, the cabling for the ADS may need to be thermally insulated or shieldeo to assure operability in the hot drywell environnent curing pretracted station blackcut event.
OUESTIONS
- 1) Are the tyces of improverents described sufficient to improve ADS reliability?
- ) What are other recomendations?
- 3) What are the berefits? Fcr example, is increased esailability of the ADS system r.ccessary, arc will it eliminate the risk from cirect centainrent heating?
41 What are pctential acserse interactions between any of the proposed improverents ard cristing safety system??
- 5) Are there otrcr improvements that might also be technically feasible to preserve the MARK 1 ccntainment function in the event of ILw prGbability Challenges?.
3.6. Procecures ind Training the procNuv es anc trc101ng necestsry to implerent a containnent nitigation in the evert cf a severe accident were discussed in early 5t:"
sucgestions ter improving PARK ! containment perfor ance ['.,5). A separate l'PC staf f initla',1fe called accident managencnt is expected tc address this area elsewhere.
l i
O 57 s
3.7 References i
- 1. Memorandum frem H.R. Denton to E. Beckjord on "Proposals for Enhancing i
BWR Mark I Containments" Enclosure 4, U.S. Nuclear Regulatory Coninission i
(April 10: 1987).
- 2. J.D. Fish, M. Pilch, and F.E. Are11ane, "Demonstration of Passively-Cooled j Particle-Bed Core Retention", Proceedirgs of the LMFER Safety Meeting, {
July 19-23, 1982, Lyon-Ecully, France.
- 3. L. G. H'ulman; "Sumary of March 27, 1987 Meeting with BWR Owners Group /IDC04 oft MARK I Containments' ; U5tRC; May 6,1987; and "Surrary of MARK
- Containment Meeting" with hRC Contractor Personnel; USNRC; Fed. 24, 1987.
- 4. NUREG-1150, Reactor Risk Reference Occument, Draf t (February,1987) !
-O i 1 5. Letter, July 8,1987, Ralph G. Bird. Boston Edison Co. , to Steven j Varga, hFC.
Subject:
Filgrim Station Safety Enhancemert Program. l
< f
- 6. C6 Ridge Lateratory Analyses Presented at Rest: arch Partrers Meet.ng; Silver Scring, MD; Fall 1987. l I
- 7. NUREG/CR-EC, Vol. 3. Analys4s cf Core Damage Frecuency ficm In:ernal Events.
1 C. huREG/CR 4696 Containnent Venting Analysis for the Peach 2cttcm Atemic Pcwer Station, Decerter 1986.
l l
4 i
O ,
f e I
58 t wr :- :a. ',:, -
L. .
c.
r i . s i. .s , . . v . o i . .;
. eta c. .
u , ' : .w .
3-n.
(5. 21) ,
O Max temperature (*F) .
pressure Rise Crywell Suppression P001 Event Type (esi)
Rapid Station 160 8 350 Bl ackout Long-Term 250 15-25 15t!-350 81ackout 160 250 AT'.4 S 15 LOCA(I) 160 Large '3 ream 20 , 160 160 small Break 0-25(-) 160 (1) Assumi99 Cont 3inne9t sprays do not CCerate. A~
( 2 )' Ce ends on Operation of RHR system.
ornt t . -' ; ' < . *.
'.. ; .0 0 ; r te O
N. r.
'i..L , ,.
e- s: s: (5, 23)
Maa Temperature ('F) j 3rassu ra 3 !sef !) O r ys,e l l su;O r e s s i e r, tW@ ! ,y;Y pool
(;s'l T snstents Higa RCS Pressure 11 72 200 250 Lcw 3C3 Pressar 22.;c 203 250 LOS L3rP Pres- le-104 <2000(2) 130 St Ci Prear 17 9R ,:00 700 130 333 (1: De lew ar 39: 2:pe r ou ans c or re5 Do"'1 to typical Zr omication re:ictejgf:::Ca r m s? and ND C ("2 DCH, NRCON) analyses ees:ect: m ;,.
(i E! . !'d r es ul t s f o r a B* ARK ! fi se;uetice.
O
59
[U'T es in co
- 1. n w s = o e sc' ~-
-mu 1. : . a ,n < .vw c .:a. c e.n ce,-
t,' ; ts .i... . , . c.c. : u r. . e rir .0-Pressure, Leak area, psig sq. inches 0 0.004 62 0.004 82 12.0 94 19.0 117 35.0 -
- The unseating of the drywell head is the major Contributor to the leak areas.
Subsecuent analyses by the Argonne National Laborator Laboratories indicate that when temperature effects (yup and to Sandia 1,000*F)National are considerec, the leak areas will be considerably less because of the increased clamping force in the nead bolts under themal gradients.
a.
I
.g
..- :, ,- ;<.- ns :. . :. . . . , - ~ - n -- .w ~ % :, . . .- - .: t ,
l
. r.r.v: . . . :. ."w.-.
3< 0. c. > u s ; - d.' e : e c t
=.
- l u :. t s . ; O *,
. ..s c. . . .z e s' ' : - 2 *. . . ' .
- te r.
l
~
- y,, . .,...n
. .. : m. !
l i Leakage past operable penetrations Drywell head Pressure unseating equipnent hatcha Personnel airlockC Control rod drive removal hatch i 2 Orywell rupture i 3 Wetwell (torus) ruptured 4 Material Failure of Penetrationsb 5 Pelt-tnrough aue to core attack or. snell 5 Purge anc vent valve leakageb l 7 Electrical penetration assemblias leakagab l 3 Containment not isolated at initiation of accident '
i O D "et '4keix to occur.
Occurs under high temperature. ^
c Unaer investigation by Sandia under an NRC program; may be significant, d
Sunject to ccmplex hydrostatic and hydrodynamic load; may be significant.
60
- i 1 = n r .1 n r. is .r.cn er
' u a c. . . . -s~. :, >
. nr 1 4 1 11, .
Debris Heignt (m)
Cov ered Floor Region Covered
- Area (m:) Zero Porosity Porosity = 0.40 In-Pedestal 29.9 1.04 1.86 in-Pedestal + 50: Ex Pedestal 81.0 0.38 0.69 '
In-Pedestal + All Ex-Pedestal 132.1 0.24 0.42
~
- Volume of Whole-Core Gebris (zero porosity) = 36.8 m 3 Volume of Drywell Sumps = 5.7 m 3 e:.3 ..u. _ . u .w . _ o, re : .: e 3. n ; . i.u c p
. e . e 'c . . - . - - < o s i t . ': s :..:.
Corium Time To Axial + . Ini cKnes t.+
Run Concrete
- Temperature (X)
% of Core gg Fail Concrete Erosion (cm) of Liner Ablated (cm) g lNoMelt-1 8 1775 80 Tnrougn 3.3 0.1 2 L 1775 80 2842 1.2 3.0
. No Melt-3 E 1900 80 Througn 7.4 0.3 4 L 1900 80 895 1.5 3.0 5 B 2550 80 323 4.0 3.0 6 L 2550 80 208 1.6 3.0 7 E 2550 60 325 3.6 3.0 3 L 2550 60 226 1.6 3.0
- 8 = Basalt, L = Limestore
+ A; liner melt-tnrougn '.we.
O
61 Condticca! Precab!!!ty .
1 -
P E
- E r
~
% l
~
I /
/ wmo.tr IE1 .i
($
/
// NE_MouG-I
- b 2
2 !/>
fe /
i= 2 r Wm-i m //
- 3. NE. N //
~
- c. 9-1E-4 LATE FA1.Li:.E NO FALLP5 EA:LY FAL'J:.E 1
. w. - =u.bre
. 5 ' .: Pr' L' ($),
I l
O I
1
. ~ - - _ _ _ - - - - _ , _ ., m_ . - - _ . , _ - _ _ - .
1 l
62 1 l
I RANr( ECRESSlON ESLUS l u
O ;
u.
, u. '
Y i o
$= m. 1 4": g'g t y I!=$2-g1b a;'1! 1 g ,
g tii ,'gtj))2' s
ltI=3 ~
E' i! . . Q=l,e,d a
,1 5
[.in g a- ]iff
]},g:
,=r a 1H-e
- d,4
- 91. 5 81it, : t-y jtd!.!I__ .II iLki s ) h u- -
uw.y raumts .
RANK PIGRESSION RESUJS u 9-u<
h -
1
,u<
g i
b u. ! . ,.
l I! g g
2 y E
$ @ $'W {9 5 $j k p r . f &crCasi;I : :i M ' *.-(
q; C' 8
.g alei-il.
g stii
[3 3.);]5!3II!ja3 . t. t. e j !!] ! I : r, i n :i. s t i ! j:E E u.
1 a ys.
i r -a--
~asa l{yEs:
-]: s
=
r s s in. ag !es_u w g2.. it g'_.ca E3_L=.15 %v a.e Jn:ue g LATENT C#8CERS
-f i $1';:Ur .v 9 C O r,1s, c ; r,* 01 f.f.1 erit l
- 0 . e c- _-.. ';- 1 ' 'i ; # * ? r. . 7 r e s e ;,,, ; r; e . A c ; 4 , ,-3 ; p ,: r - wegen J..; .
,'r,.- ci ri e
- c. : r.. (g),
l 1
l l
1 1
O l e
S
63 0 1 0 ...<
UI REACTOR BUILDING 1
~
.. h eg . CONCRETE l j . BIOLOC' CAL :
- r SHIELO i
ad8 ORYWELL i
- -REACTOR
- i
') (- PEDESTAL l ORYWELL SUMPS DRYWELL SHELL PEDESTAL 9-Q ,
WINN//qn 3go gg, u l
-: Cl ' ' DOORWAY 8' AT FLOOR LEVEL
'N,'$ .
N' PEDESTAL
- O//////g/j//$. INNER RADIUS 3.09 m PEDESTAL OUTER RADIUS 4.00 m SHELL INNER l
\ RADIUS 6.96 m
.-ca ::: - > . ' :- .. '^c ce:rt - u
~ ?c e .;...
64 O
OEBRIS DEPTH:0.12m T m2100K T :400K Fo So 1000
~
000 -
1 x ,_
d NO WATER -
1000 od o 800 -
O O
2 j 700 -
800 j w w I
- M 600 -
600 s P 0.1MPa >
M m
500 P 0.3MPa y
$ P 0.5MPa 400 H H
9
- 400 -
~
9h 200 300 O 200 400 600 800 1000 TIME. sec
.= . . . u . a._.. m.... , c c 1-
- , , ., .- , .. . : . . 1 <. m . .n =>:.. stac
. . , v. 4.
O p . _ _ _ . . . . _ -
65 0 .
VARIOUS DEBRIS DEPTHS .
Tp,:2100 K T 3g:400K 900 0.12 m ' A x -
1000 *.
800 -
0.09m w g O o 800 9 700 - - 0.06 m -
8 a #
w 800 -
600 H r f M
m O -
m s00 -
400 m w
w H H H H 400 -
0 O -
200 I 300 O 200 400 600 800- 1000 TIME, sec
- _.... .. _,_,; ,
- -- r e m,i. r : z: : w . . , . , , . "
shell
... n e , . .> ...- . . cme a- ..~.. .nc e : c:22.
O
66
.. r . 2 3.1 . .m t . , n .'
'2 ':e: L 'i ' ' ' " " ~
- ' " "? :"Oco g
- t.v>e(5).
Urywell Time Max Orywell P & T (during 5 h)
I Percent Concrete Initial Deeris to Fail (min)
Core Type Temp (K) P T P T 1,77s 9 50 -- 65 411 B 1.9 00 2,550 460 -- 108 477 80 1,775 500 3 23 88 533 L 1,9 U0 2,550 133 62 145 622 Note: B = casalt, L = limestone P = pressure, T = temperature s-G
~
O
l l
67 O
- 1 6
- i 4 0 _
som -
$ INTER (BCU MARCON ISNU ,
< 800 z
- 5 CORON iBNU-g 600 ~
U g 400 w _
Em g VESSEL FAILURE l
I I 0 150 200 60 100 O
i TIME AFTER VESSEL FAILURE (Minuten j a
200 i 6 M ARCON ISNU CORCONISNu O 7 2
a INTER (BCU
[
c 3
g im p A
d w
3 M
Q I 1 l
g M 100 150 m J
TIME AFTER VESSEL FAILURE IMinutel
' ; e. p , . ; ; 4.r . ,;,
, ; g, , ., . )
i *- . r ' s i .:, , ,
.i . .
O
68 O
i, g g 31003
!, 8 I
. t3 1 Ol ;
' 0' ] O v*1 81 a l 3 M v 4 Ot ",1, 5. OI -
U Il -
-t =;
3 a' 5 bl I i Q O _p*
100
- Y 5
- 5' ' -$ .a Il OM
~a si 2 41 i J 8.e & O
=
n . El 2 og 4 ,$ $
n -
! .*i,k? # ## #
b &,
- [
i
't I Q .
- /
g
. *g -
' ?>' &
- /
- 8
/ l
,8 d
$ i y 5 -
/
,1
/
e
)
e Q,
s' ' s &
/ o- F-i
- e ,/ ) ,
/ /
s' f/
f/ / /
/ /
g $ , ,. / */
/
,/ ,
' l l
' i l -
I Jo0.0 21v.a 310.0 0 160.0 260.0 ,
60.0 110.0 10.]
Contatefent Pressure (psig) ,
s :. i . . . r . . : 1.,. -2,.3..,3 j.;,
.:' ; .. c es- .r s. .
i 1
0 9
- 69. .
(J . x. : .-. :.4 . :., t : , s .; 1 : .- i. .. :, v > . .r w e s :.. . .-+ u a : v co
.. ; e . > . . n. g . , , i+:.: : c e n t A t :: v
.,.r.3..r,-
e t i. . o _ie e o r 't e a , :$. I12 Estimated leakage rate Equiv al ent Fractioncl (Volume?, Contribution Relative Change in Holes {ze" Due to leakage Overall Accident Risk per day) (in )
0.01 0.0007 .0002 1.000
~
0.007 .002 1.002 1.0 .
0.07 .02 L O2 10.0 100.0 0.7 .2 1.2
. *For a small MARK !. ,
a.
m
~.:. 4.. - *: . : .- ;r .,d:tec ee 3reacnes: of
- c. - r . . . .:. r . - . m et.r .v S.12; (s:;. inches Estimated Likelinood Learage Rate' (per reactor year) Leak Area (% per day)
LE: !!a Else Tec n t:a! Soecification 0.25 - 0.35 0.001-0.01 1-10 times Violations all owabl e leak rate *
~
Ocen Airlocks 1x 10 6 5,000 100 to 1 x 10 5 Larga Lead 1x 10 3 28 >100 (e.g. , coen val ves) to 1 x 10 2 i CILO' Da'.a Rise BWP O.16 0.006 1 0.14 0.06 10 0 0.04 0.6 100
~
Total 0.34
- Allowable leak rate is defined as the containment leak rate that will satisfy 10 CFR Part 100 using dose guirielines using Standard Review Plan assugotions in
_ _ - . - . .a
70 Tabic 2.6.1' Timing of Key Events for Peach Bottom [2,45]
Bottom Head late Centainment Ccre Uncovery Failure Time Failure Tire Event Type Time (min) (min)* (min)
Short Term Station 30 90 270 "
Blackcut (45]
530 735 915 Long Term Station Blackout [2]
30 125 ATWS**=[2] ,
- Early centainment failure (by overpressure er liner meltthrcugh) is expected to occur at or shcrtly after RPV failure.
"Estirated based on Peach Bottom calcuation for TB2 (2)
"*For corc-vulnerable TC containrent fails before core urcovery (for Pecch Bottcm TCl the containment f ailure time is 85 minutes).
I el l
9
71 L/
APPENDIX A '
FEB/NAR lET MARK I QUESTIONS
- l. What core melt accident sequences may be expected to be significart in SWRs with MAP.V, I containments?
- 2. Do current analyses indicate centainment failure preceding core melt?...
anc causing core melt?
- 3. What are the aoproximate time scales for significant sequences? e.g.,
time to core uncovery, time to core melt, time to melt threugh, time to containment failure. Is tnis generic or very plant specific?
- 4. De high pressure melts (ADS failure) have a significant effect on the physical behavior of the core' melt in a BWR?
- 5. Do current models indicate substantial differences between PWRs and BURS in meltdown times?...in nieltthrough times?
- 6. Are the physical properties of the "core-on-the-flocr" for a EUR expected to be significantly different than for a PWR? e.g., thermal conductivity, viscosity, etc.
O 7. in a typical MARK I, initiation of drywell spray before meltthrough can
(,/
cover the crywell ficer with up to I foot of water before core material begins to drep. Is the presence of such a water layer beneficial.
- 8. In a typical MARX I the drywell spray can distribute up to 20,000 gpm in the area outside of the reactor pedestal area. If this spray is operating at the time of meltthrough can it inhibit corium mcVement toward and attack of the outer wall of the drywell? Would success be prcportional te water ficw rate?
- 9. Giver the cre:ence of dr>well spray, would a short civersion barrier wnien could coubie or triple the path length to the outer wall significantly reduce the likelihocd of liner meltthrough?
- 10. If a substantial barrier of retractory character could be provided to hold most of the corium in the reactor pedestal area, would this be preferrec? Would attack of the reactor vessel pedestal be a significant concern?
- 11. Is any release attenuation expected frem the biological shield tu-rour. ding the MARK I crywell?...is it treated in current models?
- 4. In a typical MARK I containment available or practically acaptable vert oaths have an effective diameter of abcut 10-12 inches which is sufficient to nass aater vapcr at 1 to 11 times design pressure ocuivalent to 1-2%
cenav hqat. What effect on sicr.ificant accident sdouer.Ces can De expected O it there ere >=s"ree mee"s to caea this vert Peth?
- 13. Calculatiens new available indicate that althcugh ncble gas doses can be i hign, celiberate release of those gases appears to be
72 l
better to avcid the far greater releases that might occur with an uncontrolled release. Do present rodels indicate that deliberate venting g
of noble gas activity may not be justified?
- 14. To what extent could reliable containments spray alone, without venting, substantially reduce containment failure in the station blackcut sequence?
- 15. Is there any other practical change to the MARK I containment system which can significantly improve its performance in core melt?
O O
", $ ( f QUESTION 5, PAGE 40 1Eh 'A O SEeARAre EFFECTS exeERiMENTS O WE NEED TO DEVELOP VALID SUCCESS CRITERIA FOR STABILIZING CORE DEBRIS. I IN-VESSliL FX-VESSEL 1
1 0 SUCCESS CRITERIA MUST COME FROM ANALYTICAL MODELS OF THE ;
PROCESSES INVOLVED.
O FULL OR LARGE SCALE TESTS TO VALIDATE THESE MODELS ARE NOT PRACTICAL.
O THE CRITICAL ELEMENTS OF THE ANALYTICAL MODELS SHOULD BE ADDRESSED BY SEPARATE EFFECTS EXPERIMENTS.
O O THeaE iS NO AreARENT SYSTEMATIC EFFORT TO CARRY OUT SuCH A PROGRAM FOR STABILIZATION OF EXVESSEL CORE DEBRIS AT THIS TIME BY EITHER INDUSTRY OR THE NRC.
O
i I
OPERATOR ACTIONS l (S) 0 WE HAVE FOUND THAT THE USE OF CONVENTIONAL RISK ASSESSMENT MODELS AND ASSUMPTIONS CONCEAL THE TRUE DOMINANT RISKS TO THE PLANT AND CAN DIVERT ATTENTION FROM THE CRITICAL AREAS NEEDING ATTENTION IN PROCEDURES AND TRAINING.
O IF THIS DIVERSION UF ATTENTION IS PERMITTED, THOSE CONVEf;i10NAL MODELS MAY BECOME SELF-FULFILLING PREDICTIONS OF PERFORMANCE.
O WE BELIEVE THAT CONTEMPORARY BWR RISK ASSESSMENTS SUFFER FROM THIS DEFICIENCY AND CONSEQUENTLY PRESENT AN IMPROPER PROFILE OF DOMINANT ACCIDENT SEQUENCES.
O Tills IMPROPER PROFILE INVOLVES:
AN INADEQUATE ACCOUNTING FOR STOPPING CORE DAMAGE
() PROGRESSION BEFORE REACTOR YESSEL FAILURE.
EXCESSIVE MAGNITUDE, RATE, AND ENERGY OF CORE DEBRIS I POURS FROM A FAILED REACTOR VESSEL ONTO A DRY DRYWELL FLOOR.
O THIS IMPROPER PROFILE RESULTS IN A GREATLY EXAGGERATED ESTIMATE OF THE THREAT TO THE CONTAINMENT OF THE DOMINANT ACCIDENT SEQUENCES.
l l
0
QUESTION 6, PAGE 40
([)
PHENOMENA INTERACTIONS l l
0 THE HIERARCHY OF ACCIDENT RESPONSE OBJECTIVE SHOULD BE: l AVOID LOSS OF FUEL CLAD INTEGRITY r
AV0lD LOSS OF REACTOR VCSSEL INTEGRiiY CAUSED BY CORE HELT.
AVOID LOSS OF CONTAINMENT INTEGRITY CAUSED BY CORE MELT AND REACTOR VESSEL FAILURE.
l l
0 FUEL CLAD DAMAGE THE DIVERSITY AND REDUNDANCY IN BWR SYSTEMS RESULTS IN A NEGLIGIBLE PROBABILITY OF SUCH DAMAGE IN THE ABSENCE OF
() MASSIVE COMMON. MODE FAILURE SUCH AS STATION BLACK 0UT OR ATWS.
l l
0 REACTOR VESSEL FAILURE l
THE DAMAGE PROGRESSION MODELS OF BWRSAR INDICATE A VERY I
HIGH PROBABILITY OF AVOIDING REACTOR VESSEL FAILURE.
O CONTAINMENT FAILURE
^
INDUSTRY AND THE NRC HAVE GIVEN LITTLE ATTENTION TO CRITERIA FOR SUCCESS HERE. RESEARCH IS NEEDEC.
i l
l l
l
($ &..
/
NI O E Pennsylvania Power & Light Company Two Nonh Nintn Street
- Allentown, PA 181011179
- 2151770 5151 Harold W Keiser Vce PrescentRuc! ear Ocerations 215/770 7502 February 17, 1988 Mr. Hulman, Chief Severe Accident Issues Branch Office of Nuclear Regulatory Research United States Nuclear Regulatory Commission Washington, DC 20555
Dear Mr. Hulman:
PP&L has reviewed the document "BWR Mark I Preliminary Issue Characterization" which was transmitted with your letter of January 23, 1988 inviting Paul Hill to attend your BWR Mark I Workshop to be held February 24 through .
February 26, 1988. We believe that these issues will ultiastely affect the O- _BVR Mark II containment, and so we believe that it is appropriate that our views be considered in the course of resolving the issues for the Mark I containment. For this reason we have prepared a response to each of the questions posed in the document cited above. These responses reflect the results of PP&L efforts in the area of severe accident issues over the past several years and are consistent with the various prior documents and presentations by PP&L which have addressed severe accident issues. PP&L has devoted a considerable effort to addressing sesere accident issues, and we believe your consideration of our findings and our recommendations is appropriate.
Paul Hill, who requested presentation time as a self nominated speaker on your agenda, will address the topics presented in response to your Questions 1) through,7) on pages 39 and 40 of the document cited above. This limitation is necessary in order to keep within the time constraint to be. observed by the self nominated speakers. This topical limitation should not, however, be taken to imply any less interest or concern on the part of PP&L for the additional issues addressed by the questions in your document. We have focussed on these first seven questions on the basis that the concerns and recommendations relating to these questions must first be resolved before the remaining issues can be effectively and properly addressed.
O v
February 17, 1988 1 1
I hope that the information we have provided will have a beneficial influence on your resolution of all concerns over the adequacy of the BWR pressure suppression containment concept.
Sincerely, I
(0:;.1 :; !!. ',l. "E!!ER H. W. Keiser Attachment PRE /laj j bec: W. E. Barberich A2-4 M. B. Decamore A2-5 i
P. R. Hill A2-5 C. A. Kukielka A2-5 l
t
ATTACHMENT l
Response to Questions from BWR Mark I Preliminary Issue i Characterization, January 21, 1988 Pages 39, 40
- 1. Question 1)
Are the discussions of phenomenological issues adequate? Are there other important mechanisms which can challenge Mark I containments involving phenomena not encompassed by the identified issues?
Response
The presentation of phenomena seems unbalanced. There is much discussion about phenomena, such as steam explosions, which are ultimately dismissed as unimportant or not significant, risk contributors. In the specific case of steam explosions, for example, a decision must be made on whether such a risk is present or not. This is necessary because the operator response to the potential for core melt.and vessel failure should M to flood the drywell floor if it is clear that steam explosions would not threaten containment integrity.
This decision must be made at the time when the Emergency Operating Procedures (EOPs) are developed. The issue here is not "What is a conservative estimate of the risk from steam explosion?"
in the classical PRA sense, but rather 1
"What is'the optimal strategy to respond to a threat of core melt and l reactor vessel failure?" l so that the E0Ps and operator training vill assure the most favorable reaction for minimizing plant damage and risk to the public and plant personnel.
This perspective seems to be lost in the NUREG-1150 approach and in the ;
bulk of other contemporary risk assessment studies. To repeat, the focus l cust be on how to structure procedures and training to minimize damage and l risk, not how to derive a conservative estimate of damage and risk. It is essential that recognition of this difference in approach occur and that it be adopted by the NRC and nuclear utilities.
These two approaches frequently lead to very dif ferent views of a particular accident sequence and can lead to quite different views on the appropriat.e response to the threat which the sequeace presents.
l
- 2. Question 2)
Are the magnitudes of the parameters representing potential containment O
challenges all included in the ranges identified, or can credible sequences be postulated with values which exceed those included in the issues as discussed?
Response
There is an inordinate tendency in the discussion provided in Section 2 of this document to focus on extending the range of conditions which can be tolerated by the containment. As a result, the accident sequences and phenomena presented in this discussion are excessively conservative and severe.
Operator action to preserve the containment function should be very much a part of the assessment of containment capability. 7f more attention is directed to operator action to avoid these severe sequences or reduce their severity, the actual risk associated with operation of the plant is almost certainly much less that if primary attention is focussed on the overly severe sequences identified in NUREG-1150 and other contemporary risk assessment documents.
The PP&L experience in performing a realistic risk assessment for the Susquehanna Steam Electric Station has been that attention to the details of the dominant cont'ributors to plant damage has reduced the expected frequency of damage by two decades and the frequency of severe containment g challenges which could represent a significant public risk by W approximately three decades. This result has been achieved, however, by careful attention to the specific vulnerabilities of the plant and making certain that procedures and training vill assure a very high probability of correct operator actions to achieve these reductions. This process invc'.v.s more detailed attention to the spscifics of the event transient timing and the potential for operator failure to accomplish the necessary actions in a timely fashion. This assessment requires consideration of o the time available for operator actions, o the quality of symptomatic information available to the operator to prompt the action, o the extent to which training has sensitized the operator to the critical situations associated with specific plant vulnerabilities.
The result of such studies and attention to procedures and training we believe represents a real reduction in the expected frequency of the very severe accident sequences which can be postulated. We believe it is f ar ,
better that a nuclear utility devote resources to this process before any j major revisions to containment capability are considered. This process j will be far more productive and cost effective for true risk reduction. l can strongly impact the cost of appropriate revisions to improve containment capability, and can also strongly influence their relative g
[
effectiveness.
W l I
! l l
1 l
I O 3. Question 3)
The manifestations of the phenomena can be affected by human actions both '
prior to ,the accident and through errors in accident diagnostics and management. Do the identified phenomenological issues imply undue reliance upon favorable maintenance, operation or mitigative behavior. l Conversely, are there reasons to assume that any of the phenomenological issues can be further narrowed by reliably predictable human behavior?
l
Response
In general, PP&L believes that the perceived importance of the l
- phenomenological issues identified are a result of improperly discounting the ef fectiveness of operator actions backed by effective surveillance and .
maintenance programs. This approach has, in our opinion, resulted in a failure to recognize the importance of a 'well trained and disciplined operations staf f backed by clear and ef f ective procedures and training specific to the plant, and has inappropriately focussed attention on less effective or overly expensive methods for improvement of containment performance for the expected dominant accident sequences.
PP&L believes that much can be done to influence the phenomenological issues which should be considered. PP&L has identified twelve important operator actions in severe accident response and four analysis assuspeions l normally used in BWR risk assessments that severely distort the view of i che dominant severe accident sequences which should be considered in EOP development and operator training. These sixteen items were identified and discussed in a letter to E. S. Beckjord, Director of Research, in November, 1987. We suggested etiat these items should be given explicit attention by the NRC and BWR utilities before any direct consideration is given to containment performance related plant modifications. The reason for this preferential consideration is that attention to and accommodation of these sixteen items not only results in a drastic reduction of the calculated frequency of plant damage, but it also has a dramatic impact on the profile of dominant risk sequences. This impact can influence the relative benefit, and cost, of the various containment related modifications that should be considered for operational risk reduction.
We strongly recommend that these sixteen items, and perhaps other similar or related items be given consideration before making any decision on modifications intended primarily to improve containment performance.
- 4. Question 4)
Can any of the phenomena be either precluded or limited in magnitude of challenge to the containment by a feasible addition to or removal from any region of the containment cf materials, or by alteration of current maintenance or operation procedures? Are there other phenomena which diminish the challenges?
Response
It is not clear at this time whether or not addition or removal of materials in the containment, or any other containment performance related modification would have a significant favorable impact on overall containment performance. We believe the answer to this question is very strongly plant specific and that it can only be answered by a specific
l I
}
I consideration of the detailed vulnerabilities of the plant. We believe these vulnerabilities are very much dependent on specific design features g
of a plant with properly developed procedures and training as discussed in l answers to the preceding questions. Aside from the sixteen items {
discussed in the answer to Question 3) above, the presence of steel liner l plate on the drywell floor is believed to have a potentially major effect on the threat from core-concrete interactions. The presence of this liner will potentially inhibit the initial release of water vapor from the concrete and result in a drastic reduction in the chemical reaced.on energy I that can be released in the early period of the core debris pourt. This i delay can, in turn, provide a greater time period for quenching the debris I to sufficiently lower temperatures so that these chemical reductions will not occur to an extent sufficient te defeat the quench process.
l The presence of this liner may be sufficient to result in a very large reduction in the conditional probability of a core debris pour that cannot be quenched prior to severely threatening containment integrity.
! In the same way, more attention to terminating the core damage progression prior to failure of the reactor vessel can very sharply reduce the ,
conditional probability of the core on the floor situation and therefore (
l greatly reduce the threat to the containment. These two issues, and l others not discussed here, we believe will result in a drastic revision to {
the current perception of severe accident risk involving loss of I containment integrity.
l l
- The influence of the various items involved, however, are believed to be highly plant specific so that a definitive evaluation should be carried g
I l out for each plant.
- 5. Question 5)
Are there separate effects experiments which have the potential for demonstrating low likelihood or limited effect of any of the phenomena?
l
Response
There are certainly a number of analyses which could directly contribute l
co the development of procedures which offer the highest likelihood of having a favorable influence on accident sequences to either terminate the sequence or alleviate the severity of the sequence. These analyses include:
I o calculations to determine limits on drywell spray operation for a spectrum of severe accident sequences to reduce drywell temperature, flood the drywell floor, or cool core debris on the drywell floor, o calculations to determine the criteria.for success in arresting core damage sequences p'rior to reactor vessel breach for a wide spectrum of core damage sequences, and I
o calculations to deter 1aine the criteria for quenching and stabilizing core debris on the drywell floor for a wide spectrum h
I of reactor vessel breach sequences.
l It is important, however, that the accident sequences take realistic credit for effective use of available plant equipment by the plant operators. These calculations of sequences should assume that optimal procedures have been de'veloped for the plant analyzed and that operator training is thorough and effective so that failure to utilize plant facilities is dominated by equipment failure rates or realistic time constraints for operator action and not by operator error in following or executing procedures. This approach is essential if proper conclusions are to be drawn regarding optimal response strategies to the most probable severe accident sequences. Development of procedures based on the accident sequences resulting from present day conventional risk assessment assumptions and practica could lead to improper conclusions regarding optimal respons,e strategies.
The analyses recommended do involve a wide variety of physical and chemical. phenomena that are poorly understood in terms of hard experimental data to demonstrate the validity of the analytical models and physical data used in the calculations. It would not be justifiable or economically viable to conduct full or even large scale tests to develop the necessary empirical evidence needed to demonstrate the adequacy of current methods or to idensify the improvements necessary to current I l
methods. Nevertheless, it should be possible to greatly improve the credibility of current models by careful examination of the analytical representations and fundamental data used, to define separate effects experiments which could be used to reduce the potential for deficiencies in our models and data which could lead to a defective view of the event x progress. ions and the influence of operator induced changes in the I conditions which determine the event progression. ;
At the present time PP&L sees very little indication of a systematic (
effort to implement this type of evaluation to develop a credible approach !
to savste accident management on the part of either industry or federal organizations. While such an approach may not produce confidence in the appropriate nature of the response strategies developed in the short term, we believe that it is important that this approach be taken. In the short term we must use the models and data currently available as effectively as we can. It is essential to realize, however, that conservatism in developing an estimate of public risk on the frequency and nature of plant damage may lead to an inappropriate or non-optimal choice of accident response to procedures. Recognition of this distinction is essential in achieving the goal of minimizing the risks associated with nuclear plant operations.
- 6. Question 6)
Are any of the phenomenological issues related to one another or combinations? If so, how are they related?
Response
There are many complex and direct phenomonological issue relationships which must be considered. As an example the degree of success of stabilizing a core damage sequence prior to reactor vessel failure has a very direct and powerful 19 fluence on the importance of core-concrete interacties phenomena if a high degree of core stabilization can be achieved. PP&L believes that optimal Emergency Operating Procedures
backed by offective operator training and relatively low cost plant g modifications can result in a very high level of success in this regard w for most BWR plants. At Susquehanna ve believe the success rate should be I well in excess of 90%. l There are nu=erous other similar relationships which should be considered J in the development of optimal E0Ps for a plant. These include.
I o the relative probability of ex-vessel steam' explosion versus l drywell floor flooding to reduce core debris threat to the -
containment, i 1
o the relative probability of loss of containment integrity from )
drywell spray operation versus the need to reduce containment I temperature or flood the drywell floor, I o the relative negative impact of wetwell vent closure before re-introduction of non-condensibles versus the added release associated with extended vent opening, o and others.
It is these types of interactions which should receive primary attention as opposed to consideration of adverse influence of uncertainties in l I
mod.els and data for individual event sequences. The analyses should be done with consistent and realistic models unless it can be shown that' conservatism in treating a given phenomonological uncertainty will not lead to a selection of an unfavorable procedural response to accident g )
i sequences.
l
- 7. Question 7) l Is the containment response (failure modes, locations of leakage paths and tear locations) to overtemperature/ overpressure challenges evaluated for i Peach Bottom (44] likely to be similar for all Mark I containments' !
l
Response
l While PP&L cannot address the adequacy of analyses for facilities other l than Susquehanna, we do believe that dominant severe accident sequences and'the specifies of the damage state are surely highly plant specific. i The results for one plant should be used only as general guidance for l examination of a different plant even when the superficial design features I of the plants are identical. The details of construction, equipment installation and configuration, and support system design can have a dramatic impact on the magnitude and nature of dor inant accident sequences.
Page 47
- 8. Question 1)
Do the periods of deinerted operation allowed by technical specification (24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> before shutdown and 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> upon startup) present a sufficient g
vulnerability to necessitate a reduction?
Response
This issue was not addressed in the Susquehanna IPE (January 1986). We intend to examine the potential impact of this LCO in the first revision to the Susquehanna IPE which is tentatively scheduled for completion in 1988. Our engineering judgement at this time is that this LCO does not represent a significant contributor to increasing the ' severity of consequences from severe accident sequences.
- 9. Question 2)
Are the nitrogen supplies provided at Mark I plants sufficient to maintain an inerted atmosphere for long term station blackout scenarios?
Response
Specific analysis has not yet been performed for Susquehanna, but this issue is not expected to represent a problem at Susquehanna,
- 10. Question 3)
Are the risks of oxygen addition from compressed air supplies used for primary and back-up operation of valves and instrument lines too high? If so, should nitrogen back-up supplies be substituted for oxygen supplies?
() Response l
Susquehanna uses a nitrogen system for these purposes, and so PP&L has not l addressed this issue.
Page 49
- 11. Question 1)
What are the potential negative impacts of blocking most of the containment spray nozzles (for example on RHR system operation), and providing a crosstie to diesel fire pumps? How can these impacts be ameliorated?
l
Response
PP&L recognizes this issue as a potential problem area. Our recommendations for this issue are presented in the answer to Questions 5) and 6) on page 40.
- 12. Question 2)
Fission produce scrubbing is a known advantage of spray use. What quantifiable advantages also could b6 realized with respect to shell and debris cooling (separately, and in combination with venting, or in combination with venting, core debris control and use of fire sprays in
() the reactor building)?
Response
The PP&L views on this question are also covered by our ranponse to Question 5) and 6) on page 40.
i 1
1 1
i Page 52 llk l
l 13. Question 1) l A key question is whether or not there is a net benefit to plant safety I derived from venting. Can specific procedures and hardware modifications be developed to demonstrate that a complex severe accident at a Mark I plant is understood enough such that an adequate venting strategy may be developed?
Response
PP&L evaluations of the potential benefits of wetwell venting have shown a significant reduction of containment overpressure failure frequency if wetwell venting capability is provided. In addition we have found that assuring that the PP&L defense in depth criteria are satisfied may require wetwell venting for some accident sequences, although our evaluations and conclusions on this aspect of wetwell venting are not yet final.
On the negative side, however, we find that is is highly likely that modifications to existing equipment will be required in order to assure that vent operation does not result in potentially detrimental loss of equipment or loss of access to critical plant areas. Furthermore, even the most limited modifications to achieve an acceptable level of certainty that vent operation will not result in such loss are expected to be quite expensive and co':ld result in a significant increase in outage time.
Finally our preliminary studies to determine appropriate wetvell vent functional design criteria and operational constraints have indicated that these issues are quite complex and considerable effort to develop a demonstrably acceptable design and the necessary procedures for its use may be required and may involve considerable effort and expense.
Our preliminary findings on this subject for Susquehanna were presented to an NRC team headed by Dr. W. R. Butler during a viste to Susquehanna on Feb ruary 11, 1987. We recommend reference to the information provided to Dr. Butler and his associates during that visit.
- 14. Question 2)
Can the potential negative aspects of venting be minimized? How?
Response
The negative aspects of venting can almost certainly be sharply reduced, if not completely eliminated, by providing appropriate design features and by imposing appropriate operational constraints. The specific mix of dominant accident sequences for a plant can have a major impact on the nature of the design features and operational constraints required. The design features in turn can have a dramatic influence on the cost of the wetwell vent and the operational constraints can have significant impact on operator training. It is important that ebe dominant accident g sequences be properly identified in advance so that the financial impact W of providing a vent will not be needlessly high and so that the procedures and training are appropriate to the accident sequences judged to be most likely for the plant.
)
l l
t Question 3)
Another issue of major concern centers around the pre-accident decision of who has the authority to cause venting to take place? Should the operator be vested with that authority, the utility management, or should a passive system be designed which will provide absolute, invariate action?
Response
PP&L has not yet resolved its views on this issue. At the present time our inclination depend on operator action to open the vent in accordance with approved procedures for the action. It seems very probable that a manual enable function should be provided to allow subsequent vent operation regardless of whether the subsequent operation is passive-automatic or operator initiated. I
- 16. Question 4)
From a design perspective, the question arises as to the need for a safety-grade system. Can a modified vent system that includes non-safety grade components be relied upon just as a safety grade system could?
Response
The interface of the vent system to any existing safety related system must satisfy the existing requirements for such interfaces. It is f]
% entirely inappropriate, however, that the usual requirements for safety grade systems, such as redundancy or seismic capability, be imposed on the vent system itself.
There should be assurance that good design and construction practice has been utilized, and there should be some provision to assure that the j system would have a high probability of proper operation should it be required during the licensed life of the plant. PP&L has not yet generated a position on what the nature of the provisions cc assure these features and to demonstrate adequacy should be.
1 page 54
- 17. Question 1)
Are drywell curbs technically feasible?
Response
PP&L believes that the effectiveness of dryvell curbs is highly plant specific, at least for Mark II plants. For Susquehan.a. drywell curbs, which could provide additional assurance that critical containment functions would give additional protection against core debris, are believed to be feasible. We have not yet determined that such protection provides significant additional protection against containment failure caused by core debris. We believe that such a finding could only be made af ter execution of the investigations discussed in Questions 5) and 6) on page 40 have been carried out.
I At this time we believe primary attention should be given to core stabilization (prevention of reactor vessel failure) and core debris h 1 quench (pre-flooding of the drywell floor). If these actions can be l achieved in a high percentage of accident sequences for Susquehanna, then we would see no true benefit from provision of such curbs. Further, we have identified no probable sequences for which defense in depth would require such curbs based on cur current expectations for resolution of core damage and core debris phenomena.
In the worst case, we would expect to identify a need for only greial protection from curbs to serve the function of providing additional margin against uncertainties as opposed to a full protection requirement
. providing primary protection against loss of containment integrity resulting from core debris attack on critical containment components. ,
- 18. Question 2)
Are torus room curbs with or without drywell curbs feasible?
Response
Not applicable to Mark II containments.
- 19. Question 3)
Would curbs in either location interfere significantly with reactor g operations? In addition, would curbs in the drywell alter the outcome of W accidents within the original design envelop with respect to such topics as blockage of vents and the drywell heat attenuation capability?
Response
This issue has not yet been investigated for Susquehanta with regard to operations. It-is highly unlikely that the containment design basis would be in any way influenced however.
- 20. Question O ,
4 Several means of protecting drywell concrete from debris attack, for example by covering the floor with a water-porous pebble bed of thoria and I alumina, have been suggested. What would be the efficacy and practicality l of such measures? Is water in addition to a curb necessary to protect the curb?
Response
At Susquehanna the drywell floor is covered by a steel plate which serves I as the leakage barrier between the dryvell and the watwell air space. We believe that the presence of this steel plate will serve to' prevent immediate decomposition and release of water vapor or carbon dioxide from the concrete for the debris pour rates expected for accident sequences at Susquehanna. This would prevent the energy of chemical reactions from g inhibiting debris quench by water due to film boiling or critiet.1 heat w flux heat transfer limitations. Further, according to current ORNL models, which we believa have considerable merit, the initial debris pour vould include relatively little fission product decay heat energy. For
. _ m J
l 1
this reason the quench process would be required to remove only the latent
(]'
heat of fusion and sensible heat carried by the core debris. The subsequent period of the core debris pour which would contain UO, in significant amounts would then be separated fre- the concrete not only by the steel plate, but also by the initial pour of debris which has been quenched.
This, we believe will also then permit effective quench of the UO 3 ,
material and prevent any significant attack by the melt on the dryvell concrete.
For this reason we view water on the drywell floor as the primary means for protection against core debris phenomena and consider barriers to be only a devise that may be needed for partial protection to provide additional margin. The studies which we will carry out, if necessary, to establish the validity of this expectation are only in the preliminary planning stages at this time.
Page 56
- 21. Question 1)
Are the types of improvements described sufficient to improve ADS reliability?
Response -
O Susquehanna already has in place additional2N bottles to provide for extended operation of ADS. In addition, PP&L is currently in the process of installing cables and connector provisions to allow a mobile 90 Kva generator to be brought in and connected to feed power to the critical 125 VDC battery chargers to assure indefinite availability of critical DC power during Station Blackout sequences.
It is not certain at this time that additional protection for critical electrical cables is required. These cables are already qualified for the extreme conditions which can occur during the design basis accident. PP&L ,
calculations indicate that these conditions will not be exceeded during i the accident sequences which are dominant for Susquehanna. There is some possibility that the duration of high temperature conditions could exceed the qualification time limit for some accident sequences or that local temperatures could exceed the calculated values of mean drywell temperatures in critical areas.
The PP&L view on these possibilities is that the appropriate response will be to utilize drywell sprays to reduce drywell temperatures for such cases. At the present time, we are not certain that the specific procedural guidance na _ ' to detect and avoid such conditions in all cases currently exist
... will attempt to develop additional detailed information to address this issue in our planned IPE revision.
O
O
- 22. Question 2)
What are other recomendations?
Response
See the response to Question 1) on page 56.
- 23. Question 3)
What are the benefits? For example, it increased availability of the ADS system necessary, and will it eliminate the risk from direct containment heating?
Response
For LOCA or reactor vessel failure sequences (direct containment heating) there is no concern over ADS since reactor pressure control is no longer an issue. The reactor vessel vill operate at or near dryvell pressure regardless of ADS operation.
Concern over ADS operation is related to reactor vessel pressure control to permit vessel injection at lower pressure, to reduce the heat source to the dryvell when dryvell cooling is lost, and to avoid the possibility of reactor vessel failure at high pressure.
- 24. Question 4)
What are potential adverse interactions between any of the proposed improvements and existing safety systems?
Response
The only issues which may represent problems are:
o adverse offacts of vetting dryvell equipment, and o excessive pressure dif ferentials between vetvell and dryvell or containment and the reactor building.
We believe critical containment equipment is designed to withstand wetting from two phase blowdown in LOCA and that differential pressures beyond the containment design basis can only occur if venting has significantly depleted the containment design basis non-condensible gas inventory. This last issue must be addressed in design of the vetvoll vent and the development of appropriate operational constraints.
- 25. Question 5)
Are there other improvements that might also be technically feasible to preserve the Mark I containment function in the event of low probability g challenges? W
l Response ,
PP&L believes that explicit and plant specific attention should be given to assurance that all plant capability to avoid plant damage sequences or minimize their consequences will be utilized, and that the plant specific E0Ps and operator training will assure a high probability of successful exploitation of such capability. This assurance nust be a first priority in evaluating adequacy of plant performance in severe accident events which are dominant for the plant.
Only af t.er this assurance has been achieved should attention be given to major plant modifications to achieve a reduction in frequency of potentially severe consequence sequences. In all cases, the primary criterion for adequacy should be defense in depth as defined by PP&L.
Assurance of defense in depth is believed to be the most effective and credible means of assuring adequacy of plant capability against accident sequences having potentially severe consequences.
O ,
O
W5 1 7 yurd QUESTION 3, PAGE 39 O OPERATOR ACTIONS o THE CONTRIBUTION OF HUMAN ERROR TO THE OCCURRENCE OF l INITIATING EVENTS AND EQUIPMENT FAILURE DOES NOT IMPLY THAT l OPERATOR FAILURE IN RESPONSE TO A SEVERE ACCIDENT MUST BE A j SIGNIFICANT CONTRIBUTOR TO THE SEVERITY OF THE EVENT. )
i 0 CLOSE ATTENTION TO PROCEDURES AND TRAINING CAN REDUCE ADVERSE I CONSEQUENCES FROM OPERATOR ERROR IN RESPONSE TO A SEVERE ACCIDENT TO A NEGLIGIBLE LEVEL.
O PPal HAS IDENTIFIED 12 OPERATOR ACTIONS IN RESPONSE TO THE DOMINANT SEQUENCES FOR SUSQUEHANNA WHICH SHARPLY REDUCE THE FREQUENCY OF SEVERE CONSEQUENCES, (LETTER TO E.S. BECKJORD, l NOVEMBER 13, 1987.)
1 Q 0 PPal HAS FOUND THAT THE DOMINANT RISK CONTRIBUTORS TO SUSQUEHANNA ARE VERY SPECIFIC TO THE DETAILS OF THE DESIGN OF PLANT SUPPORT SYSTEMS.
O PPal IS CONDUCTING A SERIES OF SIMULATOR EXPERIMENTS.TO VERIFY OUR ANALYTICAL MODELS FOR OPERATOR PERFORMANCE. TO DATE, THE RESULTS SUPPORT OUR MODELS.
O
1
- OPERATOR ACTIONS l (2) 0 WE HAVE FOUND THAT THE USE OF CONVENTIONAL RISK ASSESSMENT MODELS AND ASSUMPTIONS CONCEAL THE TRUE DOMINANT RISKS TO THE PLANT AND CAN DIVERT ATTENTION FROM THE CRITICAL AREAS NEEDING ATTENTION IN PROCEDURES AND TRAINING.
O IF THIS DIVERSION OF ATTENTION IS PERMITTED, THOSE CONVENTIONAL MODELS MAY BECOME SELF-FULFILLING PREDICTIONS OF PERFORMANCE.
O WE BELIEVE THAT CONTEMPORARY BW3 RISK ASSESSMENTS SUFFER FROM THIS DEFICIENCY AND CONSEQUENTLY PRESENT AN IMPROPER PRO' FILE OF DOMINANT ACCIDENT SEQUENCES.
O THIS IMPROPER PROFILE INVOLVES:
AN INADEQUATE ACCOUNTING FOR STOPPING CORE DAMAGE O PROGRESSION BEFORE REACTOR VESSEL FAILURE.
EXCESSIVE MAGNITUDE, RATE, AND ENERGY OF CORE DEBRIS POURS FROM A FAILED REACTOR VESSEL ONTO A DRY DRYWELL FLOOR. l 0 THIS IMPROPER PROFILE RESULTS IN A GREATLY EXAGGERATED ESTIMATE OF THE THREAT TO THE CONTAINMENT OF THE DOMINANT ACCIDENT SEQUENCES.
C:)
l QUESTION 6, PAGE 40 l
(])
PHENOMENA INTERACTIONS 0 THE HIERARCHY OF ACCIDENT RESPONSE OBJECTIVE SHOULD BE: 4 l
l AVOID LOSS OF FUEL CLAD INTEGRITY l
~
AVOID LOSS OF REACTOR VESSEL INTEGRITY CAUSED BY CORE MELT. '
AVOID LOSS OF CONTAINMENT INTEGRITY CAUSED BY CORE MELT AND REACTOR VESSEL FAILURE.
l 0 FUEL CLAD DAMAGE THE DIVERSITY AND REDUNDANCY IN BWR SYSTEMS RESULTS IN A NEGLIGIBLE PROBABILITY OF.SUCH DAMAGE IN THE ABSENCE OF
() MASSIVE COMMON MODE FAILURE SUCH AS STATION BLACKOUT OR ATWS, i
O REACTOR VESSEL FAILURE THE DAMAGE PROGRESSION MODELS OF BWRSAR INDICATE A VERY HIGH PROBABILITY OF AVOIDING REACTOR VESSEL FAILURE.
O CONTAINMENT FAILURE INDUSTRY AND THE NRC HAVE GIVEN LITTLE ATTENTION TO CRITERIA FOR SUCCESS HERE. RESEARCH IS NEEDED.
O
QUESTION 5, PAGE 40 O SEPARATE EFFECTS EXPERIMENTS 0 WE NEED TO DEVELOP VALID SUCCESS CRITERIA FOR STABILIZING CORE DEBRIS.
IN-VESSEL l
Ex-VESSEL O SUCCESS GITERIA MUST COME FROM ANALYTICAL MODELS OF THE PROCESSES INVOLVED.
O FULL OR L/RGE SCALE TESTS TO VALIDATE THESE MODELS ARE NOT PRACTICAL. 1 0 THE CRITICAL ELEMENTS OF THE ANALYTICAL MODELS SHOULD BE ADDRESSED BY SEPARATE EFFECTS EXPERIMENTS.
O O THERE IS NO APPARENT SYSTEMATIC EFFORT TO CARRY OUT SUCH A PROGRAM FOR STABILIZATION OF EXVESSEL CORE DEBRIS AT THIS TIME BY EITHER INDUSTRY OR THE NRC.
O
N\SO.\b ks% NSED % y h h%ca_s
.> wwa hostt%s ,
O NI .
- }7 j p l
l ErFEC-S OF W A- ER l l
l i
i
)
O
- DOES WA TER A FFEC T THE RADIONUCLIDE SOURCE TERM ?
9 I
l I
k O
4 O
~~
4 :: EX 3 E R:: E \1C E
~
MEL TING OF THE CORE FORNER A T TMI CA USES DOUB T THA T L INER MEL TING CAN BE PREVENTED B Y WA TER O
i 1
O u
_ _ , , . . . _ . -,----,---,---m. ( y y
1 i
O l
l
, WA-ER :: N V ENTORY ,
- L IMI TED IN THE DRYWEL L BY OPENINGS TO THE TORUS TO A DEPTHOF1 TO2 FEET O
- COULD BE REPLENISHED BY THE .
O 1 . -- -- . - .
- O l COMBINED CORE DEBRIS-CONCRETE-COOLANT INTERACTIONS
- CON TEN TIONS
~
O .
WA TER QU8NCHES DEBRIS WA TER QUENCHING LEADS TO ENHANCED HYDROGEN PRODUCTION WA TER SIMPL Y BOILS S TEAM EXPL OSIONS l
O
- n. _,
l l
O
. DATA BASE ON CORE DEBRIS-CONCRETE-COOLANT INTERACTIONS I i
- A CM TES TS l ONE EXPL ODED & ONE DIDN ' T
- FRA G TES TS O -
NO VIOL EN T INTERA C TION NO ADDED HYDROGEN DEBRIS-CONCRE TE IN TERA C TION NO T IN TERRUP TED
- SWISS TES TS l DEBRIS-CGNCRE ~E IN TERA C TION NO T IN TERRUPTED 1 NO ADDED HYDROGEN BOilIt\!G WAS NO T A T CHF BARBO TA GE ENHANCEMEN T OF FILM BOIL ING POSSIBLE
- l. .-
A T TENUA TED A EROSOL S _
O .. . . . - -. .
08D8 8
- 8 Go
~
00 -
M e
. ~
Ce
- O M
- I
- ' o . i O 5 a
o
- o*. -
=
L eO .
- 2 o- ,
f=W o gf o
S
_ E5 - m
/
O9 R ; ; ;
O (W3) NOISoW3 gg -W E
t HEAT FLUX O COOLANT TEST SWISS-2 O !
- 2. 5 ....,....,....,....,....,....,....,....,.... _
~
- 2. 9 X DATA _
i -
a
" 1. s :.
N -
=
z x -
j v -
t x -X 4
3
~
XX g 1. s gy#
l x :
I x%***x%x / x *xgx * .
tsJ -
l I - X '
i
\
E5 -
X e
~
....i....i....i....i....i.... ....i....
l g. g i
s is is as as as as 4e
.l :
4 TIME '041NUTES) i G
1
4 O
CO NCLUS :: 0 N i
THE OA TA BASE ON COMBINEO CORE DEBRZS-CONCRETE-COOL AN7
/NTERAC7/ONS ZS VERY L ZMZTED O
O
_.y., . , _,___ _ ..,.. . ..___, _ _ _. . __,_,, ,
O CO N C _US :: ON WE CANNOT PROVE THA T WA TER WZLL PREVEN 7 OR DEL A Y S/GN/FZCANTL Y L /NER MEL T/NG O BY CORE DEBRZS .
1 O
l I
I O
ErrEC-~S Or S3 RAYS i 1
l
- FORM WA TER POOLS OVER THE CORE DEBRIS TO TRAP AEROSOL S DEMONSTRA TED IN THE. !
SWISS TEST SERIES MOS T EFFEC TI VE IF THE O POOL IS SUB-COOLED ,
- TRAP AEROSOLS IN A TMOSPHERE \
l
- COOL THE A TMOSPHERE 8 THE DR YWEL L i
O i l
1
lll J
~
M o uN A N E O T T I I A
E O sA I
S T T D N 2 - s I E
- N P D 0 A N 0 S N O wE OD R I
8 N
S I
W O
I T
I I
T t N P
U R
sO C
O[
C 1
S D R I D E T
A N T R I
0 S E R 0 E T Ei W
I 6
1 T A O W P N R I OI E T T "
F A A M 0 0
R I 2
O E N e.
- e. 4 2 1
)
o - S E o (
G - o_ - o_
E N M L O I I
T O T I 0 S E
~
D I 0
R D O O A 1
R F =
E E S
R .
A E I 0 T .,e ,. 5 A
W 0 s 2
- i_ 1- ._
gQ5 O{$
O
' f lll
O O O e
=
i i i i i i x,
u - .
w -
I z 8 T ,
ay 10 r :
v E NO SPRAY :,
4
- 7 ,
> 10 r -
o - -
< ~
6 a 10 r 1 i
m -
t o .
400 um SPRAY -
W 5 i
O 10 r 1 z -
l w - .
n_ ~
U) l a
! m 4 ,
10 r
[ . .
\
2 g g I I I I E i
600 700 I
l 100 200 300 400 500 l i a
! TIME (MINUTES) i i
l O
O
- - O N
- . 5
)
O 1
- - O i m
- O l c
- . O 4 m I
- . O W '
O I
- . z ,
O O 1
- . O U 1 E O W l g g . T W '
E-N 3 0 4 -
C W CT) g e O Z O OO @
- m O "
4 oo n
tn N u n H
"D a > 0 l a . O \
E 5 i N 3 . N l E-N >
3 0 4 .
04 O
- C O O O O CD
- O O LD .
- :.7
- U n .
= O k -
h
. s O il '
. , l _ t 111 m '_ _ , , ,,,,I , , ,
O O O O O O O O O O O O O O T N O CD Q v N
~ m e j 1
1 (6H) SSYH 030N3dSnS 1 0 '
I
O
't CONC _US :: 0 \1 H
DR YWEL L SPRA YS A UGMENTED B Y SPRA YS IN THE TORUS ROOMS COULD A TTENUA TE RADIONUCLIDE RELEASES BOTH BEFORE AND O
AFTER L INER FA IL URE l
M l
O
~
O S D ECUi A-~:: ON DR YWELL SPAA YS MA Y AL SO A TTENUA TE THE REVAPORIZA TION RELEASE POSTUL A TED IN SOME ANAL YSES FOR MARK I BWRs O
O
7 !
l 7A .
O i
A METHODOLOGY FOR CALCULATING DEBRIS SPREADING ON THE BWR MK I DRYWELL FLOOR USING ,
CONTAIN r
l l
C. R. HYMAN B0! LING WATER REACTOR SEVERE Acc! DENT IECHNOLOGY PROGRAM .
O OAx RioGe NATIONAL LA ORATORv !
i i 1
1 PRESENTED AT BWR MARK 1 CONTAINMENT INFORMATION EXCHANGE WORKSHOP l 8ALTIMORE, MARYLAND
- i FEBRUARY 24-26, 1988 I
'l O
l
I
. O i
i PURPOSE OF PRESENTATION
- i 4
DISCUSS METHODOLOGY DEVELOPED TO DETERMINE ;
COMPOSITION AND THERMAL CONDITIONS OF DEBRIS ADJACENT TO BWR MARK 1 DRYWELL SHELL i
PRESENT RESULTS OF SHORT TERM STATION BLACKOUT i SCENARIO AT BROWNS FERRY 1- .
. O *
\
DISCUSS DEFICIENCIES IN PRESENT METHOD
! l t
.}
l I
l 1
)
l0 4
INVESSEL ANALYSIS POSTULATES THE FORMATION OF LOW MELTING TEMPERATURE EUTECTICS LEADING TO TIME VARYING DEBRIS POUR TEMPERATURE a
4500 0 I J
2soo
~
. 4332 r O -
230o g
' ssa _ _ _, _ _ _ _ _ _ _ _. _
O \ 3363*r -
2000 j sooo -
82 24 ""
f oo r
~
==o t C 5.4 m - _., _ _ _ _ ____ _ _ _ - _ _ _ _ ___- __ _ v 140o 0: 200o .
ca w
2100*r[
Q isso -
tion o
$ 4040 - -
soO a
F y Soo -
600 w
.a .._._n.. .
h w
0 -
zoo ano see n
44o sto soo
. .a sao veo a4o s .
s2e sooo H !
l TIME (MINS)
O O O
1 l
l l
ORNL-OWG 87 4444ETD O rTYPICAL VESSEL POURS FOR SHORT TERM 3 STATION BLACKOUT I l
5400
.a 40 J
4800 -
I g .
35l g 4200 -
, yy 3'" '
5 .
j2000 25=3 g
- a E2400 -
N5 m ~ 1800 - h 15 E4
$ 1200 d 600 -
5 d I
4000 n'A L \ .1 -- m -
n" ,a 3000 l
- ~
. 3. .
24 g Q
l g_2800 -
205 '
e -
w*j2400 2000
@~
' N x 8 1600 l '
12y 1200y g
ag l e : I I d Ni % 4 n 'ft i 1
, a g 2700 - '
'ly.
y 2400 -
1sy g 2100 -
() . 3, 3 g r is00 : V 3 5-gI g g'1500 1200 f '
sg-
$h e
'~ :
L-200 260 320 380 440 500 560 620 680 740 800
< ~
2 0 e O r= -
< om!>
ORML-DwG 87C-4446 E1D 3
MOLTEN DEBRIS FILLS SUMPS, IN-PEDESTAL REGION, AND EX-PEDESTAL REGIONS IN SUCCESSION q ---
~~,, ,,
IN-PEDESTAL , ,, ..
5, DOORWAY 2
REGION (30 m )
k,, s ,
s
~% h
, ,. ^\
EX-PEDESTAL REGION (102 m2)
- .' l::
/ :.'
(/ :::,.'.
g .
.... i l n 7p,,,,,3, )/ \ /
DRYWELL \ ',
SHELL *
), PEDESTAL mu -__ __mW
( >
e O O
O o O :
ORNL-DWG 87-4642 ETD
~m r
l CONTAIN PROVIDES THE FLEXIBILITY TO PERFORM MULTICELL DRYWELL CALCULATIONS i
e SEVEN-CELL DRYWELL MODEL ,
REACTOR VESSELS ,
4 1
/ ~\
n t
b 7 ;OLOGICAL SHIELD lA CORE <
f
-DRYWELL SHELL 7 7 Y) '
4 !!b 3 f!I 4 REACTOR 2 5 .
j PEDESTAL / 6 ) 1 J 6
~* 5555; hf!E"'
FLOOR
.t l
A STEPPING PROCEDURE HAS BEEN ADOPTED
! TO CRUDELY REPRESENT TIME DEPENDENT l SPREADING ON THE DRYWELL FLOOR l
- ~ m ,,' i
- i .
i
,,N
,fYf ##// ,
04, (
? .
.: :l.?: *.:l i.ill :iii:.i:l:'~: :i. .
~ll@@ill;}
~
qi an I J ':lll@Q!!:.L i
'k,,,,, ,lp '#'*N , :I!'!
l :!I?::I!: '_
l I r i
SUMP FILL IN-PEDESTAL IN-PEDESTAL &
1 EX-PEDESTAL l i
t > l
- e e e i
~
1 I
4 ORNL-OWG 87-4648 ETO O 3 4
. e i
IN--PEDESTAL CALCULATION D IS i ONLY AFTER DEBRIS IS MOLTEN 3000 l
C 2900 -
l 9., I 5 -
1800 -
g 2600 -
I N I g h 2#L [1589 K
- d1800$
g 2200 -
N l
$ 2000 - I $
,-IN-PEDESTAL w ONLY
' IU
.a
.a t
~
< i 4 syyp IN-PEDESTAL +
t g plLL 18,5 m2 IN-PEDESTAL +
g 1600 - 37 m2 EX-PEDESTAL 1200Y lt' EX-PEDESTAL $
h 1400 - E 2
[356 min 566 min
' ~
i 1000!-
' i >
200 260 320
.l ,
380 440 500 560 620 680 TIME (min) 74 O( .
onni j j i
l ..
ORNL-WSC-1210 ETD C h l
! PROBLEM IS REINITIALIZED :AT THE TIME THE DEBRIS REMELTS; THE DEBRIS GEOMETRY 18 CHANGED 1
TO REFLECT FURTHER SPREADING i
i
- e DEBRIS IS DISTRIBUTED SO THAT DEBRIS DEPTHS ABOVE I FLOOR ARE EQUAL e ABLATION CAVITY VOLUMES ARE CONSERVED l
e DEBRIS IS SPREAD SO THAT MIXTURE COMPOSITION IS l UNIFORM IN BOTH IN-PEDESTAL AND EX-PEDESTAL REGIONS i
- EX-PEDESTAL AREA CHOSEN SO THAT DEBRIS DOES NOT RAPIDLY FREEZE
- CALCULATION IS EXTENDED IN TIME UNTIL DEBRIS REMELTING OCCURS l
t .
OfRI>
G G G
O. CORCON ANALYSIS ALONE IS INSUFFICIENT TO DETERMINE THREAT TO DRYWELL SHELL CORCON'S RADIAL STRUCTURE IS CONCRETE, NOT
~
STEEL i i
NO TEMPERATURE PROFILE IN RADIAL STRUCTURE l NO THERMAL INTERACTIONS WITH RADIAL STRUCTURE
!F DEBRIS LESS THAN ABLATION TEMPERATURE l O .
O
O CURRENT ACTIVITIES COMBINE THE ONG0ING EFFORTS OF SEVERAL GROUPS I Sill, BNL MODEL DEVELOPMENT IN-VESSEL ANALYS!S,
. EXPERIMENTS EXPERIMENTS K. CERGERON, D. POWERS L. OTT, G. PARKER, G. GREENE ORNL i
MCCI 'NALYSIS A
C. HYMAN, ORill i
I lf DRYWELL SHELL ANALYSIS T. KRESS (ORNL)
K. BERGERON (SNL)
G. GREENE (BNL)
O
. < i
O r 3 2
CONTINUED TEMPERATURE INCREASE WITH 37M EX-PEDESTAL AREA INDICATES THAT FURTHER SPREADING WOULD OCCUR 2900 - 1850
- g f2810 -
1800g b2720 -
E 1750E 5
5 e.
g2630 -
17003 w -
O
@ 2540 .
g 4 ,
1ssog j
b 450 2 -
[1589 K
- A , -
18 " s W
gvv
?
2360 -
- '* 3$
m m
E2270 - y 1500 w 18.5 m2 37 m2 h _,
W 21# EX-PEDESTAL AREA, g w EX-PEDESTAL AREA
- - 1400 300 350 400 450 500 m 600 650 700 750 TIME (mW O
e orni>
O DEFICIENCIES IN CURRENT METHODOLOGY COUPLING BETWEEN CORCON AND BWRSAR DID NOT CONSERVE ENERGY e HARD-WIRED SOLIDUS AND LIQUIDUS TEMPERA-TURES AT BWRSAR EUTECTIC MELTING TEMPERA-TURES .
NO CONCRETE OUTGASSING PRIOR TO CONCRETE ABLATION 9 IMPACTS CHEMICAL, ENERGY PRODUCTION AND DEBRIS'SUPERHEAT UPON SHELL CONTACT g SPREADlNG ANALYSIS NON-MECHANISTIC e BASED ON MELTING / FREEZING DEBRIS BEHAVIOR, NOT FORCE BALANCES O
i O O ~
O i
PEACH BOTTOM WITHOUT DEPRESSURIZATION SHORT TERM STATION BLACKOUT BEST ESTIMATE EUTECTIC MEIffS d
In 4400 FEBRUARY 9,1988 En N
y 4100 -
2600 ,
o I
%. ~
'3800 -
Q -
400 N .
d a 3500 -
(1 -
2200 N
g h 3200 -
g -
2000 v Q
CD y 2900 -
~
h f 1800 ,
m 2600 -
M -
l3 -
k 2300 -
1600 N -
fl.
2 2000 i ,
^
200 200 212 218 224 E 230 236 242 24a 254 TIME (MINS) 4
PEACH BOTTOM WITHOUT DEPRESSURIZATION 4 SHORT TERM STATION BLACKOUT M
BEST ESTIMATE EUTECTIC MELTS FEBRUARY 9,1988
$ 64 3 _
28 58 -
l -
2.
{ 52 -
6 -
22 46 - '
? -
4
& 18 :
? **
m
~
Q f 2 .
1, 3 34 m -
g su 3
~ E a
2a -
13 O
Cal 22 -
10 4
C4 16 -
]s -
7 b 10 200 280 360 440 520 600 880 780 840 920 1000 TIME (MINS)
G G G
O O O PEACH BOTTOM WITHOUT DEPRESSURIZATION SIIORT TERM STATION BLACKOUT BEST ESTIMATE EUTECTIC MELTS FEBRUARY 9,1988 d
Co 4400
~ -
2600 p
y 4100 - _
O f 3800 -
S 2200 N
(L 3500 -
~
2000 Z _
CD
.g 2900 -
IS00 A -
O g 2600 -
C4 -
D ] 1600 k
2300 -
$ 2000 ' - ' . , ,
M 520 600 760 6 .0 920
- 200 280 360 440 680 h TIME (MINS)
e e e o e en.
e e e 1 4toF.P 4 P O t O O O N 0 8't e as e e e O e e o e e e e o e e e ee e e e e 0 == 4 if it em = O e O Es O @ t N
- to 8 7 8 e 4 1 3 m to e b e e B C I e e" I e 3 e T 4 9 ed 8 e e
& e to 9 0 o
e b sP. e q, & e e e an as e e a 9e e 3 e w- e e e es e e .e e H e en e Pe tre d se ao e d N M e- g M i && e 4 e y eD e e e e e e o e e e e e i e e a= e e == ei e d se e e e p 89 4 me e ee e d e e se e f4 mi e e1 M J e4 N O w e 3 e e 0 6 Cs P= e N *= M & P G en e
- e e e 3 e we e c N e, o e up ,
h e e 8 a'e se O dee e e e s pN e e e r e e e o e e
,a e e a e e e
e e e e i aeeeeaemeeeM 4 e e e eg e e e e e e o e e e e e 0 e e e e e e o e e e Sie e e e e e e e d e d e 6 3 se 9 me o e e e G 1 e e ee 4 e e e e. I
- Ge e si e I e
.e e e e e eW e se e C en
- e e e o e 6 en
- b b e le e #
- 3 e an e g e e e4 e e e e .s e e m e e e m I e e
> e es 4 ee e e e e e e o e e e e e e e h 6 0
- e e' e 8't O si e G 4 'e O ** 8 O e
% b 4 =* o 4 0 ** e N ge e eO u e e se e et em 0 e e Pm I ee e se b e gd se 4 e e e e D t 9 e e er e to e 3 e D &
- t4' 3 e" e C 9 e e
r 2
- a a e eseneNemeeseee N d es d e m d e h ee m e e e e e 3 3 0 g e o e e e e e o e e o e9 e ed e e e e o e e e ee e e e9 e e T e ea 89Sh seN e edeMPwdMe N d 6 si ( M N e t e e e e e e se e e e e en e I pm as iN e de 8 3 ee 4 ee me w9 e i e e se e i W $ em O N se gi e as @ N fe N 9 es e N b 8* 8 e es W% W1 e P NM e' e e G h 0 w 0 Mefw N M 0N
- m J A N l 0 ee si e4 e es u e e e 4 e =e @ e me N e C b. 3 a se ** e e b es me e a e, ee e a en s. .a e e= e eI G fe e si N si G e me 89 e d 9 e e N Pm N se N Ee p ee se se e e d i Po C e e
== an 8 3 e e o e e e a e e o e e e e as en e e e o e e e a e e o e e a e == e e u 0 en e es se N e M e te e e 85 e0e ee d N N e e & e e d d W) @ N 0m e e en e f se er. og se e e5 8 O ee & e e1 se Pm J ei si e F N ee W G em % e i
i se G 4 e es se se ee eg Po es en N e as #1 se O Pe & e C e 8 e 89 tfl 0 .lB en e e P M ee e e0 pm e i 6 e 0 m e) 9 se to e N mo se e8p a l b . e t 4 es e .e e et e !
6 le e M 4 J e I en er e qe r e '
e e e e
- e. b o e# e i b a e N I
. - e s,e - e. .
A en e e as e* e e e g l A e g4 p @ #3 e se e e e e $ e g d p se gi pg g N e @ en e e N g g O e p e ey 8 4 e o e e e o e e e e o e 6 e ed es e e e e o e a w e e e e e8 e se e 9 e 4 ** an a ge f*se d e 4 e 4 e e 1 P ee ededeM@NJeeaPmd 6 se 4 e O C e & I 3 =e N 0 A es e e e ed e f1 ee e e Po et se e i e e e se en e G 8 y I Sg P es d se se e se e se se 9 e e 4
== C e me 4 es i es e em /% ao e se @ J$d e G e op i es 9 & se O N Po 0N te e A e G ** I 6 e B N
- e e a g I ,e e d 8 eW oe on =* e W 5 e me C e* =* .8 e. e e o*
- er o 6 en a e 4 ee e e en es e a De e
- f. O 4 e ee e e D e e e to mi es en 4 e P e D eO N M & S @ ** A N e se P W 4e 3 e se e e d e e9 ee e ed e e e N 6 89 e se e e se 0 9 e e e e e o e e e e e e6 e e e e o e e e e e e e e o e e6 e e es e le e g 6 e eieien go e p M 4e a e pg c N es ee e o o es e a w se g rt og a e e eog seePo g.#14 y) e e# N e as te U e == I ed s8e er se I se e
=e GG e i O O ei se Pm el in) d 4 ** em 8 Me e e es a e se I a e) e el s e es e) 0 F se t > e se se N 4 e t as se 4 e e e t 8 e e e 3 0 0 g *= g a e C en 4.3 e & A A 9 9 e G
==
ea 9 .e as se e e w es C e e me to, 4 e
- e b 3 en e e e e b :. es e e e e. as t e i
t
> en k e 9 3 3e e 3e e e er e me se e e O *" # # 4 44 e b es e e == a e es en e e e's e == e. e si es
- to e 81 as t 86 e e D em e b er e ee si g => ep .g eG WI e re e e to 1* e a es W eOMN OON en A en 66 e9MN O4N er E 48 e 3 ye ma
- W } b en b == eb ub b ** e e 3 e te e h er h as e 4. to t 4.
- Pe 3 C de k e 4 e e e as = w C 4 se en == w C 3 3 en D e es en gp g to a se en en te g A 3 en es v e
O O O
( GANL-WSC-stSO ETO 3
COMPONENTS OF CORE DEBRIS WILL LEAVE REACTOR VESSEL AS THEY BECOME MOLTEN CONSTITUENT / MELTING TEMPERATURE EUTECTIC F f SS/B/Zr 2100 SS/Zr 2400 33 2560 Zr/B l 1
3200 '
Zr(o)/UO2 No.1 3366 Zr(o)/UO 2 No.2 4362 ZrO2 4900 UO2 6060
(
- )
PEACII BOTTOM WITIIOUT DEPRESSURIZATION SHORT TERM STATION BLACKOUT BEST ESTIMATE EUTECTIC MELTS FEBRUARY 9,1988 60000 -
450.07 A -
~
r.n -
400.07 s
p 48000 -
350.07 e
.O 42000 -
Ez -
300.07
[4 ^Z 36000 -
r] -
250.0G
> N230000, g
O CD -
200.0b
( d 24000 -
~
~
18000 -
w '
E4 Q 2000 _
100.07 A -
h s
6000 - -
50.07 0 ' ' ' *' ' '
200 280 360 440 520 800 880 760 840 920 1000 TIME (MINS) 9 9 9
ORNL-0WG 87-4649 ETO O ( 3 18.5 m 2 EX-PEDESTAL AREA RESULTS IN METALLIC DEBRIE BEING "SLUSHY" UNTIL ABOUT 566 min METALLIC LAYER BEGINS TO SUPERHEAT DUE TO ACCELERATING Zr/(CO2 , H2O) REACTIONS N -
1850 f2810 '
1800 e g
~
M -
1750E 1 -
5
~
-1700 0 * .J I ) - 1660$
a i 5
2450 -
1600 y l v' lE J N ~
1560Y N
m $
w 1500h I 18.5 m2 37 m2 52180 ' 4 EX--PEDESTAL AREA EX-PEDESTAL AREA .
1450 2090 -
[ - 1400 300 400 450 500 550 6 650 700 750 TIME (min)
O
( .
oml)
l ORN'.-0WG 87-4651 ETO 3
O CHEMICAL ENERGY PRODUCTION DRIVES DEBRIS THERMAL BEHAVIOR IMMEDIATELY AFTER 566 min, NOT DEBRIS HEAT LOSS t 1
(X 108 )
(X 106 )
g 89.6
$ CHEMICAL ENERGY $
80.6 -
SUR F HEAT LOSS - 23.9 71.6 -
9 z Z
$ 62.6 -
$ _ _18.5 m2 EX-PEDESTAL AREA 37 m2 , . 17,9 {
z EX-PE DESTAL--. E
@ 53.6 - AREA S 14.9 $
$ 44.6 -
b ,
5 '
5 z ,. -
11.9 2 j 35.6 -
< :h <
' ~
26.6 -
l i 3 5 li i ,7 5 N 17.6 5.9 d j ' b' !O t
!<tli h:T.!lQil;Tmd ll , ,
, l 8.6 -
4q -
2.9 d f W
- 0.4 __ _ . > ' , , , ,
300 350 }
400 450 500 550 600 650 700 750 l TIME (min) {
)
1 l
gwa . e aa-au-. . --
6m .n,ap .m-- m &m.- -* e - 1.sa n - .m a4a .
-! M 4
4 ,.
6 > 8
'ag f ..
s Y[
p 1 4 -
$c. ~ ,i ? I f Y M m
'1' N g : {
n
's,, d % '. ,,,,,
'% gy ,w i ,
}
.g L .n . ,. .,;
l4 3' lgy:.4 !*; " . pe- .;;
3, .h '
W... A l .' l v :q ,:, p
' ng yy' %, *p e'. , y%W
=
y-b' $_
sn tut,#
M Mi .
[*e. . -
l !
I N. If.f.'lk' 4
i g1 ; s. me A. j[
l [ h.h. h
- ' 6 % k M,22W 4
',~ ,.y %# isa b
. ~
4 y n'i'LN :
~
Qt'
- % $ ^y N ,
Ra!$ge:
? *K ~ s'" :.'i en Jf, osae .,
%nr.,ca os -
ka ot m :
-:M. .i x x.
- 5 . A 'M t^
' BG M40 AR* p p y %q.n; MC o,
p4 ,~ e- - - 'c.
33 t. W ys,jty
'"y e ,;' gs e f - t?ifi p/ C O . ;*,
- 4. .,
Zi y : . -M a',' *4.
3
$ h N h*
~
go, 2 Mg YI R !!
/%j? W f a t y s ed,"op, lff l s %g:j
?
Q p sp , g 3- gx p3 [l 7 f 4(g ' ** =w "")QQ n+pg
, jQQjk 1 y; 3 yg g;
[ge3r;[9
- 7
. pg ,
g;;wy,y7;; _gp 4
,gd h_ g yk 3
~
,_ . a*
h
- n. h ;;
.e n n ,. *. ~ .
4 ..
w .
g.
it ; ,
w p p _
- b. Q ht' Y g ,
$h !.;
mg ge a v .
- g. ,;
.p
- ^Y 7 s .
f 'l m ggpm wa>p . , +ew 7
t can, m wpx .
e ,+ (' n; M y $g*}%. j, 'b * '
_g , , t,L : ;
.e -
,k t gi N '
gghdhdCM"Q$$ g%jk
+
fg w$T d _%3 WSW,4$2h4#g ~ ~ "4.MB#(W .e. di, i_
i l
O 1 BWRSAR PREDICTS FORMATION OF IHREE LAYERS OF DEsats IN THE BWR REACTOR VESSEL BOTTOM HEAD 0 BOTTOM LAYER COMPRISED'0F DEBRr'S ORIGINALLY ACCUMULATED ON CORE PLATE CONTROL BLADES CHANNEL BOXES .
O MIDDLE LAYER FORMED BY COLLAPSE OF RADIAL COLUMN (S)
($) -
FUEL PELLETS CLAD AND CHANNEL BOXES CONTROL BLADES j l
0 TOP LAYER FORMED WHEN CONTROL ROD G'UIDE IUBES FAIL FUEL PELLETS
. CLAD AND CHANNEL BOXES CONTROL ROD GurDE TUBES AND CONTROL BLADES O
t
- , _ . -- _ y _ _ .
1 HEIGHT OF DEBRIS (POROSITY 0.40) IN LOWER PLENUM DEPENDS ON STATUS OF CONTROL R0D GUIDE TUBES.
I
!Tnllkl 1 I
!all8ls t I 200 l2000hhi;IkOh d)h55OhE0hdh- .
P NE E @@ E 4IiIiIIiiiiiIiIi Ii{ liitil i i Il 11i A A A A i i i
- ij ,
d i i ' tF ." h .t" 7 li:
<u,..;
control Roo Gutoe Tuses CONTROL Roo Gutos Tuses INTACT INCLuoED IN DEBR[S O
I ^
's. ^^ ^^ ^ ~
O THE DEBRIS BED FORMED IN THE BUTTOM HEAD WOULD CONSIST OF: ,
CORE CONSTITUENTS TOP GUIDE (NORMALLY MELTS) 15200 LBMS CORE PLATE 20500 LBMS STRUCTURES IN VESSEL LOWER HEAD 195500 LBMS
\
i O CORE MASS INVENTORY (BFNP WITH CONTROL RODS FULLY SCRAMMED)
ZlRCALLOY CLADDING 81550 LBMS CANISTER 56250 LBMS FUEL 380300 LBMS STAINLESS STEEL 36000 LBMS l
BqC 2500 LBMS O
i I
1 1
CANI.-owG 37 4646 ETD
[ )
- BWR LOWER HEAD DEBRIS BED MODELS
- DEBRIS 8ED OLSCRET1 ZED INTO RADIAL AND I P AXIAL NCOES - !
Imu e LOWER HEAD MODELED AT EACH OEBRIS
'e-- -SHROUD
)
NCOE IN CONTACT WITH WALL )
e HEATGENERATION ASSOCIATEDWITH ;
PRESENCE OF FUEL - l e ALL HEAT THANSFER MCCES CONSIDERED I
, )
GiM e MASS sat.ANCES TRACM SPECIES AS THEY IN I MELT, MIGRATE, REPREE25. EGRESS FRCM VESSEL Q* ,) ',\
l 4 CHEMN:AL REACTION OF STEAM C JET PUMP g* y' AND OEBRIS OlPFUSER.--e> l '
l M i v l H w W
, 1 l
m 1
---_ -Q p p _ cy e l e I BAFFLE l e e /e l LAYER 3 l 1 1 I
/ !
H~m- j LAYER 2 l e l*Ie T ~ $ ~/
I i I li */
b-9--+- ,
N LAYER 1l e I*
4trucac5 L Wi INCOR E INSTRUMENT HOUSINGS i l I hh A < ' l
< < <, < \ i 1
d si < d
<, <, L i ,, -, ,,
g> Q
n .o h(**
- ^
rj a - * .' I'e F- ' ' ' Y- O j. ' . ! M f - - . - , . . .- c . ^'Q h
- 's ., . '.':y '*.y.[
'X}2
.. \
f '.... : .L ' . '. **
Wc-.- I.'cl:Pht . . . C+i. ',y.:
')=..' 3 - < a - - '; .. . - . : : .: ^ .:'
. sv.-; .
Q s .. n _ . .m ,s 'e r . ' . . . - ; <-k' .-' -~ - -
'. **;..,.\ - * ~ ~ ~
.;s. ..A- l, _ . - . ' *
- e. s s . ~,? ? . .I '
1*
_O p .;.2 1- ^ (' : W. .: . . - ~ 'Y 'l</ ' ' g] h:.l - - Q , . --; ..
j ;;,.:: -}.Q. 3 .8.6 -
. . f. f - ,* .- p.y .: d_.
~
.V :.;,. .. ..
. .c. e .. ] ,4 .
N.
- T V: :' g ; '.V~i.'-i-Q.'*
i L ,,. . ... . f-y;, -- f. ... ...:.,=n . p' . :s;,.g: .K.L?Qgh.&. .
._ :: . ..Q q ~- .f,; ih&.m ~;-e9.; .k .:.. . . : :-f . . . - ,;.n. - . ...- .; y :..-;y.,
?.. ?: - .3.5 y- ~;-,s;f!*C; y.:.
- .*
- .; y. . =~?..
- c p- *
,_V-m -1 ~ v. . - e:: e ; .
3..(+f. s . . 4.. n. . a, w.-u -- p;, M. s. * :a.;.f w' g. --
j Q: .. ~.. :-.r *;..x. ,..s.- ... ._ p. . Q .s. .r q:-.u.: . 4.;:;:&. ...:_-;. .;. . , .,}.l - xu.. .,_.y. :s.v. a. .a.p..w.. ,,,,,q ,
m.. %; ;.. .p.,.:y
.. m . .: p. . .;..-. : a.
u.. ... .. .'.
w
. .e, . 3;; w: ,. ., ,a.. . m. , . , ., .a. <c.- e.,..
ar
. . . . .. .. . . . s~ :. ,, ,
. ,.n. . . y.' .. m. < ..>:..-. . .: : . ,:,.:.. ,... ... p. -
.. . . . .n -
r y .. .c., .: .n. . .,
. Qg.3 .g.... .. pm. a. a_c _ , . ,
,. . . ... m:
- : . . np .a ...
, . - :.. ... . .. .u~~ .g-g:. ,
.o .
s%.k$~b: W
..p, :, 2.t -;:9,4.n.. .
.? .. .. .
- e. ., s. x . ' .O,. ,v..,,<y . . ts: . . - .~
y .29 'e.,;;; .. .. . 3. 1.
. : m :.7 *..
- _' - . .
- .. .*1., : .
=r--
. 7 '7...; c. z's'k. . .. . ....,. ....? T *'NT: :: :* Q. . / .e.. ,.$ ....
3 ',y ~
.,d, . ' %. .,3.
. . . . . . . .,. . ;/V.li.4. . :,. ' . ;'. ley' +', ::J?.:. s ;; . . :.(': g,,[ j ,$: 's 3 .
f :y .,.l*: ,.-
- a. ~;
.. 3". g: ~
. n.:,4[. :. . .%l. ; ..*p .'? . . . . . . : l .:, .
> ;;i i . . . % .A _f,:).~$. v ; . ;,.s - VP. l ie: .*. [.: ;Y.$ ?..... ; ' . g. .y -'
.e *g g*# i 4 J^
(,;>f
'e . f.k,,:':'A .. . .
~r.; . ,5,' * ,. A .:"
- .j ' ' f *2 -(. f - }*3 {hg 1.
. , * ';.'. ,[..:** ?';..u; .;., l[A f' $..,,3. -[. y-%-9,$ .
' * - Rt . ,0 ' hc;.L,. g . _qs
.t . .
. + 3,,g . * . , d $ : C ;_e g?:4 . .-
i .: T. -
- 4. g. .?
.'4 ,
.Y. . . .* ' ?. .%'.. ii.'..*; '.N \$ ' ' ~ N;? 6 W"D..*1
- a .7% . f t w..
" l-t[.} '.
- .:5.%,,s\),l.'
w i .y: %... .'*n .$: ::t. ' <
- 7. ;:.u .y s ,. f ;.~ NpfS,W*&w@.y*'**b.'i1lW""? .
f.ei zs y; Y.4 . s. . f.. . . . . , e ;fi. wAw .v,,o
.,... -.1 -. . . .- <<...,/,:4.
' . ._... ... ~.Vg M ,an..
4g ?p f ,
- Y .4.% q . ;. p w . p> [* . ,- -
Q. ':* g .*
- a n. r - pjy:.iy.3.
'*g, e.... .f , '.. -.
- .: y
- si~.6 . "3. 4,.,7 3
. . :S ;A,Y. g :@ W:s: f. .,;
2.1l %,..wgs.4-i
.x q;p .eyo.s. p :
- . : ; w. c. .,y
.. ~ ... .* y. ,w s ..w9, sn.-
. < . . ..g y .
a %. ,J' r. u. .~g. c..-- .
f g . . ,
9c.:; .[ WlQ,. ; e. t..i. .y ; - w. ,1Ml.S.**gg&,% Z w 2' iJ 9 .
? . L. r- .w?
- gw A ' ,. 1
- b. . , . >.c. , . , .p
' .' f g$$ '
m,.,3 u. 9.y. . . . . ..: . . ,: e .
%. -..?. ee . e s. ,3, ; v, w. .- ,-
m c.,,$ -
6e - + *.g 4.. >
x;9;. :
, . i- . .
w . . :. :.. .2 .
y . L . .,,.s, : ~'-t i .~s.:
,:,.. ,r 0 - ; d >n y-
.: ?.n y.. s..*i- + : g. ..q. s. . ' . . ,o,b ,e;;ra *x ;W.q.l .,,.
- .y
- C
. J,, q gg p, .
',n.
%a , . .,,-g
. $,b,. *. ? M *k*7
~
, k'.*.
- IM A,. M:'h*
- 8. .* r. [ ;. ? . .* ir t- p( [U I -
. wz. M.,g. Ar
.c
?. .c~.%.J W:j.f'W: M, . .3 4;..v.~;+ , N,.
cAJT a 4.,y..g .w v-VcW&m:+.9._ e.
- .,; w;; -. , .
. .f
- g,,, ,zC . ~ ~ :.q v.c q r.. q y y . W;,.. .s,..
w b .
- lh. ..
V. N W
[h .ff?N;.y29 JN O N. 6 . M-fMg& ". : h ihhktfC,
- .pW p
-' .&., &c_ .%*. &...
- ~ ? .k i .y? T', -
3.g :e.4 f4 j.Q p' e ._%
.-.m 7 6.$%w:*
.. e a.
- a . 4 -
M.;,.
M::w7
- 5
.f.
,.m.
z- ..- 9 AJC.i.
. . .m W 4,;
- E , :4 D.3,g.
- . . ;g g g;. e .
- yl qe 3 d ;f.
1,g . - ' - . ' .a - & -. y;.g
". ,). J
. ,;4.;p ., . mW =
- , y *3 - m
- .- * , . _M .;a.g,g, ., Q *. 3v;q& -
1;;,,;, p,u,;g f.q ,.; J:v&.: =
.), ~4:'~
r;. w?:%.v . : . : . s .:.,;,4 ~ ..
- v. . ,a. : . ..
..s.f <**:g: > ~%w us w ,
.'.g'.M.i 4;:: '; . yby n ; i}.m_ J c.;
1
.t ', . p$* Y. 5 ;;'t. . A. j.g,$.~;.. . '
~~ ~ p : ,
.~. 9 .w-.r 'f. -
- g. : n. n;.4 ~
~
.?; , ,, >:_~ -
I~' Ib '
,[.hk g $*d. ,,bgr w:.k af n.s.w.;f
- u. : m .a. . - ,y . +
5'Y'. 3 l -
.yh? .'l f -
~:W,'*:[A;fL::g-f
.o t .Q W *i.?..MY
.- .. ] <. ,
- l ?
w.p. . .
n,.
.? W ; f %.:n ; .: ? :
. m. . -s : e -~~. w. 4 : .
i 'e A; a ., ;W~
t ._.4 c, 4 ,- ;.'
l1...j 7 ..
.; %ih
- 4.'.f MV;'*l** g '
- Q
.ffy ; . g. 'e - l : } . 'i,,. '
. i.
b .
j? - - . p -. y . * .-;y;.'**t{ i
,.;Q., ..g p.rgam, ^y r~.p g Q._-, ve.:-; g s
. 3 .. ... -
z ,,,,,l.,.. :
- -' .j
.*. ~e**:K f...W- j)~f . ,
- ' ,; Q. ? '.f ~._;g Y'R
-~ " .: ' % -
., f.~ : ' ;; ; . . .L
_ 5._ ' - -
p
~
. . .l-
. - ; *p ./
- '. . : i..
- l. ; . _. __..~.
- q. - q+. y.c.. .
. ..; g g . ' ' - f. f . ,% .. _
9- + q.y,,,
., o.; .. - ....._..'....-..6.~; . m, -.y. -[.: Q f m'.,~: .
. ... .. ..%..- . .-m. - A 4.<::.
_.. y
, ;. . . L; f, :;., 4 .y
. -. 1
'*3- <..,,~,,,*g...;..... - q a , ; .; 7 ; ..
. -l..
. . _m. g
' ' . . * *7,+._ l,....? .y ~- 2 &,
^^ '
- s. ..
- s. f ;.. .
, . _ . : ,.;; &),- .'
x .b, ? -
a:AJm
. k ; r;;;1- ...? . ; .j,_c. , .:p[. ( . y:3; .W.f, g._ :"dy .. ;:-Q*
-eeng,,.y r :.
, :: 7 7-,
- -p-g..
,. , g. . ~ . - .....,2, y g,, o. .
. ...g
, .g .. .
Q - [ *._ ;
- ~ ' -
.y .
., .f ,.! . ; $ m ' ~~ - .n ;f f..f.;i ; .~,..g:f.Q;q,,,,,,,
- ::. G,.r;
_ &, ..-_ . s . . .p .
-4 %u, . . -r. .
.Q',,_..,,.m . .. -
.. . _ ' . i .x . p. .. : E,e
. _ , . , . 5 .,
.gs:,. '. ; ..~ . $ ;.
- o..* _
.:? ,-- .. y: > : . .* ra.
. ;: f.',,
.9.- .
..? ..3-*L*'.
.. r,.
. t;j _ ; so .... . ., ' Q..G.;,.q Keq
. - n . - .- -
4: 4, , .g..% <-- -e., gT , y q.a - w, E ,,
PEACH BOTTOM O SHORT TERM STATION BLACK 0UT WITHOUT DEPRESSURIZATION FEBRUARY 9,1988 1300
~
0*7 Core Place Bottom Head Dryout Dryout 1100 -
I M
\\
~
. Bottom Head . 8,9 Penetration Fails 900 -
/
W
/S team Generation By Debris Relocation 6
h -
into Bottom Head
$I@j 700 -
5.1 ^
[b a d 4.2 0
500 -
3.3 2.4 300 -
L.S t
100 O 30 60 90 120 150 180 210 240 270 TIME (MINS)
O
l
($)
PEACH BOTTOM SHORT'IERM STATION BLACKOUT l BEST ESTIMATE EuTECTIC POURS FEBRUARY 9, 1988 EVENT IIME (MIN)
BOTTOM HEAD DRYOUT 178 8 CONTROL ROD GUIDE TUBE STRUCTURE FAILS; REMAINING STANDING PORTION OF CORE FALLS INTO LOWER PLENUM, MAKING UP THE THIRD 192 2 LAYERJ GUIDE TUBE MASS ADDED TO DEBRIS IN ALL LAYERS Os .
PENETRATION FAILS IN LAYER 2J REACTOR 220 4 VESSEL BLOWDOWN THROUGH 0 1091 FT2 gotg 126,300 LBS DEBRIS HAS EXITED VESSEL, 230 0 97 5% METALS PENETRATION FAILS IN LAYER 1 746 6 CALCULATION TERMINATED; 624,200 LBS .1000 0 EXITED, 208,800 STILL IN VESSEL O
...--.e
e e
e e n e e O e e I e 2 4 4 ** e @ f1 O O M N M l d e e a 6 "5 e *
- e e *
- e *
- e *
- 4 e e e o e == 4
- N @ E, 8m .3 C, e e o en a e en e e e e 3m( Es em a ** e
- a e e e O se d ei *= wg e f% e t* e e == N es se au 9 e e t e 0 e
- t. e et e.
e s e e a k e se e b b e c e m as e e e eW e
== e en e, e e e e e e me e
- e me I f o N O @ e d O O e a se m I P e C e eeS e e e o e e e e e e e e e e e e
- e t
- P m @ m e ense e d p g. m p g en e S e 6 as & e # P* @ O 8* 4 P 4 m Cs e4 e 2 e
- 9 O se fm se m . O m ng et N ge e e e e e e I e m to E e Nw a9 0 e w e e O POeN e E O S e e e 8 N m6 9 e e e 4 e e
- e e es e e E e e e e e e e 8 SeOOSeOGeeOee1 O e e p 1 e e e e e e e e e e e e e 0 e e e 4 == S e O O O e e e u e e e O 6 S e N e 6 3 0 e 5 e e F 0 e e e e* 6 e -
== e e av e e se e me 8 e P e e e e o se e es e C es e e e e w e en se e T h e & a e as 3 e an e
% e e 4 i N sese & O se # O Pu>= d a d 4 M e 1 e es e O e o e o e e e o e o e e e I e e n w e e e 4 m m e *= O e m C P= O N 4 e e
% to e me f 3 e, N e se e p (g c8 4 e to Q e 4 0 m as e 4 ee 81 eO N e M
- 4 eaOMe e se i e e
=8 b se e N se me g 4 e e e se .3 e se I p1 e e e er e W e D e 4 h e 4 7 ** e C 0 e e
t C e e 3 e e e e d =e e M89 e e e N e e 4 e 3 0 t D e e e e e e e e e e e e e e e e D 8
- 4 e s e e o p u o e e m pu e e se e e ae+ e 3 ** e a S p= 6 m e en ap I O P= d o s= e) 0 N e 3 & 8 e8 N as ao se ' 8 G e b to 8 me a$ e as == a 81 N e 4 e N e G P m O e e
.w e 4 es e se ** G e h b en e C en b e e en en b e e e 4 e e S 4 4 t e e ao e se e C N me 8 e e en e
- 0 9 e o e e e e e e e o e e e 9 e e e as e
e e s= m == P= e e m ** n s.a e se e e o e e, 4 e. en el b= 5 4 ee & wt 4 *i me se $ Pi e e 3 e aNNPP O 1* N e se A eu e C e e ee0 4 *) em 9 3 e g e e O C ** e 81 0 e e e, .4 e e 4 8 e e b la e e
- == e e e e a w b o e w e e ** e Q ** e D e g
- e e e b e 6 0 PSO G r5 O d O G O O 0 0 0 09 e e d a 3 e e e e o e e e e o e e e 4 e e g e 0
- O O O 8* 8* O O O O G O f5 0 0 O 4 g C e D I 3 I e
.e e e 0 8 & 8 e en 8 e en 9* e e c, e == 0e 4 e e e o =* 0 9 e e .)
- 4 g e eu e e, C e .a C se e es (*
- e o e s gp e 8 e e b es e 3 O e e b e e
== N e en k e (e fe e 3 e4 fe 4 O @ m f to se se p e 4. . 4 Pg e e e me 0 O e o e e e e e o e e o e e 0 e e 64 4 WW e
e
- C l8 **=8 f WI he N O $ C 84 C 7* P1 *e SW I W e J N ** 4 ( e t e te e se N fe I d e me aO e $ @ at d e 8* Of ep 4 8e e e = es e a e e 4 e ** O se se ee e se es f1 e se e N -e e ti e e e e 0 9 e
= ** @ e e e 4H e se e
.O . e en e e == w e e 6 3 *= e e e
, :. e o e e g as b e ) e J e e e .e e 1* e ne =
w
- e e == se e g ae e e- w ee es e e O me e h w he S 9 m 4 e b 9 **
- re me W O 4 *4 4 00.4 as e 3 b 4 e W 4 6 le b* eh en ip h* .,c G e
$ & O e D G me
- W 3 te A *e
- go te C 4 3 es e e
O O O !
The inttial debris pouring rate is accident-sequence dependent.
80 9 i o
Y
- 50 -
CA U) ~~~~...______
CO 40 - ,,,, ~~~~~... - -
m o *', '.
- f. '
E4 30 -
& ,a-g .-
5 ,-
M l to 20 -
l A
h30 _j SHORT TERM STATION BLACKOUT h ,! _ _ _ _ _ _ _ _ _ _ _ _ LONG TERM STATION BLACKOUT 19 !
is l ' ' ' ' ' ' ' ' ' '
0 0 50 100 150 200 260 000 360 400 450 500 TIME AFTER PENETRATION FAILURE (MIN)
. -- _ - - - - - - _ --- - _