ML20071C702

Forwards Response to Questions Re Thermal Margin Beyond Design Base,Design & Analysis.Overly Conservative Assumptions Made by NRC Consultants Re Treatment of Sodium Aerosols Inappropriate

ML20071C702
Person / Time
Site:	Clinch River
Issue date:	03/04/1983
From:	Longenecker J ENERGY, DEPT. OF, CLINCH RIVER BREEDER REACTOR PLANT
To:	Grace J Office of Nuclear Reactor Regulation
References
HQ:S:83:219, NUDOCS 8303070061
Download: ML20071C702 (14)

Text

.

Department of Energy Washington, D.C. 205'45 Docket No. 50-537 HQ:S:83:219 MAR 01 M Dr. J. Nelson Grace, Director CRBR Program Office Office of Nuclear Reactor Regulation U.S. Nuclear Regulatory Commission Washington, DC 20555

Dear Dr. Grace:

THERMAL MARGIN BEYOND THE DESIGN BASE (TMDBD)

Enclosed are the CRBRP Project's responses to questions raised in recent discussions regarding our TMBDB design and analysis. For your convenience, we have labelled the responses with designators (e.g. "Al") which we understand was originally established by your Brookhaven National Laboratory consultants.

Overall, the Project considers several of the overly conservative assumptions made by the NRC Staff's consultants with regard to treatment of sodium aerosols to result in an extreme departure from best-estimate analyses, and are inappropriate for assessments of beyond design basis accommodation features. While the Project believes it is feasible to design to accommodate these extremely conservative conditions, we expect to demonstrate in the OL review that conditions are within the Case 1 evaluation shown in the enclosure.

On a related matter, while the Project has not reached a final decision regarding the specific composition of limestone concrete l to be used f or the reactor cavity floor, we have discussed this

with the staf f and understand that either calcitic or dolomitic l limestone concrete would be acceptable from a TMBDB perspective, I

based on available experimental data.

If you have any questions, please call Wayne Pasko (FTS 626-6096) of the Project Office's Public Safety Division.

l Sincerely, g John R.'Longenecker b p 1 830304 Acting Director, Office of l A 05000537 Breeder Demonstration Projects PDR Office of Nuclear Energy

Enclosure:

As stated cc: Service List Standard Distribution Licensing Distribution

~

.~ . .. . . .

[ .

l AL - NRC consultants have recently completed thermal analyses of the CRBR containment for the TMBDB scenario taking into consideration the potential insulating effect of sodium aerosol deposits on the containment wall and dome. The project had earlier performed

! analyses of such an effect and discussed them with the staff at the April 16, 1982, TMBDB meeting. The two analyses were performed with significantly different assumed thermal properties of the plated aerosol (see Table Al-1) and, as a result, the analyses predict significantly different containment conditions (see Table Al-2, cases 1 and 2; the Base Case, with no aerosol i

insulation eff ect, is shown for comparison) . The project has reassessed its earlier calculation and concludes that it still represents a best estimate bound suitable for beyond design basis l analyses. Further, the project concludes that even conservative assumptions enveloping all potential uncertainties would be significantly less severe than the staff consultant's calculations.

The project has been reviewing the bases for the differences in the two estimates of the conductance of the aerosol layer. This review indicates that the primary dif f erence between the two l analyses is in the amount of aerosol predicted to be plated on the wall and dome. The walls and ceiling are emphasized because i.

this is the heat transfer path to the annulus cooling system; aerosol deposits on the floor and other horizontal surfaces have j only minor impact on calculated containment temperatures, except that once deposited there they are not available to insulate the wall or dome. The form of the deposits on the wall and dome, particularly the porosity and the resulting conductivity, is also important in determining the effectiveness of the aerosol ae an i insulator. Finally, other approximations in the evaluation, particularly the timing of the deposits, contribute to an unrealistically severe estimate by the staff's consultant of the thermal effect of the deposits.

I The amount of aerosol assumed to be plated on the walls and ceiling in the NRC staff consultant's calculation (3.9 g/cm2 in i

Table Al-1) was obtained by adding 50% more to the value orginally calculated by the staff's consultant at Battelle I

Columbus. It is our understanding that the parameters selected 1 for use in the model for thermophoresis, as the driving force which encourages aerosol to deposit on walls, led to the high mass per unit area in the calculation. To assess the reasonableness of the staff consultant's calculation, the project compared the resulting values against the latest available empirical evidence. The selected standard for comparison is the ratio of material deposited on the walls and ceilings (" vertical" surface in the analysis) to the material dep> sited on the floor and other horizontal surfaces. Comparing deposited mass per unit area, (mass loading) to allow for different geometries, the Battelle calculation of 2.66 g/cm2 involves a mass loading ratio of 7 to 1 in favor of deposition on horizontal surfaces. In all of the experiments performed at HEDL to date, comparable measured values have always been in excess of 100 to 1. In fact, in the

l most recent experiment (AB-5) which included thermophoretic l

conditions prototypic to CRBR TMBDB, the measured value was 180

to 1 (reference Al-1).

The project is convinced that these recent data prove the staff.

consultant's original mass loading calculation to be excessively l conservative. This viewpoint has been discussed with the staff l and consultant. The Battelle consultant has at least partially

~

concurred and has recalibrated his thermophoretic model parameters based on the newly available AB-5 data. The recalibrated model, including the project's latest value actual horizontal surface in containment, yields 0.38 g/cmforas the total wall and ceiling deposit, a value which the Battelle consultant estimates may still be high by a factor of two. The project considers the final adjusted value, 0.19 g/cm2, to be a conservative upper bound (the resulting ratio of horizontal to vertical deposits per unit area is approximately 80 to 1) . The project's best estimate bound in Table Al-1 is still only half as much or 0.10 g/cm2, implying a ratio of 160 to 1. This value is

quite close to the 180 to 1 mass loading ratio measured in the i

AB-5 test.

The form of the aerosol deposits, including porosity and i conductivity, must also be determined, before the thermal impact of the mass loading can be calculated. The staff consultant at Brookhaven who performed the thermal analyses assumed a porosity of 60%, but judged the value nonconservative. The staf f's Battelle consultant has observed that an earlier study at Battelle (BMI-NUREG-1977) supports this value. That study was a Millikan cell analysis of individual sodium oxide agglomerates at various humidities which yielded an average porosity of 60%. The l project has again tried to base its own evaluation on the available experimental evidence. Porosities in excess of 954 (very light " fluff") have been observed in settled deposits on horizontal surfaces in experiments at HEDL. The characteristics of wall deposits have not been measured and reported at HEDL, however, since thick wall deposits (> 1/8 inch) have never been observed in sodium aerosol tests; therefore, wall deposition has not been considered to be a significant sodium aerosol effect either by the experimeters or by analysts, for whom the ,

experiments are being performed. There is thus some uncertainty with respect to a quantitative upper bound for wall porosity though qualitatively the project is convinced that direct application of horizontal data would not be justifiable. The project's judgement is that 604 porosity is a reasonable best estimate value but given the lack of direct measurements approximately 90% was used for the Case 1 analysis. In conjunction with this porosity value, the project has concluded that the conductivity value of 0.05 BTU /hr.-f t-oF suggested by the staff's consultant is an appropriate one.

Given mass loading and porosity, one can determine average deposit depth. Given the depth and the conductivity, one can '

determine the overall conductance which is the principal input to

-__ .___,_m . _ . _ _ _ __ _. __ _ _ _ _ . _ _ _ _ _ _ _

~

i i

thermal evaluation of the aerosol impact. Enghinations ni l parameters PIpdRGlRS .thf SAEg overall conductang will 21mid the  !

same Answer ID .the thermal _ calculation- Thus the project's Case l 1 analysis would also apply approximately for alternate parameters such as 0.20 g/cm2 and 75% porosity or 0.39 g/cm2 and 60% porosity, though the determination becomes somewhat more complex if the dependence of conductivity on porosity is factored in.

While mass loading and deposit form are the principal considerations, the project has also identified other related approximations in the analysis. The most significant ones ares

. The NRC thermal analysis conservatively assumed that all of the aerosol was plated on the containment walls and ceiling instantly at the start of the TMBDB scenario even though most of the aerosols are not

formed for many hours. Since this assumption is not realistic, the project has made modifications to the thermal analysis code to obtain a realistic, time-dependent aerosol deposition rate and repeated the analysis with all other parameters unchanged (see Table Al-2, Case 3). This evaluation indicates that even for these extreme assumed aerosol plateout conditions, there are not significant increases in containment temperatures (370F) and pressures (1 psig) prior to containment venting and purging and that the i containment conditions after venting are significantly j less severe than without the time dependence (compare j Case 2 in Table Al-2).

. The horizontal surface area utilized for the aerosol analysis only includes the floor area. If all the horizontal surface areas in containment are included (as is done for experimental data reduction), the settling area increases by a factor of approximately 2 and the settled mass becomes greater than that predicted (and, hence, the vertical mass loading is reduced accordingly). The increased horizontal surface was considered in the latest Battelle mass loading calculation, but not in the one used for the thermal analyses (Cases 2 and 3).

. The derivation of parameters for the analysis considered that only sodium oxide was present, and that all sodium hydroxide would be formed by the water release from the floor and would remain at the floor.

Experimental programs, like the analysis, have not included airborne sodium hydroxide. In a CRBR TMBDB event approximately 1/2 of the water vapor formed is formed by the hydrogen burning in the RCB atmosphere.

This water vapor would react with the sodium oxide in the atmosphere and convert it to sodium hydroxide. The

presence of this sodium hydroxide is significant since

the sodium hydroxide would melt at the temperatures of interest (the melting point of sodium hydroxide is approximately 6000F) and this would decrease the thickness (and the porosity) of the plated wall aerosol layer. The Project calculates that 25% of the plated wall mass loading will be sodium hydroxide.

. The driving force for thermophoresis is the temperature gradient across the laminar boundary layer at the RCB wall, and this temperature gradient is proportional to the heat flux at the RCB wall. )

If an aerosol layer <

does build-up on the wall, it will further reduce the heat flux (at least temporarily while the atmosphere temperatures increase) which in turn will reduce the thermophoretic effect. In addition, the calculations performed by the NRC consultant utilized project supplied temperature gradients which are 15% above actual calculations for the Base Case scenario. The Project estimates that these factors would decrease the ,

plated mass loading through boildry in Case 2 (Table '

Al-2) by approximately 25%, assuming all other parameters and assumptions were not changed. The estimate is based on comparing calculated heat fluxes between Case 3 and the Base Case values used by the ttaff consultant.

. The thermal analysis assumed that all the plated aerosol remained on the wall and ceiling. This assumption is conservative for two reasons. First, the better the insulating properties of the plated aerosol (i.e., the higher the porosity), the lower the strength of the deposits will be, such that the deposits may fall off the wall under their own weight. Secondly, approximately two-thirds of the surface area for plating is on the containment dome and the aerosol must cling to an inverted surface. In reality, it would not be expected that all the plated aerosol would remain on the wall / ceiling if such large quantities ever could plate-out. (The NRC analysis assumes approximately 800,000 lb. of aerosol clinging to the wall and dome.)

In the sodium aerosol tests, no build-up of plated aerosol has ever been observed which is comparable to the NRC predicted value.

Summary And conclusions

-The Project's review of the NRC consultant's analysis has indicated several areas where overly-conservative assumptions were made. For example, correcting just one assumption so that the amount of aerosol plate-out in the calculation is deposited '

in a realistic time-dependent fashion, the TMBDB analysis is virtually unchanged for the first 50 hours5.787037e-4 days <br />0.0139 hours <br />8.267196e-5 weeks <br />1.9025e-5 months <br /> of the scenario and a decrease in the vent times is not required. The project concludes that these calculation do demonstrate the importance of

, I determining appropriate aerosol parameters and assumptions, but l that the staff consultant's initial calculation itself is neither l a best estimate nor even a suitably conservative upper bound. l Based on all of the above, the project believes that the CRBRP analysis (Case 1, Table Al-2) represents at minimum an appropriate best estimate and more likely an upper bound on the effect of sodium aerosol deposition on TMBDB thermal conditions. ,

While the available experl. mental data support this project l position, the data base at present permits some residual '

uncertainty. The project will pursue these uncertainties to

! ensure appropriately conservative design conditions for TMBDB features in the operating license application.

In the interim, to bound the potential impact of uncertainties on the CRBR TMBDB design approach, the project has run an additional analysis (Case 4 in Table Al-2) including time dependence and a relatively small (50%) increase in the conductance of insulating aerosol layer used in the NRC consultant's calculation. The

project considers this thermal conductance value (a factor of seven below the value the project deems appropriate) to be excessively conservatively and inappropriate as a licensing basis for TMBDB features. Nevertheless, the project has determined the feasibility of meeting established TMBDB criteria with reasonable modifications to present feature designs for Case 4 conditions.

Specifically, the polar crane was analyzed for 14000F (predicted Case 4 maximum crane temperature is only 12250F) to verify

acceptable strains on the containment shell. Cleanup system operation at 15000F was demonstrated in HEDL tests (reference Al-2); Case 4 conditions would require redesign for the quench tank sizing and interconnecting piping. Hydrogen sample line filtering is feasible with fa11 backs contemplated in the ongoing l test program. Finally, instrumentation for in-containment i temperature and pressure measurement is available for the higher Case 4 temperatures.

Overall, the project concludes that potential sodium aerosol insulating effects will never be important prior to venting (first tens of hours into an event) . While it is feasible for TMBDB accommodation features to be designed for upper bound aerosol impacts based on conservative interpretations of currently available data and uncertainties, the project considers so extreme a departure from best-estimate to be entirely l inappropriate for sound engineering of beyond design basis accommodation f eatures. The project expects to demonstrate in

, its operating license application that conditions are well within j j the Case 1 evaluation throughout a TMBDB scenario, j

.I Reference Al-1 HEDL-SA-2854

" Aerosol Behavior Code validation and Evaluation (ABCOVE) Program - Preliminary Results of Tests AB-5" R. K. Hilliard 12/15/82

'i

Table Al-1 Assumed Aerosol Parameter NAll add Ceiling Surfaces Project BBC Mass Loading Density (g/cm2) 0.10 3.9 Total Mass Deposited (1bs) 20,875 798,000 Porosity 90% 60%

Thickness (inches) *0.167 1.80 Thermal Conductivity (Btu /hr-ft OF) *0.05 0.05 E

The? mal Conductance X (Btu /hr-ft 2_oF) 3.6 0.33

• These values have been revised based on the current data review; the mass loading and, most importantly , the overall thermal conductance are not changed from the 4/16/82 values.

Table Al-2 Summary of CACECO Analysis Thickness / Conductance, Time Case Inches Btu /hr-ft2 _or Factor H(t) P(psig) T(OF) Tsax(OF) Tave(OF) Beas (l)

Base ..

Case - - -

4.5 13.1 620 917 750 4.0 '

1 0.167 3.6 No 4 .3 16.0 696 1215 900 4.0 2* 1.8 0.33 No 4.5 25.8 976 2188 1850 5.2 3 1.8 0.33 Yes 4.0 14.1 657 1770 1350 4.0 4 -

0.50 Yes 4.0 13.4 632 1550 1180 4.0

• This analysis was performed by an NRC staff consultant.

1 l

-- -^

... ~ ..J. .

~

JQL - We do not agree thet the error in treating sodium aerosols is on the l "non-conservative si de." What is neglected here is a possible time lag l In transfer of heat in the aerosol mass when the RC8 atmosphere temperature is incrossing and IIQm the aerosol when the temperature is decreasing. Thus there is a slight non-conservatism when the temperature is increasing and a slight conservatism when it is decreasing. In the base case TMBDB the ahnosphere it oscillating up and down throughout the sodium boIIIng period so the not ef fect will be to average out the error.

Al- The head leakage estimates are based upon the results of the Structural Margins Beyond the Design Base (CRBRP-3, Vol.1). SMBDB is currently the subject of a separate, Intensive review by the NRC which will resolve the issue of the primary system response to an HCDA. If this review results in a change of the head leakage, this change will be f actored into the TMBDB analysis.

Ad. - The heat transfer calculation done by the CACECO code determines both the heat load and the average thermal response of the structures. The TRUhF calculation was a parallel thermal analysis used to determine .

detailed thermal responses thereby yielding a 2D thermal profile of the structures as opposed to the 1D reponses calculated by CACECO. Care was taken to ensure that the same amount of energy was input to both models. The TRUMP results were checked against the CACECO results to verify that the cases were parallel.

Both CACECO and TRUMP models accounted for the change in thermal effects caused by the Initiation of the annulus cooling system. CACECO accomplished this by adding a heat sink which simulated the ef fects of annulus cooling. TRUMP did this by a gas flow in the annulus, in both cases the heat transfer coef ficients were changed to reflect the transition from no annulus cooling to annulus cooling at 36 hours4.166667e-4 days <br />0.01 hours <br />5.952381e-5 weeks <br />1.3698e-5 months <br />.

A1 - At the rated make-up flow (150gpm), approximately 90,000 lb. of sodium could be pumped into the RCB over a period of approximately 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> l (compared to 1,100,000 spilled initially). Adding this sodium to the RC would improve containment conditions at the time of venting and purging, since the additional mass of sodium would also increase the length of time to sodium boiling and, therefore, decrease the total amount of aerosols generated and energy transferred to the RC8 at the time of venting. The ef fect of additional sodium drained into the RC is shown for cases 3 and 4 in Table F.4-1 of CRBRP-3, Vol. 2. These cases assumed that 37,100 gallons were drained into the RC from the i

PSOV (Note: This amount of sodium is larger than the 16,000 gallons that can be in the Make-up Tank for 3 loop operation.). As can be seen, this additional sodium increases the time to hydrogen Ignition and improves slightly the containment conditions at 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />.

fu. - The Project has proposed a " Reasonable Upper Bound" (RUB) case of 7"/hr for 20 minutes followed by 1"/hr thereaf ter. This case bounds the rates and amount of reaction of all sodium-concrete tests performed to date, it is not reasonable to extend the 7"/hr from 20 minutes to I hour since there have been no tests to date which have shown rates of 7"/hr extending beyond 20 minutes.

.. .L .

BZ - The Project plans no more sodium-concrete interaction tests with dolomitic limestone concrete. The tests on dolomitic-concrete characterization (water release, heat of reaction, etc.) are almost completed and the results will be made available to the MtC. If it is decided that the dolomite data base is not suf ficient to support the Project position on sodium-concrete reactions, calcitic limestone will be used in the RC floor and pipeway cells.

B1 - Dif ferential Scanning Calorimeter (DSC) methods were used to supplement the DTA results. However, the small sample size and temperature limit (50000) of the DSC equipment ilmited the usef ulness of that method.

DTA equipment was available that was capable of handling the desired samples and temperatures. DTA methods can also give quantitative data from which heats of reaction may be calculated f rom Integration of temperature curves, as was done for the sodlum-concrete work. The method provided results which give reasonable agreement when compared with fl e known heats of reaction for the expected chemical reactions.

Additionally, as noted in item B4, this " sodium-concrete reaction" energy does not include the energy of CO2 and water reacting with sodium, and these latter two reactions provide for more energy input to the sodlum pool than the " sodium-concrete reaction" [ Note: there is some concern that a portion of the CO2 reaction is also counted in the sodlum-concrete reaction, but this is obviously conservative.3 Therefore, this issue does not appear to be a significant issue with respect to the overall TE DB scenarlo.

EL - The reviewer's theory that hydrogen burning explains the absence of hydrogen in the Series E experiments overlooks the f act that, in the same experimental series with otherwise Identical test conditions, when there was an abrupt the mole fraction of H O 2 exceeded that of 02 appearance of hydrogen.

E2 - A major source of water vapor in the R W is the burning of hydrogen.

This water vapor and that from Cel1 105 are in Intimate contact with the finely divided sodium oxide aerosol being produced by burning sodium vapor in the atmosphere, so it is appropriate to assume they would react fIrst as modeled in CACECO. CACECO does not calculate that any water is held up by absorption on sodlum hydroxide.

El - This question was asked during the 1978 independent design review of l TEDB and resolved as part of the 1979 Key Systems Review 1

"Olis, paints, insulation materials and flammable coolants were evaluated in terms of the maximum energy release assuming complete oxidation (sodium would be gottered by oxygen thereby precluding a sodium / hydrocarbon reaction). The calculated energy release in containment above grade was found to be less than 3% of the total chemical reaction energy during the scenarlo. The 35 value is conservative since complete oxidation of the addressed materials would not occur (i.e., the insulation and cable materials would smolder and char, also the oil considered is well protected f rom the ignition source in contalment)."

~. .

i

" Conservative first order calculations below grade Indicate that gas temperatures would rise by approximately 500F assuming Instantaneous j complete oxidation."

"The fiberglass media filter materials present in containment would not

' be expected to contribute as an energy source to the scenario (via chemical reactions). This Is because glass fiber materials are i

non-reactive in the presence of NaOH, end as a consequence other material Interaction with the filter media would not be expected.

Furthermore, these materials are scheduled to be removed prior to reactor startup."

E1 - The Project modeling of hydrogen flammability is based upon tests i

performed by the Project on hydrogen flammability in the presence of sodium. The LWR hydrogen Information is not applicable to this case since this Information does not address the offact of the sodium.

There is no disagreement that the 65 hydrogen concentration criterla for purging and venting is conservative, particularly with sodium j l present. -

i

^

There is a concern that carbon monoxide is not considered in the post-boll dry period of the T>eDB analysis. The Project does consider j carbon-monoxide in the post-boll dry period (see item 20 on page 3-25 of CRBRP-3, Vol. 2). Additionally, a presentation was made to the NRC on August 17, 1982 In which the generation of CO and hydrogen in the post-boll dry period was addressed and the enount venting and purging required to control these combustible gases was discussed.

l DZ - We agree that only trace amounts of nan exist in the sodium pool at the time of incipient pool bolling (9-10 hours in the base case scenario).

As presented in the April 1982 NRC technical exchange meeting this i trace amount, when multiplied by the 1.2x106 lbs. of sodium at 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br /> l would "tle up" enough hydrogen, if released, to increase the calculated l 4.4% hydrogen in containment (CACECO calculation with NaH) to 5.2% If l

that amount were not tied up in NaH. This difference of 0.8% is all i

that is tied up, and it does not amount to a large enough quantity to significantly change any TMBDB conclusions.

IEL - The Project currently has underway, and has reported to the C on, an experimental program to qualify the operation of the filte. for the containment air sampling l'..es. This item should, theref ore, not be an i Issue.

jyt - The only error in CACECO we are aware of was In the energy modeling, ,

which was identified by BNL in 1979, and corrected by HEDL at that L i

time. This is discussed in HEDL-TME 79-2, " Analytical Validation of CACECO," dated August 1979.

10L- We agree that the minimal gain in accuracy may well be,over-balanced by the Increase in cost. No action is planned on this item. ,

l

. .. . . . . . . . c. . .

fi - The case where the RCB floor was sub-divided to account for the slower water release from the RCB floor is the TEDB Base Case. In naming the resulting heat structures, one was called the "RCB floor" and the other war called the "HAA wall" which may have conf used the reviewer, but they both are in fact the Rm floor. Their area split was contrived so as to give the correct water release to the R2 from the floor for the first 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> of the scenario, such that it is not true that all the RCB floor water went to the drain and was forever removed from the calculation. This time period of 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> was chosen originally because of the NRC Imposed 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> no-vent criterion and the analysis is realistic for this time period.

The Base Case was known to be non-conservative in this regard beyond 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> and work is underway at W-WM to assess the Impact based on new WATRE results recently obtained from HEDL. It appears from the BNL analysis that it is a smalI of fact, i.e., a two hour decrease in vent time from the base case 36 hours4.166667e-4 days <br />0.01 hours <br />5.952381e-5 weeks <br />1.3698e-5 months <br />. We expect to f urther confirm this when our results are available.

D. - The purge system does not contain a check valve; rather, narrow range pressure Interlocks will be provided which will close the purge line Isolation valves whenever the containment pressure exceeds the .

atmospheric pressure (see page 2-25 of CRBRP-3, Vol. 2). Additionally, the hold-up time in the purge system will be approximately 5-10 seconds at f ulI purgo fIow (8,000 scfm), such that smalI snounts of backfIow S-will not cause an unfiltered release. It should be noted that only atmospheric air will pass through the purge line Isolation valves.

i fil . This issue has been resolved with the NRC as part of their containment review (rae letter HQ:S:82-156, J. R. Longenecker to P. S. Check,

" Agreements and Commitments from the December 21, 1982 Meeting on SER Open items," dated December 29, 1982.)

! fd - (The reviewers refer to page A-19; this should be page 4-9, since this t

Is the page where the reference to sodium carbonate is made.). The statement regarding the dif ficulty of filtering sodium carbonate (,oes not Indicate a lack of confidence in the scrubbing system, but, rather,

reflects a technical concern based upon the limited solubility of sodium carbonate in water. This issue is discussed in more detail in

! Appendix E.7 of CRBRP-3, Vol. 2 and an up-date of this section was recently provided to the NRC (see letter HQ:S:82:140, J. R. Longenecker to P. S. Check, " Submittal of Information on TEDB," dated December 7, i

1982.) Basically, this update states that the TE DB Air Cleaning Tests demonstrated that the sodium carbonate removal of fIclency is comparable i to those for other aerosol products (99% or greater) when suf ficient quantitles of water are used. This issue should, therefore, be

} resolved.

I 111. - The comments on the formation of a coolable bed reflect a misunderstanding of the RV failure modes and the design of the RC. The best estimate RV f ailure mode for the Base Case is a creep rupture f ailure of the RV at the bottom head of the RV and a subsequent creep

rupture failure of the GV. For this f ailure node, the core debris will j exit the vessel, folIowed immediately by the sodium in the vessel. For i

,---- .n.-----~- . ~ _ - , . - , _ _ _

- - . - - ~ _ - -.--. - --__ __ _ _ ___. --

.m , . . _ _ _ ._

. . . .. .. 3 . . . .L . .

t purposes of analysis, all the debris is assumed to exit the vessel at this time, since this will maximize the energy input to the sodium pool. The comment on the ef fect of the vessel head f alling before the upper vessel structures is conf using. Failure of the upper vessel structures and upper vessel head will not af fect the local distribution of the debris. If the head were to f all, any material would escape to j the RCB, not to the guard vessel. However, core debris would not be expected to escape from the RV through the head.

Ist - The distribution of core debris within the RV prior to the RV f ailure is not significant with respect to ex-vessel debris bed coolability.

The analysis assumes that all the core debris is at the bottom of the vessel and that a non-coolable bed forms in tne bottom of the RV, which is the cause of the RV failure.

Even if the "in-vessel" bed stratlfles by melting of the high power '

assemblies, the ex-vessel bed would not be expected to stratify, due to the mixing Inherent following RV f ailure and subsequent sodium spill into the cavity.

JUL - Burning 1000 pounds of sodium in air will not result in an RC8 pressure of 6.6 psig. Conservative calculations using spray code analyses have -

Indicated a maximum pressure of approximately 2 psig. In addition, the cell liners have been designed for external pressures of 5 psig. The liner anchorage is capable of resisting significant tensile loads, equal to the strength of the studs in tension. It is expected that the a

actual capability for external pressure at the outset of the accident is much higher since both the liner and the supporting concrete are at low temperature.

LL - The reactor cavity liner was represented by models of restrained strips or panels and not by a model of an infinite unconstrained cylinder.

The actual boundary conditions imposed by the physical constraints of the system were prescribed in the analysis and no superposition was

'u se d. The models and methods of analysis are described in Section 3.2.2.5.1.1.1 of CRBRP-3.

It appears that the reviewer is not refering to the Reactor Cavity liner, but to the Reactor Cavity concrete wall. In this case a model of an infinite cylinder and superposition were used. Subsequently extensive analysis of the Reactor Cavity concrete wall has been ,

performed (not incorporated in CRBRP-3) using detailed finite element models that include the actual physical constraints of the system with no superposition. The results of these analyses support the conclusions in CRBRP-3.

a0L - We are not attempting to calculate "the complicated heat transfer processes, phase changes, and chemical reactions" during the post-bolidry phase with the TRUMP code. We chose, rather, to bound the

, problem of the melting attack into the basemat by forcing 90% of the available heat downward and assuming the lowest eutecfic melting temperature, without attempting to calculate now that much heat could I

L -

. .. .._. . . . ...o..-..u-.. .-. - -. . - - - . . . - -

O be forced downward. Since this reduces the problem to heat transfer (i.e.'where does the heat end up when 90% Is forced downward?), TRUMP, -

a heat transfer code, was used.

.JZ - We agree that our model of molten pool attack bounds the problem of post-bol1 dry basemat penetration.

.J3.- The molten steel is not considered in the thermal modeling since its thermal resistance is insignificant. However, it is considered as a source for the post-boll dry generation of hydrogen and carbon monoxide (see the response to DI).

While the density driven inversion of the molten steel and oxide layers may be 'signifIcant with respect to a realistic evaluation of the melt '

front, the T>BDB analysis was performed permetrically by varying the downward heat transfer form 10% to 90%. The density driven inversion phenomena would be of Interest only if it could cause the downward heat transfer to exceed 90%.

E1. - The Project program for demonstrating survivability of the TEDB Instrumentation has been presented to the NRC. With respect to the concerns expressed regarding the accuracy, time response, and time of qualification of the pressure and temperature sensors, the requirements specifled by the Project are consistent with the purpose of these sensors, which is to monitor the state of the RCB to aid the operator in a determination of protective action. This supports the overalI philosophy of T>BDB to provide an extra measure of protection to the public health and safety, i.e, there is no requirements for a

" diagnostic" capability for "Beyond the Design Base" events. [ Note:

. This diagnostic capability does exist with margin for DBE's as described in the Accident Monitoring Program.] Any additional increase in response time and accuracy is not feasible or necessary for the hydrogen monitoring system.

~,.,..--,.,_w-_ - - . , , _ _ - - . - - ..--.m. - ._ _ _ ~.-- r. _ _ , . . , -.-e