ML20112H242
| ML20112H242 | |
| Person / Time | |
|---|---|
| Site: | Mcguire, Catawba, McGuire, 05000000 |
| Issue date: | 03/29/1985 |
| From: | Tucker H DUKE POWER CO. |
| To: | Adensam E, Harold Denton Office of Nuclear Reactor Regulation |
| References | |
| NUDOCS 8504020184 | |
| Download: ML20112H242 (18) | |
Text
..
p
-t DUKE POWER GOMPANY P.O. DOX 33180 CIMHLOTTE, N.C. 28242 HALH. TUCKER 7stgy,,oy,
J" ".", """>o-March 29, 1985 Mr. Harold R. Denton, Director Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C.
20555 Attentio1: Ms. E. G. Adensam, Chief Licensing Branch No. 4 Re: Catawba Nuclear Station Re: McGuire Nuclear Station Docket Nos. 50-413 and 50-414 Docket Nos. 50-369 and 50-370
Dear Mr. Denton:
Attached herewith are twenty (20) copies of Revision 12 to Duke Power Company's report, "An Analysis of Hydrogen Control Measures at McGuire Nuclear Station". As noted in Revision 9, this report is
_ applicable to Catawba Nuclear Station. This revision provides responses to the questions submitted to Duke Power Company by letter dated October 3, 1984 (E. G. Adensam, NRC/NRR, to H. B. Tucker, Duke Power Company) and is responsive to License Condition 14 of Facility Operating License NP?-35 for Catawba Unit 1.
Included in Revision 12 are the previous Revision 11 pages. These pages have been changed to correct the page numbering. This information should be inserted in Section 7.0 of Volume 3.
On January 25, 1985, the NRC published the final rule on hydrogen control (50 FR 3498).
In accordance with 550.44 (c)(3)(vii)(A) each applicant for or holder of an operating (v) cense subject to the, and (vi) is required to li requirements of paragraphs (c)(3)(iv),
submit a proposed schedule for meeting the requirements of the Final Rule.
This is to advise that the design features of the McGuire and Catawba Nuclear Stations as described in the report "An Analysis of Hydrogen Control Measures at McGuire Nuclear Station", currently meet the requirements of the Final Rule.
Therefore no schedule need be submitted.
Very truly yours,
-l k pl Hal B. Tucker ROS: sib Attachment b0 V,\\9
\\
8504020134 850329 PDR ADOCK 05 g9 P
_s;
.5;
- Mr. Harold R. D:nton, Director March 29, 1985 Page Two cc: Dr. J. Nelson Grace, Regional Administrator U. S. Nuclear Regulatory Commission Region II 101 Marietta Street, NW, Suite 2900 Atlanta, Georgia 30323 NRC Resident Inspector Catawba Nuclear Station Robert Guild, Esq.
'P..O. Box 12097 Charleston, South Carolina 29412 Palmetto Alliance 21351 Devine Street Columbia, South Carolina 29205 Mr. Jesse L. Riley Carolina Environmental Study Group 854 Henley Place
-Charlotte, North Carolina 28207
1
\\
Response to questions submitted by letter from NRC (Elinor G. Adensam) to Duke (H. B. Tucker) dated May 8, 1984.
1.
With regard to the CLASIX code, the staff has previously requested clarification of the structural heat sink heat transfer models.
The following pertinent points have been derived from the responses:
i) Heat transfer is based on a temperature difference determined by (Tbulk - T,,j)).
ii) Heat transfer coefficients for degraded core accident analysis are determined from a natural convection (stagnant) correlation appli-cable to condensation heat transfer.
iii) CLASIX does not explicitly model mass removal due to condensation heat transfer.
Based on the description of the CLASIX structural heat sink model, it appears that the CLASIX model differs dramatically from generally ac-cepted approaches and is not, as is claimed, consistent with standard methods such as those used in CONTEMPT.
The differences are related to the treatment of the three items cited above.
By comparison, previously accepted approaches are characterized by the following:
i) Heat transfer is based on (T atureoftheheatsinkisleggtthan,i)j),whenthesurfacetemper-
-T sat;
- ' wall < sat
- ii) Heat transfer coefficients are based on condensation only when Twall < Tsat*
iii) Condensed mass removal is based on condensation heat transfer with provisions for revaporizing a small fraction of the condensate.
A more detailed description of accepted practice is containeJ in NUREG-0588 and NUREG/CR-0255.
The effect of the CLASIX models would appear to be the de-superheating of the atmosphere too rapidly thus reducing gas temperatures and possibly altering the combustion characteristics.
Considering the above discussion, provide the results of analyses, with acceptable models to determine the effectiveness of deliberate ignition for the Catawba plant.
The analyses should address the effects of hy-drogen combustion on containment integrity and equipment survivability.
Furthermore, the analyses should be performed to address a spectrum of appropriate degraded core accidents.
Specific items that should be addressed include:
a.
Model input and analytical assumptions; b.
Calculated compartment atmosphere pressure, temperature, and gas concentration transients; c.
Equipment temperature response profiles; 7 n-136 Rev. 19
~
s d.
Differential pressure transients between compartments which will 4
allow for an evaluation of AP effects on interior structures and mechanical components (e.g., doors, fans); and e.
Considering the capability of the containment shell, crane wall,
.and the operating deck,' perform an analysis to determine the maxi-l mum concentration of hydrogen which could be accommodated in a de-l flagration.
Your estimate should consider realistic initial conditions and approximate combustion parameters.
i
Response
I A justification for the use of the heat sink models in CLASIX was pre-sented to NRC when this question was first posed to Duke in Elinor G.
Adensam's letter of August 18, 1983.
That response appears on pages 7.0-129 - 7.0-133.
We have reviewed that response and continue to i
support the case that it makes for the adequacy of the original analysis.
Our conclusion is that no additional CLASIX analysis is required to
' justify the results of our original work.
We note, however, that the additional CLASIX analysis requested by the staff was performed by AEP using heat transfer models which were in accordance with the staff's request that the models conform to those of NUREG 0588 and NUREG/CR-0255.
The results of this analysis were reported l
to NRC by M. P. Alexich's letter dated march 30, 1984.
These results are very interesting in view of the theoretical arguments presented previously by Duke Power Company in support of the original CLASIX heat transfer i
models.
In.their work, AEP compared directly the original heat transfer models with those requested by the staff using identical geometries, initial conditions, and release rates.
The AEP results indicate:
[
1.
Pressure and temperature profiles are generally similar for the two sets of heat transfer correlations.
2.
The original CLASIX analysis tends to underpredict the temperature in containment at the peaks associated with the hydrogen burning by about 100*F.
3.
The original CLASIX analysis tends to overpredict the baseline containment temperatures (the temperature of the containment between hydrogen burning).
This indicates that the original CLASIX heat sink models remove less energy from the containment atmosphere in the period immediately following a hydrogen burn and therefore provide a conservative baseline containment temperature profile.
4.
Further evidence of the conservatism of the original CLASIX heat sink models can be found from examining the containment pres:;ure response.
In every case, pressures during the hydrogen burn period
{
were higher for the original CLASIX analysis than for the analysis i
using the " corrected" heat sink models.
This indicates again that the original CLASIX heat sink models remove less energy from the containment atmosphere per unit time than the heat sink models based on NUREG-0588 and those used in CONTEMPT.
I 7.0-137 Rev.12
r 3
i In summary, analysis performed by AEP wherein a head-to-head comparison of heat sink models was made supports the position taken by Duke Power in its previous sub'ittal concerning the question of CLASIX heat sink models m
~
(Revision 10).
These models have been shown to be conservative from both a theoretical and an analytical standpoint.
The higher peak-temperatures during hydrogen burning predicted by the " corrected" heat sink models are of no consequence to the analysis of equipment survivability as our surviv-ability analysis used the adiabatic flame temperature (1400*F) rather than a lower temperature predicted from CLASIX results.
4 The ability of the hydrogen ignition system has been shown to be effective in controlling the concentration of hydrogen to levels less than 8.5% by volume in CLASIX analyses, small scale testing, and more recently, in the large scale Nevada tests.
Our structural analysis has consistently shown considerable margin in the containment design in its ability to withstand the pressures and differential pressures associated with hydrogen burning at this concentration.
To seek some maximum theoretical higher concent-ration which could be tolerated represents an unrealistic extension of our previous work and, at best, can be considered of academic interest only, and of no consequence in proving the adequacy of the concept of deliberate ignition.
Further support for the adequacy of the CLASIX code is presented in reference (a), wherein CLASIX is compared with HECTR.
For identical input conditions, and in spite of considerably increased technical comp-lexity in many of the HECTR models, results from the two codes are nearly identical. We conclude that the models contained in CLASIX are suitable for use in analysis of beyond design basis conditions, and that further discussion of CLASIX is unlikely to affect our confidence in it as an analytical tool for the study of deliberate ignition in ice condenser containments.
2.
Provide a complete evaluation of fan (both air return and hydrogen skimmer as applicable) operability and survivabilicy for degraded core accidents.
In this regard discuss the following items:
a.
The identification of conditions which will cause fan overspeed, in terms of differential pressure and duration, and hydrogen combustion events.
b.
The consequences of fan operation at overspeed conditions.
The response should include a discussion of thermal and overcurrent breakers in the power supply to the fans, the setpoints and physical locations of these devices, and the fan loading conditions required to trip the breakers.
c.
Indication to the operator of fan inoperability, corrective actions which may be possible, and the times required for operators to complete these actions.
d.
The capability of fan system components to withstand differential pressure transients (e.g., ducts, blades, thrust bearings, housing),
in terms of limiting conditions and components.
l l
7.0-138 Rev. 12
s I
- e a
Response
This identical question was submitted by letter from NRC (Elinor Adensam) to Duke (H. 8. Tucker) dated August 18, 1983.
It was answered in Revisions 8 and 10.
3.
Provide an analysis of the pressure differential loading on the ice condenser doors created by hydrogen combustion in the upper plenum and upper compartment.
Describe and justify the assumed or calcu-lated door positions.
Provide an evaluation of the ultimate cap-ability of the ice condenser doors to withstand reverse differential pressures.
Discuss the probable failure modes and the consequences of such failures; including the impact on a) adjacent equipment and structures, b) ice bed integrity, and c) flow maldistribution.
Response
Referring to previous CLASIX results for measures of the intercompart-mental differential pressures results in unrealistically conservative answers.
This result is caused by the manner in which CLASIX models the lower inlet and intermediate deck doors.
The dynamics of door closing contains no inertial term; therefore the doors close instantaneously whenever the net force in the closing direction is greater than zero.
For example, as soon as an upper plenum burn is initiated and upper plenum pressure increases, the intermediate deck doors closed instant-aneously.
The pressure rise in the upper plenum will therefore be conservatively high as venting into the ice bed will be precluded.
This effect was noted in the comparison of CLASIX analysis with similar analyses using HECTR and COMPARE reported in reference (a).
In addition, reference (a) states:
"During burns, CLASIX predicts fairly large pressure differentials between the compartments, which we would not expect to occur, given the large flow areas connecting the compartments.
HECTR predicts rapid pressure equilibration, and only small pressure differences between compartments.
As shown later, COMPARE also predicts rapid pressure equilibration".
Based on the discussion above, differential pressures obtained from CLASIX might be considered a gross upper bound for the differential pressures which would be developed in an actual hydrogen burn situation.
A review of previous CLASIX analysis reveals the following results.
For an upper plenum hydrogen burn initiated at 8.5% by volume, and a flame speed of 6 feet /second, the maximum indicated differential pressure across the intermediate deck doors is 1.2 psid.
As reported in an answer to a previous question, the reverse differential pressure capability of the intermediate deck door is 6 psid.
There is therefore substantial margin in the intermediate deck to withstand the reverse differential pressure associated with an upper plenum burn, even under the bounding conditions of an analysis using CLASIX.
l 7.0- 139 Rev.12
- (
3 r
For an upper compartment burn, which is shown to be precluded except under the most extreme assumptions, the pressure rise time is relatively slow due to the length of time it takes for the flame to propagate throughout this large compartment.
Results of the EPRI Nevada large scale tests show that hydrogen is reliably ignited by top ignition at 6%
by volume in the presence 'of sprays or fans, and that the corresponding flame speed is less-than 10 ft/sec.
Pressure rise times are less than one psi /second generally for the cases where typical plant conditions have been modeled.
We conclude that upper compartment burns cannot exert large differential pressures across the top deck doors, even if the doors are assumed to be fully closed.
In an actual hydrogen burn, the differ-ential pressure would be minimized by the increase in flow area caused by dislocation of the top deck blankets during the early portion of the accident.
4.
Identify the essential equipment needed to function during and after a de-graded core accident.
Provide the location inside containment for this equipment.
Response
This information has been furnished previously to the staff on at least two occasions.
Refer to reference (b), Section 6.2, and to Section 5.2 of this volume.
5.
In view of the recent TVA test results with Tayco igniters which indicate desirability of additional spray shielding, please discuss whether supplementary spray shields may be appropriate for the glow plug igniters.
Response
None of the glow plug igniters found by Duke Power to be required for adequate coverage of the containment is exposed to a spray environment.
The four additional igniters added to the upper compartment at the request of the staff are in the environment created by the containment sprays; however, we note the following:
1.
During the small scale testing reported in Chapter 2, there was no evidence that a spray environment had an adverse effect on the performance of the glow plug igniter.
2.
The tests performed in the large scale test vessel in Nevada, in which' ignition was started by glow plug igniters located at the center and bottom elevations (and thus in the spray) show no evidence th 4 containment spray inhibits the ignition of hydrogen by glow plug igniters.
We conclude that no further testing or modification of the glow plug igniters is required for McGuire or Catawba.
l 7.0.140 Rev.12
r I
References:
(a) Camp, Allen L., Vance L. Behr, and F. Eric Haskin, MARCH-HECTR Analysis of Selected Accidents in an Ice Condenser Containment, Sandia National Uboratories.
(b) A_n Analysis of Hydrogen Control _ Measures at McGuire Nuclear Station, Volume III, dated January 5, 1981 (this has been referred to as the " Grey Soo k").
l 7.0 -141 Rev.12
o 1
l:i -
Response to questions submitted by letter from NRC (Elinor G. Adensam).
to Duke (H. B. Tucker) dated October _3,1984.
1.
Recent containment analysis for degraded core accident sequences performed as part of the NRC Severe Accident Sequence Analysis program suggests the:need to further address:the survival of
'certain essential equipment for an expanded set of degraded core
. accident sequences. Specifically, analyses using the MARCH and HECTR computercodes -indicate that for more recent' estimates of
. hydrogen and' steam release rates for the Stu sequence and certain other equally probable degraded core sequences, the tempe'ratures.
s L
and differential pressures to which certain essential equipment may be exposed can exceed that calculated using the utility-developed release rates and combustion assumptions. The difference in calculated temperatures and pressures is due to changes in the timing, location, and magnitude of hydrogen burns as well as the mass and energy release rates for the blowdown.
- Considering the above discussion, provide the results of analyses to determine the effectiveness of deliberate ignition for the Catawba Plant. 'The analyses should address the effects of hydrogen combustion on containment integri!.y and equipment survivability, and should be based on accepted structural heat sink heat transfer models, as outlined in our May 8,1984, Request for Information.
Furthermore, the analyses should be performed to address a spectrum of appropriate degraded core accidents. Specific items that should be addressed include:
a.
Model input and analytical assumptions;
.b.
Calculated containment atmosphere pressure, temperature, and gas concentration transients; c.
Equipment temperature response profiles; and d.
Differential pressure transients between compartments which will. allow'for. an evaluation of differential pressure effects ion interior structure and mechanical components (e.g., fans and doors).
Response
In considering how to formulate a response to this question, we reviewed the previously developed containment analysis performed by Duke Power, our contractors, and that performed by NRC contractors. We note that Duke Power has submitted to NRC for review twenty separate containment anayses which are applicable to both McGuire and Catawba. These submittals began with early work using CLASIX (Reference 1) in which high concentrations of hydrogen were examined in order to create very conservative results due to uncertainty in the physics of hydrogen burning in containment.
7.0-142 Rev. 12
nm
-7 These uncertainties were resolved by the' completion of 'the extensive research program sponsored by the ice condenser utilities. We followed that work with a more detailed examination of containment response (Reference 2), incorporating in our models the results of research done at that time, reducing unrealistically conservative results and more accurately reflecting the best estimate models
- and conditions -identified during the research program..Nevertheless,
'we retained ' conservative assumptions concerning the nature of the accident which ' produced hydrogen release in containment and of the burn parameters which we expect to be characteristic of deliberate
. ignition.
Included in this second submittal was analysis which showed the sensitivity of our work-to changes in the major parameters of the analysis, such as hydrogen release rate and the effects ~of steam or oxygen inerting. We also considered at that time the
- effects of unavailability of single trains of the various safeguards
' systems. These-previously submitted analyses supported the effective-ness:of deliberate ignition in protecting the containment _ building and the vital _ equipment in it from adverse effects due to hydrogen release.
NRC, in its own research program, sponsored experimental and
. analytical work in the response of the containment to hydrogen release and burning. The most extensive'of these programs
-(Reference 3) analyzed 16 different accident scenarios and 53
'different containment analyses based'on these accidents. This work by NRC contractors used a containment analysis code developed by them under NRC sponsorship called HECTR and a previously
~
developed subcompartment analysis mode called COMPARE. The effect of these two' extensive and expensive research programs is a considerable amount of information concerning the response of the containment building, and the components and structures inside it, during the ignition of hydrogen.
Ir considering whether additional analyses are required for this problem, we have~ attempted to determine by our review of the previous results exactly what specific additional work should be done.
In doing so we asked'the following' questions:
1.
'Is there any reason-to doubt the validity of previous analysis, either because of the modeling assumptions, the particular computer codes used, or lack of knowledge of the physics of the reactions analyzed?
- 2. ' Has the on-going research program produced results since the development of the work reported in references 1, 2, and 3 which contradict previous assumptions about the behavior of the phenomena involved, in either conservative or non-conservative directions?
3.
What level of confidence is required in the analysis - that is, at what point do we decide that further analysis, or further 1
refinement of the analytical methods, is useless in that the ultimate conclusion of the analysis will not be ' altered by e
l further work?
7.0-143 Rev. 12 L
.. ~. _.
y
'6.
-It'is apparent from the general scope of the first question presented by NRC that the staff continues to harbor some' doubt about the validity of earlier work. This is somewhat surprising in light of the staff's conclusions reached during the review of the hydrogen issue performed for the McGuire station (Reference'5) and the-resul.ts contained in the reports of the confirmatory analytical-work carried out-by the NRC contractor (References 3 and 4). Not only'did the work reported in Reference 3 confirm the validity-
. of the CLASIX. computer code, in the configuration used by Duke to r:
. develop the results presented in Reference 2, the work reported by Reference 3 presented no cases wherein the integrity of the containment was threatened by hydrogen, under the ground rules.for the analysis of this problem which we have been following for five years. - Specifically, unless the core melted, or containment
- minimum engineered safeguards failed in some way, containment
. integrity is' protected from the effects of hydrogen release by
- the distributed ignition ~ system... Since this point is now so clear from analysis, there does not appear to be a need to perform additional work in assessing the response of the containment building to accidents in which significant quantities of hydrogen are released into containment. Reference 4 provides analytical information concerning the possible temperature response of equipment inside containment to accidents involving hydrogen release and burning and it too endorses the conclusion reached by Reference 2 that the S20. accident - the accident on which the hydrogen mitigation program for the ice condensers is based -
1 does not threaten equipment inside containments.
l We were left somewhat puzzled after this review of the specific intent of the staff's question. Despite its general scope, it appears to be related simply to the Reference 4 results which point i
to excessive equipment temperatures for certain specific accident i
sequences examined by Sandia in the analytical work reported in Reference 3.
Reference 4 : reports results of analytical work first reported in Reference 3, concentrating on the containment heat sinks in the Reference 3 analysis which represent' equipment simulants.
Specifically, these were small-disks and plates with dimensions representative of ~small metal parts which might be part of-an instrument in containment. We note a number of problems with this analysis, enumerated below, and consider it inferior to the work we have done and reported in Reference 2.
The specific problems are as follows:
1.
The small metal heat sinks were insulated on one side, thus L
the only way they could-lose energy to the environment was f
by radiation and convection from one side. This is a very simplistically conservative look at a three dimensional L
process in which heat transfer from all surfaces, including the effects of conduction, arc also important. The small i
heat sinks have very high surface area to mass ratios, thus 4
maximizing temperature rise beyond what would be expected in I
an actual accident situation.
d
[
7.0-144 Rev. 12
~,, - -
._,_.-_,__.-_m_.-
-i 2.
A review of the data contained in Reference 4 does not support the contention by NRC in the question above that an S2D sequence produced temperatures which exceed those.alculated using.the utility methods. The highest temperature produced for any S2D sequence reported by Reference 4 was produced by the analysis of the identical input conditions that the utilities used for their analysis, that of the CLASIX base case. Since we took a much more detailed and careful look at equipment survivability for the case of the CLASIX base case than did Sandia in Reference 4, we conclude that the S2D
' sequences do not produce temperatues which threaten equipment survivability, and that the Sandia analysis produces unrealistically conservative results when compared to a more detailed analysis.
3.
The highest temperature responses reported in Reference 4, and the only accident sequences in that document for which temperatures beyond the LOCA qualification temperature are produced in their simplistic equipment simulants, are the S1 sequences. We do not consider S1 sequences to be a valid candidate for consideration in the performance of the igniter systems for several reasons. First of all, Si sequences occur with a lower probability than the smaller LOCA's for McGuire and Catawba. Second, the particular series of events which lead to degraded cores, as typified by the TMI accident, are not associated with large breaks - there is unambiguous indication of a LOCA, ECCS injection is much more effective due to the lower primary pressure, and S1 sequences quickly become core melts beyond the scope of the hydrogen mitigation system rule.
We also se~iously question the ability of MARCH r
to model hydrogen release from a large break LOCA in any realistic way. As noted in Reference 3, it is very difficult in MARCH to initiate core recovery prior to melting, and MARCH is unable to properly predict release rates of steam and hydrogen when tb core level is at the level of the break. We note also that Reference 3 reports that the ECCS injection, when supplied at its normal flow rate, stopped hydrogen generation abruptly well below a level corresponding to 75% metal / water reaction, and that it was necessary to make unrealistic adjustments to the ECCS flow in order to produce any significant hydrogen release and retain a recoverable core geometry.
In short we question whether the Si sequences are a valid consideration in the design ard operation of the hydrogen mitigation system. They certainly do not appear to be valid for McGuire and Catawba due to their lower probability of occurrence and high probability of leading to core melting beyond the scope of the hydrogen rule.
In any case, we do not agree that the analysis reported by Reference 3 of these sequences is a valid indicator of the response in the field due to the severe limitations of MARCH and the unrealistic constraints which were placed on the system response to the problem in order to produce the reported results.
7.0-145 Rev. 12
4.
As noted several times in Reference 4, the temperatures in the dead-ended compartments are well below those in other compartments in containment. All of the vital instrumentation for Catawba is located either outside of containment or in the dead-ended compartmencs.
It is the conclusion of Reference 4.that hydrogen burning in containment does not threaten equipment in the dead-ended compartments, and we concur with that conclusion based on our previous work.
As part of our review in formulation an answer to this question, we also performed a review of the vessel thermodynamic response and equipment survivability results of.the EPRI sponsored hydrogen burn test program in Nevada in the large scale vessel. We can find nothing in any of this work which indicates that the hydrogen igniter system would not be effective in protecting the containment against hydrogen burning. On the contrary, all of this work confirms the effectiveness of the ignition system, shows that the earlier analysis we have sublaitted over the years to NRC are based on conservative assumptions, and have confirmed that the hydrogen control rule recently published is an adequate set of requirements for hydrogen control in ice condenser containments.
The pressure and temperature response results of this large scale test, as specifically related to the assumptions and models used for containment analysis, are discussed in detail below.
In regard to equipment survivability, a number of conclusions can be drawn from the large scale test, specifically:
1.
In spite of repeated exposures to burning environments, equipment operability was not impacted significantly by bydrogen burning. As noted above, the vital equipment at Catawba which is temperature sensitive, such as instrumentation transmitters, are located in dead ended compartments and are not exposed directly to hydrogen burning.
2.
Cables exposed to multiple hydrogen burns showed surface damage, such as charing or blistering in some cases, but passed post test electrical tests with no significant problems.
The only failures reported were related to the manner in which the cables were secured in the test vessel and not as a result of the hydrogen burning.
It should be noted that the EPRI test series included burning at much higher concentrations than would be present in an actual accident in containment, with correspondingly hotter flames and faster temperature changes.
The equipment, where relevant, was also exposed to a LOCA-like environment containing steam and water spray. We conclude from the results of the EPRI test program, which confirm a more limited series of tests done at TVA's Singleton Labs, that hydrogen burnign does not pose a threat to the survivability of safety related equipment in containment.
7.0-146 Rev. 12
6 The EPRI Nevada test also was extensively instrumented so that the pressure and temperature response of the vessel could be measured.
These large scale experiments were designed to confirm the results of work perfomed by the utilities and their contractors, and by the NRC contractors at Sandia and Lawrence Livermore Laboratories.
The EPRI tests have provided the following significant information:
1.
Under conditions representative of the upper containment (the dome, in Sandia terminology), global hydrogen burning was reliably ignited at 6.5% or less, regardless of igniter location. Under dynamic injection conditions, hydrogen burning in the simulated upper compartment conditions began to occur as low as 3%. The turbulence induced by the contain-ment sprays caused burn completion of close to 100%, even for hydrogen concentrations as low as 6%. These results therefore render the Sandia work reported in Reference 3 invalid with respect to burning in the upper compartment. Sandia repeatedly refers to "large dome burns: occuring in various accident sequences, indicating that these "large dome burns" are responsible for the major pressure rises seen in the accident.
The Nevada tests prove that "large dome burns" do not occur, and that if hydrogen is to be burned at all in the upper compartment, it will burn at concentrations far below those which would produce adverse effects on the containment building or on the internal containment structures and components.
These Sandia results in Reference 3 can therefore be considered overly conservative up and pressure response.per bounds for containment temperature In spite of that conservatism, the integrity of containment is assured assuming capability at ASME Service Level C.
2.
In the tests involving fan only, representative of the conditions present in the lower compartment during hydrogen burning, burningis shown to occur consistently at hydrogen concentrations of 7% and below. Again during the dynamic injection tests, hydrogen burning occurred at much lower global concentrations due to the development of diffusion flames.. Not only do diffusion flames significantly reduce the pressure response in containment, the localization of burning effects reduces the threat to equipment, because the separation requirements for redundant safety equipment guarantee that diffusion flames cannot adversely affect both trains of any vital safety system.
It is also interesting to note that the utility consultants postulated in 1980 that hydrogen burning in the ice condenser upper plenum would be predominately by diffusion flames. We conclude that the assumptions used by Sandia in Reference 3, that of prohibiting any burning in the lower compartment and ice condenser until the global concentration reaches 8%, is unrealistically conservative and therefore produces results which are not representative of actual accident responses in containment.
It would be unscientific to attempt to use these results produced by Sandia to call into question previous 7.0-147 Rev. 12
i.
-work done by the utilities, particularly when that work has been confirmed by large scale experiments.
L We continue to be interested in the best characterizations of the response of the Catawba containment to accidents, both with and without significant hydrogen burning. The objective,of our effort is to develop both qualified computer analysis codes, and qualified containment models which will place in our hands the capability to address future issues concerning the performance of the ice condenser containments. Our program includes accident analyses using the RETRAN and MAAP codes (for design basis and beyond design basis accidents, respectively), and the use of a modified version of CONTEMPT 4/M005. Modifications already in CONTEMPT 4 include revised ice condenser door models, properly accounting for the effects of gravity, closing springs, and differential pressure. Future modifications include flame propagation capability. Some preliminary results from this work are available.
In an attempt to establish better the actual hydrogen and steam release rates from postulated accidents in containment, we have performed analysis using the MAAP code, Reference 5.
Analusis was performed for the small break LOCA
.and transient cases which are of the highest probability based on the PRA work performed for McGuire. The MAAP code calculates the release-of hydrogen to containment in a much more proper way than MARCH, taking into account the holdup and dilution in the primary system and the simultaneous release of water, vapor, and non-condensibles when the water level is at the break. A number of interesting conclusions were noted when our results were compared with those of Sandia:
1.
The amount of zirconium reacted in these accidents is typically very small. MAAP consistently predicts no more than about 25%
of the core undergoes corrosion. These results are consistent in a sense with the Sandia work in which only about 37% of the core could be reacted without melting the core beyond recovery.
Sandia results were then empirically extended to 75% metal / water reaction levels because of the requirements of the gydrogen rule. There was a corresponding increase in hydrogen injection rate.
It can be argued whether the 75% metal / water reaction i
specified by the hydrogen rule represents a realistic way to ensure margin in containment response.
It is certainly true that MARCH results, confirmed by MAAP, show that one cannot i
get 75% of the clad reacting with steam and maintain e.n unmelted' core geometry which is likely to be recoverable by resumption of ECCS injection.
l 2.
For the S2D cases where MAAP and MARCH can be compared, MARCH consistently overpredicts the amount of water released from the primary system. This overprediction, coupled with unrealistically conservative assumptions concerning the ignition limit for hydrogen in the lower containment used in the later HECTR analysis reported in Reference 3, leads to the incorrect results obtained by Sandia concerning lower compartment burning.
7.0 -14 8 Rev. 12
4 2.
Provide a complete evaluation of fan (both air return and hydrogen skimmer) operability and survivability for degraded core accidents.
An evaluation of fan operability and survivability was requested by the staff in Requests for Information dated May 8,1984, and August 18, 1984; however, the utility responses are incomplete and do not provide an adequate basis for the staff to reach a conclusion.
In this regard, discuss the following items:
The identification of conditions which will cause fan overspeed, a.
in terms of the magnitude and duration of differential pressures required to produce overspeed and hydrogen combustion events.
b.
The consequences of fan operation at overspeed conditions. The response should include a discussion of thermal and overcurrent breakers in the power supply to the fans, the setpoints and physical locations of these devices, and the fan loading conditions required to trip the breakers.
Indication to the operator of fan inoperability, corrective c.
actions which may be possible, and the times required for the operators to complete these actions.
d.
The capability of the fan system components tc withstand differential pressure transients (e.g., ducts, blades, thrust bearings, housing), in terms of the limiting conditions and comoonents, An assessment of whether the requisit conditions for overspeed, e.
tripping, or failure of the fan systems, will occur for each of the spectrum of degraded core sequences and the impact of anticipated fan behaylor ib the progression, of the accident.
Response
As confirmed by the results of the testing performed in Nevada, upper compartment burns, if they ever occur in a global manner, occur at hydrogen concentrations of 6.5% or less. Burns occurring at this level of hydrogen do not create sufficient differential pressure across the fans to speed them up to synchronous speed, much less cause overspeed. These results from the Nevada tests provide convincing proof that upper compartment burns are not a threat to the containment or its internal structure and components, and no further work is required on fan survivability.
3.
Utility responses to staff questions regarding reverse differential pressure loads on ice condenser doors indicates an apparent inconsistency in reported values for both reverse pressure capability of the doors and the peak calculated differential pressures.
For example, the reverse pressure capability for the intermediate deck doors was reported to be 6 psid for Catawba and 2.8 psid for D.C. Cook; the peak differential pressures across these doors resulting from an upper compartment burn was 7.0-149 Rev. 12
7
~
reported to be 1.2 psid for Catawba and 12 psid for D.C. Cook.
Furthermore, utility responses do not provide a quantitative assessment of the reverse pressure differential loads across each of the doors resulting from an upper compartment burn.
The door positions and pressure loads resulting from upper compartment burns needs to be further examined in light of the recent MARCH /HECTR analyses which indicate a greater frequency of upper compartment burns than indicated in the utility analyses.
Considering the above discussion, provide a quantitiative assessment of the pressure loading on each of the ice condenser doors created by hydrogen combustion in 1) the upper plenum and b) the upper compartment. Describe and justify the assumed or calculated door positions. Provide an evaluation of the ultimate capability of the ice condenser doors to withstand reverse differential pressures. Discuss the probable failure modes and the consequences of such failures; including the impact on 1) adjacent equipment and structures, b) ice bed integrity, and c) flow maldistribution.
Response
The problem of the development of large differential pressures between compartments was addressed by the NRC contractors during -
their confirmatory research work, and the results are reported in Reference 3.
There it is noted that the CLASIX code predicts large pressure differentials between compartments which are unconfirmed by HECTR or COMPARE results. This is not surprising in view of the ice condenser door models contained in CLASIX, and the simplistic mass flow ' equations.
CLASIX was structured to provide conservative predictions of containment overall pressures; therefore the flow models used and the assumptions about ice condenser door operation were set up to maximixe the overall containment pressure response. Using more appropriate models of flow, which accurately predict the venting or pressure between compartments, and of the doors, which close in a realistic manner when subjected to differential pressure, HECTR provides realistic predictions of compartment differential pressures.
Reference 3 notes that irtercompartmental differential pressures are very low.
Additional information has been gained form the Nevada tests in this matter, and that information was presented in the response to question 2.
With hydrogen burnign occurs at very low concentration levels, much below 8%, and venting between compart-ments effectively reducing dif.~erential pressures, we conclude that the internal structures of the containment are not threatened by differential pressure, and there is no need to do more detailed structural analysis of the ice condenser doors.
7.0-150 Rev. 12
r-The following additional information is supplied in response to the concerns of the staff over the perfonnance of glow plug igniters in a spray environment. There are four glow plugs exposed to the upper compartment sprays. These igniters were added at the request of the staff during the McGuire licensing proceedings. At that time we provided technical justification which demonstrated that mid-level igniters were not required to ensure ignition of hydrogen in the upper compartment. This has now been confinned by the results of the EPRI Nevada tests.
The EPRI test has shown that the effect of spray in the upper compartment is to generate sufficient turbulence that hydroger.
is burned at very low concentration, as low as 6%, and that theses extremely low hydrogen concentrations are ignited reliably by igniters located only at the top of the test vessel.
It was also shown that igniters at the midlevel do not contribute in any wasy to the effectiveness of an igniter system in an environment typical of the upper compartment in an ice condenser.
We conclude that igniters exposed to spray at McGuire and Catawba are not needed, and their operability under spray conditions is irrelevant.
It is also significant to note from the EPRI large vessel test that there was never any indication that the spray environment, which was scaled to that of the containment, impaired the operation of glow plug igniters, even for hydrogen concentrations of 6% and below and igniters located at the mid-level and bottom of the test vessel.
References:
1.
Duke Power Company, An Analysis of Hydrogen Control Measures at McGuire Nuclear StatT6n, Volume 2, dated November 17, 1980.
2.
Duke Power Company, An Analysis of Hydrogen Control Measures at McGuire Nuclear StatT6n, complete through Revision 11, dated May 22, 1984.
3.
Camp, Allen L,. et. al., MARCH-HECTR Analysis of Selected Accidents in an Ice Condenser Containment, NUREG7CR-3912, December, 1984.
4.
Dandini, V.
J., et. al., HECTR Analysis of Equipment Temperature Responses to Selected Hydrogen Burns in an Ice Condenser Containment (draft), Sandia National Laboratories, June,1984.
5.
NUREG-0422, Supplement 7, dated May, 1983.
7.0-151 Rev. 12
,