ML20148P421
| ML20148P421 | |
| Person / Time | |
|---|---|
| Issue date: | 03/30/1988 |
| From: | Advisory Committee on Reactor Safeguards |
| To: | |
| References | |
| ACRS-T-1656, NUDOCS 8804110203 | |
| Download: ML20148P421 (262) | |
Text
- . _ _ _ _ _ . . _
C6B /d54 1 l
CR 3 YAL o gaino arms NUCLEAR REGLI.ATORY COMMISSION y
In the atter of:
STFCCIUPX CIGI?EERPM O
"'7es:
1 thrru7h 162 Place: Culver City, Califomia Date: 'brch 30,1988
..........................................................c.
HERITAGE REPORTING CORPORATION 4
r O aman 1224 L Street, N.W., Seite 600 WanMattos D.C. 20005 8804110203 080330 (202) 028-4888 PDR ACRS T-16S6 PDR l
PUBLIC NOTICE BY THE ggg 1 2 UNITED STATES NUCLEAR REGULATORY COMMISSION'S 3 ADVISORY COMMITTEE ON REACTOR SAFEGUARDS 4
5 WEDNESDAY, MARCH 30, 1988 6
7 The contents of this stenographic transcript 8 of the proceedings of the United States Nuclear Regulatory 9 Commission's Advisory Committee on Reactor Safeguards 10 (ACRS), as reported herein, is an uncorrected record of 11 the discussions reported at the meeting held on the above 12 date.
13 No member of the ACRS Staff and no participant h 14 at this meeting accepts any responsibility for errors ,
15 or inaccuracies of statement or data contained in this 16 transcript.
17 18 19 20 21 22 23 24 25
_ _ _ _ _ _ _ _ _ _ _ _ _ .J
gggg 1 UNITED STATES NUCLEAR REGULATORY COMMISSION 2 ADVISORY COMMITTEE ON REACTOR SAFEGUARDS 3
4 In the Matter: )
)
5 )
)
6 STRUCTURAL ENGINEERING )
)
7 )
8 Wednesday March 30, 1988 9
Malibu Room 10 Pacifica Hotel 6161 Centinela Blvd.
11 Culver City, California 12 The above-entitled matter came on for hearing, 13 pursuant to notice, at 8:30 a.m.
kh 14 BLFORE: DR. CHESTER P. SIESS, CHAIRMAN Professor Emeritus, Civil Engineering 15 University of Illinois Urbana, Illinois 16 ACRS MEMBERS PRESENT:
17 DR. PAUL G. SHEWMON 18 Professor, Metallurgical Engineering Department Ohio State University 19 Columbus, Ohio l 20 DR. DAVID A. WARD Research Manager on Special Assignment 21 E.I. du Pont de Nemours & Company
! Savannah River Laboratory 22 Aiken, South Carolina 23 24 25
ACRS COGNIZANT STAFF MEMBER:
gggg 1 2 Elpidio Egne 3
NRC STAFF PRESENTERS: Page No.
4 Dan Guzy, NRC Research 3 5
6 PRESENTATIONS BY:
7 Sam W. Tagart Jr., P.E. 27 Technical Specialist 8 Nuclear Systems and Materials Department Electric Power Research Institute 9 3412 Hillview Avenue Palo Alto, California 10 William English 62 11 General Electric Company Nuclear Energy Business Operations 12 Structural Analysis Services 175 Curtner Avenue 13 San Jose, California lllh 14 Sampath Ranganath, Ph.D. 111 General Electric Company i
15 Nuclear Energy Business Operations l
Manager, Structural Analysis Services 16 175 Curtner Avenue San Jose, California 17 NRC CONSULTANTS:
19 l
Spencer Bush 20 Everett Rodabaugh 21 l 22 23 24 25
1.
1 March 30, 1988 gglg 2 8:30 a.m.
3 4 --
PROCEEDINGS - -
5 6 CHAIRMAN SIESS: Good morning. The meeting 7 will come to order.
8 This is a meeting of the ACRS Subcommittee on 9 Structural Engineering, and present today, starting on 10 my right is Paul Shewmon, Dave Ward, and we have two consultants:
11 Mr. Rodabaugh and Mr. Bush.
12 Today we will review and discuss the EPRI NSRC 13 Piping and Fitting Dynamic Reliability Program PFDRP--
lh 14 unpronounceable.
15 The cognizant ACRS staff member for the meeting 16 is Elpidio Egne, who is seated on my left.
17 The rules for participation by the public at 18 today's meeting was announced as part of the notice published 19 in the Federal Registry on March 14. It says here that 1 20 the meeting is being conducted in accordance with provisions 21 of the Federal Advisory Committee Act for the government, i '2 and the Sunshine Act, and we've received no written statements 23 from members of the public nor any request to make oral 24 statements.
25 These microphones are not working. He said 0
2c 1 he might be able to get them fixed during the break. We gg g 2 are a small enough group that I think the--if the people 3 sitting out here want to move up a little bit they can, 4 but let's just try to speak loudly enough to be heard.
5 I'm a little hard of hearing, so I may be the test for 6 the volume level, and please give your name the first 7 time that you speak so that the recorder can get it.
8 Any of the Subcommittee members or the consultants 9 have any opening remarks that they would like to make 10 at this stage?
11 (No response.)
12 Just for the record, I would like to point out 13 that we had the opportunity yesterday to visit the two lllh 14 sites at which tests are being made at Blue Tech on the 15 system's test, and at Anco on the component tests, and 16 there will be essentially no repeat of what we learned 17 at those visits. We will concentrate today on a brief 18 review, I think, of the program, but then we will concentrate 19 on the test results and the analyses and something on 20 what is being considered for changes in the ASME Code 21 on the piping.
22 We will start off with Dan Guzy from NRC Research.
23 We have both NRC and EPRI represented and Mr. Guay and 24 Mr. Tagart will lead off this morning.
25 Dan.
I i
, 3.
MR. GUZY: Let me just emphasize some of the gggg 1 2 things that Professor Chet said about the presentation.
3 We will be covering the summaries of the tests that we 4 saw yesterday of the systems and the component tests, 5 plus a little more on the specimen tests that were discussed ,
6 briefly yesterday, and Bill English of General Electric 7 will handle that.
8 Sam Ranganath of General Electric will talk 9 about our concepts now for changes to the ASME Code, but 10 the point is I will begin off with a brief talk of the structure 11 of the program, status, a little about what I consider 12 the highlights of what has happened from the programmatic 13 point of view.
llll 14 Sam Tagart will talk a little more about the 15 technical overview and perhaps give a little more of an 16 industry perspective on why they are doing this program.
17 Before I begin, I would like to point out that 18 this progran is called the EPRI/NRC Piping and Fitting 19 Dynamic Reliability Program. The reason that EPRI has 20 the top billing is--well, actually two reasons: one is 21 they are contributing more money to this; but more importantly 22 they have the lead in the planning of the program.
23 The NRC has been involved in this program from 24 the beginning--from the beginning of the testing and the 25 analysis, but the lead, in terms of structuring this program, l
I l
t
4c the credit goes to EPRI. The NRC recognizes this is a gggg 1 2 good thing to be involved with, and we got involved after 3 it had been pretty much planned out.
4 As far as the program, I would like to start 5 off by talking about what the emphasis of this program 6 is, so there is no uncertainty with what we are trying 7 to do here. The emphasis of the program is the design 8 of piping components for dynamic inertia loads. The key 9 words are design--we are talking about design rules, not 10 so much inspection rules. Piping components, we are looking 11 at the stress rules, or the rules for designing elbows 12 and tees of the piping system, not so much supports and 13 say nozzles, but that would be considered in our design 14 roles.
15 Also inertia loads: one of the chief objectives 16 was to provide a more rational set of rules for dynamic 17 inertia loads because that seemed to be an area of concern.
18 We will address other types of loads though, too, such 19 as anchor motion loads.
20 The objectives of the program have been from 21 the beginning to identify clearly what the dynamic failure 22 mechanisms and failure levels are for piping systems under 23 dynamic loads. It is important to know what the level 24 is for the large cycle failure and how do they fail so 25 that we can develop more rational rules for preventing
I 5.
g 1 the failure.
2 Also, we are interested in gathering high level 3 response information so we can know more about what happens 4 in the area regime of a failure, in terms of parameters 5 such as damping, ductility, deamplification, we are lacking 6 information in this area, and this program is providing 7 some valuable information for things not only in this 8 program, but for future programs, as far as bench marks 9 and data that we can use later on.
10 And, the key, final product of this program 11 will be a recommendation for changes to the ASME Code.
12 We are talking about changes to the design rules themselves, 13 as given in subsections MB, MC, and D, stress allowables llh 14 for Class 1, 2, and 3 piping.
15 CHAIRMAN SIESS: Dan, you use the term non-16 linear response in there. This may be partly semantics, 17 but it is the way that I think about it.
18 When I look at this, I am really looking at 19 the inelastic response. Now, I admit that inelastic is 20 non-linear but nonlinear is not necessarily inelastic.
21 MR. GUZy: Right.
22 CHAIRMAN SIESS: And, I think the inelastics, 23 that's where your large damping comes from, not from a 24 nonlinear.
25 MR. GUZY: Okay.
O
6.
1 CHAIRMAN SIESS: But, that is your thrust. It gggg 2 is the inelastic.
3 MR. GUZY: Yes.
4 We lack information in the very high levels.
5 Maybe the emphasis should be on high levels rather than 6 nonlinear, but that is where we didn't have much information.
s 7 I have got a list of the cast of characters, 8 or some of the cast of characters that are involved in 9 this program, and many you met yesterday. I would like 10 to highlight the people that you didn't meet yesterday.
11 Y.K. Tank is also from EPRI. He is program manager for 12 the tests, for the Anco and the systems test for the EPRI, 13 and of course Sea Tagart is overall program manager for A lot of credit for the development lh 14 the EPRI program.
15 of the program comes from--the credit should be given 16 to Sam.
17 I am the NRC person, Dan Guzy, and responsible 18 for the program in terms of what research programmatic 19 responsibilities are.
20 From General Electric, I have listed some of 21 the people--not all of the people, but the main program 22 manager is Bill English, who you will hear from later.
23 A person who is not here today, but has been heavily involved 24 in the analysis is Henry Hwang. Sam Ranganath, who you 25 will hear from later on, is involved in developing the
70 Code rule changes, and Ed Swain is another General Electric gggg 1 2 person who has been heavily involved in these Code changes 3 also. There are other people from GE that I have not 4 listed.
5 From Anco Engineers you have met Kelly Merz, 6 who is here today; also Paul Ibanez you haven't met, at 7 least at these meetings, and is involved in the program.
8 ETEC, you met Mr. Devita yesterday, Ron Johnson 9 and a cast of thousands, I guess, at ETEC yesterday.
10 A key part of the program, but not represented 11 by an individual today is the specimen tests. The main 12 person for that has been Roy Williams, formally of General 13 Electric of Schenectady. Now he has his own company called 14 Material Characterizations Lab, and he is the 15 one responsible for the specimen tests, and Bill English 16 will talk about those tests some today.
17 There also have been several consultants involved 18 with the program who have reviewed it and have given a 19 few suggestions for changes in the program, and these 20 included Everett Rodabaugh, who is here today; Bob Kennedy; 21 Don Landers; Bob Cloud; Doug Munson; Stan Moore from Oakridge; 22 Bob Bosnak has also served, he is from NRC and has served 23 as a consultant; and Verne Severud.
24 (Slide]
25 Okay, the program is structured into eight tasks,
80 gggg 1 and I will just briefly go through what they are. We 2 had a Task 1--and who is involved in it--Task 1 is the 3 program plan development. This has been the primary responsibility 4 of General Electric in San Jose.
5 The pipe component test, which you saw yesterday, 6 is at Anco, and they are responsible for all of tha 41 7 tests that they have run.
8 The pipe system testing has been split into a 9 number of organizations. The main seismic and hydrodynamic tests 10 have been conducted at ETEC. you saw the results of these 11 yesterday. The point I would like to make is the system 12 1 and the system 2 tests, the red and green tests that 13 you saw yesterday, have been in integral part of this
!h 14 program; however, some of the earlier tests, the demonstration 15 tests, have been part of the NRC's contribution to this 16 program, although it is not formally a part of the program.
17 There is a distinction between the tests--maybe it is 18 just a paper distinction.
19 The water hammer test that you saw yesterday, 20 of course, was being conducted at ANCO, so that all of 21 these together consist--comprise of Task 3.
22 The other tasks--okay, the specimen tests at 23 Schenectady are a separate set of tasks which we will 24 hear about today.
25 The remaining part of under GE's responsibility.
9.
gggg 1 They have done the analysis of the tests. They have taken 2 the data and ETEC and ANCO and MCL have supplied them 3 and looked at this data and summarized it, and you will 4 hear more about that today. They have also been charged 5 with developing--to identify and justify new design rule 6 changes based on these test results.
7 And, the final reports are General Electric's 8 responsibility, although I think ANCO and ETEC have reports--
9 okay, so that GE is in charge of the final reports and 10 they will be draft reports that will be supplied by GE 11 to EPRI. The final reports will be EPRI reports not General 12 Electric, however the NRC has information, all of the 13 data, and we just--the burden of publication is not on lh 14 us for this one.
15 (Slide]
16 Okay, as far as the status and schedule, the 17 program itself, in terms of doing anything other than 18 program planning began in the spring of 1985, three years 19 ago. All the testing now has been completed except for 20 the retest of System 1 which you saw ready to go at ETEC 21 yesterday, so all of the component tests have been completed, 22 the water hammer test, and the specimen test, so having 23 been completed--some of these very recently have been 24 completed, but they are all finished now.
25 The process of evaluating this data and developing
10.
Code rules ongoing you will hear about where we are today, g 1 2 but it has not been finalized yet, so we are still working 3 with the data and drawing conclusions to make recommendations.
4 The program itself will formally end in June 5 of this year. General Electric being the main contractor, 6 that is when their role is over. Other than writing reports, 7 most of the other subcontractors are completed.
8 Okay, then when the final recommendations are 9 made to the program, of course they will be reviewed by 10 EPRI and NRC--at least in NRC Research, and then will 11 begin the Code revision process through-the various organizations 12 that will be involved in the Code changes.
13 And finally, as I mentioned, the reports will llh 14 be published probably sometime this year, I imagine. That 15 is the EPRI reports.
16 (Slide) 17 Some of the key points that I would like to 18 make from perhaps more of a programmatic point of view.
19 This is a formal EPRI/NRC research program. We have a 20 formal agreement on it. There have been five review meetings 21 that have been held with the program managers and consultants.
22 The most recent one was less than a month ago. This is 23 our way of getting input, by getting everybody together, 24 getting input on direction and what the results mean, 25 and it has had an impact on--particularly in the component
11.
gggg 1 tests with--there have been some changes made, suggestions 2 as we reacted to data as it has been coming in.
3 All along there has been interactions with the 4 ASME and the PVRC standards groups, in a number of ways.
5 First of all, we have been giving presentations to everybody 6 at the meetings, and also a number of the members, the 7 people who have been involved with this program directly, 8 are also in this core group, so there is direct involvement 9 by many of the members.
10 CHAIRMAN SIESS: Excuse me, Dan.
11 Does PVRC write standards?
12 MR. GUZY: Well, they--what do they do? Write 13 recommendations?
llh 14 MR. BUSH: They write recommendations basically.
15 CHAIRMAN SIESS: And, where are they implemented?
16 In ASME?
17 MR. BUSH: Yes.
18 What we do, for example, is I would write a 19 letter and transmit it to say Roger Reedy, the chairman 20 of section 3, suggesting an implementation of a given 21 action. That is the mechanism. We are not a formal standards 22 writing body as such. PVRC is the reason they do it...[ voice 23 fades out of hearing range)...certainly do it, but it 24 ends up going directly into the code.
25 MR. GUZY: There has been some activities that i
12.
1 have had an impact on the Code, the code cases. As a--
gggg 2 using the data from the tests we have to date, a Code 3 case--a Class 1 Code case effecting BBOB allowables 4 has been approved through the Code system, essentially 5 this gives relaxation for inertial load requirements 6 on OB and overloads at B level--
7 CHAIRMAN SIESS: What does it do to the change?
8 What dominates the design? OBE versus--
9 MR. GUZY: --this would--if implemented, this 10 would make the SSE dominant, essentially of less importance 11 than OB.
12 There is a similar Code case--class--that should 13 be code case, not class--for Code case, not class, for lllh 14 class 2 and 3 piping that is up to the main committee 15 now in the ASME codes. I believe that section 3 has one 16 more committee, or two more committees.
17 CHAIRMAN SIESS: Help me a little again.
18 I don't think class 1, 2, 3. The only thing 19 I think, from what I deal with, is seismic category 1 20 or not seismic category 1.
21 Does 2 or 3 make any sense in that classification?
22 Or, is it something else?
23 MR. GUZY: They are both subsets of that classification.
24 They are all seismic category 1. Class 1 frankly rego. ires 25 a more rigorous fatigue analysis--
13.
gggg 1 CHAIRMAN SIESS: Okay, but--
2 MR. GUZY: --so you don't have that in the--
3 CHAIRMAN SIESS: --2 and 3, in this sense, is 4 category 1 piping?
5 MR. GUZY: Yes.
e 6 CHAIRMAN SIESS: It is just a different type 7 of analysis?
8 MR. GUZY: Right.
9 CHAIRMAN SIESS: Okay.
10 MR. GUZY: And, there are different rules that 11 have to be changed. The class 2 and 3 rules are pretty 12 much identical, as far as the design part. Class 1--
13 CHAIRMAN SIESS: In terms of the plant, how lllh 14 do you decide whether something is class 1, 2 or 37 15 MR. GUZY: Class 1 has to do with the pressure 16 boundary--primary system pressure boundary--
17 CHAIRMAN SIESS: Primary system pressure boundary.
18 MR. GUZY: --main loops in the surge lines and 19 the recirculation loops.
20 Class 2 and 3 are other piping than category 21 1. The distinction between 2 and 3, I think, is more 22 of an inspection--maybe that is sort of an arbitrary type 23 of thing.
24 CHAIRMAN SIESS: So there is an isolction valve 25 between class 1 and class 2 and 3?
14.
1 I
gggg 1 MR. GUZY: Right.
2 MR. BUSH: However, for clarification, the utility 3 has a considerable say in the pipe. There is one utility 4 that has a system they call class 2, and it doesn't necessarily 5 say it will be the same--[ voice fades out of hearing range]
6 CHAIRMAN SIESS: Mr. Bush, we can't hear you.
7 MR. GUZY: I think the point here is that most 8 of the piping that would be required is class 2 and 3.
9 We pay a lot of attention to class 1 and sometimes, even 10 in this program--
11 MR. BUSH: Don't say that, Dan. That's not 12 true.
13 CHAIRMAN SIESS: In a PWR--
llh 14 MR. BUSH: Most of the piping--you say safety 15 related, because don't say "safety related" because that 16 is not true.
17 MR. GUZY: Yes, yes, that is what I meant. I 18 am sorry.
19 MR. BUSH: Okay.
20 Most of the piping now is--
21 MR. GUZY: What I am trying to say is there 22 is a lot more class 2 and 3 systems than class 1 systems.
23 CHAIRMAN SIESS: All right, but in a PWR, steam 24 lines are whac?
25 MR. GUZY: Class 2.
15.
1 gggg CHAIRMAN SIESS: Class 2.
2 MR. GUZY: Well, PWR is class 2.
3 CHAIRMAN SIESS: Well, PWR is steam lines, and 4 are class 2.
5 MR. GUZY: All right.
6 CHAIRMAN SIESS: And a BWR, what is class 1N?
7 MR. GUZY: It would be the--
8 CHAIRMAN SIESS: Like a turbine stop valve?
9 No?
10 MR. GUZY: Isolation valve.
11 MR. BUSH: Inaudible.
12 COURT REPORTER: Mr. Chairman, Mr. Enge, I can't 13 hear Mr. Bush at all. I can't even see him.
lllh 14 CHAIRMAN SIESS: That's all right. It wasn't 15 important.
16 MR. GUZY: If there is any point to be made 17 here, it is we are concentrating--there is a lot of emphasis 19 on class 1 piping, and even programmatically we have GE 19 who is our main contractor; however, we plan to address 20 class 2 and 3 piping,.ind this is probably where we will 21 get the biggest relief in snubbers and snubber reduction, 22 et cetera.
23 Other than the Code cases there is also an activity 24 that just started with the PVRC, and is a task frequent 25 functionality critoria. NRC has a requirement on piping
16.
functionality which we think that the results from this ggg 1 2 program can support changes to, and that will be something 3 that PVRC will make a recommendation on, and perhaps the 4 NRC itself will take care of the standard change on that.
5 Publications, there has been a number of papers 6 that have been presented, and will be presented, in like 7 the Pressure Vessel Piping Journal and SMRT. There have 8 been four semi-annual progress reports and you all should 9 have received the last one from this program, ar.d that 10 will be the last progress report. The next set of reports 11 will be the final reports, and again will be issued by 12 EPRI.
13 CHAIRMAN SIESS: Dan, let me ask you a slightly llh 14 unrelated question, as far as this particular thing is 15 concerned: the National Research Council Report on the 16 NRC research program placed a considerable emphasis on 17 peer review, and in the response to that report the NRC 18 research seems to have said that the way to get peer review 19 is to publish in referce journals.
20 Now, I notice here that you have got papers 21 and journals--I assume that SMPT is more or less referring, 22 although I question it sometimes, whether they over threw 23 anything out, but you have your panel of consultants.
24 Which serves as peer review in your mind?
25 MR. GUZY: Personally, I think the consultants
17.
do a better function of peer review than having it p::blished; gg 1 2 however, it gets to a wider audience by having it published 3 in the journals.
4 CHAIRMAN SIESS: Do you think publishing in 5 a journal is a oeer review? It is sort of after the fact, 6 isn't it?
7 MR. GUZY: I pertonally--I think it is valuable, 8 but I think it is more valuable to have the right people 9 review it, in a more formal setting.
10 CHAIRMAN SIESS: Well, once you have said "the 11 right people" I won't ask you whether you think this is 12 a peer review setting.
13 MR. GUZY: As far as consultants, I am convinced llh 14 that these are the best people we could get to review 15 the program. I mean, nobody is not on that list that 16 should be on the list. I think it is an impressive list 17 of peopic involved--
18 CHAIRMAN SIESS: Now, whose job is it to see 19 that they are listened to? Yours and Sam's?
20 MR. GUZY: Yes, sir.
21 CHAIRMAN SIESS: Okay.
22 MR. GUZY: Let's see.
23 Maybe since we are talking about this, there 24 is planned to be three--I think three main papers that 25 will come in the near future that will try to summarize
18.
gggg 1 and explain what this program has been about, and what 2 the findings are, and that will be--and there are three 3 important papers that are planned that will be, you know, 4 supported by the reports but this is our way of introducing 5 to the world what we are really doing.
6 CHAIRMAN SIESS: Very good.
7 MR. GUZY: At the end of the program, we realize 8 that there is a lot of information here that is valuable 9 that we don't even know about yet, and we plan to do the 10 best we can of storing this information. It is all available 11 to the NRC, but we would like to make it available to 12 everybody else and not throw anything away, and so there 13 are some plans to archive those ANCO and ETEC tests at
' 14 the NE Center in Charlotte, I believe, also the information, 15 the data, we will try to do the best we can to save all 16 of that.
17 (Slide) 18 one more slide I would like to present on is 19 just talking in terms of what the NRC's perspective of 20 this program has been.
21 The main thrust of why we are in this comes 22 from the activities of the piping review committee of 23 many years ago. The piping review committee looked at 24 piping design and they identified a number of concerns 25 about overdesigning for dynamic loads, especially inertial i
l
19.
g 1 loads.
2 In their reaction to this they made a number 3 of recommendations; however, because of the state of the 4 knowledge or information data at the time, recommendations 5 were mainly in the response areas, and of these the most 6 significant, tne one that has paid out perhaps the best 7 has been the damping.
8 So, their immediate recommendations for changes 9 were addressed more to response, because they didn't have 10 data; however, the realized that they needed failure data 11 and one of--the highest, the A category priority items 12 that the piping review committee recommended for research 13 has been to do pipe tests and this program was mentioned h 14 by name as something we should be involved with, so NRC 15 research and NRC is involved in this program because our 16 piping review committee recommended it to us, and I think 17 it was a wise thing to do.
18 We've had a number of interactions with the 19 NRC staff and we will continue to have tnat. There has 20 been much information sent informally. There has been 21 presentations and video tapes to staff, to people interested 22 in what was going on. There has also been meetings, formal 23 meetings, on other subjects where results from these tests 24 have been quoted in terms of what's happening with piping?
25 How is piping going to fail? In particular there was
20.
gggg 1 a formal meeting for the staff and people from standards 2 group on damping code case N-411--
3 CHAIRMAN SIESS: When you say "staff," you are 4 staff. You mean "other staff."
5 MR. GUZY: I mean the whole staff, and not just 6 research.
7 CHAIRMAN SIESS: You mean NRR--
8 MR. GUZY: I mean NRR and the licensing people 9 and people that now inspect the projects are involved 10 in piping, too--the people outside of the Nicholson Lane 11 Building.
12 There was also a presentation on this class 13 1 code case that I mentioned before, and we presented lll 14 information from this program in support of these other 15 changes.
16 We've given presentation at our information 17 meetings at Gettysburg year to year, and then perhaps 18 the first really formal presentation of staff, solely 19 on this subject, was given last September when we gave 20 detailed briefing of the results and where we were heading 21 at the time and criteria development.
22 Today is the first meeting with the ACRS. We 23 are interested in your comments on the program, and any 24 suggestions or conments you may have on the program once 25 you 'icar us out today.
l
21.
llll 1 We also plan to have future meetings with the 2 NRC staff, particularly licensing people, and particularly 3 in terms of the criteria of changes that we probably will 4 be recommending. Many of the--all of the standard groups 5 will be involved, the ASME representatives from the NRC.
6 We like to have feedback from the staff before--get some 7 direction from the staff before we actually start becoming 8 involved as an NRC representative to the ASME, so we plan 9 to have meetings this spring with the staff to present 10 what we have developed, 11 In terms of how the regulatory cnanges go, the 12 Code case, such as the stress allowable Code cases are 13 endorsed formally through revisions of R.G. guide 1.84--
h 14 CHAIRMAN SIESS: That is the one that periodically 15 comes up?
16 MR. GUZY: Yes.
17 CHAIRMAN SIESS: And, updates the reference--
18 MR. GUZY: And, the status now of the Code case, 19 the class 1 Code case N-451 is published and will be treated 20 in the next revision, revision 26. That is R.G. guide 21 1.84, so that will go through the formal NRC endorsement 22 process for that Code case. The other Code case will 23 probably be in the next revision to R.G. 1.84.
24 (Slide]
25 However, the changes we are talking about today t
l g
e .o '1
\ \.
~ - -
22.
1 will be to the Code itself, and the way the NRC endoIsos g
2 the ASME Code changes is through 10 CFR 50.55A and we 3 essentially incorporate specific addenda and revisions 4 to the Code as they come about.
5 CHAIRMAN SIESS: knd, that again, you do periodically?
6 MR. GUZY: Do that periodically, and it will 7 be done through the regulations, so that is the formal 8 process for endorsing the changes we will be hearing about 9 today.
10 Also, there are changes that we may make to 11 the standard review plan, particularly in its functionality 12 area, also I think the information from this program will 13 have an impact on many other things we do in the piping lllh 14 area; perhaps not in a--as a more explicit way for providing 15 backgrounds, supports a lot of conclusions people have 16 made, for instance in seismic margin studies, or PRA's, 17 seismic inertia loads are generally not considered important 18 if we don't review them.
19 In contrast to the way that we may have reviewed 20 plants in the past--
21 CHAIRMAN SIESS: Wait a minute.
22 You said in seismic margins they are not considered 23 important?
24 MR. GUZY: --not considered important. Piping 25 inertial loads are generally not even--in piping systems--
l
23.
are not considered important because of the experience, (ggg 1 2 the piping experience--
3 CHAIRMAN SIESS: Oh, okay, simply based on piping 4 experience and--
5 MR. GUZY: --because of the piping experience, 6 plus findings from this kind of a program. This supports 7 piping experience data--
8 CHAIRMAN SIESS: Oh, okay, yes.
9 MR. GUZY: --showing that your margins are much 10 greater than other things in the plant that will--
11 CHAIRMAN CIESS: Yes, okay, but now you just 12 mentioned fragility in passing, and that is a big area 13 of business these days, making PRAs, and certainly there llh 14 somewhere they have got to put a fragility in there, don't 15 that?
16 MR. GUZY: They--
17 CHAIRMAN SIESS: Or, do they just--
18 MR. GUZY: --they do in the piping area, but 19 they only do in at very high level earthquakes. Sometimes 20 you can dismiss that without developing it in the piping.
21 CHAIRMAN SIESS: Okay.
22 MR. GUZY: Seismic margin areas, I was going 23 to contrast to--in the old SEP program you had to reevaluate 24 everything, a lot of the effort was involved with piping 25 analysis. In the seismic margin approach now has taken
24.
1 from the SEp program, although they recognize that all gggg 2 of this piping reanalysis is not necessary. They are 3 concentrating on things of importance such as the seismic 4 ankle motions or systems interactions, and not worry so 5 much about seismic inertial loads.
6 This program supports other data and experience 7 showing that we can, in the global formal safety scenario 8 we can downplay seismic inertial loads.
9 CHAIRMAN SIESS: That explains why--
10 MR. WARD: Well, it is--
11 CHAIRMAN SIESS: --couldn't find any failures.
12 MR. GUZY: That's right.
13 MR. WARD: --but a lot of that has been taken lllll 14 credit for already though--
15 CHAIRMAN SIESS: Yes, that is what he is saying.
16 MR. GUZY: We are supporting--
17 MR. WARD: --and this is confirmatory, actually.
18 [ Slide) 19 MR. GUZY: Okay, ! have one other slide in my 20 package on piping resource that I would like to hold on 21 for later.
22 CHAIRMAN SIESS: Okay.
23 MR. GUZY: I would like to, at this point, take 24 any questions.
25 Yes.
250 MR. SHEWMON: I am very pleased, as a metallurgist gggg 1 2 who makes his living sort of on the fact that metals are 3 different f roro glass , seeing these things designed to 4 take credit for some of the plasticity that is inherent l l
5 in the metal that we've paid for and built it out of, l l
6 but, as you do this I am some concerned about the fact 7 that there is no consideration of castings and the fact 8 that they might have different plastic properties than 9 the wrought material and the test consisted only of wrought 10 material.
11 Is there a basis on this that it was just so 12 much more convenient to work with wrought material? Or 13 the Code says that castings always must have appreciably khlBI 14 lower stresses? Or that the Code has been able to ignore 15 it because they do elastically and the plastic properties 16 don't enter, or what?
17 MR. GUZY: I think--maybe somebody else would 18 like to speak to this, but--
19 MR. SHEWMON: I ask you, but you can pass it i 20 off. That is your advantage to--
l 21 MR. GUZY: --it is my understanding--okay, but l
22 my understanding is that, you know, we are looking for l
23 the majority of the piping in the plant, and the majority 24 of the piping, to my understanding, does not use cast 25 fittings.
i 26.
1 MR. SHEWMON: Well, but it is not just the piping.
gg 2 It is the components in the elbows--
3 MR. GUZY: That is what I was talking about.
4 MR. SHEWMON: --and the valve bodies and the 5 pu np housings, these things often are, and the fact that 6 if you design your plants with the strongest elements, 7 it is the weak link that is going to rise up and bite 8 you, so saying the ma]ority of it is piping, probably
, 9 isn't the right way to look out for failures.
10 MR. GUZY: Maybe somebody else can address that, 11 but it is my understanding that the majority of the fittings, 12 elbows, and tees, et cetera, were wrought and not cast.
13 MR. SHEWMON: There is a lot of cast that comes out there. I don't know whether the majority of it is lh 14 15 or not, but if you change the Code--
16 MR. GUZY: The Code will not address that at 17 this time, because we don't have the data right now to 18 do--
19 CHAIRMAN SIESS: The Code will be limited to 20 wrought material?
21 MR. GUZY: Wrought material, yes.
22 MR. SHEWMON: So the Code will distinguish between 23 wrought and cast--
24 MR. GUZY: Yes.
25 MR. SHEWMON: --in this case?
27.
MR. GUZY: That's right.
gggg) 1 2 MR. BUSH: Are you sure of that? I don't think 3 that. I am not quite sure of that.
4 MR. GUZY: Well, maybe later on we will talk 5 about this, but our data, since we have not tested, particularly 6 the ratcheting specimens, we have to limit our recommendations 7 now to things we've tested or somehow address this later.
8 CHAIRMAN SIESS: Can you use the MCL type materiai 9 test to make a bridge between the material properties 10 and the observed behavior of components and systems?
11 MR. GUZY: Yes.
12 MR. TAGART: We have not, at this point done 13 that.
lllh 14 We had a long discussion on this point at our 15 last review meeting, about potential restrictions of the 16 first round of rules that we are going to recommend for 17 the Code. We do expect to have caveats, restrictions, 18 whatever you want to call them, with respect to the application 19 of these rules to the materials.
20 We have not completed the process of identifying 21 exactly which materials that will not be specifically 22 included; however, comments like the ones that you are 23 providing will be helpful in our--as an input to that 24 consideration.
25 CHAIRMAN SIESS: Well, there are two ways to l
28.
gg 1 limit materials: one is by naming them; and the other 2 is by tying it to properties, measurable properties, and 3 I gather trom what you say it will be by name rather than 4 by the other.
5 MR. TAGART: Yes.
6 We would plan to do it by name, because we have 7 identified four materials in the materials testing work, 8 and one of the primary purposes of the materials test 9 is to bridge the gap between room temperature tests and 10 elevated temperature tests.
11 None of the tests that you saw in the last two 12 days were run at elevated temperatures, and the materials 13 tests were intended to bridge the gap between room temperature lllhI 14 tests and the elevated temperature tests, for the important 15 failure phenomenon that occurs in the materials, fatigue 16 ratchet.
17 MR. GUZY: Anything further?
18 CHAIRMAN SIESS: Any other questions?
19 [No response.)
20 No, let's proceed.
21 MR. TAGART: Good morning, ladies and gentlemen.
22 I welcome this chance to provide the introduction and 23 overview of this program.
24 The things that I am going to talk about for 25 the next few minutes involve a brief history of the Code
29.
glll 1 rules relative to piping. I am going to summarize what 2 we believed we knew in 1985 when we started this program. j 3 I would like to take a few minutes to explain a very simply 4 way of understanding why piping is so resistant to dynamic 5 and seismic type loadings, and I would like to summarize 6 what we know now near the completion of the program.
7 Also, I would like to discuss the challenges 8 and the opportunities that we think are available as the 9 result of this program.
10 (Slide]
11 This is not a complete history of piping, but 12 I wanted to highlight some of the things that I think--
13 what we've done in perspective, h 14 The earliest date here is 1952 when the Markl 15 fatigue tests were introduced into the B31.1 Code. The 16 basis of these tests were semi-static, that is they were 17 slowly fatigued to determine when various kinds of piping 18 components would fail in a leaking manner. They became 19 the basis for the detailed rules and the B31.1 code.
20 In 1963 the nuclear pressure vessel rules were 21 introduced into the SME Code where static and fatigue 22 type loads were considered. The Markl work was the forerunner 23 of the static treatment and the low-cycle heat treatment 24 of loads for pressure vessels.
25 In 1968 nuclear piping rules were introduced
30.
into the ASME Code, which involved static, dynamic, and lllg) 1 2 fatique loads.
3 To put a footnote on dynamic--and I want to 4 emphasize that the basis for the dynamic loading was that 5 the effects of dynamic loads were handled by static failure 6 criteria; that is, it was recognized that there would 7 be dynamic loads, but the criteria for failure under those 8 loads would be the same as if those loads had been applied 9 staticly. That was a simplification, made at the time.
10 If we had had the results of this research available then 11 we would have done that differently.
12 In approximately 1975, Japanese research, which 13 was aimed at confirming whether--particularly whether llk 14 the D-level stress levels in the ASME Code, whien goes 15 somewhat beyond the elastic limit, were acceptable and 16 safe. In the process of evaluating these D-levels stress 17 limits, they identified large dynamic margins; however, 18 their focus was not to find out exactly how large they 19 were but simply to establish whether the ASME rules for 20 level-D were acceptable and safe.
21 They also identified fatigue ratcheting, the 22 swelling of the pipe, as an important part of the failure 23 mode.
24 In 1982, a PVRC program under the leadership 25 of Spencer Bush was initiated to improve nuclear piping,
31.
g 1 and one of the major things that ccme out of that was 2 the Code case N-411, which allowed people to begin removing 3 snubbers.
4 In 1985--Dan has already mentioned- the 1061 5 piping recommendations, and simultaneously with that was 6 the beginning of this program.
7 Now we are about to complete these dynamic tests 8 and we will have a basis for new rules for the ASME Code.
9 [ Slide]
10 To summarize what we knew in 1985:
11 No. 1. We knew that dynamic margins were large, 12 but we were uncertain as to exactly how large they were, 13 so the emphasis of these tests was to take them all the 14 way to failure, and there was considerable thought and 15 effort put into selecting test facilities that would produce 16 failures in a relatively few number of load applications, 17 and our target was no more than five time history earthquakes 18 being applied to the specimen to produce failure; and, 19 at the same time we were planning to use more or less 20 normal pressure loads. Pressure loading was not exaggerated, 21 only the dynamic loading.
22 No. 2. The fatigue failure mode for reversed 23 dynamic loading is ratcheting and fatigue, not static 24 collapse. We knew this from the Japanese research.
25 MR. SHEWMON: Is static collapse what Chet would i
32.
gggg 1 call "not section collapse" sometimes? Or, what is static 2 collapse?
3 MR. TAGART: Static collapse--
4 CHAIRMAN SIESS: It is collapse.
5 MR. TAGART: --is the--
6 MR. SHEWMON: It is not failure. It is tye 7 walls coming together?
8 MR. TAGART: No.
9 What I mean by static collapse here is that 10 if you plot the load deformation behavior of the structure, 11 the deformation starts becoming large with small increases 12 in the load. It is well beyond the elastic--
13 CHAIRMAN SIESS: The curve you showed--somebody h 14 showed yesterday--went down versus the one that went up.
15 MR. SHEWMON: Went up, yes.
16 MR. TAGART: The ASME Code has some definitions 17 of collapse that suggest collapse occurs before you start 18 going down, so it depends on whose terms you are using 19 as to what it means. I think it means something slightly 20 different to civil engineers.
21 CHAIRMAN SIESS: Would that be different if 22 you left static out?
23 MR. TAGART: It think it is not quite as clear 24 without the word ' static" and I am suggesting static collapse 25 as a term used in this program to distinguish it from i
33.
1 an incremental collapse. Incremental collapse, as you ggggg 2 saw in the films yesterday, could involve step-wise collapse 3 of the structure. We want to distinguish static collapse 4 from a one-application of load to cause collapse, between 5 that which occurs with many applications of load where 6 it moves slewly.
7 CHAIRMAN SIESS: I am still trying to understand.
8 Frequently we take repeated loadings and find 9 that they can be enveloped by a single monatonic static 10 load.
11 Are you saying that the static collapse would 12 be that monatonic loading, and that the dynamic collapse 13 that you are talking about would not be enveloped by that?
llllh 14 Do you visualize what I am talking about? You know, I 15 will draw you a curve, and it will look like this--(drawing 16 curve in the air)--they look like this, a static curve, 17 and a monatonic loading would be right in the upper bound 18 of it, and this is in certain types of things, not piping 19 necessarily, but things that I know about. Is that a 20 distinction you are making?
21 MR. TAGART: It is very hard to make a general 22 distinction in that way; for instance, in our materials 23 tests we clearly see what you are talking about. We plot 24 on a diagram what happens to the reverse loading, and 25 at the same time we can look what the comparative material
34.
ggll) 1 behavior is of a uniax).a1 specimen just being pulled.
2 There one can see a relationship between a incremental 3 collapse and a singic collapse.
4 CHAIRMAN SIESS: Are you saying that you do 5 not get that kind of relationship in piping?
6 MR. TAGART: Well, we see it, yes, but there 7 is a complication in piping, because piping is a fairly 8 complicated structure, even for example, understanding 9 an elbow--
10 CHAIRMAN SIESS: Yes.
11 MR. TAGART: --the elbow is one of the more 12 difficult things to understand because it is a complex 13 structure and each material in it behaves in a certain 14 way differently.
15 CHAIRMAN SIESS: But, I couldn't envelop the 16 dynamic incremental collapse with the static collapse 17 curve?
18 MR. TAGART: I don't know.
19 CHAIRMAN SIESS: It would be wonderful if you 20 could!
21 MR. RODABAUGH: Isn't the answer "yes," Sam?
22 CHAIRMAN SIESS: I don't think you can.
23 MR. RODABAUGH: The answer is "yes" because that 24 is a much bigger envelop than--
25 CHAIRMAN SIESS: No. I mean match it.
35.
1 MR. BUSH: No.
2 CHAIRMAN SIESS: I mean, if I draw the envelop 3 for the ratcheting of the incremental collapse that that 4 would agrec. I think it would fall well below the static 5 collapse.
6 MR. TAGART: It is not identical, though, and 7 I think the behavior of the structure--it is not even 8 identical for the materials test.
9 CHAIRMAN SIESS: Okay, then I think I understand 10 what you are saying there.
11 MR. TAGART: All right.
12 Now, the third point here is a conclusion 13 that we reached in 1985 as a result of some preliminary 14 thinking about this program. We felt quite strongly t' it hh>
15 fatique ratcheting was the mode of failure, but we didn't 16 feel that that was well known in the industry, and we 17 felt that we could support that kind of conclusion analytically 18 but we felt such a demonstration would not be convincing, 19 and we concluded that experimental evidence, plus an engineering 20 understanding of those experiments would be necessary 23 to effect a change at this point in time, and of course, 22 the observation that many people came to was that nuclear 23 plants have too many snubbers.
24 Now, I would like to take a few minutes to describe 25 a simple explanation of why piping is so resistant to
36.
I aynamic loads. I am going to talk about a picture that appears in that fourth simi-annual report, and I'll show 3
you what the picture looks like first, and then go back 4
to this diagram.
5 It is this page in the fourth semi-annual on 6 page 3-208, where we are describing an amplification versus 7 The standard kind of thing that appears in frequency.
8 textbooks on force vibration analysis.
9 But, before I discuss that I want to discuss 10 the assumptions here that go into one of the curves. This 11 diagram is the force on a single degree of freedom spring 12 versus the deflection of the single degree of freedom 13 spring.
14 So, this is the deflection--(referring to the 15 drawing]--this is the stiffness times the deflection, 16 or the force. And, this is the simplest model showing 17 complete pl.asticity, assuming that there is some value 18 at which the structure become elastoplastic. At this 19 point it unloads, become clastic again, goes into reverse 20 plasticity and absorbs energy through this loop.
21 The component tests are very close to a single 22 degree of freedom system. The ma]or complication that 23 could be added to better understand it would be to put 24 a slope on this curve right here, and to make it clastic 25 and then strain hardening in both areas. But, for the
37.
gg 1 moment, let's look at perfectly plastic.
2 One of the simplest models that we can use, 3 which is an approximate dynamic analysis--
4 CHAIRMAN SIESS: That has a mean stress on it, 5 obviously?
6 MR. TAGART: - - r.o .
7 CHAIRMAN SIESS: Then it is not at zero.
8 MR. TAGART: This is adjusted so that the diagram 9 centers around this point here, but there is no mean stress 10 in this. Sam Ranganath will talk a little later about 11 what happens when we add mean stress to this.
12 MR. SHEWMON: This is an A cycle taken after 13 you have reached a steady state.
llll) 14 MR. TAGART: This is a steady state behavior 15 with no mean stress.
16 CHAIRMAN SIESS: Okay.
17 MR. TAGART: Now, then we--
18 CHAIRMAN SIESS: I guess that I am still having 19 a problem.
20 How do you get to this? You have to start at 21 zero when you go up on that slope and get plastic, and 22 then it settles down to this loop then?
23 MR. TAGART: Okay, let me describe the problem--
24 CHAIRMAN SIESS: This is the Nth cycle.
25 MR. TAGART: This is sinusodial exicitation
38.
(glll 1 of a single degree of freedom system after it gets through 2 its transient. It settles into some steady state.
3 CHAIRMAN SIESS: Okay.
4 MR. TAGART: And, it is driven hard enough that 5 it becomes elastoplastic.
6 CHAIRMAN SIESS: In both directions?
7 MR. TAGART: In both directions, right.
8 CHAIRMAN SIESS: Do your component tests become 9 plastic in both directions?
10 MR. TAGART: Yes.
11 CHAIRMAN SIESS: Then I don't get a ratchet 12 out of it, do I?
13 MR. TAGART: In this case, no.
14 CHAIRMAN SiESS: Okay.
15 MR. TAGART: If you add a mean stress, you will.
16 CHAIRMAN SIESS: Okay.
17 MR. TAGART: In this model, there are two things 18 done to the simple equation of motion: one is to approximate 19 the damping by the energy absorbed in this loop.
20 The second is to put a reduced stiffness in 21 the single degree of freedom, which is the slope of this 22 line, and this nakes the solution nonlinear; that is, 23 you now don't know the deficction before hand, and the 24 equation that will solve this for a sinusodial 25 motion has to be solved by some trial and error technique.
I
390 gg 1 It is not an exact solution to the problem because 2 the motion is not sinusodial, purely sinusodial 3 when it goes to the clastic plastic, but this makes a 4 very simple explanation of what goes on, so that is the 5 assumption that goes into the clastic plastic model, and 6 the diagram that is in the report shows the results.
7 The solid curve is a curve at two percent damping--
8 CHAIRMAN SIESS: I am having a problem.
9 That is a response spectrum, in effect?
10 MR. TAGART: Yes.
11 It is a steady state response spectrum for a 12 single degree of freedom system--
13 CHAIRMAN SIESS: And, it is the amplification 14 of what?
15 MR. TAGART: It is the amplification of mass 16 relative to the ground.
17 CHAIRMAN SIESS: Its acceleration?
18 MR. TAGART: It is the--either the displacement 19 or acceleration. The assumption here--
20 CHAIRMAN SIESS: What's "A"? It says C over 21 A in there?
22 MR. TAGART: Okay.
23 "A" is the amplitude of the input motion. See 24 in the box there--
25 CHAIRMAS SIESS: Okay, okay--
f
40.
ggggg 1 MR. TAGART: -
"A" is the amplitude of the 2 sinusodial displacement.
3 CHAIRMAN SIESS: --I've got it. All right.
4 I've got it now. That is a displacement versus--
5 MR. TAGART: Right, it is a displacement response.
6 CHAIRMAN SIESS: It is a displacement response, 7 yes.
8 MR. TAGART: Right.
9 This curve is for the two percent danping case
, 10 if it were elastic.
11 The assumption for the picture that I just described:
12 it is labeled "Tagert's Model Lambda =Zero" racaning no strain 13 hardening, is this dotted curve right here, and Stat it llllI 14 tells us is that we get a frequency shifting from this 15 point back to this point, the softening effect, and an 16 enormous lowering of the peak, as you know. This peak 17 goes way up here, so you get an enormous reduction.
18 This particular case here is pictured for five 19 times the yield stress--
20 CHAIRMAN SIESS: Wait a minute, wait a minute.
21 Five times the yield stress? Or yield strencth?
22 MR. TAGART: Yield strength, yield strength.
23 CHAIRMAN SIESS: Okay.
24 MR. TAGART: I'm sorry.
25 CHAIRMAN SIESS: Does that frequency shift correspond
42.
gggg) 1 to that dotted line you had on the previous figure?
2 MR. TAGART: Yes.
3 CHAIRMAN SIESS: Okay.
4 MR. TAGART: An exact solution of this same 5 problem, not an approximation, is also shown on the diagram 6 by what's called the numerical solution for Lambda to equal 7 :cro. It is this curve right here.
8 Now, that solution is conservative compared 9 to the exact solution; however, when you put a little 10 strain hardening into it, in the exact solution it brackets 11 the approximation, so the approximation is very close 12 to the reality of what goes on in this single degree of 13 freedom test, and here we see the offect of the energy llllh 14 absorption which completely chops off this high resident 15 peak, and it shifts the response to the left on this diagram, 16 Another very interesting thing that one can 17 observe from this diagram--which we did not strongly observe 18 in any of our tests--is that there is a region in here 19 where elastic analysis will underpredict the response 20 regardless of what damping you put in it, and that's one 21 of the reasons why--
22 CHAIRMAN SIESS: It is the frequency shift.
23 MR. TAGART: --yes, it is because of the frequency 24 shift. That is one of the reasons why we were less enamored 25 with the idea of making changes by controlling just the
43.
ggggg 1 damping. The damping certainly has an important effect 2 because it chops off this peak, but the frequency effect 3 is also more important, and we think probably the easier 4 way, the more straight forward way, is to handle both 5 of these effects in linear analysis by changing the allowable 6 stress, rather than by trying to deal with the damping.
7 MR. BUSH: Sam, how do you handle the strain 8 softening aspects, or do you simply--
9 CHAIRMAN SIESS: A little louder, Spence.
10 MR. BUSH: I was asking how you handle the strain 11 softening aspect, or do you simply cut it off at Lambda 12 equals :cro?
13 MR. TAGART: The lambda equal zero model is llllh 14 a--do you mean if the slope were actually to go down rather 15 than up?
16 Well, strain softening, there are two ways to 17 think about strain softening. There is strain softening 18 in a single cycle, or there is strain softening where 19 successive cycles may have lower stress range than they 20 had in the earlier cycles.
21 We've addressed that by our materials tests.
22 We selected four different materials in the materials 23 test: one to be strain hardening: one to be strain softening; 24 and two to be more or less neutral.
25 We are concerned about the problem of predicting
43.
1 the ratchet. I think the subject of st.alu hardeaing ggggg 2 and softening is most important in the area of how much 3 ratche-ing will occur in each cyclc? And, I think Sam 4 Ranganath is going to cover this in a little bit of detail 5 to explain what conclusions we've drawn to date on the 6 ratcheting. I hadn't planned to talk about that at this 7 part of the discussion.
8 CHAIRMAN SIESS: Finc.
9 MR. TAGART: What I think is useful about this 10 diagram as that there is a simple physical way to explain 11 why under the steady state response--and I think you saw 12 yesterday in these experiments--that you reach some kind 13 of a steady state. Now, most of those tests were seismic lllll 14 inputs. We only had a test where we've done sinusodial 15 inputs, and we will be able to more directly compare the 16 applicability of this model to those sinusodial 17 inputs, but it is a relatively easy thing for us to put 18 a non-sinusodial input into this single degree of 19 freedom model and get comparisons, and we have done that, 20 so we've made a lot of progress to understanding those 21 compliment tests by simply looking at a single degree 22 of freedom clastoplastic model. That is the point that 23 I wanted to get across with this diagram.
24 (Slidel 25 I would like to give an overview of what we
440 We llllh 1 think we know now, as the result of this pregram.
2 believe that we know why static collapse does not occur--
3 it does not generally occur, in the dynamic loading situation.
4 The previous diagram is an clantoplastic system. If any 5 system could collapse as a singic degree of freedom system, 6 that one would collapso and it generally does not, and 7 there is a good explanation as to why it doesn't.
8 We know also that there are certain types of 9 dynamic loads that can collapse the piping. As we saw 10 yesserday, we were abic to produce large deformations 11 in the water hammer cases where the load holds up long 12 enough to allow the pipe to mold, so we don't want over 13 generali-ing results, our objective is to make Code changes 14 but not to overstate the case relative to certain kinds 15 of dynamic loads.
16 CHAIRMAN SIESS: So, the type that can cause 17 it is one that isn't too dynamic?
18 MR. TAGART: That's right. It behaves more 19 like a static load.
20 We know how to approximately predict the component 21 results from first principles. I haven't discussed the 22 ratcheting, but that will be discussed.
23 We know that there are some limitations of linear 24 dynamic analysis, that previous diagram showed the key 25 area where there may be some concern about that.
45.
(glll 1 We have clarified concepts of apparent damping.
2 I haven't discussed that much, but if we look at the damping 3 that is available, the equivalent damping that is available 4 in that single degree of freedom system, it is vety, very 5 large, on the order of 40, 50 percent is available in 6 that single degree of freedom system, with sinusodial 7 inputs. But, we need to distinguish between the what 8 we call--Henry Hwang has termed true damping, and apparent 9 damping--and this word should be true--(referring to tbc 10 slidej--I am sorry for the error here, instead of "time" 11 this should be true damping.
12 piping systems are fundamentally resistant to 13 scismic and other dynamic loads because the true damping 14 is very high at ductility of as low as three, and as a 15 matter of fact, if you will look at that little degree 16 of freedom model, which is not an exact solution, it 17 actually maximizes the damping at a value of three.
18 MR. SHEWMON: Can you tell me what a dynamic 19 ductility of three means to a stress strain curve?
20 MR. TAGART: It means if the single degree of 21 freedom system were a mass hung on a tensile bar, then 22 the yielding of the structure and the yield of the material 23 would be the same, and if we had a clastoplastic material 24 and the deformation which occurred was three times the 25 collapse load, that is what I mean by ductility of three.
l l
i i
t
46.
lllll 1 CHAIRMAN SIESS: Not three times the yield?
2 MR. TAGART: Well, in thet case it would 3 be the same.
4 CHAIRMAN SIESS: That is--
5 MR. SHEWMON: This is three times the strain 6 of the yield: is that right?
7 MR. TAGART: Well, this is three times the 8 deformation at which the structure collapses.
9 CHAIRMAN SIESS: On this curve, isn't it the 10 ratio of that distance to that?
11 MR. TAGART: Yes, yes, 12 CHAIRM'N s SIESS: The ratio of--
J3 MR. SHEWMON: Thank you, 14 MR. TACART: We believe that we understand ratcheting.
15 Ratcheting very simply is a lack of symmetry through the 16 cycle, and the presence of a maan load means that when 17 you have half of a cycle, a mean load is adding to one 18 directicn of plasticity and in the opposite direction 19 it may be subtracting, and therefcre there is a net accumulated 20 plastic strain, or deformation, when one completes a cycle 21 in the presence of mean loading.
22 CHAIPMAN SIESS: Now, you talk about mean loading--
23 MR. TAGART: Mean stress.
24 CHAIRMAN SIESS: --it is mean st; ess--
25 MR. TAGART: Eight, f
r
470 1 CHAIRMAN SIESS: Because that mean stress can 2 be a pressure induced stress--
3 MR. TAGART: Yes.
4 CHAIRMAN SIESS: It doesn't have to be a load 5 induced stress.
6 MR. TAGART: That's right.
7 CHAIRMAN SIESS: Okay.
8 MR. TAGART: All right.
9 Typically, in the--and most of the applications 10 we are talking about pressures is the dominant behavior, 11 although weight is also an important consideration, and 12 we've seen this in our experiments. The influence of 13 the weight is a very strong effect in how much ratcheting 14 will occur.
15 CHAIRMAN SIESS: You mean simply the weight 16 producing a longitudinal stress in the pipe, or whatever 17 you test?
18 MR. TAGART: Yes, yes.
19 CHAIRMAN SIESS: Just gravity.
20 MR. TAGART: Yes.
21 (Slide]
?2 I would like to spend a couple of minutes on 23 the opportunities and challenges that present themselves 24 as the result of ccmpleting this research. We will have 25 a significant Code margin reduction proposed, as a result
48.
of this program, that's ;:ot necessarily going to be easy.
llgg 1 2 As I have discussed this with Dan Gu:y, he tells me we 3 shouldn't expect that this is going to happen one week 4 atter we make our croposal. It -I almost hesitate to 5 say this--but it may take a year or more f or these results 6 to get into the Code, and some of the things that we see 7 that the regulator will have to look at is managing the 8 prior and future Code changes.
9 A couple of important things, the Code case 10 N-411 which got us going relative to snubber reduction--
11 CHAIRMAN SIESS: What is 411?
12 MR. TAGART: Code case N-411 is the one that 13 increased the damping to five percent--
14 CHAIRMAN SIESS: Okay.
15 MR. TAGART: --At low frequency, and two percent 16 at high frequency.
17 CHAIRMAN SIESS: Okay.
18 MR. TAGART: The Code case N-451 which was just 19 recently passed, which toolc the operating basis carthquake 20 out of equation--yes.
21 "R. SHEWMCN: Would you put some words on significant 22 cod margin reduc.4cn? Would that come later?
23 MR. TAGART: Okay, it will come later, but a 24 think it is a good point to tring it up now.
25 We are thinking of offe:tively increasing the I
l w_---______
s N 49.
.5 1
llggg 1 ajLiwable stress in the Code somewhere in the order of "I
.. 2 50 to 100 percent, and we haven't decided exactly how q ,' ..
, l j :' < a much that should be right now.
4 MR. SHEWMON: Double?
5 < . q MR. TAGART: Up to doubling it, at least 50 X\
/ 6 p?u.'ent higher than--
7 , MR . CHEWMON: That is the seismic component 8 Or the Code?
Iofthestress?
s, 9> MR. TAGAPr: The dynamic--the total stress equation
( 's
\ lb that cavers t;;.e cembination cf pressure stress and the b
11 inertia strass from dynamic loading that is of a reverse 12 type, either seismic or roversea other type that behaves
) ,13 like~ seismic.
lh 14
(. ,
\ CHAIRMAN SIESS: B .i t ,. af you are going to increase
/ IE ,
0.h c Io.a1 allowable,howar: you gotng.ro take care of s
16 tne range and ratio between seismic anc c ther stress?
'h 17, ' An elei.ent that has very little seismic stress, do you .
l s
is harp same other equation that, governs te.at?
. s 19 MR. TAGART: Ycs.
. i 20 Cl%IRMF.( sIESS: Okay.
\
, \ \
' i 21 MR. TAGART: There dre otScr equations that -
t i i 1 22 CHAIRMAN SIESS: 3 TIhis wou'.d be the onlr equatic::
N '
23 that incaddes the seismid stress?
24 MR. TAGahT: Right. i q( ' 'I( 2 5 CHAI"MAN SIEF3: Load combination--
' 4.j s s
(
,,~.a- ,t 7,' \
>\ y 9 ,
i i
\
,t \ ' '
s
\. \
50.
lllll 1 MR. TAGART: Seismic or other dynamic stress.
2 CHAIRMAN SIESS: All right, I got it.
3 MR. TAGART: One of the activities that is going 4 on is handling the independent support motion with SRSS 5 and it is currently the subject of research.
6 We have simplified static methods that art now 7 being seriously considered by the Code. We have nct -
8 linear methods which we know are going to be coming along, 9 seismic anchor motion modifications, and I mention here 10 possibly designed by rules. We believe that the results 11 of this research program are conducive to eventually producing 12 design by rules in piping, as opposed to design by analysis, 13 particularly for the seismic effects. We are not recommending 14 that at this time, but we think it is a fruitful potential 15 improvement in the future, and our approach, or our recommendation 16 will be to work with the current rules that are designed 17 by analysis, make those changes which are appropriate 18 as the result of those rules, and at some point in the 19 future explore and examine the possibility of great simplification 20 in the piping design process for nuclear plants. That 21 is for the future.
22 I would like to show you something that addresses 23 a method to optimize piping design. I think that is one 24 other real opportunity here. In the past the approach 25 to making changes to the Code has involved trading one
510 1
conservatism against another, and that's a viable approach, llggg 2 but it doesn't allow us to optimize the piping system, 3 and I would like to show you an approach that we have 4 examined at EPRI in the last year and what I've shown 5 here is a probabilistic approach to optimizing piping 6 design, and this is what decision analysts call "an influence 7 diagram."
8 [ slide]
9 A circle on this diagram represents something 10 that we are uncertain about. A square represents something 11 that we have control over, that we can made a decision 12 about. So, at the top of the diagram we focused on snubber 13 reduction efforts.
lh 14 What we did in this program was attempted to 15 say if we temporarily removed all requirements for piping 16 design, and we could trade off the pluses and minuses 17 relative to how many snubbers we would put into a nuclear 18 power plant, how many would we put in. We had complete 19 freedom to do it, and we could make the decision on the 20 basis of cost, safety, or both--
21 CHAIRMAN SIESS: Knowing what you know now.
22 MR. TAGART: --knowing what we know now, yes.
23 CHAIRMAN SIESS: The early plants didn't have 24 any--
25 MR. TAGART: Yes.
l
52./53.
lllgg 1 CHAIRMAN SIESS: --knowing what they dic then.
2 MR. TAGART: Right.
3 This model is an ambitious model, and it was 4 done by one of our contractors with the help of a second 5 ane who was very familiar with probabilistic risk analysis, 6 so we have things in here about the core melt. How does 7 the pipe failure influence the core melt? And, here you 8 see in this original diagram, we had things like water 9 hammer, the seismic loading--
10 CHAIRMAN SIESS: Sam.
11 MR. TAGART: Yes.
12 CHAIRMAN SIESS: What's the significance of 13 the two circles with a lot of arrows out of them that 14 don't do anywhere?
15 MR. TAGART: It means they connect to a lot 16 of other ones, and--
17 CHAIRMAN SIESS: Okay.
18 MR. TAGART: --for instance--
19 CHAIRMAN SIESS: And, you don't know which ones.
20 MR. TAGART: We could connect them to almost 21 all of them.
22 CHAIRMAN SIESS: Okay.
23 MR. TAGART: It would make the diagram so confusing 24 to put all of those arrows in there, we are saying the 25 design and construction errors can--are pervasive through
54.
gglll 1 the whole diagram. You can have things go wrong, with 2 what you know about anything going on in this diagram.
3 Similarly, we have this influence of NRC regulations 4 on very many parts of this diagram, in other words there 5 are regulations that effect how each one of these things 6 are done, and we don't know how those regulations are 7 going to change in the future.
8 The study that we did simplified this diagram j 9 a little bit, and although it would take too much time 10 to discuss the complete implications of this study, I 11 would like to show you some of the results.
12 CHAIRMAN SIESS: Sam, it seems to me that if 13 you are going to do a true optimization you have got to 14 know almost everything about everything.
15 MR. TAGART: That's right.
16 CHAIRMAN SIESS: It is obvious that you don't.
17 MR. TAGART: That's right. You are very 18 uncertain. That is why these are put in circles.
19 CHAIRMAN SIESS: Yes, so it has to be a 20 probabilistic optimization?
21 MR. TAGART: That's correct.
22 And, what we attempted to do here is a picture 23 as close to reality as we know it today. We recognize 24 that some parts of it are very uncertain, but we want 25 to give the best expected values for each one of the variables i >
55.
in the diagram.
l llllh 1 2 CHAIRMAN SIESS: Okay.
3 MR. TAGART: The results of that study, en a 4 very simple system, extrapolated to an entire plant, and 5 the costs of an entire plant is a diagram that looks like 6 this.
7 [ Slide) 8 This is where we stand right now, and typically 9 this might be a plant with, let's say, a thousand snubbers.
10 We are plotting here the number of snubbers removed, versus 11 expected lifetime costs of thr. plant.
12 CHAIRMAN SIESS: Percentage.
13 MR. TAGART: And, for different costs involved 14 in maintaining snubbers, we have different curves.
15 This is a very high cost snubber, $7000 per 16 snubber for the lifetime, annual maintenance cost for 17 the snubbers.
18 This is a considerably lower cost, and this 19 is a very low cost.
20 And, what you see here, of course, as you can 21 well imagine--and we have put the safety costs in this 22 picture, as well. You will notice down here, we said 23 if we get a core melt there is some large cost associated 24 with that core melt. For example, in this case, $20 billion, 25 and in this case $5 billion, and of course it changes
56.
g 1 the picture.
2 So, this is one diagram that puts the safety 3 picture and the cost picture on the same diagram. It 4 shows, for example, that we should be, regardless of what 5 the cost of snubbers are, we should be moving in the direction 6 that we are moving, and that is to remove snubbers.
7 If the cost is very, very low, we see a diagram 8 that reaches some minimum point here and then starts coming 9 back up again. If the cost is very high, and we werc 10 optimizing on cost, it would come down.
11 (Slide]
12 Now, I would like to show you a diagram that 13 shows what happens if you forget the cost, and simply optimize on the safety question. Here is a diagram that lh 14 15 says the lifetime probability of core melt due to pipe 16 failure, and here is where we stand right now with a large 17 number of snubbers in the plant. It says that--and or 18 two probabilities or earthquake--
19 CHAIRMAN SIESS: What carthquake is that, that 20 you are putting the probability on?
j 21 MR. TAGART: This is the probability of the l
22 SSE.
23 CHAIRMAN SIESS: Well, that won't cause any 24 damage--well, I guess it is some probability.
25 MR. TAGART: Okay.
l
570 ggggg 1 The reason that it causes some problem is that 2 this model says I could have degraded piping. I could 3 have pining with cracks in it before the earthquake comes.
4 CHAIRMAN SIESS: Well, now, suppose you did, 5 and a t.hree SSE at 10 to the minus 5--two SSE at 10 to 6 the minus 5, what would it look like?
7 MR. TAGART: Well, it would change the picture 8 and I don't want to speculate about how it would change 9 it.
10 I want to tell you about the results that we 11 hr <e .
12 CHAIRMAN SIESS: What pipe failure frequency 13 are you taking for that earthquake? You said degraded lllh 14 piping, so you must be some way putting in--
15 MR. TAGART: We have a very crude model for 16 the degraded piping.
17 What we have done, essentially, is look at the 18 piping as if it can fatigue all the way to a failure point, 19 with a very simple model. We have a very simple fracture 20 mechanic's model. The input is not the failure rate of 21 the piping. The input is part of the equation to describe 22 the lifetime of the piping as a function of the thermal 23 loads, the seismic loads, and how big a crack it has.
24 CHAIRMAN SIESS: And, you are assuming that 25 the earthquake is 1/22nd cycle like you have? No after
58.
1 shocks?
2 MR. TAGART: We are not into that level of detail.
3 CHAIRMAN SIESS: But, you are assuming some 4 kind of an earthquake, at so many cycles, and on the basis 5 of what you know that that combined wit h other things 6 will cause fractures.
7 MR. TAGART: Yes, yes.
8 CHAIRMAN SIESS: Okay.
9 MR. TAGART: So this model admits the possibility 10 that the earthquake--and in fact a very crude assumption 11 was made here. For example, we said if a pipe can get 12 to a failure point by leakage, that 15 percent of the 13 cases would break before leak.
lllh 14 CHAIRMAN SIESS: Okay.
15 MR. TAGART: Here again, we see that the right 16 direction to move is to remove snubbers, and these curves 17 turn up sharply, which was a bit of a surprise when we 18 first saw that. We thought that perhaps the diagram would 19 look like this, and come to a point which was below this 20 point so that if we removed all of the snubbers we'd be 21 better off than if we kept some in.
22 This is because--and it is dotted here, and 23 I would emphasize that this is very tentative in nature--
24 it is because of the degraded pipe questisn, and we have 25 another program, the IPERC program which is coming along
590 to examine degraded pipe. We focused on sound pipe. The gglll I 1 2 degraded pipe program tells us--answers questions about 3 the uncertainty of unsound piping, and I believe the long 4 term optimization of piping depends on our looking at 5 both of these programs.
6 I think it is clear at this point that very 7 large numbers of reduction will both improve the costs 8 and improve the s..ety.
9 CHAIRMAN SIESS: It would seem that--let's take 1
10 not the 80 percent point, but your 90 percent point--
11 it would seem to me that it would make a difference as {
12 to which snubbers were included in that 10 percent window?
13 MR. TAGART: That's true, and what we are assuming 14 here is that we take out the right ones, too. We are 15 taking them out in some systematic way-- ;
16 CHAIRMAN SIESS: So, the 20--
17 MR. TAGART: --so we take the right ones.
18 CHAIRMAN SIESS: --percent you are leaving in 19 are some of the most crucial ones--
20 MR. TAGART: Yes. l 21 CHAIRMAN SIESS: --and when you start taking 22 those out the risk goes up.
23 MR. TAGART: Right.
24 CHAIRMAN SIESS: Okay, 25 MR. TAGART: And, if one wanted to think about
60.
ggggg 1 further optimizing it, some of the passive restraint devices 2 could be substituted for those few that really do us some 3 good.
4 MR. RODABAUGH: Sam, what is it about--is it 5 possibility of snubber malfunction that makes this curve 6 go down?
7 MR. TAGART: Yes, that's it, exactly.
8 The snubbers themselves can malfunction, they 9 increase the stress, and that is why these curves go down 10 there.
11 MR. BUSH: And, that depends on the types 12 of snubbers.
13 MR. TAGART: Yes.
lll h 14 MR. WARD: Then are these curves--let's see, 15 the last figure, are those consistent with the previous 16 figure?
17 MR. TAGART: Yes.
18 MR. WARD: I mean it is the same?
19 MR. TAGART: They are the same.
20 MR. WARD: So the break off is what you see 21 on the previous one?
22 MR. TAGART: Exactly, right.
23 MR. WARD: What would thac previous one look 24 like with, if you are talking about a new plant capital 25 costs of snubbers? The same?
l
61.
Which curve would we pick?
lllll 1 MR. TAGART:
2 MR. WARD: Yeah.
3 MR. TAGART: It probably is this middle one 4 for the cost is more like what we would expect.
5 MR. BUSH: Sam, a question on that curve.
6 I don't know what the maintenance cost are that 7 it has, because one of the problems, obviously, with the 8 snubbers is the effect on outage time.
9 MR. TAGART: Yes.
10 MR. BUSH: Which has a, spread over several 11 years, it doesn't take many days of outage time to increase 12 to bias the costs considerably.
13 MR. TAGART: Yes. I think we have not considered 14 a lot of outage time in this, and if outage time got to 15 be a big factor, these number of course would move more 16 in this direction.
17 CHAIRMAN SIESS: This is sort of routine maintenance, 18 then.
19 MR. BUSH: You are talking about $100 million 20 roughly, as a zero baseline on that conservative model.
21 MR. TAGART: Yes.
22 CHAIRMAN SIESS: And the snubbers that you can't 23 test, I assume you are leaving in?
24 MR. TAGART: Pardon me?
25 CHAIRMAN SIESS: The big snubbers that you can't
62.
ggg 1 test, I assume those are some that you are leaving in?
2 MR. TAGART: Well, here we are talking about 3 piping.
4 Some of those big snubbers are on--
5 CHAIRMAN SIESS: Oh, that's right--
6 MR. TAGART: --peak generators.
7 CHAIRMAN SIESS: --they are on peak generators.
8 MR. TAGART: With that, I would like to--I guess 9 we are at the break?
10 CHAIRMAN SIESS: Yes.
11 MR. TAGART: Anymore questions?
12 (No response.]
13 CHAIRMAN SIESS: Any questions for Sam before 14 we take a break?
15 (No response.]
16 Okay, let's take abouu 10 to 15 minutes for 17 a break, and the audio systems man may be able to fix 18 the microphones.
19 (Recess: 9:50 a.m. to 10:10 a.m.]
20 21 CHAIRMAN SIESS: All right, you may proceed.
22 MR. ENGLISH: My name is Bill English. I am 23 from General Electric.
24 I would like to take just a couple of minutes 25 to discuss with you the portion of the program that I
63.
llllh 1 will be talking about today, in a few moments, to describe 2 some of the objectives that we focused on through the 3 program, but the bulk of the time I would like to spend 4 talking about the component test, system test, specimen, 5 and analysis avoidance test.
6 CHAIRMAN SIESS: Please keep in mind that we 7 have had a pretty good description of the tests themselves.
8 MR. ENGLISH: Right, right, okay. I will try 9 to go through those portio 1s of the presentation very 10 fast, and if I am going too slow, you just speed me up.
11 CHAIRMAN SIESS: All right.
12 Well, I may ask you to just skip some of these, 13 since we've seen them.
14 MR. ENGLISH: Okay.
15 Today I will be talking about the component 16 tests of Anco, the system tests at ETEC and Anco, the 17 specimen fatigue ratcheting tests at MCL--actually started 18 at the General Electric turbine technology lab with Sumu 19 Ykowa [ sic.J and Roy Williams, and GE got out of the materials 20 testing business at that particular location, and the 21 lab was transferred over to--Roy Williams actually bought 22 the equipment and continued the tests, so we didn't lose 23 much time on part of the program.
24 And, then task 5 was the analysis et test and 25 design rules, which was done with GE in San Jose.
64.
g 1 (slide]
2 As you've probably heard a number of times, 3 a major objective of the program was to try to take 4 advantage of current dynamic margins that we've seen in 5 metal piping systems, and to devise new, more realistic 6 ASME Code rules, but in addition to that we wanted to 7 determine what the actual failure mechanism was in piping.
8 It is not believed to collapse at the same point now as 9 fatigue ratcheting.
10 We wanted to measure pipe damping as a function 11 of different stress levels, with the system configurations 12 and frequency inputs. We wanted to determine how big 13 an earthquake the piping systems could tolerate without 14 failure.
15 (slidej 16 We would like to develop a lab specimen to quantitatively 17 predict fatigue ratcheting, and ultimately plan to suggest 18 changes to the standard review plans, regulations and 19 codes to account for the margin that we have in piping 20 of the dynamic loading, and ultimately we would like to 21 be able to simplify the piping dynamic analysis, i
22 (slide) l 23 We felt the focus of the testing program was I 24 really on the component tests. The most severe loading 25 was in the component tests. It had the most instrumentation, 1
l
65.
llllh 1 28 to 30 to 32 channels of instrumentation, as compared 2 to only 80 channels on the system. The component behavior 3 for a number of different components could be demonstrated, 4 and we could determine what the actual failure modes were, 5 show that functionality was not compromised. We could 6 use the component test results to predict the system test 7 behavior and we could calibrate the design rules.
8 The main function of the system tests then was 9 to confirm what we've learned in the component tests, 10 to confirm that a single component doesn't actually collapse, 11 the load redestributes, that the mode of failure is not 12 collapse, that it is fatigue ratcheting, some kind of 13 incremental kind of form of failure. It confirms the l 14 functionality of the piping system, that the pipes actually 15 get bigger in diameter rather than smaller, and tend not l
16 to restrict the flow. It helps to design rules and margins 17 and it provides a lot of benchmark analysis for benchmarking 18 some of the computer programs used in piping aralysis.
I 19 The specimen tests, on the other hand, are a l 20 very simple method of demonstrating ratcheting, enabling 21 us to evaluate many different materials at minimal costs l
l 22 as compared to the component systems tests, and enables 23 us to determine the effects of temperature.
l 24 MR. WARD: Bill, why aren't those--I mean l
i 25 Paul shewmon asked earlier about the--
l 1
l
66.
llllh 1 CHAIRMAN SIESS: Speak up, please.
2 MR. WARD: --Paul asked earlier about 3 the, you know, seems to be a remaining kind of major uncertainty 4 with the cast materials.
5 Why wasn't the answer given that the specimen 6 tests are going to give a lot of information about that?
7 Won't they?
8 MR. ENGLISH: Well, part way through the program 9 this issue about cast materials came up, and we discussed 10 it in the context of advising the component test makers, 11 and at every review meeting we discussed the component 12 tests matrix to decide what we would like to change, and 13 it seemed at those meetings when the question of castings 14 came up that there weren't enough of them to warrant changing 15 the component test program to include a casting, and maybe 16 you couldn't draw a significant conclusion from one cast 17 component.
18 The obvious way to look at this in some detail 19 would be in the specimen test program, but that was already 20 established, and maybe at some later date we can get at 21 cast materials.
22 As Sam pointed out, we would probably restrict 23 the rules at this point in time to exclude casting.
24 MR. B'JSH : I may comment that if that is what--
25 (voice fades]-- on the basis that it would just make it l
67.
llll) 1 too difficult from a design point of view.
2 MR. ENGLISH: Just briefly, you probably had 3 most of this i.nformation on the component tests, but the 4 major objectives were to show once and for all that collapse 5 was not the failure mode of the components, to measure 6 the ratcheting and cycles to failure.
7 We had 41 components in the program, all six 8 inch in diameter, scheduled 10 through 40, with various 9 combinations of elbows, tees, reducers, and in the original 10 plan it was devised with Dr. Kennedy's help, was to put 11 the peak of the input at about a half hertz below the 12 component natural frequency, such that the component plastic 13 input would be driven up to a higher peak value, and at lk 14 Anco we were able to drive the sleds at the maximum capability 15 of the sleds.
16 We wanted to get the fatigue ratchet crack to 17 develop in two to three of these 20 second seismic inputs, 18 any longer than that we felt would be pretty much just 19 a fatigue test, so this was a target. Ultimately we eliminated 20 schedule 80 because it took too many inputs at the Anco 21 table capacity to generate a crack, and originally we 22 had planned to do some schedule 160 testing, and that 23 became obvious early in the program that we wouldn't be 24 able to crack those components, so we focused on schedule 25 10, 40 and 80.
68.
llllh 1 CHAIRMAN SIESS: In selecting the two or three 2 inputs, you gave no consideration to what the probability 3 was of a seismic input of a given duration--
4 MR. ENGLISH: No.
5 CHAIRMAN SIESS: --of frequency?
6 MR. ENGLISH: No.
7 We selected--I'll show you later. We just picked 8 a typical seismic time history in a BWR. In a BWR--
9 CHAIRMAN SIESS: Yes, I know what you did. I was 10 just asking, 11 MR. ENGLISH: No.
12 We reviewed that with Dr. Kennedy and we selected 13 one time history and we used it for all of the tests, 14 so we had one common basis--
15 CHAIRMAN SIESS: No, but my question you didn't 16 understand.
17 When we have an earthquake it is ona time history--
18 MR. ENGLISH: Right, 19 CHAIRMAN SIESS: --or D earthquake. It may 20 be longer than 20 seconds, depending on where it is--
21 MR. ENGLISH: Right.
22 CHAIRMAN SIESS: --and there may or may not 23 be after shocks, and the question of now what happens 24 to a second earthquake when it has been damaged by the 25 first one? And, I wondered if that was a consideration
69c llllh 1 in the two or three--
2 MR. ENGLISH: No.
3 CHAIRMAN SIESS: --or just getting enough cycles?
4 MR. ENGLISH: No.
5 We just put in the same earthquake two or three 6 times, with no consideration that the subsequent ones 7 might be different in frequency counts.
8 CHAIRMAN SIESS: Okay.
9 [ Slide]
10 MR. ENGLISH: The next two or three pages provide 11 a capsule summary of the 41 component tests, and I didn't 12 want to go over all of these with you, but only to describe 13 what these headings are, and you can refer to them and 14 ask any questions as we go along.
15 we show the number of the test here, the type 16 of component, the material--as you can see, it is either 17 carbon steel or stainless steel--and the schedule of the 18 pipe. The residual strain, it was a cumulative ratchet 1
19 strain that we measured in the component at the completion 20 of the test.
21 In some cases we have no data because early 22 in the program the high elongation of the gauges tended 23 to come off of the component before the test was completed, 24 and later we put scratch marks on the components so to 25 be sure of getting a measurement of the cumulative strain.
700 lllll 1 The pressure is the internal pressure in the 2 component, and typically the pressure is selected such 3 that it creates a stress of 1 S of M loop stress in the 4 component which is the ASME Code limit. Some of the 5 lower stresses were to get immediate data points that 6 were less than the Code limit.
7 The load direction is either in plane or out 8 plane, typically . The ratio here of the dynamic moment, 9 that is the measured moment in the test compared to the 10 static limit moment, and you can see in most cases that 11 we actually exceed the static limit moment without causing 12 collapse of the component.
13 There are three types of loads: SSE is a seismic 14 load. We used mid-frequency loads on a couple of components, 15 and water hammer loading on a couple of components, and 16 we had two static tests, but most of them were seismic 17 tests.
18 The peak-to-peak cyclic strain is the maximum 19 strain that was measured on the exterior surface of the 20 component, at what we believe to be the maximum strain 21 location, or high stress location of the component.
22 Now, the input times level D, this is the number 23 of times the ASME Code faulted limit that the input represented.
24 That is, if we took the input from the sled, did a linear 25 response spectrum analysis with two percent damping, 15 l
l l
710 lllll 1 percent broadening--just like you would have to do in 2 a typical piping analysis--and calculated the stress in 3 the elbow: in case 1 it would be 15 times the ASME Code 4 allowable.
5 The r. umber of times histories is the number 6 of times histories that were input--20 second times histories 7 at full amplitude of the sled before failure was in induced, 8 or before we stopped the test.
9 Typically, we would not go more that five inputs.
10 NF means that we got no failure. FR means there was a 11 fatigue ratchet failure.
12 Without going into any detail on these different 13 tests, there are a couple that I might point out that 14 were significant. If you look at 6, 7, and 8 on the first 15 page, that was an attempt to determine the effect of mean 16 stress on fatigue ratchet failures. You see that they 17 are all stainless schedule 40 clbows, and test 8 has zero 18 internal pressure, test 7 has a 1000 psi, and test 6 has 19 1700.
20 And, if you will look over to the far right, 21 under the number of time histories required to fail the 22 component, you can see that as the pressure increases 1
23 it took less input times to fail the componeitt, so in l 24 effect, as we all knew, mean stress had some effect on j 25 the actual failure of the component.
72.
1 There was one other component that I might point 2 out that we introduced as a result of recommendations 3 from Ev Rodabaugh. It took some convincing to convince 4 everyone that we couldn't collapse the piping as the result 5 of dynamic loading, and Everett suggested that one component 6 test, Test 37, which would probably convince him, would 7 be if we took a very low frequency component--in this 8 case it was 1.4 hertz--and as you understood I am sure, 9 the lower the frequency you get the closer you get to 10 just a static collapse test--so we tested a component 11 at 1.4 hertz with zero internal pressure--and with no 12 internal pressure you don't get any stiffening effect 13 from the pressure--and with that test we felt that was 14 the bounding test of this whole program to show that collapse 15 was not a credible mode of failure.
16 And, in this test it was somewhat excessive, 17 in the sense that we had a very large weight stress--
18 10,000 pai in this particular component, not the typical--
19 in the normal operating reactive plane, and because of ,
20 space limitations at Anco it ca ldn't--to get the frequency 21 low, they had to put a large weight on the inertia arm, 22 rather than bringing the inertia arm along, so we had 23 a very large weight stress, and even with that we could 24 not induce a collapse failure. We got what is called 25 ratchet buckling incrementally because the large weight
73.
lllh 1 stress it tended to bend significantly.
2 CHAIRMAN SIESS: Did you convince Mr. Rodabaugh?
3 MR. ENGLISH: I hope so.
4 MR. RODABAUGH: Yes, that did it.
5 CHAIRMAN SIESS: Good.
6 MR. ENGLISH: And, the final page of this series 7 is just more or less a legend that indicates what the 8 terminology is.
9 [ Slide]
10 I am sure you probably have seen this. Kelly's--
11 at Anco yesterday, he probably showed you this, where 12 we use the interia arm to--
13 CHAIRMAN SIESS: Yes, that we heard about.
14 MR. ENGLISH: --okay.
15 [ Slide) 16 This was a tirtic history that is input on almost 17 all of the tests. It is representative of a RPD steam 18 nozzle, typical PDW plant.
19 [ Slide]
20 You can see the response spectrum of that time 21 history as we tune the compcnent just slightly to the 22 right of the peak, with the input such that it softens 23 as it is driven up.
24 [ Slide]
25 This is the mid-frequency input. It is representative
74.
lllh 1 of a safety relief valve discharge on a BWR. It is a 2 two-second duration. It is interesting that there is 3 a small 7 hertz peak in this response spectrum, when the 4 bulk of the responses we've seen in the components is 5 the result of this peak, and practic;11y none from the 6 30 hertz portion of the spectrum.
7 We found that if we tilt it out at this small 8 peak here, and only applied true mid-frequency input to 9 the component, we have negligible response.
10 [ slide]
11 This is just to give you an example of the kind 12 of data that we get from Anco's use of four channels cut 13 of 28 that we would be examining. On a typical component 14 test we spread the time histories out considerably more 15 than the Anco data. We are able to look at the cycle.
16 I show this one just to show tnat you can see 17 that as the relative displacement from the top of the 18 sled to the bottom increases, those are these peaks, and 19 you can see the ratcheting actually occurring in the elbows.
20 When the peaks are small there is no ratcheting. When 21 the peaks increase again, the ratcheting is back up to 22 a point where, in most of these cases, it will shake down.
23 (slide) 24 Now, this slide shows that as you increase the 25 sled acceleration for these component tests, the cyclic
75.
llllh 1 strain initially increases relatively fast, and *. hen tends 2 to level of f and become asymptotic somewhere between two, 3 three, four percent cyclic strain. We believe that is 4 because the component becomes so plastic up in this region 5 that the plastic energy is absorbed by damping in the 6 plastic deformation.
7 CHAIRMAN SIESS: How did you draw that curve?
8 Whoever drew it didn't make it become asymptotic.
9 MR. ENGLISH: Well, maybe I shouldn't use the 10 word asymptotic. It does tend to show that it doesn't 11 increase--
12 CHAIRMAN SIESS: Well, I suspect that it might, 13 but that looks like that these squares fit.
14 MR. SHEWMON: Is what you have there, is cyclic 15 strain the same thing as the rise in the mean strain increases?
16 MR. ENGLISM: No, it is a cyclic peak-to-peak 17 strain.
18 MR. SHEWMON: Okay.
19 MR. ENGLISH: We distinguish between cyclic 20 strain and cumulative strain. Cumulative is the rise, 21 and I'll show you that in the next slide.
22 MR. SHEWMON: Okay.
23 MR. ENGLISH: Okay.
24 (Slide]
25 This is the ratcheting strain or the cumulative l
76.
1 lllll 1 strain that we measure on the enterior surface of the 2 component, and these are the number of high input runs 3 that tne component was subjected to, and it shows that 4 the majority of the ratcheting strain occurs in the very 5 first time history input. It also shows the effect of ,
6 pressure increasing, the internal pressure, the cudelative 7 ratcheting strain tends to increase.
8 CHAIRMAN SIESS: Does the end point represent 9 failure?
10 MR. ENGLISH: Yes.
11 MR. SHEWMON: What is the Code allowable pressure
)
12 in this piping?
13 MR. ENGLISH: 17--well, these are two different 14 kinds, two different schedules. This is schedule 40; 15 this is schedule 10. The Code allowable is 1J00 for this 16 schedule 40, and 800 for the schedule 10. ,
17 MR. SHEWMON: Okay, yes.
18 MR. ENGLISH: This is a plot that,shows that 19 as the cyclic strain on the exterior surface of the component 20 increases the damping increases.
21 We are talking, the SSE level D is down in this 22 low region here. These are much higher strains than would 23 be permitted to have at level D. But, also--even though 24 we don't have great deal of data--it appears that the 25 damping--f or a given strair., the damping is greater in i
77.
llllh 1 these thicker pipes than in the thin scheduled 10 pipe, a
2 which implies that ma*/be more energy is being absorbed 3 in the thick-walled pipe for a given surlace strain than 4 in a thin-walled pipe.
5 Now, these damping values here are strictly 6 material damping, and you have a lot greater damping in 7 a piping system of insulation, gaps, sliding, friction.
8 [ Slide]
9 This is a typical hysteresis loop that in this 10 case just shows moment versus displacement for one of 11 the components, and you can see that the curve flattens 12 out which would tend to indicate that you have reached 13 scme kind of limit moment, but the curve reverses before 14 the displacement can actually cause physical collapse 15 of the component.
16 CHAIRMAN SIESS: What is that vertical line 17 over on the right?
18 MR. ENGLISH: It bas no significance.
19 MR. RODABAUGH: That's their plotter.
20 MR. ENGLISH: That's the plotter, right.
21 CHAIRMAN SIESS: That's the plotter, okay.
22 MR. ENGLISH: Yes.
23 CHAIRMAN SIESS: And, they forgot to put tne 24 zero in--
25 MR. ENGLISH: Okay.
r ,
J
.'s >\
. ^i 8 .
E '
s
.x ,I
, 1 <,.,
g[ 1 [ Slide] \
2 A ,Now the nuxt couple of slides show the effect i 3 of the--or the comparison of the linear elas;1, analysis
\
4 that we do with two percent, five percent., damping-*
\ ,
S CHAIRMAN D ESS: Excuse me. 1g
/
5^
is 6 MR. ENGLISH: --yes. i
/
s 7 CHAIRMAN SIESS: Could you go naak v.o that previous 8 clide for just a minute.' a s 9 MR. ENGLISH: Yes.
10 CHA'. AMAN U.ESS: That icithe van:tbat I was
\ 11 looking for.
Y 12 MR. ENGLISH: Okayi
, 13 CHAIRMAN SiFSS: Where does it start?
14 MR. ENGLISH: Somewhere down in here. I am 15 not sure that it is very clear.
16 MR. RODABAUGH: I think it is a little bit above 17 the and of that diagonal line, right acrcss :ran zero.
18 You had yot . pencil almost on it.
19 MR. ENGLISH: Down in hurei 10 MR. RODABAUGH: Down a little bit.
2.1 MR. ENGLISH: !fGe .
22 MR. RODABAUGH: Now, up that diagcnal, there.
23 CHAIRMAN SIESC: What about that little peculiar--
24 MR. ENGLISH: Well, it strain hardens as you 25 go along.
l 4
79.
llllh 1 CHAIRMAN SIESS: What is that little peculiar--
2 MR. ENGLISH: This was one of the better looking 3 ones. The rest of the looked real strange. I don't think 4 we can--
5 CHAIRMAN SIESS: I put the rero axis on it.
6 I couldn't figure out which one it started at.
7 MR. ENGLISH: Yes.
8 CHAIRMAN SIESS: The first small loop is practically 9 all over in the upper Icft quadrant.
10 Okay, go ahead.
11 MR. ENGLISH: All right.
12 (Slide) 13 As Sam indicated earlier, the clastic analysis 14 can be non-conservative at ratios of the peak of the seismic 15 input to the natural frequency of plus of one, and so 16 we investigated that and at some of the component tests 17 found that for two percent, five percent, damping with 18 peak broadening we were able to conservatively predict 19 the moment, compared to the measurements, so I think the 20 peak broadening used in the calculations probably insures 21 that the clastic calculations that we've done on these 22 component tests are conservative.
23 CHAIRMAN SIESS: No, I'm sorry.
24 This is peak?
25 MR. ENGLISH: This is the peak of the input--
80.
lll 1 the frequency of the peak of the input divided by the 2 natural frequency of the component.
3 CHAIRMAN SIESS: Okay, I see.
4 MR. ENGLISH: And, this is not linear. These 5 are just three data points.
6 CHAIRMAN SIESS: I see.
7 MR. ENGLISH: Probably could be better represented 8 with a bar chart.
9 (Slide) 10 This is a similar calculation showing the displacement 11 versus the frequency ratio.
12 (Slide) 13 From the component tests we--
14 MR. RODABAUGH: I would like to make a point, 15 I think for the benefit of--
16 CHAIRMAN SIESS: Speak a little louder, will 17 you please.
18 MR. RODABAUGH: --some of the other committee 19 members.
20 In some of your other tests, don't you have 21 measurements that would not be conservatively predicted 22 by two percent damping, or two percent broadening--displacement 23 is what I am thinking about.
24 MR. ENGLISH: Yes.
25 The displacement is the one that--the moments
81.
g 1 are--
2 MR. RODABAUGH: The moments are okay.
3 MR. ENGLISH: --always conservative, yes, but 4 some of the displacements were a little bit screwy at 5 times.
6 MR. RODABAUGH: Well, I think that is an important 7 point.
8 I didn't see it on the graph here, is what I 9 am saying.
10 MR. ENGLISH: I had to cut down the presentation 11 to something within the time constraints.
12 MR. RODABAUGH: Yes.
13 MR. SHEWMON: Well, is your point, Ev, that h 14 the designer also specifies the maximum displacement, 15 and that this is used and important?
16 MR. RODABAUGH: The displacement could be important 17 in the sense that--as we were discussing yesterday--you 18 have a motor operated valve with some cable, a certain 19 amount of slack, now as you took off snubbers, for example, 20 the displacement would increase. You would like to know 21 what that displacement is so that you can look at your l
22 cable, electrical cable, and see whether it has got enough 23 slack in it.
24 MR. SHEWMON: Okay.
25 MR. RODABAUGH: And, there are many other examples l
l l
82.
ll 1 of that type of thing--
2 MR. ENGLISH: Well, as Sam pointed out that 3 is an area of non-conservatism down there and some of 4 the calculations showed that we were okay, and others 5 it was questionable.
6 CHAIRMAN SIESS: Fut that last figure back on.
7 MR. ENGLISH: Okay.
8 CHAIRMAN SIESS: It is interesting, and I don't 9 quite understand it.
10 This is the ratio of the applied frequency to 11 the natural frequency--
12 MR. ENGLISH: Yes.
13 CHAIRMAN SIESS: --and that I understand, and 14 that is the displacement.
15 MR. ENGLISH: Yes.
16 CHAIRMAN SIESS: Now, one of those curves is 17 the measured displacement?
18 MR. ENGLISH: Right.
19 CHAIRMAN SIESS: And, then the others are the 20 computed displacements--
21 MR. ENGLISH: At different dampings.
22 CHAIRMAN SIESS: --elastic analysis, single 23 degree of freedom--
24 MR. ENGLISH: Yes, that is it.
25 CHAIRMAN SIESS: --at different damping, okay.
83.
llll) 1 Now, I understand, thank you.
2 MR. ENGLISH: Okay.
3 We are not implying that we are--that clastic 4 analysis is good across the board. As Sam pointed out, 5 that's an area of concern which we are still trying to 6 decide whether to use peak broadening, or what is supposed 7 to be done. In these couple of cases, it looked like 8 we had done a pretty good job. We haven't looked at every 9 singic case by any means.
10 [ Slide) 11 So the observations from the component tests 12 are that the dynamic load reversal, we believe, is what 13 prevents collapse. The seismic loads behave more like h 14 secondary than primary, and the ratchet failure loads 15 are much greater than the SSE. The ratcheting doesn't 16 impair functionality, the diameters tend to increase, 17 rather than decrease.
18 The damping for large dynamic loads, bigger 19 than the SSE is certainly greater than the reg guide would 20 permit. The amplified high frequency SRV loads cause 21 negligible response to the component.
22 So, the bottom line we were trying to show from 23 these component tests was that failures were not collapse 24 type failures as the current Code indicates they might be, 25 but rather fatigue and fatigue ratcheting types of failures.
l 1
84.
1 (SlideJ 2 The system tests main objectives were more confirmatory 3 to confirm that the failure mode was was as observed on 4 the component tests, using either three or four sleds, 5 and confirmed the effects of low- and mid-frequency loadings.
6 To determine a system damping for a number of different 7 kinds of configurations we looked at systems with balanced 8 stress high stress everywhere, unbalanced stress, different 9 time histories inputs at different sleds, with or without 10 snubbers and struts.
11 We also wanted to confirm that the functionality 12 was not violated, compromised, and conformed the design 13 rules and margins that we'd observed--
14 CHAIRMAN SIESS: Now, when you say "functionality" 15 you are limiting yourself to the pipe?
16 MR. ENGLISH: Yes.
17 CHAIRMAN SIESS: And, not the valve operators.
18 MR. ENGLISH: Yes, yes, that is right.
19 CHAIRMAN SIESS: Okay.
20 MR. ENGLISH: Okay.
21 [ Slide) 22 The major focus on this program has been piping, 23 and it has been indicated that it would be nice to look 24 at supports, too, but--
l 25 CHAIRMAN SIESS: And, the valves were there l
85c lllh 1 simply to provide loadings.
2 MR. ENGLISH: And, to get a little bit more 3 information--it was kind of a, you know, piggy back type 4 test.
5 CHAIRMAN SIESS: Right.
6 MR. GUZY: I think we should point out in that 7 system 1 test, there was a valve that we operated--
8 CHAIRMAN SIESS: Yes, yes.
O MR. GUZY: --and we do have limited information 10 on that.
11 CHAIRMAN SIESS: That was the piggy back.
12 MR. GUZY: Right.
13 MR. ENGLISH: So, you have seen this. There 14 is an operational hanger and an operational valve. This 15 pressure vessel has a vesselette that simulates--
16 CHAIRMAN SIESS: We saw that.
17 MR. ENGLISH: --and it has an R over T ratio 18 that is typical of a reactive pressure vessel, even though 19 it is very small.
20 (Slide]
21 I think you have seen this stress summary--
22 CHAIRMAN SIESS: No, we haven't.
23 MR. ENGLISH: The only thing I wanted to--
24 CHAIRMAN SIESS: No, we haven't seen that.
25 MR. ENGLISH: --oh, you haven't seen this?
86.
lllh 1 Okay, this was a pre-test calculation, mainly 2 to try to identify the location of where a failure would 3 likely occur, and to make sure that the stresses in this 4 particular system are relatively high, and relatively 5 uniform throughout the system.
6 This was a high-stress system. Stresses were 7 calculated using ASME Code techniques, and we identified 8 this short radius elbow as the highest stress location.
9 The 1 means the highest stress; 2 means the next highest 10 stress location.
11 So, the Code calculation, in fact, was successful 12 in predicting the failure location for this particular 13 test.
14 MR. BUSH: Bill, can I ask you a question?
'S MR. ENGLISH: Yes.
16 MR. BUSH: It is kind of a follow-up on something 17 that I said earlier.
18 Now, in this instance you had a--I think this 19 is the one where they had a valve--
20 MR. ENGLISH: Yes, correct.
21 MR. BUSH: --okay, and on the basis of either 22 a pump or a valve on the inherent thickness--or inherent 23 stiffness, it is necessary to provide the function. In 24 other words, ;ou can't have the body of the pump weaving 25 all around or it won't pump water, and the valve the same
87.
lllll 1 thing is true, so as the result they are much, much thicker 2 than would be calculated on the basis of code allowable.
3 It would seem to me that on a simple analytic 4 basis you could pretty well establish that the body of 5 a pump or of a valve, whether cast or wrought--and most 6 of them are cast, and whether stainless or forritic--
7 probably is not a factor--
8 MR. ENGLISH: Because the stress is so much 9 lower.
10 MR. BUSH: --because the stresses are so low 11 that you can basically--it is just like a rigid object 12 sitting there, and you could probably, in a relatively 13 straightforward fashion dismiss them, which would remove I 14 about 95 percent of my concern with regard to the system, 15 because I worry if you kind of ignore pumps and valves, 16 because there an awful lot of valves in this pipeline system.
17 MR. ENGLISH: I might point out that this valve 18 did not lose pressure integrity through the whole test, 19 and it was identified as a high stress location. And, l
20 in any event it saw G loadings well in excess of what l
21 its rating is.
l 22 [ Slide]
23 These are the various runs that I am sure that 1
! 24 Spence discussed with you yesterday. We looked at uniform l 25 input, and we looked at independent support motion input.
l
88.
1 (Slide) 2 These are the tables, all the way up to full 3 table capacity.
4 (Slide]
5 And, I think you probably saw the fail component, 6 the crack initiated in the elbow bent center, and propagated 7 around the elbow, which is where the ASME Code calculations 8 had predicted it would occur.
9 (Slide) 10 This is a summary of system test 1 at that short 11 radius elbow. You can see, depending upon whether you 12 assume two percent damping or five percent damping, the 13 stress was a large number times the total allowable fault 14 conditions. The stress at the SSE input was about half 15 a level D limit. There was no ratchet strain at that 16 level. At three to five times the level D limit we only 17 had a quarter percent ratchet strain, and no ratchet displacement.
18 The staff was questioning us back in September about how 19 much ratchet displacement in the piping system as the 20 result of these large dynamic loads, and it wasn't until 21 we got up to very large loads, up to half the table capacity, 22 that we got any significant ratchet displacement in the 23 piping system.
24 CHAIRMAN SIESS: Why was the ratchet strain 25 larger at half than full?
89.
llllh 1 MR. ENGLISH: This is the strain during that 2 run, and the component failed early.
3 MR. SHEWMON: What are the units on these first 4 set of numbers up there?
5 MR. ENGLISH: These?
6 MR. SHEWMON: Yes.
7 MR. ENGLISH: They are non-dimensional, just 8 a multiple times level D.
9 CHAIRMAN SIESS: So it is a multiple of level 10 D, okay.
11 MR. ENGLISH: Yes.
12 CHAIRMAN SIESS: So, the SSE was 8<10ths?
13 MR. ENGLISH: Yes, 8/10ths of level D, right.
14 MR. SHEWMON: Now, then I guess I am--the inputs 15 known, and if you assume a five percent damping, then 16 it is 24 X D--
17 MR. ENGLISH: Right.
18 MR. SHEWMON: --and if you assume at two percent, 19 it is 42--
20 MR. ENGLISH: Right, yes.
21 MR. SHEWMON: --okay, I see.
22 MR. ENGLISH: It is how you calculate it, really 23 determines what the ratio is.
24 I think the other significant thing here is 25 that at these relatively low stress levels the damping
90.
llllh 1 wasn't any higher than the N-411 would permit you to use 2 today, even though, again, this is material damping, and 3 damping at the plant would be considerably higher than 4 that.
5 CHAIRMAN SIESS: What would you call relatively 6 low? The SSE? Well, even the five SSE--
7 MR. ENGLISH: Yes.
8 CHAIRMAN SIESS: Yes.
9 MR. ENGLISH: These two.
10 CHAIRMAN SIESS: Yes.
11 MR. ENGLISH: This is not relatively, it is 12 relatively low compared to the full table.
13 I think the one message from these system tests 14 that it just takes one hell-ef-a-lot of load, seismic 15 load, to break a piping system, or even to damage it.
16 (Slide) 17 I did have a couple of slides here that again 18 Spence may have shown you yesterday. This is the strain 19 gauge on the elbow that failed, and it shows that the 20 half table run, the ratcheting that occurs, and in fact 21 the ratcheting continued on up after the sensor departed 22 the scene.
23 (Slidej 24 And, this is the accompanying ratchet displacement 25 at the top of the piping system. You can see these large
91.
lllll 1 swings of 14 inches, or so, peak to peak, the top of the 2 piping system, and when it finally came to rest after 3 a half-table input, the final displacement was only about 4 an inch from where it started initially, so that ratchet 5 displacement of these piping systems doesn't seem to be 6 of concern.
7 The zero is right here in the middle.
8 CHAIRMAN SIESS: Computers are funny that way.
9 MR. RODABAUGH: That one ratcheting figure here 10 that you showed us, this is gauge failure?
11 MR. ENGLISH: Yes.
12 MR. RODABAUGH: Then the sensor apparently kept 13 saying it was a high strain, even though it failed?
k 14 MR. ENGLISH: That is what it looked like.
15 MR. RODABAUGH: Yes, okay.
16 CHAIRMAN SIESS: At least it shows a high resistance.
17 (Slide) 18 MR. ENGLISH: This is again an attempt to show 19 that the linear clastic analysis that we do in piping 20 gives conservative results, especially conservative at 21 the high input levels. We calculated the moments. The 22 moments are what we use to calculate the allowable stresses--
23 or calculate the stresses in the piping system.
24 Once again, as Ev has pointed out earlier, the 25 displacement calculations don't have that much conservatism
92.
llllh 1 in them, but they are a pretty good approximation of the 2 displacement at low inputs, and somewhat conservative 3 at the high inputs--at least with the spector that we 4 use.
5 MR. SHEWMON: Now, is moment, the calculated 6 moments that you have here is something you get from a 7 stress--a strain you would get out of a strain gauge, 8 or what?
9 MR. ENGLISH: These moments are calculated--
10 yes, they are.
11 MR. SHEWMON: Okay.
12 MR. ENGLISH: They are an adjacent part of the 13 component that remains clastic. We had a big thick component 14 adjacent to the elbow that stays elastic.
15 MR. SHEWMON: I see.
16 MR. ENGLISH: Okay.
17 (Slide) 18 A system two was a system of unbalanced stress, 19 and it used a fabricated nozzle rather than a forged nozzle, 20 and it had a snubber, and all of the valves in this system 21 are simulated there, they are no operational valves.
22 Four sleds were used in system 2, rather than 23 three, and we used stainless steel as the material. I 24 don't know whether they mentioned to you yesterday that 25 we were attempting to simulate 316 nuclear grade material,
93.
lllh 1 and we did that by selecting 316 L with 316 mechanical properties 2 which is the least expensive way to try to simulate 316 3 nuclear grade.
4 MR. BUSH: (out of hearing range]
5 MR. ENGLISH: Well, it was low carbon and high 6 strength. It was hand picked for the components.
7 MR. BUSH: yes, but did you use ELC grade? Or 8 did you use the 316?
9 MR. ENGLISH: The 316 L, hand selected--
10 MR. BUSH: Okay, I misunderstood, so what you 11 really had was what is in the upper part, the right hand, 12 so you are closer to what you'd expect to get--
13 MR. ENGLISH: Right.
14 MR. BUSH: --in the increased strength because 15 of that.
16 MR. ENGLISH: yes.
17 (Slide) 18 This is a stress summary for a system 2, and 19 you can see this nozzle was the area of high stress. It 20 was a factor of two higher than anyplace else in the piping 21 system, and again the ASME Code calculations correctly 22 predicted the location of the failure in this system.
23 Again then--
24 MR. BUSH: And, that no::le must have been a 25 fairly stiff one so that the load was pretty much translated
94.
llllh 1 to the interface between the nozzle and the vessel?
2 MR. ENGLISH: yes, yes.
3 (Slide) 4 These are the inputs that are essentially the 5 same as we used in system test 1, with the exception of 6 the sine sweep input, and it turns out that the fatigue 7 usage introduced inco the nozzle was significantly increased 8 by this sine sweep input, as compared to even the full 9 table input. Sinusodial input has a much more marked 10 effect on the tee uses than the random nature of the seismic 11 inputs.
12 CHAIRMAN SIESS: you get more maximum cycles.
13 MR. ENGLISH: And, in fact, I am sure they indicated 14 to you yesterday, they started to see surfacc cracking 15 right after the sine sweep input.
16 (Slide) 17 And, this is just to show you that the cracks 18 actually occurred on the side of the fabricated nozzle 19 out at this--adjacent to the well, the vessel interface.
20 CHAIRMAN SIESS: What was ISM correleted?
21 MR. ENGLIiH: That is--
22 CHAIRMAN SIESS: To the input side?
23 MR. ENGLISH: --independent, oh, okay.
24 MR. ENGLISH: Inputs in the different sleds, 25 and we had an inface, outface.
95c 1 CHAIRMAN SIESS: Okay.
2 MR. ENGLISH: Again, I wanted to show you this 3 relationship.
4 (Slide) 5 The relationship between the local ratchet strain 6 and the high stress component and the net residual displacement 7 in the piping system, such that even though we get a large 8 residual, local strain, ratcheting strain on the component, 9 the piping system itself comes back to its initial starting 10 position in this particular case.
11 CHAIRMAN SIESS: Go back one.
12 [ Slide) 13 MR. ENGLISH: This is the ratchet strain in 14 that nozzle that failed.
15 MR. ENGLISH: Okay.
16 Now, what is the big step in there?
17 MR. ENGLISH: You have this high displacement, 18 you immediately get large ratcheting, local ratchet strain, 19 but that doesn't translate into a large displacement of 20 the piping system. The piping system, even though it 21 is vibrating, it could be 16 inches peak to peak, it still 22 comes back to rest because the load is redistributed and 23 there is no permanent deformation of the piping system.
24 CHAIRMAN SIESS: Okay.
25 MR. ENGLISH: Okay.
96.
1 (Slide]
2 This again is an attempt to show that lineer 3 elastic analysis is very conservative at the high load 4 level and reasonably good at the low load level, again 5 with peak broaden in all of these calculations. That 6 is a displacement in the moment calculations.
7 (Slide]
8 So the observations from the system test, I 9 think, could be summarized as followed: We confirmed 10 that fatigue ratcheting again was the failure mode, as 11 we expected. Failure loads are much greater than the 12 SSE. The functionality was not impaired. Damping was 13 very large, much greater than R.G. at these levels that 14 we obtained failure.
15 MR. SHEWMON: Now is the Reg Guide the same 16 as what was called the Code case N-411?
17 MR. i:NGLISH: No, that is five percent damping.
18 They are a little different.
19 MR. SHEWMON: Oh, okay.
20 MR. ENGLISH: The damping is also bigger than 21 the Reg Guide--
22 MR. SHEWMON: Yes.
23 MR. ENGLISH: --also than the Code case--
24 MR. SHEWMON: I was just wondering if the Reg.
25 Guide had caught up with the Code case yet?
I
97c llllh 1 MR. ENGLISH: It is catching up, but it is not 2 there yet.
3 And, the amplified safety relief valvo loads 4 are very small. When we filtered out that small seven 5 hertz peak we got practically no response.
6 I think, again, the bottom line is that piping 7 is the result of--in socing these tests--just extremely 8 difficult to fail, and I am sure you all realize by now 9 that the building will fall long before the piping system 10 is going to fail, 11 MR. SHEWMON: I am not so sure of that.
12 MR. BUSH: Well, the great Alaska earthquake 13 pictures.
14 CHAIRMAN SIESS: Well, I'll admit, you found 15 more buildings thac fall down during earthquakes than 16 you ever found piping that failed.
17 MR. ENGLISH: Yes.
18 (Slide) 19 I've separated out the water hammer testing 20 because we've come to somewhat different conclusions from 21 the water hammer testing than from the seismic and low 22 frequency testing.
23 We have two component tests that are really 24 two small loops, about 50 long, six inch in diameter, 25 carbon steel and they have been tested with and without
98.
I supports.
2 CHAIRMAN SIESS: Well, we saw both of those.
3 MR. ENGLISH: You saw both of those yesterday?
4 The mini-system test is considerably longer 5 and it has supports, branches, vessels, thin pipes.
6 (SlideJ 7 And, these are different kinds of loads that 8 we have conducted--Anco has conducted--simulated steam 9 hammer, hard system acoustic test, the water slug, and 10 various gravity pressures.
11 MR. BUSH: In the steam hammer, you are--
12 CHAIRMAN SIESS: Little louder, Spence, we can't 13 quite hear you.
14 MR. ENGLISH: Steam hammer is more like the hard 15 systems test, you have a pipe that is just fu'.1 of air 16 and you pressuri::e it up stream at 1000, 2000 psi, compress 17 the diaphragm and watch the reflective wave in air, rather 18 than in water. It is not steam, but it is a gas rather 19 than a liquid.
20 MR. BUSH: And, the loads frem that steam nammer?
21 MR. ENGLISH. The lowest of the three categories 22 of loading.
23 (Slidej 24 Now, you stop me if you have seen all of this.
25 CHAIRMAN SIESS: Well, we've seen the next two
990
% 1 slides, yes.
2 MR. ENGLISH: Okay.
3 (Slide]
4 okay, these are the preliminary observations 5 from the water hammer test, and I say they are preliminary 6 because the second mini-systems test was just finished 7 yesterday, or the day before.
8 But, at any rate, as the result of what we know 9 already, the water slug definitely the primary type loading 10 that can cause collapse of piping systems, and we would 11 probably say in equation 9, and in fact maybe all of these 12 water hammer loads should be in equation 9 with some kind 13 of a relief on the allowable.
14 MR. SHEWMON: Equation 9?
15 MR. ENGLISH: It is a piping design, sort of 16 a fundamental piping design equation which is based on 17 collapse of the component.
18 MR. RODABAUGH: I will write you later, but 19 I will reserve judgment on that, because I am not so sure 20 that water hammer is that much different than earthquake.
21 MR. ENGLISH: Good, glad to hear that.
22 Well, I think the steam hammer and the hard 23 test certainly behave more like secondary loading. They 24 didn't collapse the pipe, even though the limit moment 25 was exceeded and so we tend to be more inclined to categorize l
l
100.
% 1 2
those as secondary.
CHAIRMAN SIESS: Which was the loading that 3 picked that pipe up, up against the--
4 MR. ENGLISH: That is the water slug test.
5 CHAIRMAN SIESS: That is the water slug test.
6 MR. ENGLISH: Yes.
7 MR. SHEWMON: I'm interesced to see Everett's 8 letter on the water slug test.
9 MR. RODABAUGH: What I am thinking about, Chet, 10 is remember this is a loop that went all around tnat building, 11 and the end, when you got down to the end was free.
12 MR. ENGLISH: Right.
13 MR. RODABAUGHz tiow, if you compare that with 14 your elbow test, where you had that long, I think, but 15 very, very short length of pipe when compared to this 16 huge loop.
17 MR. ENGLISH: yes.
18 MR. RODABAUGH: I would like to think about 19 it a bit more before I said that there is really that 20 much difference between the two.
21 MR. ENGLISH: All right.
22 MR. BUSH: Are you arguing about the primary 23 versus the secondary, or about the test per se?
24 MR. RODABAUGH: Well, both.
25 I think other people have brought up the point l
101.
gglll 1 that if you had a pipe that long you would surely have 2 a second anchor someplace.
3 MR. ENGLISH: Yes, yes, this was definitely 4 more severe than you would expect to see in a power plant.
5 We have had that comment made by others, that in the power 6 plant the piping terminates at a heat exchanger pump, 7 or vessel, or something. It doesn't just hang out in 8 the breeze like that one did.
9 MR. RODABAUGH: Or a containment--
10 MR. ENGLISH: Or containment, yes.
11 MR. RODABAUGH: --and it breaks there.
12 MR. ENGLISH: Right.
13 (Slide) llllh 14 We also found that supports could tolerate dynamic 15 loads up to ten times more than a rated load, and it looks 16 like, based on the telephone conversations yesterday with 17 Anco, that piping can tolerate transient pressures maybe 18 one to two times the burst pressure without actually rupturing.
19 We had a section of schedule 10 pipe put in 20 the second mini-system test and we were unable to break 21 that with the loads that they tested.
22 I think, at this point, that the basic rule 23 is still to design to avoid water hammer.
24 CHAIRMAN SIESS: When you say two times the 25 burst pressure, that is the static burst pressure?
103c 1 MR. ENGLISH: Yes.
2 CHAIRMAN SIESS: And, the design pressure would 3 be at what ratio of that?
4 MR. ENGLISH: The burst pressures are, I think, 5 around 4700 psi, and design pressure would be 800.
6 CHAIRMAN SIESS: So, there is that kind of a 7 factor in there?
8 MR. ENGLISH: Yes.
9 MR. BUSH: Let me ask an embarrassing question.
10 When you consider a straight section 10, if 11 you would have put a section 10 cibow at that first location, 12 what do you think would havn happened?
13 MR. TAGART: Do you mean schedule 10?
lllh 14 MR. BUSH: I'm sorry, not section, schedule.
15 MR. ENGLISH: I don't think it would have ruptured.
16 MR. BUSH: Well, I am much less optimistic than 17 you are.
18 CHAIRMAN SIESS: Are you making calculations 19 that will enable you to answer that kind of a question?
20 or, is this strictly empirical?
21 MR. ENGLISH: I would say that it is pretty 22 much empirical at this point.
23 (Slide]
24 Incidentally, there is a typo in your handout, 25 this 1 dash was left out, and there was another typo.
1030 ggggg 1 I have forgotten where it is.
2 CHAIRMAN SIESS: What is that?
3 MR. ENGLISH: One to two. We know the pressure 4 is higher than the burst pressure, but we don't know, 5 you know, we are just getting the information over the 6 phone at this point, so perhaps we shouldn't have even 7 put that in there. We haven't seen the data yet.
8 (Slide]
9 And, on this systems test 2, the damping down 10 at the bottom of the page, there are a couple of errors 11 there, particular, the five and a five and a 22, that 12 should be 5.235, 23 (Slido) lllll 14 okay, the specimen test, as I indicated earlier, 15 that the objective was to develop a lab specimen that lo could give us some quantitative evaluations of fatigue 17 ratcheting with mean stress.
18 We also wanted to be able to correlate the behavior 19 specimen with the components, and extrapolate conclusions 20 from the four test materials--the four materials that 21 were tested to other piping materials, and as Roy indicated 22 in one of the progress reports, thst two of the materials 23 are non-strain, softening or hardening. One is strain 24 softening, and one is strain hardening, and two are stable, 25 and we wanted to investigate the fatigue ratcheting effects
104.
at 550, which you could really on do practically with ggll) 1 2 specimen tests.
3 (Slide) 4 This is a matrix of the specimen test--excuse 5 me, I'm sorry.
6 MR. SHEWMON: All those materials that you used 7 for strain hardening in the tensile tests, so when you 8 say they are strain softening, ycu are implying after 9 a certain number of cycles and certain kind of a test.
10 Would you detine that a littic bit better? How 11 may cycles?
12 MR. TAGART: It is a cyclic fatigue test.
13 MR. SHEWMON: Yes.
lllh 14 MR. TAGART: You do a low-cycle fatigue test, 15 and the stress range increase for a given strain range 16 as you progress--
17 MR. SHEWMON: And, this is a distress with--
18 low cycle is anything under 1000 cycles of failure, as 19 I recall.
20 So, you are in that range with a lot of cycles, 21 is that l. t ?
22 MR. TAGART: Well, it varied, depending on what 23 strain range you put on it initially, but they are all 24 characterized--when we say softening we mean that after 25 some initial hardening it begins to lower the stress range
105.
+
gglll 1 with the number of cycles.
2 CHAIRMAN SIESS: Now, these are controlled strain 3 tests? -
4 MR. TAGABTt Yes.
5 CHAIRMAN SIESS: So the stress--
6 MR. ENGLISH: Now there are 140 component tests t
7 here, of which 48 were- excuse me?
8 CHAIRMAN SIESS: Forget it, go ahead.
9 MR. ENGLISH: The majority of the tents were 10 two-bar tests.
11 (Slide) 12 This is a two-bar milimodel where you have two 13 bars with rigid ends, you apply a fixed-mean load and lllh 14 you cycle one of the two bars, let's say by heating, such 15 that that bar is in compression with other hars in tension, 16 and it turns out in this testing program the bar that 17 is being heating is simulated by a computer and the testing 18 is done on a single-uniaxial specimen.
19 Most of the testing in the program, as you can 20 see, is /one using this two-bar technique, and then the 21 verification test dcne with beams and pressuri:cd pipe, 22 to show tnat the two-bar tests are conservative.
23 CHAIRMAN SIESS: Now, the two-bar test are not 24 two bars.
25 MR. ENGLISH: It is just one bar.
106.
1 CHAIRMAN SIESS: I see.
2 MR. ENGLISH: Ic is just this bar with the testing 3 machine, computer controlled to sim*> late the effects of 4 the second bar on this first bar.
5 CHAIRMAN SIESS: What is the ooiect of the two-6 bar test, originally?
7 MR. ENGLISH: It is r. simple way of measuring 8 ratcheting, mean strest.
9 And, thdi two-bar test was verified with--
30 )
CHAIRMN7 SIE.3S: .Getore ycu had computer controlled 11 resting machines?
'/ 12 Mr.m LNGLISH: Yes.
7 13 (S'lide) 14 , The ret;:ar.gle beam specin en looks like this 15 and it is t.vsted with an applied axial load and an alternating i
\
16 bending lodd, at:) ,tl.at fits f.ntcgthis machine, in this 17 rehionhere. these springs apply the static load and then
\
t 18 the full-point bending is done by the actuator here.
19 .
CHAIRMAN SIESS: Is that a strA2n control?
\
s 20, [,)ieneraldiscussionbyseveralspeakers.)
. \
(' x dl ! MR. RODABAUGH: Chet asked the question: is this i
' 22 strain controlled, and he said, "Yes."
\,
3 CHAIRMAN SIFSS: A little louder, Ev, please.
/
24 MR. RODABAUGH: You asked the question: is it
. ( ,
25 strain controlled? And--
l -
i
['" <
N._.)
t <
x
,1
1070 gg g 1 MR. ENGLISH: I guess it--is it load control?
2 MR. RODABAUGH: --no, it is some of both, is 3 the point.
4 MR. RANGANATH: It is initially strain controlled.
5 MR. RODABAUGH: The bending is strain controlled, 6 but see those big springs on the end--
7 MR. RANGANATH: That is prime mean load.
8 MR. RODABAUGH: --yes, it is load controlled.
9 MR. ENGLISH: So, it is attempting to simulate 10 the ratcheting.
11 (Slide) 12 The observations from the specimen tests to 13 date are that the two-bar test is a conservative representation llll 14 of the ratcheting on cyclic life. The beam and pipe specimens 15 confirmed that the two-bar test is conservative. We found 16 that if you have controls on the cumulative ratchet strain, 17 that is if you don't permit significant large ratcheting 18 to occur, you get in fact a fatigue type failure, and 19 that with controls on the ratchet strain: the mean stress, 20 then the temperature doesn't effect the cyclic fatigue 21 life.
22 And, we observed some cyclic creep in these 23 tests that were done at MCL, and the creep occurs in the 24 testing whic is done at about two minutes per cycle.
25 It is a very slow test. We've identified some creep
108.
in these tests, but we don't think it is going to be present gg g 1 2 in the seismic test.
3 CHAIRMAN SIESS: This is creep in the axial 4 direction?
5 MR. ENGLISH: Yes.
6 MR. SHEWMON: Are the basis for these conclusions 7 written up in a paper that has been submitted, yet?
8 MR. ENGLISH: No.
9 MR. TAGART: No, not ye.t.
10 MR. ENGLISH: The whole program will be written 11 up soon.
12 MR. TAGART: This will be one of three papers.
13 MR. SHEWMON: Okay, I would like to see this ll 14 one. i 15 CHAIRMAN SIESS: What rate are these run at?
16 MR. ENGLISH: These are run at a rate of 1000 17 times faster frequency--higher frequency than these tests, 18 so we think the cyclic creep that has been observed in 19 these low frequency tests probably won't be present in 20 any significant degree in the seismic tests.
21 CHAIRMAN SIESS: How many cycles?
22 MR. ENGLISH: Well, these tests are two minutes 23 per cycles, and the seismic tests are like eight cycles 24 per second.
25 CHAIRMAN SIESS: And, now many cycles did you
109.
1 get per failure?
2 MR. ENGLISH: In these tests?
3 CHAIRMAN SIESS: In these specimen tests?
4 MR. ENGLISH: Oh, gosh, it depends on the strain 5 range, but hundreds, thousands.
6 CHAIRMAN SIESS: Hundreds, yes.
7 MR. BUSH: You don't distinguish whether 8 you are talking about--on that last bullet, on room temperature 9 or elevated temperature, or both?
10 MR. ENGLISH: Both.
11 MR BUSH: It is both. I thought it might 12 be, but I wasn't sure.
13 MR. ENGLISH: Okay.
l 14 MR. RODABAUGH: Bill, you have one other type 15 of test, the pipe test, pressurized pipe test.
16 MR. ENGLISH: Pressurized pipe test?
17 MR. RODABAUGH: The last in f our list of specimen 18 tests.
19 MR. ENGLISH: Oh, oh, yes. I didn't show that 20 picture. I don't have a cross-sectional drawing, but 21 it is just a small piece of steel pipe, and it failed 22 at the grips where the pipe is being held by the testing 23 machine.
24 MR. RODABAUGH: I was mentioning the rubberband 25 aspect of it.
110.
gg g 1 MR. ENGLISH: Oh, yes.
2 It turns out in the specimen test, the pressurized 3 pipe test, the rubberbands on the exterior--oh, you have 4 already heard about that?
5 CHAIRMAN SIESS: No, no, go ahead. I would 6 like to hear it again.
7 MR. ENGLISH: I am not sure--
8 MR. RANGANATH: It is on page 374.
9 MR. SHEWMON: On 374?
10 MR. RANGANATH: On 3-74.
11 MR. ENGLISH: It is pretty interesting to see 12 that a rubberband could actually cause deformation in 13 a pressurized pipe.
CHAIRMAN SIESS: Well, is that good or bad?
llklh 14 15 MR. ENGLISH: I'm not sure.
16 (General discussion]
17 MR. BUSH: It is surprising.
18 MR. TAGART: Damping is very difficult to predict 19 exactly, and ratcheting is very difficult to predict exactly.
20 CHAIRMAN SIESS: Does that conclude your presentation?
21 MR. ENGLISH: Yes.
22 CHAIRMAN SIESS: Are there any further questions?
23 (No response.)
24 MR. ENGLISH: Sorry to run over.
25 CHAIRMAN SIESS: No, no, you were right on schedule.
111.
gglll 1 We started late. We are doing real good, unusually good.
2 Okay, Mr. Ranganath.
3 MR. RANGANATH: Thank you.
4 (Slide]
5 I am going to briefly cover potential design 6 rule changes that we may look at as the result of the 7 data that we have generated here.
8 I want to emphasize the word "potential" because 9 we are still evaluating it, and what I am going to talk 10 about, it kind of gives you a flavor of the direction 11 we are going in. We may in fact change some of the exact 12 numbers that we have, either recommendations, that I will 13 explain here.
lllh 14 Now, I am going to talk about three items, kind 15 of like what were the conclusions from these component 16 and system tests, and regards to design rule development?
17 What did we learn from the specimen fatigue and ratcheting 18 tests? And, I also am going to give you a little more 19 of a description of the single degree of freedom model 20 analysis that we did as it relates to the inelastic dynamic 21 response as well as ratcheting.
22 And, from that I will try to make some conclusions 23 relative to the elastic analysis, make some proposals 24 on rule changes, and briefly remark on what the long-25 term goal might be in terms of this proposed design rule i
112.
I changes.
2 CHAIRMAN SIESS: Let me interrupt you for just 3 a minute.
4 MR. RANGANATH: Yes.
5 CHAIRMAN SIESS: Could you at some point give 6 us a little picture of what tne design rules on now, for 7 example, what equation 9 is, and how it relates to equations 8 1 through 8?
9 I am sure that some of the people at the far 10 end of the table are quite familiar with this, and some
- 1. in the middle may be.
12 MR. RANGANATH: Yes, I am going to do that.
13 [ Slide]
lh 14 Now, this has been said many times: the component 15 system test, we just couldn't fail the thing, even though 16 we went to well in excess of what the level D allowables 17 might be, and in fact we could not have any collapse effect, 18 so no limit load--
l 19 CHAIRMAN SIESS: Level D is faulted?
20 MR. RANGANATH: --level D is faulted--
l 21 CHAIRMAN SIESS: That is everything you can 22 think of?
23 MR. RANGANATH: Right, right.
24 CHAIRMAN SIESS: Thrown in there?
25 MR. RANGANATH: Right.
113.
CHAIRMAN SIESS: That is the highest level?
ggllg 1 2 MR. RANGANATH: That is correct.
3 CHAIRMAN SIESS: It doesn't get any worse than 4 D, then?
5 MR. RANGANATH: Yes.
6 CHAIRMAN SIESS: Okay.
7 MR. RANGANATH: We didn't see any limit load 8 failures, and again maybe this is the time to talk about 9 the equation 9. Equation 9 just says that in effect we 10 can call it without confusing the issues. There are some 11 constants, M over Z, less than some allowable, call it 12 here 1.5 S of M.
13 What it is then, this is the pressure stress, llh 14 okay, this is the actual pressure stress, this is the 15 bending moment that includes the seismic weight and so 16 on, and these are generally multiplied by a figure which 17 was determined in large part by some tests done by Markl 18 in the beginning, and EV Rodabaugh himself lead to a lot 19 of these indices. These are called B-2 indices.
20 CHAIRMAN SIESS: Are they factored greater or 21 less than one?
22 MR. RANGANATH: Some are less than one, but 23 most of them are greater than one.
24 CHAIRMAN SIESS: Now, just to let you know what 25 level you are talking to, what is S of M?
114c MR. RANGANATH: S of M is what's called as a gglll 1 2 design stress intensity that established by the ASME Code.
3 It is generally quoted to the lower of two-thirds of the 4 yield strength, or one-third of the ultimate.
5 So, anyway, it says you have a factor of three 6 margin on pressure for ultimate strength.
7 CHAIRMAN SIESS: Well, they call it a stress 8 intensity, and the units are stress?
9 MR. RANGANATH: That's correct. It is nothing 10 to do with the stress intensity factor.
11 CHAIRMAN SIESS: It is a stress?
12 MR. RANGANATH: Right.
13 CHAIRMAN SIESS: It is an inelastic strain.
14 MR. RANGANATH: Right.
15 So, and again these are all done on a pseudo-16 clastic basis so if you look at--
17 CHAIRMAN SIESS: Then 1.5 S of M, might be the 18 yield stress?
19 MR. RANGANATH: For carbon steel it is. For 20 stainless steel it is a little higher than the yield strength, 21 but as a rule it is--
22 MR. RODABAUGH: Since Chet asked about level 23 D, specifically, you have been talking about SSE, why 24 don't you put a 3 out there?
25 MR. RANGANATH: Right, okay.
1150 This is for what is called a design condition, ggll) 1 2 and 3 S of M is what is called as the level D.
3 CHAIRMAN SIESS: Design, in terms of seismic 4 would be OBE?
5 MR. RANGANATH: That is correct.
6 CHAIRMAN SIESS: But, there are some other things 7 also that are in there, design and--
8 MR. RANGANATH: Right, yes, it will be the design 9 pressure plus all of the loads.
10 CHAIRMAN SIESS: Yes.
11 MR. RANGANATH: Now, when we did get failure, 12 failure was predominantly due to a combination of fatigue 13 and ratcheting, so that accounts for emphasizing that llllh 14 we ought to be focusing on those.
15 But, even when we did get failures by fatigue 16 and ratcheting, we were able to have the number of cycles 17 it took to cause failure was well in excess of what you 18 would, for example, expect in one level D event, like 19 many of the tables that Bill showed you had two, three, 20 four times full cycles of this level D transient, so that 21 shows that you have got a lot of margin even there, too.
22 Analysis of the test shows that elastic prediction 23 are generally conservative for response spectrum analysis 24 with peak broadening for up to five percent damping. You 25 know, you recall that Bill showed you this comparison
116.
llll) 1 of displacement and moment, and--
2 CHAIRMAN SIESS: I am not sure that I buy the 3 two or three cycles, the two or three imputs, as well 4 in excess of an SSE.
5 That could be an SSE plus one good after shock, 6 or two afcer shocks, and if I recall we've had earthquakes 7 where the second shock was much greater than the first, 8 and I think there have been three well up in there. Now, 9 there can be considerable time between the first and the 10 second--
11 MR. SHEWMON: A year.
12 CHAIRMAN SIESS: --not this one, as one was 13 in December and one was in January.
14 [ laughter]
15 Now, if we had the first earthquake, obviously 16 the plant would be shut down, but the second one could 17 bust some pipe and they may be required to move the KE 18 except there would be a lot less, but again we frequently 19 get an after shock within a day or two, and sometimes 20 within hours. It usually is smaller than the first one.
21 But, again, I am, you know, we are sort of--
22 MR. RANGANATH: Yes, but the key though is that 23 we didn't get--all of the core calcul.ations are based 24 on primary loads. If you apply the' the same load enough 25 number of cycles, yes we can expect cracking, but the
117.
lllll 1 magnitude of the loads themselves you can tolerate numbers 2 well into excess of--
3 CHAIRMAN SIESS: Well, if you talk about that, 4 yes, the magnitudes.
5 MR. RANGANATH: Yes.
6 [ Slide]
7 You also heard Bill's conclusions relative to 8 the fatigue ratchet cycles, the testing that was done, 9 and ratcheting occurs when you have a combination of primary 10 means stress, and cyclic dynamic stress.
11 And, here are some conclusions that we can get 12 from Dan Miller's original model which was originally 13 intended for thermal stresses. It turns out that these
- h 14 are applicable even for mechanical loading with seismic 15 inputs, but the two-bar and the bent tests showed in addition l 16 to the normal ratcheting predictions that you would get, 1
l 17 some time dependent behavior, which caused us some concern 18 in the beginning when we looked at it. The cycles were, 19 as Bill pointed out, relatively very low frequency when 20 compared to the earthquake's and since then we have done 21 some additional work--Roy Williams has done some additional 22 work at higher frequencies, and what he found was that 23 indeed the extent of the time difference in cyclic deformation 24 was much lower, so that gives us reason to believe that 25 the time different type of behavior was more symptomatic
118.
lllll 1 of the very low frequencies, and is the kind of behavior 2 that you would not expect at high frequencies.
3 MR. RODABAUGH: How high is higher?
4 MR. RANGANATH: He has done this at .5 cpm, 5 and he went to ten times that frequency, which is five 6 cycles per minute, and whereas the real earthquake you 7 wou]d be looking at five to ten hertz.
8 CHAIRMAN SIESS: What are you calling time dependent?
9 What he called creep?
10 MR. RODABAUGH: Sam, I think there is 11 something wrong with your first bullet equation, and your 12 combination of mean stress and cyclic, dynamic stress, 13 your bounds are not complete, but I'll write you a letter 14 about it.
15 MR. RANGANATH: We can talk further about that, 16 okay?
17 MR. RODABAUGH: Okay.
18 MR. RANGANATH: All right.
19 (Slide) 20 Failure is either by fatigue or excessive ratcheting.
21 He observed some situations--
l 22 CHAIRMAN SIESS: You are still talking about t
23 the specimen tests?
! 24 MR. RANGANATH: --beg you pardon?
25 CHAIRMAN SIESS: You are still talking about l
L
119.
I the specimen tests?
2 MR. RANGANATH: I am still talking about the 3 specimen test, where he had, after he had enough number 4 of ratchet cycles, he found that in fact he could neck 5 this specimen, even though he was doing fatigue testing, 6 when he had enough ratcheting accumulated the failures 7 became more like a rupture failure in a tension test, 8 and so he had kinds of failure.
9 One, was where the classic symptoms of some 10 nice, fine, smooth, shiney surface of fatigue failure, 11 whereas the others were loss of ductility, rupture type 12 of failures, okay.
13 So, what he found was when he excluded all of 14 the data points where the failures occurred by necking 15 and loss of ductility, he found that the data points fell 16 very nicely on fatigue curve that is close to the mean 17 data curve that is used in the ASME Code.
18 So, what it is saying is that as long as you 19 don't allow your ratchet strain to get out of control, 20 whereby might get failure by necking and excessive deformation, 21 the ratcheting in itself does not have substantial impact 22 on fatigue cracking.
23 CHAIRMAN SIESS: The ASME curve is a high cycle 24 fatigue?
25 MR. RANGANATH: It is a combination of a low l
l
120c gglll 1 and high cycle, but they were all done with zero means 2 stress and certainly did not have ratcheting.
3 So, this is an important conclusion because 4 what it is saying is, well, ratchet is something to be 5 worried about, as long as you have reasonable confidence 6 that the total ratchet strain is small, you can still 7 use the fatigue rules for prediction.
8 (Slide]
9 I am going to very quickly go over the single 10 degree of freedom model analysis.
11 [ Slide]
12 Some of it Sam has already covered, so I really 13 don't have to go into in any detail. This is the standard lllh 14 single degree of freedom system, and we put the structural 15 damping here, and if you were looking at elastic behavior, 16 the kind of response spectrum that most of you--all of 17 you are probably aware of--is what you'd get here.
18 CHAIRMAN SIESS: That is the one that he had 19 to cut off down there by five.
20 MR. RANGANATH: That's correct, right.
21 Now, what we then did was we said the components 22 are well described by single degree of freedom system, 23 so what would happen if you do elastic plastic analysis 24 and here is the elastic plastic analysis.
25 (Slide]
121c 1 This is where we now did an analysis based on 2 the elastic perfectly plastic behavior. This is the--
3 Sam Tagart described it, it is his modhl--and essentially 4 what we did was we came up with an effective stiffness 5 and an effective damping based on a hysteresis slope like 6 this, and we performed the evaluation of the response 7 of the sy stem in an elastic manner.
8 To compliment that, we also did an elastic plastic 9 model. This was a numerical model where we modeled the 10 material by a bilinear spring, with two slopes. We assumed 11 kinematic hardening, and we also later on applied a static 12 force, in o; der to simulate what would happen in a ratchet 13 type of condition, so this is what you would call as a--
14 with the limitation of bilinear stress strain curve, and 15 the assumption of kinen;atic hardening what you call as 16 an exact solution to the problem.
17 CHAIRMAN SIESS: Your "p" is his lambda?
18 MR. RANGANATH: That is correct, that is correct, 19 yes. Lambda is a non-dimensional slope.
20 And again--
21 CHAIRMAN SIESS: Is kinematic hardening, of
~2 what English was talking about as a strain hardening fatigue 23 curve sort of a behavior, or what?
24 MR. RANGANATH: The kinematic hardening is 25 what I mean as the unloading and subsequent loading. It
122.
g 1 operates between these two lines.
2 MR. TAGART: It is the conservative description 3 of hardening when talking about--there are two models 4 talked about--isotropic hardening, and kinematic hardening.
5 Isotropic hardening assumes that the whole--
6 MR. RANGANATH: The isotropic--
7 MR. TAGART: --expanse increases with cycles, 8 and the kinematic ones does not increase with cycles.
9 MR. RANGANATH: The kinematic--
10 MR. SHEWMON: So, there is no work hardening?
11 MR. RANGANATH: --well, there is work hardening, 12 but what it is what you gain on the tensile portion of 13 the curve, you lose--in effect, it accommodates the--
It is strain hardening, but it l
llhll 14 MR. SHEWMON:
l 15 doesn't change with number cycles.
16 MR. RANGANATH: Right, that is correct, so that l 17 is the assumption that was made in the elastisisa most 18 materials, maybe after about ten cycles the initially l
19 show hardening, where you get increase in both tension 20 and compression, but after a few cycles you don't get 21 this continuous increase, and this is what you would have.
22 [ Slide) 23 So, again then to very quickly go over--Sam 24 already went through this thing. What it is is here we 25 are plotting the relative amplification. In non-dimensional
123.
gglll 1 terms this could be viewed as a strain in the spring, 2 and this is dimensional frequency, and one thing we see 3 very clearly is if you account for clastic 1 tstic behavior, 4 there is substantial reduction in the amplification. The 5 peak strains are significantly lower.
6 And, what you do see, though, is the sift in 7 the peak and that goes with the fact that the slope corresponding 8 to plastic behavior, the effect of stiffness, is somewhat 9 lower so you can see that there is a shifting to the left.
10 And, depending upon the extent of deformation 11 that you allow, or in other words the ductility that you 12 can tolerate, you see that if you go to higher ductility, 13 in fact, the peaks are lower, but there is a region in l 14 the response spectrum where the elastic analysis can be 15 somewhat non-conservative, and this is something we have 16 talked about earlier.
17 CHAIRMAN SIESS: What is that less than one 18 by clastic?
19 MR. RANGANATH: Well, what he is saying is he 20 is non-dimensionalizing this with the--
21 CHAIRMAN SIESS: Oh, okay, I see.
22 MR. RANGANATH: --and he is saying that is less 23 than one.
24 CHAIRMAN SIESS: Right.
25 MR. RANGANATH: So, that kind of gave us a good l
124.
understanding, Sam's model couple with an exact analysis, gg 1 2 and now we feel like we understands what happens in terms 3 of the elastic behavior, inherently the higher the strain 4 the more the damping, and therefore it acts like a self-5 limiting type of behavior.
6 Now, the second thing we did was to see if we 7 could predict what would happen in terms of ratcheting.
8 [ Slide]
9 You can see we used the same model with the 10 clastic plastic behavior, you know, and in relation to 11 that we applied a static force to simulate the mean stress,.
12 if you will, say due to pressure or dead weight in the i 13 system.
14 Again, we found, for example this is typical lh 15 response here, this is the support motion. This F = .5 16 means that we applied a force equal to one half of the 17 yield force on the spring, and L = .1 meaning the slope 18 of the plastic portion is one-tenth the slope of the elastic 19 portion.
20 So, with that, you can see the strain--dimensional 21 strain, in fact, goes up very rapidly, and after awhile 22 you end up just cycling once you obtain a mean value.
23 Now, if you look a the same thing--
24 CHAIRMAN SIESS: What was the mean value here?
25 MR. RANGANATH: This is what--this is steady
125.
1 ratchet strain.
2 CHAIRMAN SIESS: No, you said you applied a 3 mean stress.
4 MR. RANGANATH: The mean stress was one-half 5 of the yield.
6 CHAIRMAN SIESS: And, so this went into yield 7 in both directions when you applied the cyclic?
8 MR. RANGANATH: Yes, I think it will show you 9 here.
10 (Slide) 11 CHAIRMAN SIESS: Okay, sure.
12 MR. RANGANATH: This is che--this shows, for 13 example, it is kind of a--well, we felt like we understood llllh 14 the model much better once we started looking at it from 15 the simple spring mass system.
16 What it is, is the mean stress act like an asymmetric 17 on the stress strain hysteresis load, so all you do, as 18 long as you shift the hystaresis load so that it can support 19 this mean stress, and now you have a symmetrically heavy 20 load over the--so this hysteresis load in fact moves up 21 along the cyclic stress change--
22 CHAIRMAN SIESS: You are only yielding in one 23 direction, now?
24 MR. RANGANATH: There is yielding in the reverse 25 direction, also, you can see.
126.
Some of these data points--
(lllh 1 2 CHAIRMAN SIESS: Very little, though.
3 MR. RANGANATH: There is some yielding, yes, 4 but as you can see, more of it, of course, is on this 5 side.
6 CHAIRMAN SIESS: This is dimensionless? What 7 is it? Ratio to yield?
8 MR. RANGANATH: This is dimensionless, right.
9 CHAIRMAN SIESS: But, what is the ratio?
10 MR. RANGANATH: This is ratio to the yield strain.
11 CHAIRMAN SIESS: So, I look at the bottom part 12 of that and I am not reaching yield. It looks more like 13 a--
14 MR. RANGANATH: Well, this would be the elastic 15 unloading and then--
16 CHAIRMAN SIESS: Well, how far does it go? That 17 is what I can't tell.
18 I take one of the early cycles--
19 MR. RANGANATH: Un-huh.
20 CHAIRM>li SIESS: --and when you went up you 21 moved over about a foot on that screen, and when you came 22 down you essentially went very little--
23 MR. RANGANATH: Yes, yes, there is less yielding 24 on the compression side than on the tensile side but there 25 is yielding.
127.
1 CHAIRMAN SIESS: Even though the stress is not ggggg 2 yield?
3 MR. RANGANATH: That is correct. What it is, 4 is you can yield on the compressive stress at a lower 5 stress level because of this Bauchelder [ sic.) effect that 6 we are talking about.
7 CHAIRMAN SIESS: All right, when I get over--
8 MR. RANGANATH: Yes, kinematic.
9 CHAIRMAN SIESS: --to this right-hand side, 10 is that one big loop there.
11 MR. WARD: Trace the loop on the right-hand 12 side.
13 CHAIRMAN SIESS: I couldn't tell whether--
llll 14 MR. RANGANATH: I think it is kind of hard to--
15 I think it is hard because it goes up like this and then 16 back--
17 CHAIRMAN SIESS: Okay.
18 MR. RARGANATH: --and after awhile it just starts--
19 you can see these dark lines are when you no longer had 20 a shift in the hysteresis loop it kept revealing itself, 21 so this is the final hysteresis loop.
22 CHAIRMAN SIESS: The other thing that bothers 23 me is why, if you have an applied mean stress, you still 24 got your line going through the origin--
25 MR. RANGANATH: Now, this is what--here is the l
t
1280 where we--here is the applied--
ggggg 1 2 CHAIRMAN SIESS: Okay, that is your static point--
3 MR. RANGANATH: Right.
4 CHAIRMAN SIESS: --I see.
5 MR. RANGANATH: Right, right.
6 CHAIRMAN SIESS: You got up to there and then 7 put the other ot.e in--
8 MR. RANGANATH: And, now we are cycling this--
9 CHAIRMAN SIESS: And, what is the one that doesn't 10 come down all of the way?
11 MR. RANGANATH: Here?
12 CHAIRMAN SIESS: The first unloading.
13 MR. RANGANATH: Oh, I think that was more--
lllh 14 it is a plot and--we use Lotus to pick up points, so it 15 is an aberration caused by the--we didn't pick up every 16 point because there are so many points to calculate.
17 CHAIRMAN SIESS: That one should come down a 18 little.
19 MR. RANGANATH: Right, that is correct.
20 So what it is showing is, this nice, simple, 21 strain-stress model can in fact--we can understand ratcheting 22 as it happens in a structure when you have primary mean 23 stresses in seismic type of Icading.
24 (Slide]
25 But, we did find that the predicted ratchet--
l
129.
ggflk 1 CHAIRMAN SIESS: Excuse me, excuse me.
2 Go back, because you have got something defined 3 on that previous one. The diagonal line, sloping line, 4 would be E of P.
5 MR. RANGANATH: Right.
6 What we found out was that the--you can make 7 a prediction of the cumulative ratchet strain by just 8 taking your applied mean stress and divide it by the plastic 9 slope of the stress-strain curve.
10 CHAIRMAN SIESS: Is that the slope of the--
11 MR. RANGANATH: Right, this is the plastic slope 12 of the stress-strain curve.
13 CHAIRMAN SIESS: Oh, that is the plastic slope ll 14 of the stress-strain curve.
15 MR. RANGANATH: Right.
16 CHAIRMAN SIESS: It is almost parallel to the 17 envelop.
18 MR. RANGANATH: It is parallel.
19 CHAIRMAN SIESS: It is parallel?
20 MR. RANGANATH: It is parallel.
21 CHAIRMAN SIESS: Okay.
22 MR. RANGANATH: So, we found out this was kind 23 of a phenomenal observation that we could come up with 24 a prediction for the hysteresis and it turned out that 25 the Dan Miller Model which was based on thermal stresses
130.
1 gave us essentially the same kind of prediction.
gg gg 2 CHAIRMAN SIESS: So, if I look at the top line 3 there that would be what I would get in a static monatonic 4 test?
5 MR. RANGANATH: This would be--yes, yes, if 6 you used a, you know, a cyclic stress strain.
7 CHAIRMAN SIESS: Okay.
8 MR. RANGANATH: Okay.
9 (Slide]
10 Now, when we--
11 MR. SHEWMON: You might have to run it through 12 several cycles.
13 MR. RANGANATH: Right, that is correct.
14 You see, because you may have a yield material lh 15 that has somewhat lower yield strain.
16 [ General discussion]
17 Right, but anyway, these are some of the conclusions 18 then from the single degree of freedom system model. We 19 felt that the single degree of freedom system model was 20 somewhat conservative because we applied axial loads of 21 tremendous stress, the real piping is subjected to more 22 lending behavior, and the whole answer is, of course, 23 a strong functional, how you compute this plastic slope 24 in a bilinear model, depending upon which part of the 25 stress-strain you match you can get different answers, 1
1 l
131c but the usefulness of this is in terms of understanding gglll 1 2 what's happening in the ratcheting condition.
3 CHAIRMAN SIESS: Does this thing get very difficult 4 when you take tri-linear or multi-linear curves?
5 MR. RANGANATH: I think--we did all of this 6 on a personal computer, so I think we can--
7 CHAIRMAN SIESS: You have an equation, then 8 you can--
9 MR. RANGANATH: I think so. We can do them 10 much better.
11 And, I said, well, what do we do in terms of 12 predicting ratchet strains for all of the component tests.
13 (Slide) 14 So, what we did was we wanted to plot all of 15 the data that we had where we had ratcheting, and in many 16 of these cases we didn't quite get the ratchet strain 17 but we have enough number of data points that we use to 18 predict the condition under which ratcheting would occur.
19 (Slide]
20 The next plot--it is kind of a busy plot--shows l
i 21 then the ratchet threshold. This was for carbon steel.
22 Let me take a minute to explain it.
l l 23 Here we are plotting the mean stress, that is l 24 the pressure stress, plus the strain times the Young Smartings 25 (sic.), you know, okay, so this is, in a sense, if you l
(
132.
gglll 1 ignore the mean stress, 's almost all the seismic 2 stress, or seismic strain. f you will, and these were 3 several runs, so the axis really doesn't have any meaning.
4 It is more like a bar chart.
5 CHAIRMAN SIESS: These are several different 6 specimens?
7 MR. RANGANATH: Several different tests, or 8 in some, several different runs on one specimen, you know.
9 And, what the engineer who did this analysis 10 did was he looked at the actual records, and find the 11 point when he got ratcheting, so these p2ases are shown 12 just the onset of ratcheting. Below the plus would mean 13 he didn't see any ratchet; above that, you know, he saw 14 some ratchet.
15 Then, the squares are points where--
16 CHAIRMAN SIESS: That is the strain at which 17 he saw?
18 MR. RANGANATH: That is right, that is correct.
19 So, the squares are where the was rateneting.
20 All we wanted to do was, we wanted to find ;ut--not only 21 wasn't it enough to just say no ratcheting, we followed 22 and concluded that we can tolerate some ratchoting, so 23 what we said was--I took arbitrarily ten percent. I concluded l
24 that ten percent was the strain level t as I could live 25 with, for example, in a level D faulted effect. For a
1330 lllll 1 level B normal operation I may settle for this one, ' hat 2 is, I don't want to see ratcheting under normal conditions--
3 CHAIRMAN SIESS: Ten percent of what?
4 MR. RANGANATH: Ten percent strain.
5 CHAIRMAN SIESS: Oh, ten percent strain?
6 MR. RANGANATH: Ten percent strain--ratchet strain.
7 CHAIRMAN SIESS: Oh, I pee.
8 MR. RANGANATH: So, this would be a faulted 9 condition. The idea being if you have a faulted event, 10 a one-time occurrence, then it is okay to have some ratcheted 11 strain.
12 CHAIRMAN SIESS: And, the pluses are much, much 13 less than ten percent?
14 MR. RANGANATH: The pluses, we could not get 15 any--just the onset of ratcheting.
16 ,
The open squares are less than ten percent, 17 and the solid squares are greater than ten percent--again, 18 ten percent is somewhat of an arbitrary number that I 19 just use for the purpose of the discussion here.
20 And, I concluded that the ten percent I could 21 live with it, if you had a one-time event.
22 And what the cycles show is they tell you how l
23 much ratchet increment occu~. ocr cycle. I made the assumption ,
24 that in a typical SSE cvent a very conservative number 25 is to say there are 50 cycles, so I said 10 cycles of
134.
s I the highest--ten. seconds at the highest stress _ strain
'; 2 and 500--sa I came up with 50 cycles a:; the maximum number 3 of peaks cycles that you could get. I think it is a conservative 4 numoer, but, that is what I use in making this. plot.
5 So, then this told me that if ! wanted to look
\
o 6 at condition--first of all, we didn't get any limit load 7 failures, so all we have concern about is ratcheting, 8 and if iou allow your stress to be below 2 Sm Y for 9 example, then you would not get ratcheting, say for level 10 b type of conditions; and if you allow your' stress to 11 be less than say 4 sm Y then you would gat some ratcheting,
.\
' 12 but the tatchet cf strain would not be latte enough to 13 causa any concerns on structural integrity due to the 14 iaulted evont.
15 And, again, the numbe: n 2 and 4 may in f act 16 change over tim 9, but this is the concept that I am proposing.
17 1HAIRMAN SIESS: Now, I don't recall having 18 a*.en :he number of cyclos in the componcnr tests. It 19 is always giving the nwnber of inputs. Did any of the 20 componerce f ail at less tha' 50 cycles?
21 MR. E.'AGART: Well, they are different amplitudes.
22 I think 50 is a very largs number for any of those component 23 tests, 24 I would say that probably it is 50 is the upper 25 limit of what we've seen in the component tests. We *.cok 4 g
135.
five component tests and said you had ten maximum amplitudes gglll 1 2 equivalent cycles in those tests. That adds up to--
3 CHAIRMAN SIESS: Is that about what you got?
4 About ten for an input?
5 MR. TAGART: Close to the maximum limit. I'd 6 say ten was the upper limit.
7 CHAIRMAN SIESS: That is what I didn't know.
8 MR. TAGART: It is more like six, would be a 9 better number.
10 CHAIRMAN SIESS: Yes, I see.
11 MR. BUSH: Sam, would you interprct your horizontal 12 axis. I don't understand it.
13 MR. RANGANATH: All right.
lllh 14 The horizontal axis really does not have any 15 meaning. They are all--that was plotted as bar charts, 16 you know. These were all different runs, that's all.
17 This was one set, where he ran it at half sled range, 18 and then full range, and so on, but in itself the X axis 19 does not have any meaning.
20 CHAIRMAN SIESS: Are those all of the tests?
21 MR. RANGANATH: We, in many cases, could not 22 get meaningful data because the strain gauges came off.
23 He looked at all--many of them, and picked those 24 where he had good data that went all the way to the higher 25 str- _,
136.
CHAIRMAN SIESS: Each plus is a different test?
ggll 1 2 MR. RANGANATH: Each test maybe a different 3 run in a test.
4 Remember, they kept increasing the amplitude 5 at one elbow, with a lower forcing function and they kept 6 increasing it, so each point is a different run there.
7 CHAIRMAN SIESS: Okay.
8 MR. RANGANATH: Now, I did the same thing.
9 (Slide]
10 This was for carbon steel, but I express it 11 in terms of Sm Y--
12 MR. RODABAUGH: Sam, before you say it, for 13 the benefit of the subcommittee, when you divided by Sm Y, ll 14 what Sm Y were you using?
15 MR. RANGANATH: I use the Code minimum Sm Y, 16 okay. This material probably has Sm Y levels that were 17 higher than the code minimum value.
18 To that extent it can be construed as the somewhat 19 non-conservative, but when you also recognize that when 20 this thing is going through the cyclic behavior it doesn't 21 take a whole lot of time before it strain hardens itself, 22 so I don't know that that distinction is that critical.
o 23 MR. RODABAUGH: I think that you would be better 24 off to use your best estimate of Sm Y from your material 25 mill test reports.
137 0 lg 1 CHAIRMAN SIESS: Would that vary with the different 2 tests?
3 MR. RODABAUGH: Yes.
4 MR. RANGANATH: Yes, it would vary for different 5 tests, and--
6 CHAIRMAN SIESS: If the materials were enough 7 different that the actual yield stress, compared to the 8 specified minimum would vary?
9 MR. RANGANATH: I would say most cases the specified 10 minimums were much lower than the actual--
11 CHAIRMAN SIESS: No, that is not my point.
12 If you went in with each of these specimens, 13 and used the specimen--used the actual yield stress, measured ll k 14 yield stress, was the ratio of measured yield stress, 15 the specified minimum--or Code stress as you call it--
16 would that be fairly constant, or would these different 17 heats--
18 MR. RANGANATH: I would guess it is constant, 19 but I will let Kelly answer it. He probably has a better 20 idea.
21 MR. MERZ: Some of the specimens were taken 22 from the same heat, okay. The mill test is for that heat, 23 okay.
24 In terms of elbows, tees, pipe, I would say 25 there is probably 20 different heats, okay, all with
138, 1 probably about 10 KSI above the specified minimum, approximately.
2 CHAIRMAN SIESS: Mill test.
3 MR. MERZ: Yes, above the mill test.
4 Now, that is the mill test, of course, is for 5 the specimen before it is forged, not after it is forged, 6 really. It is for the piece of pipe that it is forged 7 from, usually.
8 Now, Mr. Rodabaugh can probably answer that 9 more correctly.
10 CHAIRMrd SIESS: Since all you ever know, in 11 designing, is the Code specified minimum, the question 12 then remains is this a representative sampling of the 13 variations you could get between the specified minimum 14 properties and the actual properties?
15 MR. RANGANATH: All right--
16 MR. RODABAUGH: Yes, that is the question, Chet.
17 Their materials, rather typically, were about 18 20 percent higher yield strength, for example--20 or 25--
19 but if you look at statistics on the materials, like stainless 20 steel, you will find some small percent, one percent, 21 is right at the minimum, even a little percent below minimum.
22 CHAIRMAN SIESS: What I was thinking is, that 23 suppose you went back and took say mill tests, which are 24 not the right answer, they are not the properties the 25 material is formed into the elbow, but supposed you took
139.
the mill tests that did fluctuate, and you did this and glll 1 2 your scatter decreased? Let's just be real optimistic 3 and say, you know, you have got all of these in a very 4 narrow band, what would you do with them? You still wouldn't 5 be able to use--
6 MR. RODABAUGH: I think, Chet, the very next 7 slide is going to be the key to my thought here.
8 Eventually these stress limits are going to 9 be shown in terms of Sm Ep. Now, that is a minimum. If 10 it turns out that--
11 CHAIRMAN SIESS: What is that?
12 MR. RANGANATH: S of M is a design stress intensity.
13 MR. RODABAUGH: Design stress intensity, which ll 14 may be either based on yield strength, or ultimate strength.
15 In general, the Code philosophy is to base your 16 design on the minimum properties.' Now we have got some 17 tests which are tests of components that had higher than 18 minimum properties. A rather straightforward completely 19 defensible way to evaluate the data is to adjust it to 20 the difference between what you tested and the Code values, 21 then you have a very straightforward four-story and you 22 don't have to go back and say, " Well, if we had a minimum, 23 we still think we are in our stato--
24 MR. RANGANATH: You are right, and in fact one 25 of the action items that I have is to redo these results
140.
1 in terms of the actual--
ggll) 2 CHAIRMAN SIESS: Then you can decide whether 3 you want a five percent cutoff or a ten percent value.
4 MR. RANGANATH: And, again that is--the ten 5 percent that I took was arbitrary--
6 CHAIRMAN SIESS: Not on the stress, the yield 7 stress.
8 MR. RANGANATH: Yes.
9 Now, here is the same plot for carbon steel 10 that was based on S of M.
11 (Slide]
12 And, again this is showing about 6 S of M may 13 be okay for Level D; maybe 3 S of M is okay as opposed lllh 14 to the current limit for level B, which is 1.8 S of M, 15 you know, so again it indicates there is some room for 16 improvement.
17 [ Slide) 18 I'll show you--
19 CHAIRMAN SIESS: You mean .he solid squares 20 then to represent level D?
21 MR. RANGANATH: Rather than ten percent, right--
22 no, solid squares are the stress amplitude that I have 23 to have--
24 CHAIRMAN SIESS: Yes, but I then you said you 25 thought 6 was good enough for--I thought you said level B?
L,
141.
ll 1 MR. RANGANATH: Level D, right.
2 CHAIRMAN SIESS: Level?
3 MR. RANGANATH: D, faulted.
4 CHAIRMAN SIESS: And, 3 for what?
5 MR. RANGANATH: Normal, level B--no upset requirements, 6 upset, upset.
7 So you can do the same thing again. For stainless 8 steel we kind a somewhat fewer points on stainless steel, 9 and again the same comments that Ev brought up apply here, 20 too.
11 This is expressed in terms ot S of Y, and here 12 we have the same thing expressed in the terms of S of M.
13 MR. SHEWMON: Tell me again what you have been ll h 14 telling me for ten minutes.
15 I don't understand--we've gotten an S me n, 16 plus some small, I hope, adjustment to the clastic treatment--
17 MR. RANGANATH: Oh, .o.
18 All we did was S mean is the pressure stress.
19 MR. SHEWMON: Okay.
20 MR. RANGANATH: The strain amplitude is the 21 strain amplitude that was measured, from constrain gauges,
! 22 you know. It was multiplied by E, right, to get the pseudo-23 elastic stress.
! 24 MR. SHEWMON: I understand that.
25 Then this is the ratio of that times the design l
l
142.
1 stress--
2 MR. RANGANATH: Right.
3 Maybe it would be a better way to look at it 4 would be in terms of S of Y. That means four times S 5 of y.
6 MR. SHEWMON: Well, but there is always as many 7 pluses above your lines as there are below, so I don't 8 quite see any threshold for ratcheting.
9 [ General discussion) 10 CHAIRMAN SIESS: You can get the open squares 11 for his bottom line margin, that is ten percent--less 12 than ten percent, and the solid squares for his top ones.
13 MR. SHEWMON: Right, right, h 14 MR. RANGANATH: See, we have--let's look at 15 no ratcheting at all.
16 That means I should not have any squares. I 17 did have two squares here. When we went back to check 18 it the strain was very small. It was almost close to 19 no ratchet, so I am saying--on the level B conditions 20 I wouldn't want to see any ratcheting, and that is saying 21 that maybe 2 S of Y is a good limit for level B.
22 For level D, I will say, yes, I can tolerate 23 some ratcheting, and I just arbitrarily picked ten percent 24 in 50 cycles, and that saying if I am below 4 S of Y--
25 MR. SHEWMON: Okay, fine.
4 l
143.
1 MR. RANGANATH: --I wouidn't get ten--
2 MR. SHEWMON: Thank you.
3 CHAIRMAN SIESS: Now, you know that line, you 4 could have drawn at five, couldn't you?
5 MR. RANGANATH: That is what Sam Tugart also 6 asked me, you know, and I can do that, and again I just 7 picked ten for this case. We can do that. I think it 8 would be one of those numbers that will have to check 9 with our consultants and come up with something that is 10 acceptable.
11 So, with that we sent 3 S of M for level B, 12 and 6 S of M for level D, okay.
13 [ Slide]
lllh 14 So, let's take a look at design rule changes 15 that we might propose. And, one design rule that--this 16 is the pressure, remember I told you--showed you the equation 17 PD over 4P plus M over Z, where M is the earthquake moment.
18 Now, right now, that includes the seismic loading also, 19 and these are the current limits. It is the lesser of 20 this or this, that is what we have today.
21 And, what we might change is--
22 CHAIRMAN SIESS: Now, 3 S of M, if it is two-23 thirds S Y, that is what the right-hand side governs?
24 MR. RANGANATH: Yes, close to.
25 CHAIRMAN SIESS: Okay.
1440 1
MR. RANGANATH: In some cases, you know, it 2 is a little different because it is the lower of one-3 third of alternate, or two-thirds of yield, and so this 4 changes.
5 CHAIRMAN SIESS: So that is equal really to 6 ultimate or two times the yield, if I look at level D?
7 MR. RANGANATH: Yes.
8 CHAIRMAN SIESS: Right?
9 MR. RANGANATH: Yes.
10 So, that is where we are today. That includes 11 the seismic loads also.
12 What we might look at in terms of a proposal 13 for new Code limits, you don't hav to do any special 14 analysis fcr ratcheting, and so on, provided you make 15 these limits for, you know, with the new proposal, 3S 16 of M to 6 5 of M, 3 for level B, and 6 for level D. And, 17 again this is somethi..g that we would probably have to 18 reiterate on, but this is the direction we are going to.
19 CHAIRMAN SIESS: And, 6 is two times ultimate?
20 MR. RANGANATH: Right.
21 CHAIRMAN SIESS: Okay.
22 MR. RANGANATH: And, again the numbers we are 23 talking about are pseudo-elastic stress.
24 CHAIRMAN SIESS: Oh, yes.
25 MR. RANGANATH: It is measures of strain.
1450 I CHAIRMAN SIESS: I know, it is a measure of 2 strain caused by--
3 MR. RANGANATH: Yes, and the non-seismic loading 4 portion will still be limited by tha current Code, so 5 weight, water hammer, and preuure, and so on, would still 6 be controlled by this.
7 So, you know, how can t'e implement it in the 8 Code? One thing we were looking at was we have some options, 9 and this again we will have to decide ;n it with some 10 more interpretation. SLQuld bc make a blanket change 11 in the Code? Or, ehculd we 5c mole restrictive in terms 12 of a Code case approach uhare we say you ran use these 13 rules provided yot use respon.So spectrum analysis, provided 14 you use paak broadenira, provic ed you don't use damping i
15 in excess of fiva percant, p r o v d .e d y o u d e'a ' t h a v e c a s t i n g s ,
16 and so on, so chat may be one w('; to lech at it.
17 (Slide) 18 And, right n w we will not ..eec' any firm decisions 19 on which way to do, the above spells xc cut.
20 (Slide) 21 And, finally, lookin, at it f rcim the long term 22 viewpoint, where, you know, do we cjo trom h?re? Clearly 23 the current Code approach hrs been showa to be very conservTtive, 24 and it is kind of does not maka se'ise, and one of things 25 that the limits on the primaly stresses, on equation 9,
146.
2 Snubbers, and so on, very fancy analysis, and 3 we can probably reduce design costs, hardware costs, and 4 really other potential problems associated with snubbers--
5 failing and so on--can be reduced, and so I think we can 6 get a simple, more cost effective system--
7 CHAIRMAN SIESS: Now, you had a caveat before 8 about the response spectrum analysis, so you are not going 9 to change the analysis costs there much?
10 MR. RANGANATH: Oh, okay, in the long term, 11 that kind of let's me up to the third bullet--
12 CHAIRMAN SIESS: Let's stay up at the second 13 bullet.
14 I am trying--
15 MR. RANGANATH: I agree.
16 CHAIRMAN SIESS: --to visualize, given those 17 Code chang 2s in stresses, what would be different in the 18 plant? What would be different the design? What would 19 be different physically in the plant?
20 They would have fewer snubbers and steel pipe 21 supports--
22 MR. RANGANATH: Right.
23 CHAIR'4AN SIESS : Would the pipe sizes be different?
24 MR. ?WGENATH: No.
25 CHAIRMAli SIESS: The schedule wouldn't be different?
147.
MR. RODABAUGH: No.
gglll 1 2 CHAIRMAN SIESS: That is based on pressure, 3 so it would be mainly supports and snubbers--
4 MR. RANGANATH: Right.
5 CHAIRMAN SIESS: --and the maintenance that 6 goes with snubbers--
7 MR. RANGANATH: That is correct.
8 CHAIRMAN SIESS: --and the problems that go 9 with supports and so forth?
10 MR. RANGANATH: That is right.
11 MR. GUZY: It should be noted that the designing 12 of the piping system, support design, is the major factor 13 in this, so you have less support design--
14 CHAIRMAN SIESS: Yes, I know.
15 MR. RANGANATH: And, in the long run--
16 CHAIRMAN SIESS: Thousands--
17 MR. RANGANATH: --as I--
18 CHAIRMAN SIESS: --you didn't finish your last 19 bullet there.
20 MR. RANGANATH: Yes, I will get to that in just 21 a second.
22 As you can see here, if you look at the alastic 23 plastic response they are almost getting to a factor of 24 1, you know, so what this is saying is--and a lot of the 25 experts like Bob Kennedy are saying, you know, the old i
i l
l
148.
I way of doing things where we Just did it in a static manner, 2 and in fact is okay, and this clastic plastic--
3 CHAIRMAN SIESS: An equivalent static.
4 MR. RANGANATH: --analysis has shown that that 5 is okay.
6 CHAIRMAN SIESS: An equivalent static.
7 MR. RANGANATH: --right, equivalent and static 8 method.
9 CHAIRMAN SIESS: At the workshop on Appendix 10 B that was held a couple of years ago, somebody proposed 11 an equivalent static. Was it Bob Kennedy?
12 MR. TAGART: I am not sure about that meeting, 13 but equivalent static has always been in the Code. The h 14 problem with it it it has always taken a 1.5 factor peak 15 of the response spectrum as the equivalent static--
16 CHAIRMAN SIESS: Well, that is not equivalent 17 static, then--
18 MR. TAGART:--and that is just way too conservative--
19 CHAIRMAN SIESS: yes.
20 MR. TAGART: --and recently there is a proposal 21 by John Stevenson to introduce an equivalent static approach 22 that is more realistic.
23 CHAIRMAN SIESS: Thank you, that is who it was.
24 And, then somebody suggested, you know, simplify 25 that and put a little more effort into steam generator
149.
1 supports, and vessel supports, in places where the consequences ggggg 2 would really be serious.
3 It was Stevenson at that meeting, I guess.
4 Well, I tell you, there are real benefits in 5 getting back to that sort of an approach, because you 6 have got a heck of a lot better feel for what--
7 MR. RANGANATH: Bring in the understanding.
8 CHAIRMAN SIESS: --you are doing, yes, I mean 9 this stuff all comes out of the damn computer and if there 10 is a mistake then nobody has any feel for it.
11 MR. SHEWMON: Tell me what drives snubbers now?
12 Is it displacement? Or does decrease mean stress? Or 13 what?
lllll 14 MR. RANGANATH: Stress-strain reduction is the 15 one that drives, although--
16 MR. TAGART: Equation 9 right now, which is 17 a stress equation on the pipe.
18 MR. SHEWMON: And, so if you put a snubber in 19 you decrease the dynamic stress?
20 CHAIRMAN SIESS: You also change the frequency, 21 too, don't you? Shift cff of the peak.
22 MR. SHEWMON: Now, the frequency didn't come 23 into that equation, as he drew it?
24 CHAIRMAN SIESS: No, but in the analysis it 25 gang,
150.
ggg 1 11R . SHEWMON: Yes.
2 CHAIRMAN SIESS: For smaller pipes, it just 3 support i*: to get it above 33 hertz, and there is no amplification.
4 MR. RANGANATH: That is the last one, some kind 5 of a design by rule test, Sam was talking about, and the 6 whole idea is mainly these experts believe there is a 7 very simple ways of designing piping that is quite effective 8 and still maintain the safety margins.
9 CHAIRMAN SIESS: Well, again, don't we have 10 some evidence from SKRUG [ sic.) that at best was design 11 equivalent static?
12 MR.TAGART: Yes, there is a lot of evidence to 13 that effect.
llllh 14 CHAIRMAN SIESS: What causes the failure?
15 MR. TAGART: The only problems that would cause 16 failures is where you have seismic anchor motions that 17 weren't accounted for in the design.
18 CHAIRMAN SIESS: Yes, but you can do those by 19 equivalent static and probably better than anything else 20 can, and the anchors are something else.
21 MR. GUZY: Also, connections with the piping, i
22 when we talk about threaded piping, then there may be 23 a problem.
24 CHAIRMAN SIESS: Okay.
I 25 MR. SHEWMCN: To go back and try to understand l
l l
151.
these two curves, I find they come in pairs, where the ggfll 1 2 data points are the same but the answers change.
3 MR. RANGAN\TH: That is correct.
4 MR. SHEWMON: Or the numbers on the left axis--
5 MR. RANGANATH: Right, we have the same data 6 point, and in one we express it as a ratio of the yield 7 strength, and the other way we express it--
8 MR. SHEWMON: Okay, fine.
9 MR. RANGANATH: --as the ratio of the design 10 stress--
11 MR. SHEWMON: He just defines strength--
12 MR. RANGANATH: Thank you.
13 CHAIRMAN SIESS: Well, I have got on comment ll 14 about the proposed Code changes: lots of luck!
15 It is hard to change it, but I think everything 16 we kaow says that we are awfully conservative in what 17 we are doing, and not necessarily that we are getting 18 a conservative power plant as the result of it, and that 19 kind of bothers me, and if there is some way of arriving 20 at a better design that is just as safe I sure we could 21 do it.
22 Gentlemen, that concludes the presentations 23 that are being made to us, except Dan had one slide left 24 he wanted to bring in at the end. I am not sure if he 25 forgot about it or not.
l l
152.
g I MR. GUZY: It was just to talk about the rest 2 of the research that we are doing.
3 From a research point of view, I just sort of 4 wanted to put this into context along with some other 5 things that we are doing in piping.
6 First of all, we are developing a new piping 7 research plan that will include--the type of work we talked 8 about today--response design rules for piping, as well 9 as cracked piping, so that there will be a forthcoming 10 NUREG, and the pad has only addressed, like the degree 11 of piping program now, only it will be more comprehensive 12 and will include things we are doing that use to be in 13 the old cycling plant--that should be familiar to you.
O 14 In addition to this work, which 1 consider the 15 most important and certainly the major part of our research 16 in the design area, we are still doing work in the piping 17 response method area. We have done work in the past on 18 damping. We are supporting work that is being donc now 19 at Vectal (sic.) and through EPRI on the new damping criteria, 20 and there will be some forthcoming actions on the Code 21 bodies to revise this Code case, and hopefully to make 22 it easier to use. Right now only half the plants can 23 qualify for using the damping criteria--
24 CHAIRMAN SIESS: Why?
25 MR. GUZY: Because the NRC in their endorsement
1530 1 of Code case N-411 put a number of limitations on it, 2 for instance, you have to use a modern spectra, ground 3 spectra. If you use a--
4 CHAIRMAN SIESS: Oh, yes.
5 MR. GUZY: --and that can't be used in the conservative.
6 Also, there is activity now in the ISM response 7 method area, and we are doing some regulatory and Code 8 work on that, hopefully to resolve it and bring some of 9 this research to a head, a change in what our current 10 position is.
11 There is also doing some work on the treatment 12 of high frequency modes, how you can--with response analysis, 13 and better treatment of closely spaced modes response 14 analysis, and there is work in the nonlinear response 15 area, supportive work at federal. There is some--a number 16 of small projects in response to this area, so are continuing 17 to do work ir these areas.
18 I think in the response margin methods areas, 19 it is not as big as it was a couple of years ago.
20 CHAIRMAN S1ESS: Now, these two middle bullets, 21 is that going to involve experiments?
22 MR. GUZY: We can use some experimenL, a.; benchmarks, 23 but primarily analytical.
24 One of the advantages of these most recent system 25
154.
tests is that we do have physical evidence for support, gg 1 2 and in the past there hasn't been that many piping benchmarks 3 that included that, so this is good.
4 CHAIRMAN SIESS: The high frequency thing, I 5 guess I will never quite understand the concern about 6 some of the high frequency because of such small energy 7 content.
8 MR. GUZy: That's--I mean, our program, the 9 test program is showing that they are not of much concern.
10 If we could make it go away, using these results, then 11 we wouldn't have to worry about it, but on paper high 12 frequency modes do pose a problem at least in licensing 13 cases, and there may be some better way to combine--it ll ) 14 has to do with correlation of the modes and the plants, 15 and maybe there is some analytical way we can do that 16 through research.
17 Okay, my point was that piping response methods, 18 we have done work--a lot of it recommended by the piping 19 review committee--that work is kind of winding down, but 20 there still are things that we are doing.
21 We have a major effort now at Oakridge, research 22 is sponsoring now at Oakridge on noz:1c flexibility and 23 design, and there has been--it is having some impact now 24 on some of the Code activitics, and I think it will have 25 a future impact.
1550 1 There has also been work that EPRI has sponsored 2 for this area. There are some definite Code activities 3 going on now, in the nozzle area, to provide some relief.
4 The point is that when we change the piping stress rules 5 then nozzles, or supports we are talking about here, in 6 the nozzle area we are doing something, and the support 7 design will be a new area, and we are just trying to get 8 our hands around this now.
9 There is some PBRC activities, we will make 10 recommendations, and improve the support design. EPRI 11 is having a w'rkshop on support design overy month--
12 CHAIRMAN SIESS: Pipe supports?
13 MR. GUZY: Pipe support, pipe support design.
14 There is--
15 CHAIRMAN SIESS: How do they determina the force?
16 MR. GUZY: How to design what you know of the 17 forces, okay, so that this is--
18 CHAIRMAN SIESS: That is a structural engineering 19 problem that I thought the steel people had solved a number 20 of years ago, 21 MR. BUSH: They have to a degree, in fact, the 22 P8RC offort would end up looking at a package with the 23 suggestion that effectively what you do is remove NF from 24 the Code, which represents a tremendous load, particularly 25 for inspection and so forth.
156.
I It seems to me that you've llllh CHAIRMAN SIESS:
2 got two choices on support design: you design them to 3 take the loads without yielding; and the other is you 4 design them to yield and absorb energy.
5 MR. GUZY: That is the--
6 CHAIRMAN SIESS: The first one peopic ought 7 to know how to do, and the second one is a littic bit 8 more of a problem.
9 MR. GUZY: There are a number of areas that 10 are involved in support design, one is the concept of 11 supports that are used, you know, when they fail they'd 12 better fail first.
13 What I am trying to note is that we are planning
!h 14 on doing research in this area and there are a number 15 of recommendations to be made.
16 CHAIRMAN SIESS: Yes.
17 MR. GUZY: The use of piping experience data, 18 we are doing a test through oakridge in this area, research 19 is, and we are fe' lowing a project that EQE had done for 20 EPRI, which followed the project that Don Stevenson had 21 done for us. We are still trying to use the piping experience
! 22 data from SCRUG [ sic.)--essentially the same plants as 23 SCRUG [ sic.) and essentially to try and bring it into 24 more of a regulatory process.
25 MR. BUSH: That sounds like you are limiting I
1 1
157.
(llll 1 it pretty much to seismic response?
2 MR. GUZY: Yes.
3 MR. BUSH: And not just piping experience--
4 MR. GUZY: Yes, there is some element of what 5 the operating conditions were and how it is inspected 6 and if you can show that you are enveloped by industrial 7 plants, then you can feel more comfortabic.
8 In terms of degraded piping, or the IPIRG program, 9 it does have an element of--
10 CHAIRMAN SIESS: What is IPIRG7 11 MR. GUZY: --it is the International Degrading 12 Piping Program, but I don't know what that--
13 CHAIRMAN SIESS: Okay.
14 MR. GUZY: But, it will include an element of 15 dynamic testing, in fact, there is some simple tests now 16 set up for essentially dynamic failed piping with known 17 cracks in it--
18 CHAIRMAN SIESS: Degraded piping.
19 MR. GUZY: Degraded piping, yes.
20 Then the last bullet is piping liability studies.
21 Sam showed you carlier some things that EPRI has sponsored.
22 We, NRC, may become more involved in this also. We see 23 that way of integrating new piping information, like what 24 we've just talked about, the program, and say information 25 //////
1580 (glll 1 on the degraded piping will be--probably could be most 2 economically once we know the basis data on it and--
3 CHAIRMAN SIESS: Now, is that pRA related?
4 MR. GUZY: --I don't know how probabilistic 5 it will be.
6 I see us as maybe improving our full commercial 7 type program. Sam would see it as an extension of the 8 things he was ta1xing about this morning.
9 CHAIRMAN SIESS: Now, you skipped one.
10 MR. GUZY: I skipped one? Oh, cumulative effects 11 of piping criteria changes, this is sorothing that the 12 licensing staff has asked for for a long time and we had 13 a hard time getting our arms around it.
lllh 14 The way it is envisioned now would be sort of 15 a response margins approach where you would show how you 16 would trade these, and we've had a hard time getting this 17 off of the ground, and my feeling is if we can effectively 18 use the information from this test program we may not 19 have to do that.
20 CHAIRMAN SIESS: Well, I think that is very 21 important, because I think that anything we can do to 22 improve piping design, reduce some of the problems we 23 have with snubbers, is excellent, but we have gotten an 24 awful lot of comfort out of the margins we have.
25 Now, we are finding the margins are tremendously l
159.
(glll 1 large, but let's don't get the margins down to the point 2 where when somebody wants to up an earthquake hazard, 3 we get another carthquake in the castern U.S., now our 4 comfort has disappeared, and that comfort is hard to qualify.
5 It is a fairly important aspect of this, and we make all 6 of these changes and somebody needs to take a look and 7 say, "Now, okay, where are we now?"
8 1 don't know if that is easy to do, but--
9 MR. GUZY: I think the last project, the piping 10 reliability studies, which would attempt to do that--
11 CHAIRMAN SIESS: I think that will tie in.
12 MR. BUSH: Dan, could I have a comment, because--
13 CHAIRMAN SIESS: You don't have m&rgin in there, h 14 as such, but that is what I am thinking.
15 MR. BUSH: It depends on what you put in the 16 last one, because as I visualize it what you really need 17 is something--a cut across, at a minimum--three of those:
18 piping experience data, the cumulative effect of changes, 19 and the piping reliability study, and somehow they have 20 to be integrated.
21 MR. GUZY: Yes, and I think--
22 MR. BUSH: And, if they don't, if they aren't 23 integrated, you may not accomplish what you need.
24 MR. GUZY: Yes, I think that ideally, if we 25 can get something going that we should include--especially J
1600 lllll 1 the piping experience, I think that is something that 2 has to be brought in more than it has beca in the past.
3 CHAIRMAN SIESS: Would the piping reliability 4 studies take the cyclic approach?
5 MR. GUZY: I think they would take--cyclic, 6 maybe like Sam was showing this morning, maybe some frequency 7 of core melt type of--
8 CHAIRMAN SIESS: Well, yes, okay.
9 That approach just emphasized the uncertaintics.
10 MR. TAGART: Quantifies and deals with the uncertainties.
11 MR. GUZY: So that is briefly what we are doing 12 in my branch.
13 CHAIRMAN SIESS: Now, this is all.--what's current, h 14 and what's for the future?
15 MR. GUZY: Well, I would -let's see--everything 16 is sort of current, except for support design, which should 17 happen fairly soon. The cumulative effects, which we had 18 not started and the piping reliability studies, which 19 Sam has already started--
20 CHAIRMAN SIESS: And, these are all withstanding 21 the budget cuts?
22 MR. GUZY: e s .
23 CHAIRMAN SIESS: I think chat concludes the 24 presentations.
25 Are there any questions or last words?
'i62.
% 1 (No response.]
2 E would hope that Mr. Rodabaugh, who has threatened 3 you with a letter will provide us with a copy, and I would 4 appreciate anything from Spence Bush in the way of comments 5 on the meeting that you can pass on to the rest of the 6 committee.
7 MR. GUZy: Will the committee be making recommendations?
8 Or what will you do? :
9 Cl{ AIRMAN SIESS: You haven't asked fo: iny.
10 We don't now have any particular input into the budget.
11 If it comes up, cbviously, we would be prepared.
12 MR. GUZY: We've heard your concerns over cast 13 versus stainless stcol, and that will bc addresscd in 14 the program--
15 Cl! AIRMAN SIESS: What we will do is I will probably 16 make a brief report at the next full committee meeting, 17 and if they are interested in hearing more about this 18 as a c::mmittee we might ask for some presentation at a 19 future committee meeting.
20 I think it might wait until it gets to some 21 regulatory action, you know, if wo brief the full committec 22 on what's going on now, and the regulatory act. ion comes 23 two years from now, we start over.
24 I think the subcommittee should be Xtpt abreast, 25 but to try and keep the full committee addressed in i
.o sj 162.
.i
/3
\
I.
g 1 advance doesn't always work. Ten other things come up 2 with them. '!
3 Thark you, gentlemen. It has been very fine-4 presentations. '2'1 think you covered a lot of territory,'
5 and answered thb questions we had today, and I say lots 6 of luck with the Code changes.
7 8 Adjourned: 12:30 p.m.
9 10 11
///
12 fjj ,
I
, ///
14 '
15 16 t
l 17 18 19 l 20 1
21 22 23 24 25
2 3 REPORTER'S CERTIFICATE 4
STATE OF CALIFORNIA )
5 ) ss.
COUNTY OF VENTURA )
6 7
I, PRISCILLA PIKE, an official hearing reporter for the 8 State of California, do hereby certify that the foregoing pages 1 through 162, inclusive, constitutes a true and 9 correct transcript of the matter as reported by me.
10 I FURTHER CERTIFY that I have no interest in the subject matter.
WITNESS my hand this f id day of April,in the County of 12 Ventura, State of California.
13 14 NE Priscilla Pike 15 Pike Court Reporting Services 3639 E. Harbor Blvd. Ste. 203-A l 16 Ventura, California 93001
! (805)S68-7770 l 17 l
18 19 20 21 22 23 24 25
'^ *
' ~
$3_. . ~
O O T.
.i _$
PIPING AND FITTING DYNAMIC RELIABILITY PROGRAM EMPHASIS DESIGN OF PIPING COMPONENTS FOR DYNAMIC INERTIAL LOADS OBJECTIVES ,
IDENTIFY DYNAMIC FAILURE MECHANISMS AND LEVELS PROVIDE HIGH-LEVEL NONLINEAR RESPONSE DATA DEVELOP IMPROVED ASME CODE DESIGN RULES I
w..
O O O PIPING AND FITTING DYNAMIC RELIABILITY PROGRAM EPRI SAM TAGART, Y. K. TANG NRC DAN GUZY GE (SAN JOSE) BILL ENGLISH, HENRY HWANG, SAM RANGANATH, ED SWAIN ANCO ENGINEERS PAUL IBANEZ, KELLY MERZ ETEC RON JOHNSON, VINCE DEVITA MCL ROY WILLIAMS CONSULTANTS E. RODABAUGH, R. KENNEDY, D. LANDERS, R. L. CLOUD, D. MUNSON, S. MOORE, R. BOSNAK, L. SEVERUD
~
O O O PIPING AND FITTING DYNAMIC RELIABILITY PROGRAM TASK 1: PROGRAM PLAN DEVELOPMENT (GE)
TASK 2: PIPE COMPONENT TESTING (ANCO)
- 41 FAILURE TESTS OF ELBOWS, TEES, ETC.
TASK 3: PIPE SYSTEM TESTING
- PFDRP "SYSTEMS 1 8 2" TESTS (ETEC)
- OTHER SYSTEM TESTS (ETEC)
- WATERHAMMER SYSTEM TESTS (ANCO)
TASK 4: SPECIMEN FATIGUE RATCHETING TESTS (MCL)
- 140 SPECIMENS, DIFFERENT MATERIALS & TEMPERATURE TASK 5: - ANALYSIS OF TESTS AND DESIGN RULES (GE)
TASK 6: - IDENTIFICATION AND DEVELOPMENT OF ALTERNATIVE DESIGN RULES (GE)
TASK 7: - EVALUATION OF ALTERNATIVE DESIGN RULES (GE)
TASK 8: - PROJECT FINAL REPORTS (GE)
~
O O O PIPING AND FITTING DYNAMIC RELIABILITY PROGRAM STATUS AND SCHEDULE PROGRAM INITIATED IN SPRING OF 1985 ALL. TESTING COMPLETED (EXCEPT RETEST OF SYSTEM 1)
FINAL ANALYSES AND CRITERIA DEVELOPMENT UNDERWAY PROGRAM ENDS JUNE 1988 WITH DRAFT FINAL REPORTS ,.
INITIATION OF REVISIONS TO ASME CODE NB/ND/NC-3600 IN 1988 EPRI TO PUBLISH FINAL REPORTS l
~
O O O PIPING AND FITTING DYNAMIC RELIABILITY PROGRAM COOPERATIVE EPRI/RES RESEARCH AGREEMENT FIVE REVIEW MEETINGS WITH PROJECT MANAGERS AND CONSULTANTS INTERACTIONS WITH ASME AND PVRC STANDARDS GROUPS
- PRESENTATUONS AT MEETINGS
- MEMBERSHIP ON GROUPS BY PFDRP PARTICIPANTS
- ASME CODE CLASS N-451
~
- CLASS 2 8 3 DYNAMIC ALL0ifABLE CODE CASE
- PVRC TASK GROUP ON PIPING FUNCTIONALITY PUBLICATIONS
- PAPERS IN JOURNALS AND SMIRT
- FOUR SEMI-ANNUAL PROGRESS REPORTS
- FINAL REPORTS TO BE ISSUED BY EPRI ARCHIVING OF TEST SPECIMENS AT NDE CENTER
O O o PIPING AND FITTING DYNAMIC RELIABILITY PROGRAM NRC PERSPECTIVE PIPING REVIEW COMMITTEE
- RAISED CONCERNS ABOUT OVERCONSERVATISMS IN INERTIAL LOAD DESIGN
- REGULATORY CHANGES LIMITED TO RESPONSE CRITERIA (E.G., DAMPING)
- IDENTIFIED HIGH PRIORITY NEED FOR FAILURE TESTS (NUREG 1061 VOLS. 2 8 5)
PFDRP PRESENTATIONS TO NRC
- INFORMATION DISTRIBUTED, VIDEO TAPES SHOWN
- BACKGROUND PRESENTATIONS AT CODE CASE N-411 AND N-451 MEETINGS
- WATER REACTOR SAFETY INFORMATION MEETINGS
- 9/11/87 FORMAL BRIEFING TO STAFF
- 3/30/88 MEETING 0F ACRS SUBCOMMITTEE ON STRUCTURAL ENGINEERING
- FUTURE MEETINGS WITH NRC STAFF REGULATORY CHANGES
- R.G. 1.84 ENDORSES ASME CODE CASES
- 10 CFR 50.55A INCORPORATES SPECIFIC CODE ADDENDA & REVISIONS
- S.R.P. CRITERIA FOR FUNCTIONALITY CRITERIA, ETC.
- PFDRP RESULTS WILL PROVIDE "FAILURE MARGINS" DATA FOR OTHER REGULATORY ACTIONS i
mAa 1 . ..-i. .L-s4 & m. s +4 e. 4 4 4.-*.1 _ m. - = m tu aau>, 4-.
4 %
l O -
O Q i
i OTHER RES PIPING DESIGN RESEARCH l
t FORTHCOMING RES PIPING RESEARCH PROGRAM PLAN (NUREG-1222) l PIPING RESPONSE METHODS
{
! - DAMPING l - INDEPENDENT SUPPORT MOTION (ISM) METHOD i
- HIGH FREQUENCY, CLOSELY SPACED MODES l
! - NONLINEAR RESPONSE PREDICTION l
4 N0ZZLE FLEXIBILITY AND DESIGN
! SUPPORT DESIGN ,
l PIPING EXPERIENCE DATA l CUMULATIVE EFFECT OF PIPING CRITERIA CHANGES
- i. IPIRG j PIPING RELIABILITY STUDIES i
1 i
i l
i a
i l
i.- . -. . - m-..- . - _
._. . - . . ~_ - .
PIPING AND FITTING DYNAMIC RELIABILITY PROGRAM EPRI INTRODUCTION AND OVERVIEW FOR ACRS SUBCOMMITTEE ON STRUCTURAL ENGINEERING O
PACIFICA HOTEL CULVER CITY, CALIFORNIA MARCH 30, 1988 J
i PRESENTED BY 4
-SAM - W ,' TA O ' EPRI t
SWT/3828MS8v
TOPICS BRIEF HISTORY OF CODE RULES WHAT WE KNEW IN 1985 SIMPLE ANALYSIS EXPLAINING NO STATIC COLLAPSE
SUMMARY
OF WHAT WE KNOW IN 1988 THE OPPORTilNITIES AND THE CHALLENGES S
O SWT/3828MS8v
d
("3 HISTORY OF PIPE DESIGN 1952 - MARKL FATIGUE TESTS FOR B31.1 (SEMI STATIC) 1963 - NUCLEAR P.V. RULES (STATIC AND FATIGUE LOADS) 1968 - NUCLEAR PIPING RULES (STATIC, DYNAMIC
- AND FATIGUE LOADS) 1975 - JAPANESE RESEARCH SHOWS LARGE DYNAMIC MARGINS AND FATIGUE RATCHET FAILURE MODE FOR PIPING (N0 COLLAPSE)
O 982 - PVRC PROGRAM TO IMPROVE PIPING 1985 - NUREG 1061 NRC PIPING RECOMMENDATIONS 1988 - EPRl/NRC PIPING DYNAMIC TESTS (BASIS FOR NEW RULES)
EFFECTS OF DYNAMIC LOADS WERE HANDLED BY STATIC FAILURE CRITERIA (CONFIRMED BY GREENSTREET TESTS)
SWT/3828MS8v
I O
WHAT WE XNEW IN 1985 PIPING OYNAMIC MARGIN WAS LARGE BUT UNCERTAIN PIPING FAILURE MODE FOR REVERSED DYNAMIC LOADING IS RATCHETING AND FATIGUE (NOT STATIC COLLAPSE)
REDUCTION OF P! PING CODE MARGINS REQUIRED CONVINCING EXPERIMENTAL EVIDENCE PLUS ENGINEERING UNDERSTANDING MODERN NUCLEAR PLANTS HAD T00 MANY SNUBBERS O
l l
l l
9 SWT/3828MS8v
.s .-. y a -. --~s. -. a w- -
-. ra - a ,a. mau sex-u , . s sa.s e ----.-n. - s e e. +aas. n + 2 ~ua.m.- s F
9
'4
'O: 4 -
I' 9
0
~
K2 e, '
m
/
/ '
F't
/ -
/
-O E
/
/ i j 5 >
. n a
b i
M i
krJ Yo! *%g foree Em Sp. ,y L r. -g i q < /c t
I 0 !
1 i
e.=w
NEDC-31542 O: I t
f I
NON-LINEAR DYNAMIC RESPONSE 2% STRUCTURAL DAMPING MAX SPRING DISPLACEMENT 5xYlELD 3
j J,F3 5- ,
. _ . - r,
- . S //- * ~ /_ ~ /7 ! _
!E p.-- . 'Z. . . s/ 5 xS
< 2 -
l U
l ELASTIC SOLUTION e 6
---* TAGART'S MODEL (A = 0) z' **** NUMERICAL SOLUTION (A = 0) 9
-- NUMERICAL SOLUTION (A = 0.1) s < ~ .
3
.. :.s. ~ O
// *-W *~c;*4 W ' m ;
i -
//
't Xg(t) DISPLACEMENTS INPUT. Xgtt) = A sin (wt)
REGION vvHERE l
- "*+#1 ELASTIC ANALYSIS IS RELATIVE: Xstt) = lXg - X)lt) 0
/ ' '
NON-CONSERVATIVE lMl
= C sin (wt + 42)
XIt) AMPLIFIC ATION, G = C/A O.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 DIMENSIONLESS FREQUENCY (w/wn)
Figure 3.5-8. Dynamic Aiplification for Elastic and Elastic-Plastic systems 3-208
p O
WHAT WE KNOW IN 1988 WHY STATIC COLLAPSE DOES NOT GENERALLY OCCUR WHAT TYPES OF DYNAMIC LOADS CAN COLLAPSE PIPING HOW TO APPROXIMATELY PREDICT COMPONENT TEST RESULTS FROM FIRST PRINCIPLES LIMITATIONS OF LINEAR DYNAMIC ANALYSIS CLARIFY CONCEPTS OF APPARENT DAMPING O
WHY SOUND PIPING SYSTEMS ARE S0 FUNDAMENTALLY RESISTANT TO
- SEISMIC OR OTHER CYCLIC-TYPE DYNAMIC INPUTS ("TIME" DAMPING
! IS VERY HIGH AT MODERATE DYNAMIC DUCTILITY OF 3) 1 HOW TO UNDERSTAND RATCHETING
\
l l d :
L ,
SWT/3828MS8v
O OPPORTUNITIES AND CHALLENGES SIGNIFICANT CODE MARGIN REDUCTION PROPOSAL (THIS PROGRAM)
MANAGING PRIOR AND FUTURE OTHER CODE CHANGES
. N-411 -
N-451 ISM WITH SRSS SIMPLIFIED STATIC ANALYSIS i
NON-LINEAR METHODS SAM MODIFICATIONS h
l i DESIGN BY RULES g METHOD TO OPTIMlZE PIPING DESIGN FOR SAFETY AND COSTS l
t I
l T
O SWT/3828MS8v r_ .
e
,~
k COSTS fNSER WE NSPEC'Oh
\ \ _ PEnSCN AEMS
/ a CF P1 \
NUW8ER LOCAT N or SNUGBCRS NRC
% w tom X SNUBBER
,- SPECTCA 9
x N OUTAGES WNTS*
v /
~ NT NStas N CETEC CESGN CR ir
~w ERRORS tr SNVt9ER OTwfR T
,f
[ p,PE ,
TOUCwNESS C0at F AILURES CAACAS - CONQtTCN ' 'ND WELT STRENGTN db n
GZC y
PE CYC fa Su Att toca T>Envig F A TCUE CR SALA 4
$ TRESSES CRACAS d n M
AT
~
- $ TRESSES - '
LCACS '~ I(I[g'I
$TRESSE3 18 a g
FRE NCY iTGATCN PRESSURE LOACS CEWE S ST $
kWPut$rvE LCAOS iNo+aECT W A TER paga gg MAMMER t
I Figure 1. Snubber Reduction Model as Originally Conceived O
2-2
$10e
- 9
, - ANNUAL MAINTENANCE COST = 57.000/ SNUBBER ANNUAL MAINTENANCE COST = 51,334/ SNUBBER 8
i G
$107 8 -
G -
s -
G
~
l 1
~
EARTHOUAKE l FRECUENCY = 5:10"/Y E A R.
CCRE MELT CCST = $20 BILLION l
ANNUAL MAINTENANCE
- 1- CCST = 5125/ SNUBBER
{ -
EAATHCUAKE
- F A ECUINCY = 1x10"/ YEAR, CCAE MELT CO5T = $5 BILLION l
SIC * ' '
O 20 40 60 80 100 PERCENTAGE OF SNUBBERS REMOVED 1
, Figure 15. Expected Total Cost for an Entire Plant 9
1 5-16 1
1
. e; O
~
- -~
104-I .
/
j l
I w
I 5 /
4
.t
. ,/.
2
/ -
s I /
a /
2 /
E 10-5 /
/
~
8 I
' ~
j EARTHOUAKE FRECUENCY
- 5x10"fvEAR f
/
/ /
l m
1 I
/
l EARTHCUAKE FREQUENCY
/
= 1 x 10~4/ YEAR [
/
10-6 1 ! ! I f 0 20 40 60 80 100 PERCENTAGE CF SNUBBERS REMCVEo Figure 14. C0re Melt Probability as a Function of Snubber Reduction (Results Extrapolated to an Entire Plant)
O 5-14
- - - u_ _
4 sa\
O i
POTENTIAL DESIGN RULE CHANGES S. RANGANATH GE NUCLEAR ENERGY l O i
l 1
)
O
(2) OUTLINE O IAEVIEW OF CONCLUSIONS FROM TEST PROGRAM COMPONENT AND SYSTEM TESTS
- SPECIMEN FATIGUE & RATCHETING TESTS
- RESULTS OF SDOF MODEL ANALYSIS 0 POTENTIAL DESIGN RULE CHANGES
- VALIDITY OF ELASTIC ANALYSIS (2) - PROPOSED RULES AND TECHNICAL BASIS
- STRATEGY FOR IMPLEMENTATION O LONG TERM GOALS l
t I
4 i
1
g t
()
CONCLUSIONS FROM COMPONENT AND SYSTEM DYNAMIC TESTS o COMPONENT AND SYSTEM TESTS SHOW THAT SEISMIC LOADS WELL IN EXCESS OF LEVEL D LIMITS CAN BE TOLERATED.
O NO LIMIT LOAD FAILURES OCCURRED IN ANY l OF THE COMPONENT AND SYSTEM TESTS. THIS CONFIRMS THAT CURRENT CODE STRESS LIMITS PROVIDING MARGINS ON LIMIT LOADS MAY BE l
OVERLY RESTRICTIVE.
i
() 0 TEST FAILURES INVOLVE A COMBINATION OF FATIGUE AND/OR RATCHETING SUGGESTING THAT CODE RULES SHOULD CONSIDER THIS. EVEN WHEN FAILURE DIO OCCUR, THE NUMBER OF CYCLES WAS WELL IN EXCESS OF THAT IN TYPICAL SEISMIC LOADING EVENTS.
O ANALYSIS OF TESTS SHOWS THAT ELASTIC PREDICTIONS ARE GENERALLY CONSERVATIVE FOR RESPONSE SPECTRUM ANALYSIS WITH PEAK BROADENING FOR UP TO 5 PERCENT DAMPING.
T v
i
J 8
() CONCLUSIONS FROM FATIGUE-RATCHET SPECIMEN TESTS i
0 RATCHETING OCCURS WHEN THE COMBINATION'0F PRIMARY MEAN STRESS AND CYCLIC DYNAMIC STRESS EXCEEDS THE YIELD STRENGTH FOR POSITIVE MEAN STRESS.
- TIME INDEPENDENT RATCHET STRAIN DETERMINED FROM MILLER MODEL
-ER*bMEAN ! P FOR BILINEAR STRESS - STRAIN CURVE WITH KINEMATIC HARDENING 0 TWO BAR AND BEND TESTS SHOW TIME DEPENDENT RATCHET STRAIN FOR THE LOW
() FREQUENCY (0.5 CPM) TESTS RATCHET STRAIN PER CYCLE DEPENDS ON MEAN STRESS, CYCLIC STRESS, AND TEMPERATURE i
0 PRELIMINARY DATA SUGGEST THAT TIME DEPENDENT RATCHET IS LESS SIGNIFICANT AT HIGHER FREQUENCIES.
l
._ 4 -
O FATIGUE-RATCHET RESULTS (CONTINUED) 0 FAILURE IS EITHER BY FATIGUE OR BY EXCESSIVE RATCHET STRAIN LEADING TO NECKING AND SUBSEQUENT RUPTURE.
O WHERE FAILURE WAS BY FATIGUE, THE DATA POINTS FALL ON THE MEAN FATIGUE DATA CURVE REGARDLESS OF THE RATCHET STRAIN.
O THUS, AS LONG AS THE CUMULATIVE RATCHET l STRAIN IS NOT EXCESSIVE (SAY 5% - 10%)
! THERE IS NO SIGNIFICANT EFFECT ON CYCLIC FATIGUE LIFE.
O O
"'"F*-"TTM'w-T omygy,.w-% .,-._._
5-I O ,
t SINGLE DEGREE OF FREEDOM (SDOF)
MODEL ANALYSIS i
i O
l f
l l
t O
l
t 1 run mLAsTIc sDor sysTsM
() WITH INPUT SUPPORT MOTION
/
/////////// '
Input Support Motion, u
Xs(t) = A sin (u/t) -
Damper, C Spring, K l
Mass, M p Absolute Motion, Xa(t)
() Relative Motion, Xr(t) = Xa(t) - Xs(t) i Two types of Displacement to be studied:
- 1) Absolute Displacement of Mass - Analogous to the motion of piping system components.
Important in determining accelerations and i velocities for loads on pipe mounted equipment.
- 2) Relative Displacement between Mass and support -
, Analogous to relative displacements or l deflections of piping components. Important for determining strajn in piping components.
O 1
4 O 0 O RELATIVE ELASTIC ~ DYNAMIC RESPONSE .
26 FOR SDOF SYSTEM WITH 2% DAMPING .
I~
24 -
22 -
T x 20 -
N y_ 18 -
i 16 --
9 14 - ._
b 12 -
n.
k 10 -
w h 8-5 ,
i y 6-4-
!?
2-O , , i -, , ,
~ ,
O O.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
$ I DIMENSIONLESS FREQUENCY. W / Wn v'
-g .
ELASTIC-PLMTIC SDOF MODELS O
i l
Modified Elastic Model - Developed by Sam Tagart
/////////// ^
Input Support Motion, v Xs(t) = A sin (w t)
Damper, C Spring, K (Xr) l Mass, M r
p Absolute Motion, Xa('.)
Relative Motion, Xr(t) =.'Xei(t) - Xs(t)
Assume Flastic - Perfectly Plastic SDrina h
py.
Spring i I
Force K ,
I i
+
0 Xy Xr Spring (Relative) Displacement o Model Elastic-Plastic system as elastic synten O with reduced stiffness and increased damping as shown on following page.
- - ---.__.a--._.____.._, . - _ _ . . _ _ _ _ _ . - _ _ _ - _ . - _ _ - _ _ _ _ . __ _ _ _ _ _ _ _ _ - _ - _ - - - _ - - -
A-MODIFIED ELASTIC SYSTEM O
Sorina Force-Disolacement Hysteresis A
PY : ,
/ [
Spring '
Z xe i Force '
/ lXrl s'
/
- /
Spring Displacement o Stiffness reduced to account for plasticity.
Effective stiffness calculated to give yield force at maximum relative displacement.
i o Damping increased to account for irreversible work. Effective damping calculated to give same irreversible work per cycle as elastic-plastic system.
Modified Elastic Model ,
///////////
Input Support Motion, y Xs(t) = A sin (w t)
I
- Effective Damping, Ce Effective
- Stiffness, xe I .
Mass, T 0 u Absolute Motion, Xa(t)
10-ELASTIC-PIJLSTIC SDOF MODELS O.
' Exacts Numerical Elastic-Plastic Model
///////////
Input Support Motion, e Xs(t) = A sin (k/ t)
Structural Danping, C Bilinear Spring, K, Kp I
Mass, M i
i Absolute Motion, Xa(t)
Static Force, F v
\
( Assume Bilinear Sorina - Includes strain hardening d
py , _________ gp I
spring i Force K I
O Xy Xr Spring (Relative) Displacement o Assumed Kinematic Hardening as shown on following page.
o Static force is included such that ratchetting may be simulated, o ' Exact' time history solution using numerical solution.
.___~ . . _ _ _ . . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ . _ _ . _ . . . _ _ _ . ___
O O
~
Spring Force-Displacement O Cycle .
Bilinear Spring w/ Kinematic Hordening 1.6 .
1.4 -
1.2 - '_
a 1- Yield j Pfostic Loading
~
0.8 - '
c u ~~__
~'
h O.6 -
h- a a O.s - Elastic Loading Elastic
.c a O.2 - UnIooding un n O 1, n '
D
/
E -0.2 -
f
.9 /
a -0.4 - / Reversed Loading U *
/
E -0.6 - / - ~
~6 / ~~_
-0.8 - ,' ~~, ,,
, Ruersed Yield
- 1.2 - -
- 1.4 -
- 1.6 i , , ,
-3 -1 J
1 3 7 Dimensionless Spring Displacement
~
O RELATIVE EL-PLkYNAMIC RESPONSE O I FOR SOOF SYSTEM WITH 2% DAMPlNG 5 .
i i T 4-
.25.
N l
T X
ELAS11C ( < 1 )
3-
! b i b i c
! 3
- o.
i
{
w 2-2
\
f E
lxci/ xy - 10,
\
1_ -
s
\'
l I '
Reglon where linear elastic i
i ,
onalysis is non-conservative O i i i i i i
i iiiiiiiiiiiii O O.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 1 P
DIMENSIONLESS FREQUENCY. W / Wn I 6
, - . - - , , , _ - n-- ,------v --
o e t
O
~
SPRING STRAIN vs TIME .
CASE 19, Xg - sin (2 T) F -O.S. L-0.1 7 .
Ratchet t l
5 - - - - - - - - - - - - - - - - - - - - -
7
- ~ 0 ] J OJ Y UO E 3-JJ 0 V 2-m V 1-O l
-1 O
h' 20 <O TIME (dimensionless) v- v v vv. v v ve * + > - ,. -
e- # < >-r+ -m-+
noc-sts42 8
_.14.-
O
\
' \ $
s s ( -
o 5
\'s 1, s\
, 's
,O*g -' '
t D
'g
_ s' o \
\
h %
\\ - ,
s'
\ .
g 5 \
\
3 ,
,s
\
\ \ ' ~~ , %
\ \
- f T O '4 \
1
_J
\
\
s g \ \ -
- \
y e ss %
d \
4, ' s --
s, t
s Y g\ ' g s
\ U.
g \
s t
\
\
\
\ ,
~ g g , \
\ r
.. 4 e. # *
- 4 e O e, d .\
- g O O 6
- Y' r - # O.
\ g \
gg4N O
3-210
-IF-CONCLUSIONS FROM SINGLE DEGREE OF FREEDOM (SDOF) SYSTEM ANALYSIS 0 SDOF F. VALUATIONS SHOW THAT ELASTIC ANALYSIS MAY NOT BE CONSERVATIVE FOR APPLIED FREQUENCIES BELOW THE NATURAL FREQUENCY.
O PEAK BROADENING MAY BE NECESSARY TO ASSURE THAT ELASTC ANALYSIS IS CONSERVATIVE. THIS ACCOUNTS FOR THE SHIFT IN NATURAL FREQUENCY WITH PLASTICITY.
O RATCHETING OCCURS W8.jEN SMEAN + bDYN bY O CUMULATIVE RATCHET STRAIN ER*bMEAN ! P l
i
() 0 SDOF ANALYSIS INDEPENDENTLY PREDICTS THE SAME CUMULATIVE RATCHET STRAIN AS THE MILLER MODEL FOR BILINEAR KINEMATIC HARDENING.
O SDOF MODEL MAY BE OVER CONSERVATIVE COMPARED TO RESULTS OF COMPONENT AND SYSTEM TESTS:
- GROSS SECTION YIELDING ASSUMED INSTEAD OF LOCAL YIELDING IN BENDING
- VARYING E p IN THE ACTUAL STRESS STRAIN CURVE INSTEAD OF LOWER CONSTANT E p IN THE BILINEAR MODEL
- HIGHER MATERIAL YIELD STRENGTH DUE TO '
STRAIN HARDENING
()
h
~l b-()
RATCHET CRITERION BASED ON RESULTS OF COMPONENT & SYSTEM TESTS O WHERE GOOD DATA ARE AVAILABLE, THE MEASURED STRAINS CAN BE USED TO DETERMINE THE STRESS LEVEL BELOW WHICH THERE IS NO RATCHETING.
i 0 FOR TYPICAL MEAN STRESS VALUES ( 0.5 Sg)
SIGNIFICANT RATCHETING WAS NOT OBSERVED FOR STRESS AMPLITUDES BELOW APPROXIMATELY 6 Sg FOR BOTH CARBON STEEL AND STAINLESS STEEL AT I
ROOM TEMPERATURE.
()
0 THIS STRESS VALUE MAY BE USED AS THE STRESS LEVEL BELOW WHICH SPECIAL FATIGUE OR RATCHETING ANALYSIS IS NOT NECESSARY.
i i
1 4
,ea 'w ---wy-- ww-w-,-,,amw c.---,cww,,,vc,--,w,,
O O O
~
RATCHET THRESHOLD -
FROM COMPONENT TEST RESULTS
) 8 =
O 7- + + No Ratchet O Less than 10% in 50 cycid3 i + 5 Creater than 10% in 50 cycles X
" 6-4 i
N O O
- ^ O a a50E +
j d 5- m
! Z E O Ot 1
0 0 0 0
}
y +- 3- -o -D--- ##
4-
++ ' g f C - *- -B + O- -
4+
3 19 0 * + 0 4 %+S
- w + + + p #
+
c 3] +
+
++ 4.+
4 0 + R a 0 O f 0+ M ++
g E 2- _ _ _ _ _ __ + _ _. g - +
+
l 1- #
l 4
l 0 mimumm mim un m uu nn um u mi mmimn ummm unn um mm miumu nm mnuun muurm mmimummim 1
HALF FULL 6 10 11 12 57 910 11
13 1964 5 1
TEST RUN 4 4 i l
l l
. , , , , _ .~
t o' e O
~
RATCHET THRESHOLD .
FROM COMPONENT TEST RESULTS 15
'i 13 - t- No Ratchet O Less than 10 % in 50 cycles a
+ gr Creater than 10 % in 50 cycles
! E 12 -
m + O
! N 11 -
! E 10 - a 0 k 9- 5 a Os 4 0 z
f h 8- E .
b 0 0 0+
)
a 7- + 0 0 0 0 ,,
+ O lbI O @ O O
+
6 - --
+ - - - - -
,-p -
- _O b _C +- - _-
qgQ4_--
5-. +
- c
+ ++
E 4- + 0 m 0MO +
43 +
\
3- _ _ _ _ p _. _ _ +_ _ _a$- +
2- #
l
- i 1-O
- ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,n ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,',,,,,,,,
HALF FULL 6 10 11 12 57 910 11 12 13 1964 5 i
TEST RUN l
1 D
j
RATCHET HRESHOLD FROM STAINLESS STEEL COMPONENT TESTS 8
+ No Ratchet O Less than 10% in 50 cycles 7_ E Creater than 10% in 50 cycles 3 5 5 M O 6- O +
\ + Of E --+ a
( 5-
+
+ +
Z 0
& + +
& 4- ---- ----
7___--____ +
e + 0 O 0
w O +
+ 3- + 0 U
c o
+
O O e + + +
D +
g v
E 2- - - - - - - - -
-_________w+ O O
+
1- +++
O sisiiiiiiiiiii,iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiirisi 4 5 5 6 7 10 TEST RUN 1 a
I
I o~ a v
p v
, RATCHET THRESHOLD FROM STAINLESS STEEL COMPONENT TESTS 12
+ No Ratchet 11 - O Less than 10I in 50 cycles 3 Creater than 10% in 50 cycles 10 - ,
" " o E
m 9- o +
N + Ot E 8=~+ R
+ +
7- +
Z o g + + ~
y 6- ----------zy
+
+ o o a w 5- O +
+ 0 4
c 4- +
o O O f
3- --------------U O a r > - - - - - - -- d -
2- +
+++
1-i O iisiii,,isissi,i,,,,,,i,,i,,si,,iiisiisi,siiiii siiiisiissieiisii 4 5 5 6 7 10 1
TEST RUN G
~21-O POTENTIAL DESIGN RULE CHANGES O PRESENT CODE LIMITS '
PRESSURE + EARTHQUAKE STRESS (Eo. 9) LIMITS LESSER OF LEVEL B 1.8 S M OR 1.5 S y LEVEL C 2.25 S g OR 1.8 S y LEVEL D 3.0 S g On 2.0 S y ,
FATIGUE ANALYSIS FOR LEVEL B; NO SPECIFIC RATCHETING CONTROLS.
O ER0 POSED NEW CODE LIMITS ,
PRESSURE + EARTHQUAKE STRESS (Eo. 9) LIMITS LESSER OF 1
LEVEL B 3.0 S g OR 2.0 S y LEVEL C 4.5 S g OR 3.6 S y ,
LEVEL D 6.0 S g OR 4.0 S y i FATIGUE ANALYSIS FOR LEVEL B; NO SPECIFIC RATCHETING CONTROLS.
() r
-~-----,--..-n,---n--,,_.,,,_____.___
-2 z -
,4 >
STRATEGY FOR CODE IMPLEMENTATION O
SNORT TERM GOALS i
l 0 PROPOSE FOR INCLUSION IN CURRENT NB - 3000 REQUIREMENTS. WITH THIS APPROACH IT MAY BE DIFFICULT TO IMPOSE RESTRICTIONS ON ANALYSIS APPROACH.
O INCLUDE NEW RULES IN A CODE CASE; IMPOSE SPECIFIC RESTRICTIONS ON HOW IT IS USED:
- RESPONSE SPECTRUM ANALYSIS l
- PEAK BROADENING TO BE INCLUDED
() - DAMPING NOT IN EXCESS OF 5 PERCENT O USE NON-MANDATORY APPENDIX APPROACH l
l l
I i
.i
- 2' 3 -
l
() LONG TERM GOALS O TEST DATA SHOW THAT PIPING COMPONENTS CAN i
TOLERATE LOADS WELL BEYOND CURRENT CODE LIMITS 0 THE CURRENT DESIGN CRITERIA MAY BE ADDING TO COSTS SIGNIFICANTLY WITHOUT COMMENSURATE SAFETY BENEFITS
+ SUBSTANTIAL DESIGN / ANALYSIS COSTS
+ INCREASED HARDWARE COSTS, E.G., i SNUBBERS, PIPE SUPPORTS
+ INCREASED MAINTENANCE COSTS
+ POTENTIAL FOR INCREASED STEADY STATE t STRESSES IF SNUBBERS ' LOCK UP'
() O STATIC ANALYSIS MAY IN FACT BE ACCEPTABLE IN
- MOST CASES RESULTING IN SIGNIFICANT i SIMPLIFICATION 3
0 LONG TERM DESIGN ANALYSIS GOAL SHOULD BE TO l.
IMPLEMENT REALISTIC STRESS LIMITS WITH
{ SIMPLER ANALYSIS METHODS l
l 1
a a
9 O
1 4
4 PIPING AND FITTING DYNAMIC RELIABILITY PROGRAM i
e OVERALL PROGRAM STRUCTURE 4
h j e OBJECTIVES OF PROGRAM
- O e PFDR TEST PROGRAMS j es COMPONENT TESTS es SYSTEM TESTS '
! es SPECIMEN TESTS
) :
l l
I
)
i O
I l
O OVERALL PROGRAM STRUCTURE EPRI/NRC PIPING AND FITTING DYNAMIC RELIABILITY PROGRAM (PFDR) e TASK 1: PROGRAM PLAN DEVELOPMENT - GE-SJ
> e TASK 2: PIPE COMPONENT TESTING - ANCO ;
> e TASK 3: PIPE SYSTEM TESTING - ETEC, ANCO
> e TASK 4: SPECIMEN FATIGUE RATCHETING TESTS-MCL O
>e TASK 5: ANALYSIS OF TESTS AND DESIGN RULES -
)
i GE-SJ e TASK 6: IDENTIFICATION AND DEVELOPMENT OF ALTERNATIVE DESIGN RULES AND REGULATIONS - GE-SJ e TASK 7: EVALUATION OF ALTERNATIVE DESIGN RULES f - GE-SJ i
j e TASK 8: PROJECT FINAL REPORTS - GE-SJ ,
i
- O !
O OBJECTIVES OF PROGRAM t
, MAJOR OBJECTIVE:
DEVELOP AN IMPROVED, REALISTIC
- AND DEFENSIBLE SET OF PIPING DESIGN RULES FOR INCLUSION IN ASME CODE l
- 1. DETERMINE ACTUAL FAILURE MECHANISM FOR PIPING SYSTEMS AND COMPONENTS ,
i C) HIGH FREQUENCY IMPULSIVE LOADS O
i
- 2. MEASURE PIPING SYSTEM DAMPING FOR VARIOUS STRAIN l LEVELS OVER A LARGE RANGE OF FREQUENCIES
- 3. DETERMINE INFLUENCE OF SUPPORT FAILURE ON PIPING SYSTEM RESPONSE
- 4. SHOW EFFECT OF LOW FREQUENCY INPUT TO PIPING FROM BUILDINGS SUBJECTED TO LARGE AMPLITUDE ;
I EARTHOUAKES l 5. DEMONSTRATE THAT PIPING COMPONENTS AND SYSTEMS i
CAN TOLERATE EARTHQUAKES MUCH LARGER THAN SSE j O WITHOUT PIPE FAILURE ,
i L _ -_ ._ _ _..
f
O OBJECTIVES DE PROGRAM -(CONTINUED)_ ,
- 6. DEVELOP LABORATORY PROCEDURE FOR QUANTITATIVE EVALUATION OF FATIGUE-RATCHETING
- 7. QUANTIFY ECONOMIC BENEFITS OF NEW DESIGN RULES TO UTILITIES BY APPLICATION TO TYPICAL NUCLEAR PIPING SYSTEM
() 9. DEVISE SIMPLIFIED METHODS FOR ACCOUNTING FOR PLASTIC DEFORMATION, AND FATIGUE RATCHETING
- 10. SIMPLIFY PIPING SYSTEM DYNAMIC ANALYSIS l
O i
l
i l
O PFDR IESI1ROGRAMS l
l
_COMPONEN TESTS e MOST SEVERE LOADING ISPECIMENTESTS e MOST INSTRUMENTATION e DEMONSTRATES COMPONENT e DEMONSTRATES BEHAVIOR RATCHETING e DETERMINES FAILURE MODES e EVALUATES MANY e PROVES FUNCTIONALITY MATERIALS e HELPS PREDICT SYSTEM TESTS e DETERMINES
! e CALIBRATES DESIGN RULES TEMPERATURE EFFECTS
- SYSTEM TESTS [
l l e CONFIRMS REDISTRIBUTION OF LOADS e CONFIRMS MODE OF FAILURE l e CONFIRMS FUNCTIONALITY e CONFIRMS DESIGil RULES AND MARGINS PROVIDES BENCHMARK ANALYSIS DAT o . . -
l O i COMPONENT TESTS l ,
- OBJECTIVES !
e DETERMINE FAILURE MODE (S) UNDER DYNAMIC LOADING e MEASURE RATCHETING AND CYCLES TO FAILURE l e DEVELOP ENGINEERING UNDERSTANDING OF COMPONENT
- BEHAVIOR COMPONENTS TO BE TESTED (6 IN. DIA., SCH 10, 40. 80)
I ELBOWS, TEES, REDUCERS, N0ZZLES, SUPPORT CONNECTIONS O
l PLAN FOR COMPONENT TESTS e INPUT PEAK AT 0.5 HZ BELOW COMPONENT NAT. FREQ.
e ANCO SLEDS OPERATED AT HAXIMUM EXCITATION DESIRED RESULTS l e FATIGUE RATCHET CRACK IN 2 - 3 SEISMIC INPUTS i
j ACTUAL RESU1IS
) ~ e- SCH. 10: CRACKED IN 1/2 TO 3-1/2 SEISMIC INPUTS I e SCH 40: CRACKED IN 1-1/2 TO 3-1/2 SEISMIC
' INPUTS - OPTIMUM l O e SCH. 80: CRACKED IN 5 TO 9 SEISMIC INPUTS i INVESTIGATING BEHAVIOR AT INTERMEDIATE LOAD LEVELS
e O
PIPluo AhD P1TTir4 ,qiuAuft RftfARftltY PtDat&M topp'M af f f1i tummaRV i
to ffPt NAT til Pittl LOAD M LOAD P P CYC thPUT X k0 FAIL SCm sit Olt Lim som tipt sinAlu Ltytt 0 in u00 g 1 /1328 00 (1) 1 Elbow Cl 1500 1P 1.21 sli 2.51(1) 15 5 at 40 (setest) 2600 1P 1.21 88t 1.5t(1) 15 0.5 Ft ,
80 2 tibow Cs 1500 0P 1.04 Slt 1.41(i) 15 5 nr 80 (tetest) 2600 0P 1.04 Sit 1.41(1) 15 4.5 it 80 i 3 t t bow $8 3.5 400 1P 2.36 sit 2.4t(1) 21 3.5 ft 10 '
4 Elbos C8 1000 1P 1..J Sit 2.01(1) 18 2.5 ft ;
40 5 tibow Cl 13.8 1700 1P 2.06 sit 2.03(1) 21 3.5 ft 40 6 Elbow S1 16 1700 lP 2.00 888 2.01(1) 19 3.5 78 43 7 Elbow 88 9 1000 lP 1.80 Blt 2.01(1) 23 4.5 ft 6,0 8 Elbow '88 1.5 0 1P 1.80 Sit 2.01(1) 24 5 hF d
40 9 fee Fla 2 88 8 1700 0*P 2.50 Slt 2.21 21 1.5 ft 40 10 fee Fla 2 88 6.5 1000 0P 2.40 sit 2.21 21 2.5 78 40 11 fee Fla 2 SS 3 400 0P 1.00 Sit 1.91 16 0.5 ft 10 12 fee Fla 2 st 11 1700 1P 2.30 Blt 2.21 27 2.5 ft
) 40 13 Short Elb. C5 6 1000 1P 2.30 881 1.91(1) 22 2.5 H
- 40 i
i 14 fee fla 2 Cl 10 1700 0P ?.46 sli 2.23(2) 18 1.5 ft 40
, 15 teduc e r 88 18 1700 84 1.18 Btt 13t(3) 13 5 ft i 40
- 16 tedec e r 88 2.5 1700 See 1.72 Ist 3.3 30 0.5 ft O .
1
Pl#lhC AhD flf'IRC STE&N!f bill &illiff Pthee&M tcustoatti itsi sussaat? ,
F 80 ffPt RAT tts PREli LOAD M LOAD P.P CTC ltPUT I he FAIL Stu sit Die Liu 208 ffPE Sftelu Ltytt 0 in m;;I 1 /S!!! 00 -
(1) if thert Elbow CS 2.5 1000 tot s/A Sit 2.51(1) 20 3 ft 40 -
18 telmforced Feb. fee C6 1000 One sit 0.3 ft 40 19 E l bow Cl 2300 1.P lit 40 20 sessle 88 1000 itse lit 40 Culde two Cl 021 1700 C6
- 5 One $51 ft 40 Cirs. .
Nee.
22 Guide twg Il 1700 4:6 Blt 0.4 ft 40 Cirs.
see.
[
23 Stewt C8 1.5 1000 1P 2.3 Slt 2.1 t/A $ af 40 24 Ilbow C8 1000 1P 1.0 static 2 Cettepse 40 '
Closies i
25 ttbo. >!d 88 ( 800 uta 6.6 tv2 1.4 27 7 hl '
10 Mid 26 E l bow Cl 1700 1.P $1nesop 8 pa 40 i IF Tee fla.1 SS 8 1700 0P uld
- 9 bl f
40 time i 28 Waterhoemer C8 1700 fP 2.7 Solid W 2.2 3 tf 40 1000 1.P 3.1 Water slug 3 CollePse O
l
r a .
O PlPfuC AhD flfilet DTh& Wit IfLtatit!** Pto:0&M tempontet tatt tuovaev DO 1f91 ILA1 All Pttll LOAD M LOAD P*P CTC luPW1 I 40 # Alt i ICr sit Olt Liu 150s1 1TPI $1tAls Livit 0 ?> n0;t I /8tti 00 ft) 29 Waterheaper Cl 1700 1+P/ 2.7 Solid wh 0.55 3 at 40 Strut 1000 l'P/ 2.8 Water 0.40 3 at strut Stwo 30 ftboe 1.4 sa il 0 1.P 4 at $$t 1.36 3 ft 10 410 1*P 1.3 ut 2.0 31 Ilbow 4.1 53 $$ 410 lP sineswp 3.5 ft 10 e llt I
32 tlbow 88 1700 1P 3 Stette 0.5 3 er 40 Opentes 33 Pipe 15 ut Cl 1000 g/A 1.1 linesop 1.1 10 af l 40 34 Pipe 6 us Cs 1000 m/A 1.8 Stoeswp 2.16 4 ft 40 35 Elb uigh Wt Cl 1700 1P 1.65 Bli 3.4t(1) 18 5 Ft 40 le lee Ftr 1 Cl 1700 thrw re lit 0.5 ft 40 37 Ilb 1.4 at il 4 0 4P 1.03t4) 858 2.01(1) 10 i tt 10 38 fee fla 1 CB 20 1700 0P 1.92 lit 2.75(3) to 3.6 ft 40 39 fee Fla 1 88 10 0 0P 1.84 lit 3.4t(3) 'i 4 tf 40 40 ledscer $8 9 0 8s4 1.2 til 3.3 1 2 Il 40 et E l bow Cl 1000 lP lit & list 40 0
a i
4 1.L*12L1 we . weie, ee or I.p e {m* Plane 0*P e Owt of Plame Fla*1 e Single End Fine.
Fla 2 e goth Ends Ptaed b0 to e tumber of high levet imput test runs to towse felture fg a fatigwe retcheting fellwee ,
el e he f ellwee 30 e tetthet gwtkling ist a fatigwe retchetteg fellwee esd felLoved by dwetite teerleg gesidwet Strain e stesured by 2 leth stretch marks impwt I
- Level C e Celtwtete9 stress using timeer response spectrum emetysis, 21 Demping, g ill bres.eming -
- . 4 d stiwat sied in ,es, use the seitwisie. st,ess, is, er >, divided .y te.ei 0 enesteie.
' 38 to tetorales owltiple of Level D allseeble.
b R1111 (1) f or all the etbees, the meesweed strains are em the owtside surf ace. for strain om testce s.* fete.
3 mwltipty the velwes by 1.54 The inslet surfere is 1.364 times of the ewiside, princl al P strain is 1.072 time of tirtseferemtiet i
! strain and the tirewnferentist strain to 1.01 time of everage strain over page leagte 1/16". r 1.334 a 1.072 s '.01 e 1.10. For 2* stretch mark, the fetter 1.01 is increase to 1.52 ans tot j multiplicattom in 1.388 e 1.072 a 1.52 e 2.26, i f or retthet Strain en the inside surf ace, e ** tt) of 2.0 over the tabuteted valves.
l j (2) Case felted too early, there is t'most ne date, uf, previews eleiter test run date.
- <>> .nge .iis, too e.,i,. It is est e., if ,es, veiwes have ,een e.teimed.
J (4) Weight stress over 1000 pol.
4
- t 1
__ ,_ , . _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ . _ _ ~ _
l
P J
\
s t
w t o
s M u s 1 2 1 g 1 , 1 1 o t
2 ,
r a o B
s A
i T A m H w
t T u t
C s
e N a o E f t R w G t t O t s e s T T *
- t c c s w N e s e w N *e n a
c t
T V
a o
t o
s t
r 0
3 T
A o
e T*
o i
v A
o t
t U
L Aw T
LE AM T
m,"" 'sr to r E o
t n
m t
s a e R a n f s t e C E
O"AC o
.U oce R ;r T U
T-s t a o n s C
i r e i nA W8 8 s
e s n n m A s t t t t
t 2 I L 2, " oS s t
u e i i E
C t
C c i
C u s et *F ,
t't U t u
os a :
s Osw a g s D C A
C A
C A
C A Ht O "L "D o#
n EC L
r w- f P , o p
m t o o
n t C o 2 o x u
Y s
k l e M
v Y
g #
2 v 3 4 Y
g Y
2 Y
h o r u a
.m b w
o s '
l.
1 L s i
T
- v. -
- r. n i
Y. ,
- M , ' ',!_ r o
, f l
l l
l / k D
O x r_
d r
e u
n c
i
- e
_ r, ' n 3 l o e V. h r 3 T 4 9y -
s x f e
I
. r
[ -
h :.
\ A .V 4
m h i Yh' i
3 c- -
, . t n
\ 5 ~ f
_ *0 n
- a' 3
r 1 I
2 -
, I _
_Z= -
. _ -W n e
~
7 b * :_
t m %tf t a t
2 i U
Y
- g 58 M . ct 2 M e n v
Y 7
M*; i n
w z t o
=
Nll 7 9
E N P c
s a
s t
v l
Y,L I 2
g E N Y R U
s 2 3
xg f -
m s
P h
x miAr g
om
. _ _ - . . . _ _ - . . - . _ _ _ _ . . . . , - - . - - . ~ . _ - . . _ . _ - - _ . . . . . _ . ~ _ . - _ _ . _ _ - - . _ . _ .
I I
i i
l
~~
I!
g j $
I I
l 1 l
l #
! l . n
- a I I I i l l h .
N eme.s e
d i
- W b
C i
.r.
t C E
$ g b
c i
O : m..
x
+
)
ff f
3 c.
C
- e C
i C 1 l.
.I museese mm.
I . # 1 1 - h i
W
I , O
- l
- I N l
,1 I -
! I
.m -
i j
- I t I t i I f I I
_ _l C 9 o a e n o a e
- o 9' l E $ E E E, E E e
. . .E , -
I e D l (ISY O AIIOjod' NOlavmIT3339
.y l- ,l h
-.w ) t,,,.
e .I
/
i '. , '
/i, , jl
' I \ f.
( '
' I '
/<
se -- n O
e
() q l I G ,
, .O p.
!)
>2 2
w '
> r e (
g.,{ , / >
C, i ' ,., ,
i
@* s \' )
l
/
i i,
-f ,) >
? '> 'y '
)
I g
l i e b
\ , O ' u c.
v.
/ e
,7' *$
4 T
i C.
v f %
E y
D v
2 7 e
(
5
?
=
4 E
x
. < w
- i .
',d
, \ f
~
L c
.o t t >',, 0
'q
)
l
_ 0 e
ed M _f \ \
d Z "w }
I EW /
y , '
II i {-
g
/ E ,
, s l ChEg '
)
v4 E9 55 '
3 e
I , I ,
t.
@ o O r @ O g C" n $ e
- O i ,t gggm gMtH)J) NQllTd333))Y
\ O '
l i
i 6
i j /
._,,m.--- . . . - - . - - . , - . , - - - . - , , , . . . - - - - , . , , . , - - - . -,. . . , . - , - - - - . . - _ . - - - - - - - - - - - -
7.7
- Di j 4.2 k l r .
.ll p 1Y l lYnINO 'l
[hii )
o - 2.7 E -6.2 0
- 9. 7 . . -
0 0.4 08 1.2 1.t 2.0 TIME (sect Mid-Frequency Input Motion Time History (SRV)
(3 V - -
i 6
5 5.0 PERCENT DAMPING f
5 E ,
3 s
5 '
3 G ! }
s l E I l 9
., 2 i
\ i 5 l u y. l E
, /\ "
,s J \f o / ! I to3 too 101 102 FREQUENCY (HU I
Hid-Frequency Input Motion Response Spectra (SAV)
TEST 4 - ELBOW O
CARBON STEEL
! IN-PLANE SEISMIC LOADING
]F INTERNAL PRESSURE = 1000 PSI ,
o r -I i ,
l 1 g
'I i1 l
1
{ I
[! 1 l
I l' )
1, ,l
.I i tr i f(r (p j
w 1 , p 5 [ h i ; u ,! > t g
j
't l
- l l
I, ,I I
}l i :j l
n! ii i l C i i
~
I j f i ff g
e m&Yf0!"qh s L
g f' "
- j. gydb4[-)gpJ[r((M"ug
- L f "gM )/ $ W d i h% y o
i g, ,
, 7, j ", i i i i i i !
O 2.0 -
O O
/ O 1.5 -
C
! Oo f
O I g 1.0 -
O w
s
> 0 e
l 0.s o 0 O
i l
f f f f i e i O
1 2 3 4 5 6 7 SLED ACCELERATION (g) 1/2 (P.P) FIC' IRED ABOVE 12 N: 0 Mear.ured Elbow Strain Versus Input Acceleration (Test 3 Cehedule 10 Elbow)
O t
l
N += g Y
e N "
28 -
TEST NUM8EM TEE. STAINLESS STEEL, 26 -
2 -
PRESSUhE 22 -
20 -
5 j 1e -
1 m
g te -
O ,8 ,. -
12 -
- 12 to -
6 ~
4 -
,_ O O t , '
1 2 g S
HIGH INPUT RUNS (NUMBER)
Tatigue Ratchering Strain Based on 2-in. Wide Marks
, e,n, - -
I O
O 50 -
O TEST 1 (SCH 80) SCH 80 0 TEST 2 ISCH 80)
@ TEST 3 ISCH 10)
O TEST 4 (SCH 40) ,
D TEST 5 (SCH 40) 40 - > TEST 6 (SCH 40) f d a h SCH 10 l > .
p 30 -
/
O<
/
20 -
' i @
/
/
10 -
f 4-0 ' ' , , ,
. 0.05 0.25 0.50 0.75 1.00 1.20 g
MAXIMUM STRAIN (%) 1/2 (P- P)
Equivalent Damping Versus Maximum Str. tin
( ,
I l
l l
. - j I
O i t I
o
~ h y
- a s =
1 o O = 0 5
e 4
C 5 a 5 .
R 3
?
I l
, , i '
4 8 8 8 8'
, a
?
i sesysnesi - em vanen a
O' O O l T31 6 IN. ELBOW SCH 1O IN PLANE 410 PSI CALC MOMENT VERSUS MEASUREMENTS l 2000 LEGEN D 1800-~ MEAS l 2 % DAMP l 1600--
_____. 5 % DAMP 1400-- ._- 15% DAMP y
1200--
Z i /
l w 1000-- /
z /
i W ,/
$ 800--
- 2 ,
/
i 600-- / -
l
/ .....
l 400-- ,- *
,/
' ' ~~..... '
200 -- -
/ __. ..
^ ^
O 0.53 0.76 0.78 0.94 FREQ. RATIO: (PEAK SSE)/(NATURAL FREO.)'
NATURAL FREO. OF THE SYSTEM = 3.95 Hz
~
O O O' T31 6 IN ELBOW SCH 1O I P 41O PSI DISPLACEMENT VERSUS MEASUREMENT 20 LEGEND 18-- M EAS. REL E __ 2 % D 6 16--
~
Q ~~
_____ 5 % D 8 ___ 15% ARS BR 2o 12-- /
/
z /
o 10-- f n /
z g. _
/
$ l U 6-- / /
3 / /
0, 4-- ! '
2- -
g.,[_._
-i- -
O : -
0.53 0.76 0.78 0.94 FREO. RATIO: (PEAK SSE)/(NATU RAL FREO.)
NATURAL FREO OF THE SYSTEM = 3.95 Hz
~
O OBSERVATIONS FROM COMPONENT TESTS e DYNAMIC LOAD REVERSAL PREVENTS COLLAPSE e SEISMIC LOADS BEHAVE LIKE SECONDARY NOT PRIMARY e RATCHET FAILURE LOADS ) SSE e RATCHETING DOES NOT IMPAIR FUNCTIONALITY e DAMPING FOR LARGE DYNAMIC LOADS )R.G. 1.61 O
e AMPLIFIED HIGHER FREQUENCY SRV INERTIA LOADS CAUSE SMALL RESPONSE BOTTOM LIkE FAILURES ARE CHARACTERIZED BY FATIGUE AND/0R RATCHETING
- NOT STATIC COLLAPSE
- O
4 O
OBJECTIVES OF SYSTEM TESTS CONFIRM FAILURE MODE (3 OR 4 SLED INPUTS) l CONFIRM EFFECTS OF LOW AND MID FREQUENCY LOADS l l
DETERMINE SYSTEM DAMPING e BALANCED SYSTEM STRESS ,
e UNBALANCED SYSTEM STRESS e DIFFERENT TIME HISTORIES e W & W/0 SNUBBER AND STRUT CONFIRM FUNCTIONALITY CONFIRM DESIGN RULES AND MARGINS 1 i 1
~-
l
O SYSTEM TEST 1 PWR COMPONENT COOLING WATER THREE SLEDS CARBON STEEL A106B INTERNAL PRESSURE = 1000 PSI e4 h
i? .
E2 O V "
I5 Eo
- f. 3 '.
n es d -
% ,, c_: is'8gf Y fl l \ b 'k h h e
l wd e,
d' TABLE 4
\s ,
--. TABLE 2 TABLE 1 l
l l O
\ /
O y 6
- 6. 8 8 1 324 6. 0 0 x -
== 836 436 .
- t. T - - -
21 2 Wb T 4
i
=^= SSR Wb SSR '
~
wb SSR y -
A j
('
l 4 2 2 i I i i I 8 i e 2 q 1
9 8 0 8 6 6 l 1 1
- 4. 7 6 m 2 324 e
' ' = - - -
T t g
s Wb wb y SSR SSR S r
o f
y 4 4 r 0
3 0 3 a y
s a
k' I 4 a m
8 1
m
/ u r
y S t 2 3
O 284 8 588 4 s
. - - 7. 4 1 2 e xyz 224 8 r
-
- t DDD g S
y wh SsR 8
g t 52 0 I- 1
~
4 2I 2 5 - * - . -
T g W w g, 3
1 6 S SSR 3
=
8 61 9 Y I
31 1 b T
- - - SR C I
i SA -
wt , RP SSR AAC O
O D T EE 6 4 S 8 2 1 4 3 0 8 5 X 31 2 9
- 3. 9 6 t D S DE
- - 8 31 1 g 5 1 4 - .. 236 SSM g a= - ES8 t wb SSR wb g RES 1 SSR wb TR S STA GS I N t
SSR HO T S , I
- 0 6 HGC ET HA l GNT I
- 4. 3 2 WN E L A 36 EL 312 M B(P h t
TS E
S AS 4 - -
g 2 T S 1 =
CE 0
wb 6
k' 8
, y 3
YWb SSS R- N0 HS I I
SSR y O
s I ll
] 1' ;
!jll', j{l
l
~
l O '
l PIPE SYSTEM TESTING PERFORMED TEST RUNS FOR SYSTEM TEST 1 e .. UNIFORM e OBE )I ISM SSE ..
e 5XSSE, e 1/2 TABLE CAPACITY (33 SSEh I.. UNIFORM O .
FULLTABLECAPACITY(52SSE}
e MID FREQUENCY . UNIFORM, FULL TABLE l
O ,
l
.m.,el_.. .e a_m . ma_ aA .e. -A. .
- s. m h- e a a- m .4 _ m.. Aa. m .- a r- w a m .-
O l l
~
l !
l .
3-O . i
+
i
'7 s4 ,
i b s,
O
O i
l I
T i
~
1 l
u '
\\ ,
s.
1 i
'f O
O SYSTEM IESI 1 - IESI RESULTS AI FAILURE LOCATION PARAMETER SSE 5SSE HALF FULL (7SSE)- (33SSE) (52SSE)
INPUT X LEVEL D 2% DAMPING 0.8 5.3 27.0 42.0 5% DAMPING 0.5 3.1 16.0 24.0 (BROAD, SLED ARS)
MEASURED M0 MENT, 66 330 725 717 Q IN-KIP CYCLIC STRAIN, 0.09 1.2 3.7 4.4 P-P, %
RATCHET STRAIN, 0 0.44 8.5 2.4 DURING RUN, %
RATCHET DISPLACE, 0 0 1.0 4.5 IN. FOR 20 SEC RUN DAMPING, % 4.2 6.4 23 -
O
>= E U:
C' = W QZW I4 so N o #e
' ..............................................<.................................................,O w O
. . . . . . . < N N W o.
- N l . m == C
. . . rm
. to ,
gc
............... .................................4 W w ms
= .
- N%@"
4
. W
. . . . mm H :
OY
,,,,,j
. C 2
N Q. M : : :. :
. . O U :.
E : : : : : : C.J k
LAJ C : : :
W W
we I', .
M C :
- l l ; : : ~6
- : ; : : co m o SFy 1
. . . . . . . . .............................. ............. ,.CWsQ
- -gNy.
g gw: : : : :
. . s M es " W o
):
E C .
l h R gj glc : : :. : :
83 E C : : : : : l u p : :. : : J : : < C W
w
. . (f)
. ~n .- --
. g,,
HN : .
. E D :
==J y
-mur ^ __ -
- . : l l
- - T- '- 6 c6
- o, C t . .
- W N(
ms ,D e e y
l .
. . . . eemem . _
. Q ,.
p m.
"M. w, .
. . . N
- : : : ; so e
%.oM o b . . .
k i 2 o 00 ' dt. 00'ds 00'ds 00'dh ..00'ds 60S 00'da
'3060 00*dt N!881S 00 0 05"o I Otw (81508DIH) wru
_60S-3M th 03S E Eld
l l
M_ h - J ETEC UMM PL -
5/21/87 l
l TTF HERL-TIME PLOT
- l/2 TABLE CAPACITY I -. ........................... ........ ...... ... ........... ... ...................................... . .... ................ ......................
m . : : :
I : : -
i un-w I : :
- i. o : : :
z-. : : . :
-- . -..................................... -.. ......... ... :.............. 4 .............. ................. 3.....................................
i w :
j il ,
i i e : i i ! I i
! w : (; -
l a
. . . . . . . . v 1
4
! o I : .
I i
< s I .
1 1 .
j w ey i -
]li : :
~ .,
I n l y: ,
w-
............. . ,i .
l
.I . .
(
q'I
. . . . . yl
........................{:
a
~ ,
1 ' \
r I , ,
\
na ,1 .
l ,
o .
' ,g , ,
I l i .l l l
'a . , ) f. -
li
, i i
.M. I I i l .
C7 x
( ') I;' - -
,5 :
,, , e~ . .......... . , ....... ,
i:
y y' ) ,
, un s* I , ,
\ l l l .
- x -
I l :
r l g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... ...... . . .... ~ .. . ... . .;...................................j 4 x : '
, w I :
. o :
ot : : :
J I n, p :.......................... .......... ... . .
. . . . . . . . . . . . .. ........ . . . ..... -........... .. .....y....................................;
m : : :
o -
i i . : : _-
p_._ .
a i a a i I O s 10 is 20
! ELRPSED TIME (SECONDS)
ORY S/21/87 5/21/87 5/21/87 5/21/87 5/21/87 DfiY MH 16:03 16:03 16:03 16:03 16:04 HeH ISEC 40.000 145.000 50.000 55.D00 0.000 SEC l
)
O O O
SYSTEM 1 TEST DATA CORRELATION MOMENT VERSUS INPUT LEVEL 2% DAMPING SLED INPUT ARS 3000 LEGEND M EASU R.
2500-- /
j __. 2 % BR f/ ______ 2% UNBR.
//
2 2000-- //
g //
l / .I E / .i l 1500-- li fp 5 //
8 JE 1000--
/(/
//
/
//./
500-- #
4;
,.4 .-
O - -
c O.79 5.33 27.1 42.0 INPUT LEVEL (X TitAES LEVEL D)
O O O'
l SYSTEM 1 TEST DATA CORRELATION l ABS X DISPLACEMENT VERSUS INPUT 2 % DAMPING SLED INPUT ARS 20 LEGEND l
18-- MEASUR.
l
-2o D BR
- 2 16-- ______ 2% D UBR l z 14-- f
/
4 '
l d 12--
//
/ ..-
i $ ' ,/ '
j! S 10--
/
/ -
5 /
u_: 8--
4 5 /
U 6-- /
! /
vs w /
- 4 /
4 .
X .
I 2__ ~
/
f ' '
O 1SSE 5SSE HALF FULL l lN PUT LEVEL
C>sSYSTENTEST2 BWR RHR SYSTEM FoUR SLEDS STAINLESS STEEL 316L INTERNAL PRESSURE = 1000 PSI Y
ll Z
x 6.62 5 in. SCH 160 in. PLPE, 150 lb O 400 lb \
4 34 N 'IIdh 30
%32 ##,g
~
m.
0 o 24 0 'h LED 3 21
- p. 2p 6 6. 3 7 5 m.
O CD
/
%;> x, 4 ;,, ,
g,, , <
g'^e;,2 , ....
- a # s v
- ,y , o 9 to /
SLED 1 28 f pd ID 2 i
o 2" 4
4 7 i
2W 2
0 3 =
I == f y 8 6 2
=
22 BC w"
g' 0 705 B gN 1 2 927
= 3 0 1 2 1
=2- -
4 p ' C,
=
"=
T W
- 33, S,2 T
6
/
Z S R ; b S Wb R 506 SR 1 6 2
=,
SS, g-1 T
3 w3 SSR 0
1 g 395
- 5. # 0 1 I' 1 I 8 1 w*
~
=
fp 3. 2 4
" = 2
=
320 .
T u "V #
f === n 0 O 4 6 i
9 0M Wb SSR -
1 t
K 2 6 g
SgR
=
MIN 0496. 9 1
- 3.
- i 4
2 I
6
==
,f 0N
=
m 0 a 2 y
, Y, 2 r 5108 3 1 2 AAg w6 S$
g 1
=
j 4 1 a
[
= 2 i den T
Wb 3
2b g g\ o D s
SSR l u, s e
o 0 0 9 01 g %
2 3 2'y 6 O r
= ~, u. t 2
r 1
5 S
-= T 2 N W Y S g s
- 8. 6 9 e 050 t 6
3 8 === n y
, 3 g S D b Y E SSR T I
N C E 686 A D P 223 A A xwx C O
R 222 ' D 8 == E N 9 , ,=, L S
U (
1 0D0 S 3 N ' C E
R A 8 66
=
. T e s
)
. 5 21 8 6 E ip ip
% Z l
=
- . 1 T A k.. k. s 5 =
p 3 I.
6 y Z inni 2 2i A 8M w6
/
T M 30i n HE) Y SSR . W -
2
. TT( S 4 2 1 6 0 S R 8 = 9
_ EY O. .S
- S E S (1 i K E R RT MR E3 T N T S S T BOM I T 5 T S W I
( L L
( E E S R E 5 T z). 0 GH 00T 2 5 3
NG n4 4 .
I D D SHH6 a n Y W H. 1
- UH4 PG= 0 - NW3CC6 E
B = bSS /
NIHfn I
O 5 M [2 = TS .nn. in m.
I3)
I23 I8 (
[g5 bW.i S.SR644 i
a f I( fl jl! jl1 l l 'illi l
I O
TEST RUNS FOR SYSTEM TEST 2 e SSE D .. UNIFORM i
e 2 x SSE .. ISM-CORRELATED e 5 x SSE i .. ISM-UNCORRELATED O J l e SINE SWEEP l
e HALF TABLE e FULL TABLE .. UNIFORM p HID-FREQUENCY
}
O
O
. /
3/8 V T
,r 3 -
n <
12 in. SCH 40 PIPE U
4 in. 4 in. SCH 180 PIPE WELD
8 in. l PREP ON ENO FOR SCH 40 4 in. SCH 40 PIPE
/
4,+ ,
I 0 15' k m
/
/\ N BOSS k /
" 0.7 N N t 0.7 7 n /
0.375 \ T 31/2 in. o,37$ 7 0.208 a::; - 0.180 h
t U IP -; h,5 P -
9n 4:4 < < . . <,, y ) O O 010 U
h h 2.250
= 131/4 sn. ID-w 2.125
" D'A
!ss Fabricated Nozzle Test Detail O
O SYSTEM IE11 2 : IESI RESULTS AI FAILURE LOCATION PARAMETER SSE SSSE HALF FULL (9 SSE) _(18 SSE)
INPUT X LEVEL D 2% DAMPING 1.0 6.3 11.0 21.0 !
5% DAMPING 0.8 4.9 8.0 15.0 ;
(BROAD, SLED ARS)
HEASURED MOMENT, 39 119 156 235 i.
IN-KIP CYCLIC STRAIN, 0.21 0.77 0.96 2.8 P-P, %
l RATCHET STRAIN, 0.07 0.18 0.65 2.1 DURING RUN, %
RATCHET D1SPLACE, 0 0 0 2.1 IN. FOR 20 SEC RUN DAMPING, % 5.0 5.0 22 -
O e
_ . , , - - - _ - - - , , - , - - _ - , . ~ _ . . . _ , _ _ _ _ , _ _ _ _ . _ _ _ _ -- ,. ___,,, _,___,_.,_.._,._ _ __._,._, ,_, ,.. ,-___-_ , - ... _ ,_._--- .__,, ,, .- .
e-,,_, .,
7
)
N t
I f*
Y.)
u LLJ
' O "E O N
< 40 .,. e
%-C
^!i ..
. . . .................................................................................W
. . . . . N
="c
- : o g : : : :
r : : : N"aM
- eo.
J.
. i
.=*> . . .
s gm . . .
l N e c:p
- l % C
.... . ......... .... ........;.................................;...............;................ _O - e N e .
=
m" CD H : -
. : . l
%. : : J U I l i I pC : .
- i. - ,
o Q- : ae .
f.0 3
., j c: r.3 3
- . : ~
QQ M.J%Z^g_.. NmC::r .
- m:
N .. ...... - - -
... .........l....... .. . . . ..... . .. .. . ....... ...
4 -)
- - O .
!w4 (,g,j -- ! .
. (ij a
CO -
- : - O m _ . . . _
c- -=mmes _
i F .
I : : : : td g 8
,j W
- . I m
IC2 * .
- L.u a% : .
- : W IQ [ : N n l c ~
5 O LO $
g
...... ..........g......_ ....;..... ..........e.............. .t... ............ g e. ,. O, L.L.
. . \
H !. -
9 l
.! C- "m FI j
__ _ - - - - y
, H : : : :
i M : .
o (J") : ' -
~
- : : ==M" -
-l
- ~-
f5
- * .- 4:#
s n D.
N
... . ...._%_- ^_-
.m a
,. . m.
'. s ~, t. *
- i. _
~M, . . -.
. . - - - - _ .p=pr v
..=
. : : : : u r.
- : : : t e g a.mc s
1 r
. m..c c.
00021 000C! 00 0 0009 000h 0002 0 ~
I ) OIDS '3060 NIdW.lS >- :c v n f % "JM Ph M "J C ESW -
l.
~ YHC
~ R:E DHS b 7 8 0
$. -i- ! . .
,.:- :i I 5 2
/ 24 0 1
1 1
/ 3 1
- 0 3
1 Y . . .
,3 . . .
,A .
7 8 4 0
. . . . / 2 0
.:: .:- - -e 0 1 5
.::5:-
! u..
- .:.::3 - . :. ;.:::i -
~
2 1 3 .
/ 1 8 1
Y .
T .
I .
C 1.
T A 1
)
P S 7 O A .
L C .
D 0 4 00 N/
O 1 2: 0 P .
- . :- . - I i 5 1
1
' C/ 13 3 E E "M
L .
j . E1 B
,i S
OI l
. I T A .
' T .
E c -
s e8 l
t e L 2
. l'I l'
f i
$6 i M
s R /
iI I
r i
r E .
T 1 1
- i o R .
I' 7
D6 o .
I\ '
. 3 0 t - .
j s
E/1 2: 0 r . . '
0 e .
,i: .:::i::::. :.:.::j ..- : :::- :. i. :.:::3 ie0 1S1 F . [
_ . . i[1' 3 8 2 . .
P/ 1 1 T .
r R 5 T M
. .I f l I
L E
o E .
, \
.l l '
. T .
'll' .
_ S .
. 'l
.p Y .
_ S .
l 3 . 7 8
/ 3:
0 f
1 2 0
. , - - . .' . .:- : . . . ::: .:' ::- - : - - - i
. . 5 1 3 3
/
l . .
1 1
S
. l . .
I i(l '
I l
l1 .
g
_ . . . . 7 8 1 30
/ 2
. 1
- 0 i
_ t
'm.0 / 3 ra
- c 1 8
! 4
- _ i s_o~sm" n><tr u < o s2uxwoqgOo o rJ -
bs
' THC I
E mo,M Nh3~
. o o. D D i -
,Ijl , ; i 4i ) :
i' ,
O
-O ' -
~
~O '
(
'*g.
5 i DISPLACEMENT MEASUREMENTS VS CALCULATIONS EPRI SYSTEM TEST #2 f
NODE zoa (BROADENED ARS) 30 -
/
LEGEND i /
/
/ MEAS 25-- / _ _ _ _ 2 7;
/
g / 5%
- E / /
y 20-- l / ._-.. 15 %
e--e / / -
~
aca E- - /
/ '
/
z$5-- !
w -
/
e' l X / /
w ,
c) / /e '1-3 ' .
,/
a 10- e -
/ - .
l cn
-- s' / , '/ y-o i
/
/.- ,/
/
i
~
/ . /,./
./
,/
5--
~
/ '.' /
/./ ' /*-/y' f /C./
,s . ,< .
/
-^
g ._ ._ _ . .
1 2 5 0 18 INPUT LOAD (SSE) _
O
. O O' MOMENT MEASUREMENTS VS CALCULATIONS EPRI SYSTEM TEST #2 NODE 6 (BROADENED ARS) l 1400 -
LEGEND l
l 1260-- MEAS i
--- 2%
! 1120-- ,
m / 5%
l s g
980--
/
/
. _ _ _ _ _ 15%
/
- r: 840-- /
/ 35%
r) / /
z j O 700--
/
/ '
/
m / /
E-* '
- Z 560-- - /
w /
l 2 / f
@ 420-- / ,/ ,-
- ~~ / . ,
' b
140- //, '
_. ~^. ll L A , . ' '
~
O + : + +
1 2 5 9 18 INPUT LOAD (SSE)
O OBSERVATIONS FROM SYSTEM TESIS e FATIGUE RATCHET FAILURE MODE -
CONFIRMED e FAILURE LOADS > SSE -
CONFIRMED e
O FUNCTIONALITY UNIMPAIRED -
CONFIRMED e DAMPING > R. G. 1.61 -
CONFIRMED e AMPLIFIED SRV LOADS SMALL -
CONFIRMED BOTTOM LINE EXTREMELY DIFFICULT TO FAIL PIPING WITH DYNAMIC LOADS
!O ,
O i
WATER HAMMER TESTING COMPONENT TESTING e TWO SMALL LOOPS, TESTS 28 AND 29 e 6-IN PIPING SYSTEMS e CARBON STEEL, SCH. 40 e WITH AND WITHOUT SUPPORTS O SYSTEM TESTING e TWO LONGER LOOPS, MINI-SYSTEMS 1 AND 2 i
e 3-IN PIPING SYSTEMS o CARBON STEEL, SCH. 40 e SUPPORTS, BRANCHES, SIMULATED VESSEL, THIN PIPE O
l l
\
O LOADING CONDITIONS FOR WATER HAMMER TESTS e SIMULATED STEAM HAMMER TEST e HARD SYSTEM ACOUSTIC TEST e WATER SLUG TEST O
e VARIOUS PRESSURES FROM 150 TO 2000 PSI O
w_______-_- _ _ - _ __ __ .
f
. i l
- Y g GE COORD
, */
- x/N ANCO COORD y ;
l 6 in, SCH 160 i SEGMENT 1 O, Og4 ggG /, WATER LEVEL IN N SLUG TEST l
b,7.5 m. '
i
.\, 6 m. SCH 160 Ng4 h-
' s :
6 m. SCH 40 l
r O he 47 J TANK j SURST DiAPMmAGM ho M. -
l
/
CLOSIC IN SOLID TEST 2 m. OPEN IN SLUG TEST Test 28 Vater Masser Test Configuration i O l
i
[V M o
kb es f hj ~ U fit 22 E $ '
33 3f.
,y
~
48 C y
t g i
.. a "-
e". e a
E I
Ea E. -
a f il IS 5
I u
M
+--- 4 --.e.
A U-5 @
W =
2 3 3 R
- 5 o .
. j - - + ~
5 0 I *.. *%' 4
'f-t_J L la
{s$
~
2 W
>= w ,
M ^ N
> ' 3 $,
M e
.O M
= b k I
~
'~ E
~
did 5 2 ~
MF
- #'~ r
' S
, - t 3,
~a
=
S . ~- .
I A -
c 4 .3 "g
4 n
- C o
" $a,h. ; t)41 ar 8 a0 l
- u
=
l =
i 0
l
I l
- O :
ERELIMINARY OBSERVATIONS FROM WATER HAM ER TESTS e WATER SLUG CAUSES PRIMARY LOADING - CAN COLLAPSE PIPING e STEAM HAMMER AND HARD TESTS MORE LIKE SECONDARY LOADING - DID NOT COLLAPSE PIPING e SUPPORTS CAN TOLERATE LOADS 10 X RATED LOAD W/0 FAILURE O
e PIPE CAN TOLERATE TRANSIENT PRESSURES 2 X BURST PRESSURE W/0 FAILURE BASIC RULE:_- DESIGN TO AVOID WAT_ER HAMMER ,
1 0
1
O OBJECTIVES OF SPECIMEN TESTS e DEVELOP LAB SPECIMEN TO EVALUATE FATIGUE RATCHETING WITH HEAN STRESS e CORRELATE SPECIMEN BEHAVIOR WITH COMPONENT BEHAVIOR e EXTRAPOLATE CONCLUSIONS FROM 4 TEST MATERIALS TO OTHER PIPING MATERIALS e INVESTIGATE FATIGUE RATCHETING EFFECTS AT O TEMPERATURE (550 DEGREES F)
I i
- =.
O 1
1 i.._-..-.-_-._ _ _ _ _ _ _ _ _ _ . _ _ , _ _ _ _ _ _ _ _ _ _ . _ . _ _ _ _ . . _ _ . _ _ _ _ _ _ . , _ _ _ _ _ _ _ _ _ _ _ _- _ _ _ _ . _ _
O O
Modified Test Matrix Test Teap No. of No. of Tvoe F Mat'Is Tests PurDose Baseline RT 4 5 Same as original program Baseline 550 4 '4 Find high temp properties Two Bar RT 4 5 Find effect of low mean Low Mean stress on all materials Two Bar RT 4 5 Duplicate tests with high High .'fean mean stress Two Bar 550 4 4 Determine effect of tem; Low Mean on low mean stress test Two Bar 550 4 4 Duplicate temp tests for High Mean high mean stress
() Two Bar Rate RT 2 8 Investigate strain rate effects on the amount of Effects ratchetting Ten /3end RT 2 4 Verify Ratchetting on 2 smooth materials Ten / Bend RT 2 2 Test same 2 material f or notched notch effects press RT 2 2 perform four point bend Pipe tests on pressured pipe MaterialJ Tested l Material 1 A223 Grade 6 Carbon Steel l Material 2 A25J Type 304 Stainless Steel Materisi 3 A387 Grade 22 Class 2 Steel Material 4 A533 Grade B Class 3 Steel i
Notes:
- 1. When two materials are to be tested they are A333 Carbon Steel and A358 Type 304 Stainless Steel.
() 2. Number of tests are f or each material to be tested.
i l
. . _ , . . _ _ . _ _ . _ .__.m.._ - . _ . . = > . _ _ . . .
e
'e O
0.2 51/0.24 9 LN. OLA -
1.0 IN. R 32/ gogg g/
U \/
A VU O.501/0.499 IN. DtA
- 0 X 0
= 1,7 60 IN. r
= 2.234 tN r c 0.750 m LN.
C 5.220/5.216 IN. r-
/
O Uniasial Test Specimen
1 e
o O
c 10.0o w r e 3.825 m. I c 2.7 50 IN. ; C 3.625 IN. %
0.500 IN.
I J ' # t i g
- o. 5 l lll g ll l1.250 lN.
O I r g i I h .
1/4R r 1 1/4 R O o.375IN' j
T l/
j DRILL THRouGH 13/32 POLISH LONG8
/ 0 g,g f4 g g
( !
j
- g- o soo IN.e o 002 I
-h- o.750IN.
I M 1.sh m. .- ,
O
% e - c.s7s W.
o.760 W.:0.002 H l=-f 0.500sN. 0.002 d
CROSS SECTION Verification Specimen O
O O
r----"------,
O A I I I
IleII
- L rTh E A
%r I
H0H -
I, -
a h
+ t
\ _ -
~
O
- e. e 9,_ i W g&
s @
- 9 I
M Tm f _
c i L. _
+
V E 7 ,
1 I
1 wi8 .
l l
- 1 11*
O
- s 0
al>
O OBSERVATIONS FROM SPECIMEN TESTS e 2-BAR TEST CONSERVATIVELY ESTIMATES EFFECTS OF RATCHETING ON CYCLIC LIFE f
e BEAM AND PIPE SPECIMENS CONFIRMED 2-BAR TEST RESULTS O
e WITH CONTROLS ON CUMULATIVE RATCHET STRAIN, I
MEAN STRESS AND TEMPERATURE DID NOT AFFECT CYCLIC FATIGUE LIFE e CYCLIC CREEP OBSERVED IN LOW FREQUENCY SPECIMEN TESTING MAY NOT BE PRESENT IN HIGH FREQUENCY j SEISMIC LOADING O
i
_. - - __ - _- - . _ _ _ _