ML20214L862
| ML20214L862 | |
| Person / Time | |
|---|---|
| Site: | Wolf Creek, Callaway, 05000000 |
| Issue date: | 09/04/1986 |
| From: | Petrick N STANDARDIZED NUCLEAR UNIT POWER PLANT SYSTEM |
| To: | Harold Denton Office of Nuclear Reactor Regulation |
| References | |
| SLNRC-86-8, NUDOCS 8609100420 | |
| Download: ML20214L862 (15) | |
Text
$
.9 SNUPPS Standardiaod Nucleer Unit Power Plant System 5 Choke Cherry Road Nicholas A. Petrick Rodville, Meryland 20050 Executive Director September 4,1986 SLNRC 86-8 FILE: 0278 SUBJ: Steam Generator Tube Rupture Analysis - SNUPPS Mr. Harold R. Denton, Director Office of Nuclear Reactor Regulation U.S. Nuclear Regulatory Commission Washington, D.C. 20555 Docket Nos.: STN 50-482 and STN 50-483
References:
- 1. NRC letter (P.W. O'Connor) to Union Electric Company (D. Schnell) dated July 15, 1986: Request for Additional Information Related to the SNUPPS Steam Generator Tube Rupture Analysis
- 2. NRC letter (P.W. O'Connor) to Kansas Gas & Electric Company (G. Koester) dated July 15, 1986: Request for Additional Information Related to the SNUPPS Steam Generator Tube Rupture Analysis
Dear Mr. Denton:
The referenced letters requested that additional information be provided in support of the NRC review of the Steam Generator Tube Rupture Anal-ysis for the SNUPPS plants - Callaway Plant Unit No. I and Wolf Creek Generating Station Unit No.1.
Enclosed are responses to the NRC staff questions.
Very truly yours, b1544 %% _
'NIcholasA.Petrick J0C/ckc/25al-16 Enclosure cc: D. F. Schnell UE G. L. Koester KGE J. M. Evans KCPL B. Little USNRC/ CAL J. E. Cummins USNRC/WC W. L. Forney USNRC/RIII J. E. Gagliardo USNRC/RIV 0
0
\\
'~
8609100420 860904 t
PDR ADOCK 05000482 P
pag y _,
-g-,
RESPONSES TO NRC REQUEST FOR ADDITIONAL INFORMATION SNUPPS STEAM GENERATOR TUBE RUPTURE ANALYSIS CALLAWAY PLANT AND WOLF CREEK GENERATING STATION QUESTION 1 Please justify the conservatism of the break flow model utilized.
In Appendix D it is stated that 3 correlations may be applicable (a)
Modified Zaloudek; (b) Burnell; (c) Henry,1970 Model.
It was concluded that the Burnell correlation most conservatively estimated the critical flow rates for the range of conditions encountered in the SGTR event and this correlation was presumably utilized for the break flow rate, providing an initial break flow rate of 44 lb/sec (Figures 4-8 and 4-23).
However, in Appendix C (Verification of RETRAN and Scoping Code), the RETRAN break flow model was modified to use the modified Zaloudek correlation, since this correlation is used in the FSAR, and in accordance with Figure C-8 an initial break flow rate of 70 lb/sec was obtained. Please clarify this apparent discrepancy.
Note: A 50% increase in break flow rate, as suggested in Figure C-8, would be more than enough to flood the main stream lines to the ARV location and would probably completely fill the main steam system (see also Figure 4-11).
Since the margin to overfill for Case 1 is only 91 cubic feet, even a small increase in break flow would result in overfill.
RESPONSE
For each SGTR scenario submitted, break flow was calculated using the single phase, resistance-limited flow equation for both the short and long sections of the ruptured tube.
The resulting flow estimate is realistic and conservative as indicated below:
A.
Short Tube Break Flow Three critical flow correlations were described as being applicable for the short tube section in a SGTR analysis (See Appendix D of Reference 1).
This applicability was determined by comparison to experimental choked flow data with conditions similar to a SGTR.
Burnell was sug-gested as a representative correlation for conservatively estimating critical flow from the short tube of a SGTR.
Using data from the SNUPPS SGTR Stuck-0 pen ARV scenario of Reference 1, it is shown that choked flow is expected in the short tube for only a limited time (See Figure 1, attached). Prior to choking, the resistance-limited equation would provide a realistic estimate of flow.
When critical flow is anticipated, the friction-limited equation would provide a conservative estimate since the critical flow correlation places a physical restriction upon the expected flow (See Figure D-1 of Appendix D of Reference 1).
Page 2 - Response to NRC Request for Additional Information SNUPPS Steam Generator Tube Rupture Analysis Callaway Plant and Wolf Creek Generating Station B.
Long Tube Break Flow For long tube lengths (1/d > 100), the onset of choking is difficult to assess due to the competing effects of large frictional pressure drops and the reduction of fluid temperature due to heat transfer.
The Zaloudek, Burnell, and Henry correlations were derived for applica-tions with 1/d not greater than 100.
Since these correlations do not account for frictional losses, they will overestimate the break flow for long tubes.
A more accurate means of estimating the flow is to calculate the pres-sure losses using the friction-limited equation until the pressure reaches the saturation pressure (this is where the fluid begins to change phase).
Then the calculation would be continued using two-phase frictional flow equations for the remaining pipe length.
If single-phase friction-limited flow is assumed for the entire long tube length, then the resulting flow would be greater than if two-phase effects were considered.
One intuitively expects this since the expan-sive phase change would act to " choke" the flow.
Hence, the use of the single-phase equations in this situation would supply a conservatively high flow estimate.
Because of this conservatism, the single-phase equation was used for the long tube flow estimate.
C.
FSAR Zaloudek Break Flow Comparisons The RETRAN calculations in Figure C-8 of Reference 1 used the modified Zaloudek correlation for a comparison with previous FSAR analysis.
The previous FSAR analysis also considered a cold leg tube break located at the tube sheet.
The previous FSAR analysis did not, however, consider break flow contributions from the long hot leg tube.
To compensate for this, the previous FSAR analysis applied a factor of 1.5 to the short cold leg tube cross-sectional area.
If the initi al break flow rate from the previous FSAR analysis were reduced by this factor, it would yield 47 lbm/sec.
This compares with the 33 lbm/sec for the short tube contribution of the total initial f riction-limi ted break flow calculated for the submittal by RETRAN (Figures 4-8 and 4-23 of Reference 1).
The remaining differences in this initial break flow results primarily from differences between friction-limited and modified Zaloudek choked break flow predictions.
Page 3 - Response to NRC Request for Additional Information SNUPPS Steam Generator Tube Rupture Analysis Callaway Plant and Wolf Creek Generating Station As stated above, critical flow conditions are only present momentarily and thus, the use of a modified Zaloudek choked flow correlation is not appropriate.
The friction-limited ficw from the short and long tubes (without two phase flow) is a more accurate, yet conservative represen-tation of true break flow than calculated by the modified Zaloudek correlation and its associated assumptions.
D.
Margin to Water Relief The margin to solid secondary water relief in the Overfill analysis is 271 cubic feet (91 cubic ft. to the SG outlet nozzle plus 180 cubic ft.
of the remaining 680 cubic feet in the steamline which will fill the steam line to the ARV inlet).
Given that the average break flow rate was 35 lbm/sec [(85,817+4000) lbm/2583 sec] at a conservative density of 42 lbm/ cubic ft., the break flow rate could be increased 12% without predicting solid water relief.
Water filled steam line considerations were addressed in section 5 of the submittal.
It was concluded that the static and dynamic loadings of the water filled steam line would be within allowable values. Therefore, the break flow could be increased an additional 12% without increasing the consequences of the analyzed event.
QUESTION 2:
Justify the use of the scoping code for sensitivity studies since a comparison of SGTR results utilizing the scoping code versus RETRAN shows differences for break flow rates, primary and secondary pressures (See Figures C-1, C-3, and C-4).
RESPONSE
l The scoping code was useful in the early stages of the SNUPPS SGTR l
reanalysis program, before the RETRAN model was completed and verified, i
to identify the worst single failures and provide preliminary results (Reference 1, p. 1-1).
It was used in conjunction with hand calcula-tions and analytical reasoning to evaluate the effects of: (1) single active failures, (2) availability of offsite power, (3) location of tube rupture, (4) operator action time, (5) operating power level, and (6) iodine spiking models (Reference 1, p. 3-2).
For the purpose for which it was used, the scoping code is sufficiently accurate and the small dif ferences from RETRAN results, discussed below, do not invalidate the conclusions as to what are the worst cases for overfill and offsite dose. The quantitative reasoning described in Section 3.2 (Reference 1, pp. 3-2 through 3-11) supports the selection of the worst cases, indep-endently of the scoping code.
i
Page 4 - Response to NRC Request for Additional Information SNUPPS Steam Generator Tube Rupture Analysis Callaway Plant and Wolf Creek Generating Station A comparison of the scoping code to RETRAN results was presented in Appendix C of Reference 1.
The agreement between the scoping code and RETRAN results for Figure C-1, C-3, and C-4, has been questioned and further amplification is provided below.
Figure C This figure of break flow as a function of time shows good agreement between the scoping code and RETRAN when one considers that the scoping code was using only five time steps to represent the transient.
Further, the integration of the scoping code break flow over time, yields excellent agreement with the integrated RETRAN break flow.
Figure C This figure of RCS pressure as a function of time shows good agreement between the scoping code and RETRAN when one considers that the scoping code was using only eight time steps to represent the transient.
Eight time steps are used in this stuck open PORV scoping code case versus five time steps for the steam generator overfill case.
It should be noted that Figure C-3 was corrected in Refer-ence 2.
Figure C This figure of SG pressure as a function of time shows poor agreement between the scoping code and RETRAN.
Further investigation into the dif ferences has shown that the relief valve flow rate as a function of pressure was input to the scoping code in table form incorrectly only for this case.
The relief valve flow rates as a function of pressure have been corrected for this case and the revised Figure C-4 is attached.
As can be seen from this figure there is now good agreement between RETRAN and the scoping code, e
Page 5 - Response to NRC Request for Additional Information SNUPPS Steam Generator Tube Rupture Analysis Callaway Plant and Wolf Creek Generating Station QUESTION 3:
The RETRAN Model for the pressurizer utilizes a single-node non-equilibrium fluid volume with phase separation.
Potential shortcomings of such a model are: (1) the non-cquilibrium raodel tends to over-predict the pressurizer pressure; (2) the single node representation assumes full mixing of the hot resident fluid with the cooler insurging primary coolant during the SI phase, and thus ignores stratification, causing underprediction of the pressure.
It is not clear how these effects interact.
The RCS pressure in the SNUPPS analysis appears to stay considerably higher than in other SGTR analyses for Westinghouse plants available to the staff.
Therefore please justify the pressurizer model by compari son with multi-node models and experimental data.
RESPONSE
Justification for the use of the non-equilibrium pressurizer model is made by comparison to experimental data.
From the comparisons it was concluded that the RETRAN non-equilibrium model is adequate for simulat-ing SGTR primary pressure response.
A.
Comparisons to Experimental Data The non-equilibrium pressurizer model supplied in RETRAN-02 has been suc-cessful in accurately predicting pressurizer pressure for several transient events.
Two studies of interest involve comparisons to actual SGTR event data at the Ginna and Prairie Island Nuclear Plants (References 4 and 5).
In each case the non-equilibrium pressurizer model was used.
Figure 2 shows a comparison of pressurizer pressure as calculated by RETRAN against the Ginna SGTR data.
The pressure as calculated by RETRAN compares well with the data except for deviations between 12 and 20 minutes after the accident.
It was suggested in Reference 4 that this mismatch resulted from inaccurate modeling of HPSI flow.
It was noted that the mismatch disappeared when RETRAN and plant data indicated the same mass in the system.
It was concluded that the "...RETRAN code reasonably predicted the overall system behavior during the SGTR as compared to the plant data".
The Prairie Island SGTR comparison (see Figure 3) also showed good agreement between RETRAN predictions and Prairie Island plant data.
The slight pres-sure underestimation after RCPs trip was attributed to inaccuracies in the safety injection modeling.
B.
Pressure Overpredictions As indicated in the RETRAN Safety Evaluation Report (Reference 3), the pressurizer model is not without limitations.
Because the pressurizer l
l l
l
Page 6 - Response to NRC Request for Additional Information SNUPPS Steam Generator Tube Rupture Analysis Callaway Plant and Wolf Creek Generating Station fluid boundary is handled adiabatically, insurges may produce overesti-mates of pressure (as noted in item one of NRC question 3).
For SGTR this is not a primary concern, since insurges are relatively slow (See Figures 4-2 and 4-17 of the submittal).
However, even if the model were to overestimate pressure during this period, the results of the SGTR analysis would be conservative since break flow increases with increasing pressure.
For the overfill case, this fact increases the potential for overfill. For the worst case dose calculation, increased break flow will result in increased secondary activity and thus increased offsite dose.
C.
Pressure Underpredictions In item two of this question, the NRC noted that pressure underesti-mation could potentially result if stratification effects are ignored during pressurizer insurges.
Referring to the comparisons to actual plant data (Figures 2 and 3), insurges occur without significant deviations for the RETRAN pressure predictions.
Thus, the RETRAN non-equilibrium pressurizer model appears adequate when modeling SGTR insurges.
D.
SNUPPS Results Compared with Other Analyses The reasons that the SNUPPS analysis showed pressures higher that the other Westinghouse analyses may be one of several variables including assumed operator action times, initial conditions, trip setpoints, etc.
Nonetheless, higher primary pressure for SGTR analyses is in the conser-vative direction as noted in section B.
E.
Conclusions l
The RETRAN non-equilibrium pressurizer model is adequate for simulating SGTR primary pressure response. This conclusion is based on comparisons i
of RETRAN analyses to actual plant data.
_UESTION 4:
l Q
The secondary side of the steam generators is modeled as a single saturated volume. The staff is particulary concerned about the adequacy of the model for the case of the " stuck open ARV", which results in very low steam generator levels (see Figure 4-26).
Therefore, please justify the model adequacy by comparison with: (1) results of analyses using multinode steam generator secondary models which have the capa-bility to predict stratification; (2) experimental data.
l
..-----7
- - + - + - -
-,yc
- + - - -
--m--
+-
- + - - - - - - -
Page 7 - Response to NRC Request for Additional Information SNUPPS Steam
' Generator Tube Rupture Analysis Callaway Plant and Wolf Creek Generating Station
RESPONSE
A single node was compared with a multinode secondary steam generator.
Initially, the SGTR transient was calculated-with a single node second-ary steam generator model.
Boundary conditions were taken from this i
SGTR transient and applied to a multinode steam generator model.
The resultant study presented in Reference 6 indicated that the single node steam generator model adequately represents the steam generator response for a SGTR transient.
s Particularly in the limiting dose case, a misprediction of the steam generator response due to stratification is not likely.
The SG liquid 1
level is low in this case and primary liquid with flashing is entering the bottom of the liquid region, promoting mixing.
1 Single node steam generator comparisons with Ginna SGTR data are pre-sented in Reference 4.
A comparison of faulted steam generator pressure with the RETRAN prediction is presented in Figure 5 of this reference.
This figure presents a reasonable agreement between the experimental data and the RETRAN simulation.
The variation between the RETRAN and experimental data is largely attributed to the use of an equilibrium steam generator model with phase separation.
For the range of condi-tions expected in the overfill case, the single node steam generator in Reference 4 underestimates the Ginna steam generator pressures, i
This trend is conservative as a lower steam generator pressure increases break flow and decreases steam generator density, thus increasing steam generator liquid volume.. In addition, a multinode generator can over i
estimate pressure and yield non-conservative results.
1 QUESTION 5:
i.
As noted above, the " stuck open ARV" case shows very low steam generator levels (equivalent to a mixture volume of about 1500 cu. ft.).
For this case you assumed that a hot-leg tube rupture located at the top of the tube sheet would result in the greatest offsite dose.
The staff be-lieves that the worst case would be a rupture of the topmost tube at the U bend, since the U bend would probably be above the mixture level for i
l at least part of the transient, and would have the greatest potential l
for iodine transport to the steam space even when immersed. Therefore, please provide adequate justification that the break location chosen for this case is the most conservative from an offsite dose standpoint, or provide the results of calculations assuming the break occurs at the t
U bend.
i i
l
Page 8 - Response to NRC Request for Additional Information SNUPPS Steam Generator Tube Rupture Analysis Callaway Plant and Wolf Creek Generating Station
RESPONSE
For the submitted SGTR scenarios, the postulated rupture sites were taken to be just above the SG tube sheet.
For the worst case dose scenario, this assumption maximizes the flashed fraction (the fraction of break flow which flashes to steam upon entering the secondary) and also yields higher break flows compared to a U bend break.
Although break uncovery would not occur for ruptures at the tube sheet, the assumed ruptures are conservative given the dose calculation method-ology.
In the dose calculations, it was assumed that the flashed fraction of the break flow entering the secondary side would be released directly to the atmosphere through the stuck-open ARV.
No credit was taken for the
" scrubbing" of iodine contained in the steam phase as it passes through the SG liquid.
Therefore, regardless of the break location, all the flashed fraction of the break flow would be released.
Break uncovery might increase the probability of water entrainment in the SG vapor, thereby increasing the quantity of radioactive fluid leaving the SG.
However, given an increase in water entrainment, a corresponding increase in moisture carryover (MCO) is not expected.
Wolf Creek instrumented SG data (Reference 7) has shown that SG separa-tor performance exceeds the MC0 design requirments and does not diminish up to the tested level of 63% of the narrow range indication.
Assuming the moisture carryover design limit of 0.25% (as compared to measured carryover of 0.015%) and recognizing that the faulted SG level remains well below 63% during the period of time that the ARV is stuck-open, the quantity of radioactive water carried over via entrain-ment would be insignificant.
Therefore, given the dose methodology and measured SNUPPS SG perform-ance, break uncovery would not increase the calculated dose.
In fact, the assumed break at the tube sheet offers the added conservatism of maximum flashing given the dose methodology.
QUESTION 6:
Please provide the basis for the assumption that primary and secondary pressure equalization is achieved within 5 minutes after SI termination.
This is not apparent from a comparison of primary and secondary pres-sures (Figures 4-1 and 4-9).
How is pressure equalization accomplish-ed?
Page 9 - Response to NRC Request for Additional Information SNUPPS Steam Generator Tube Rupture Analysis Callaway Plant and Wolf Creek Generating Station
RESPONSE
As noted in Section 2 of Reference 1, the SGTR events are divided into five phases:
isolation of the faulted SG, RCS cooldown, RCS depressur-ization, SI termination, and transition to cooldown.
Figures 4-1 and 4-9 of Reference 1 present data to SI termination, approximately 2280 seconds; therefore, the figures do not display the 5 minute period for pressure equalization.
The time from completion of the depressuri-zation step to final pressure equalization during the transition phase has been estimated to be 8 minutes (one minute and seven minutes for SI termination and pressure equalization, respectively).
It was recognized that the most critical step in prevention of SG overfill after the depressurization step, is the time to termination of SI flow.
This is because delays in SI termination allows primary repressurization and re-establishment of break flow.
In order to be conservative in the analysis, the eight minute interval was divided into three minutes for SI termination and five minutes for pressure equalization.
The resulting mass transfer to the secondary side was 6269 lbm.
The original basis for the 8 minute operator action time is based on one minute and seven minutes for SI termination and pressure equalization respectively. These times are based upon the data from the ERG, Rev. 1, Procedure Validation and Verification Study performed by Westinghouse and presented to the NRC in February,1984.
If these values were used in -the SNUPPS analysis, the resulting mass transfer to the secondary side would have been 5187 lbm.
The SNUPPS assumed times result in an additional 1182 lbm transfered to the secondary (approximately 28 cubic feet).
Thus, the assumed times are conservative.
QUESTION 7:
Please explain whether the RETRAN model includes the pump loop seals.
This is not apparent in Figure B-1.
Also provide the elevation differ-ences utilized in the model.
RESPONSE
The loop seals are modeled in the nodalization scheme in the RETRAN SGTR model by Volumes 8 and 18 (Figure B-1 of Reference 1).
These single volume loop seal nodes for the RETRAN SGTR model are adequate for the l
single phase subcooled conditions encountered in the SGTR transient.
Junction and volume elevations are used in the RETRAN code as terms representing gr wity forces in the balance of linear momentum and as
.=
Page 10 - Response to NRC Request for Additional Information SNUPPS Steam Generator Tube Rupture Analysis Callaway Plant and Wolf Creek Generating Station potential energy terms in the energy equation.
The elevation differ-t ences from the. inlet to outlet junctions are used to calculate the work done in the kinetic energy equation while the elevation differences between volume centers (calculated from volume height) are used in the momentum equation.
Elevations of junctions and heights of volumes in the SGTR model (Figure B-1) are presented in the table below.
All i
elevations are given with respect to the centerline of the core inlet nozzles.
The bottom of the volumes are used in specifying the volume elevations.
i JUNCTION NUMBER JUNCTION ELEVATION (ft) 1, 12 2.85 2, 13, 6, 17 8.58 4, 15 34.83 7, 18 2.48 8, 19
-5.81 9, 20 0.0 10, 21 0.0 11, 22 0.0 VOLUME NUMBER VOLUME ELEVATION VOLUME HEIGHT I
1, 11
- 1. 21 4.90 2, 12 1.70 6.94
]
3, 13, 6, 16 8.58 13.57 4, 14, 5, 15 22.15 15.08 8, 18
-11.60 15.33 i
9, 19
- 5.81 8.31 10, 20
- 1.15 2.30 i
l i
REFERENCES 1.
SLNRC 86-1, Steam Generator Tube Rupture Analysis - SNUPPS, 1/8/86 2.
SLNRC 86-3, Steam Generator Tube Rupture Analysis - SNUPPS, 2/11/86 3.
McFadden, J.H., et. al., "RETRAN-02 A Program for Transient Thermal-Hydraulic Analysis of Complex Fluid Flow Systems, " EPRI NP-1850-CCM-A, Volume 1, Nov., 1984.
4.
Choe, W.G., et. al., "RETRAN-02 Analysis of Ginna Nuclear Power Plant's SGTR Accident," presented at ANS Topical Meeting on Anticipated and Abnormal Plant Transients in Light Water Reactors, September, 1983, Jackson, Wyoming, Volume 1.
5.
- Chao, J., et. al., " Response of a Westinghouse Two Loop Plant to Steam Generator Tube Ruptures", NSAC-77, July, 1984.
6.
Sorrell, S.W., "RETRAN-02 SGTR Analysis: A Comparison Between Six Volumed Steam Generator Secondary and a Single Volumed Secondary", Proceedings:
Fourth International RETRAN Conference, EPRI NP-4558-SR, May, 1986.
7.
Curlee, Jr.
N.J., Draft Report "Model F Steam Generator Field Instru-mentation Program", Westinghouse Service Technology Division, May, 1986.
e l
FRICTION & BURNELL FLOWS-SHORT TUBE FALED OPEN ARV CASE so
.7 i
i i
i I
l 50M i
I M
j 40 -
i 2:
.v'%w v ', y % y P Q"4-+- er 4 V
y 5
V
\\'
g
_y Y
I W
4 E
Y 3
M ~l
\\
l o
l N
l
- ,f
\\.
4 i
10,
I l
I o!
c r
r r
l 0
1 2
3 (Thousandel TBEleeoondel a
Bumeg
+
Friction j
&aee
/
.Q I
I, GINNR S.G. TUBE RUPIO.ME 8
l n
DATA l
R Eit,
41.. a u,
g 5~
m
^
=
l G~ R OU
- .J -
c y
,/
cr.
g.
l 'o
.x
-1 y
c0 LCO 1000
.300 0200 1000
!! 0 IIME (3EC)
Fig. 2 Pressuricer Pre sure i
n l
l i
I i
1 2500 i
i
(
l e
PLANT DATA I
i RETRAN-02 i
2300 1
3 4
LOAD REDUCTION I
2 l
@ 2:00 E
SCRAM e
e e
e e
e e
N, e
sa E
MANUAL 1900 STEAM DUMP
~
RCPs TRIP l
l SI ON 1700 i
i i
j 1500 i
O 300 600 900 1200 1500 1
TIME (SEC) figure 3 Pressurizer Pressure Volume 32 e
0
fIGUNE C 4 SIEAM E NERA10R PRESSURE I
(WORST CASE DOSE) i.2 -
1.1 L
\\
\\
i e\\3 0.9 -
1 Q
Start if Nt*dclor Coolelows Irap
(
n Sg 0.8
\\
3 r J.
,,, u O
sc5 3 se 0.7 -
1 un a i
no i.: c
[b O.s -
N O.S -
RETRAN 4
Stop Scoping Code Eooldows:
O.4 -
l i
i ARV Closed 1 r G
O.3 1
l 0.2 --
O 1
2 3
!,4 (Thousands)
TNE (SECONDS)
'W l'AULTI:D
-*- INTACT