ML20209H230
| ML20209H230 | |
| Person / Time | |
|---|---|
| Site: | Vermont Yankee, Yankee Rowe, Maine Yankee, 05000000 |
| Issue date: | 11/01/1985 |
| From: | Papanic G YANKEE ATOMIC ELECTRIC CO. |
| To: | Zwolinski J, Zwolsinski J Office of Nuclear Reactor Regulation |
| References | |
| FYR-85-121, NUDOCS 8511110037 | |
| Download: ML20209H230 (98) | |
Text
.. -
Tc'eph"ne m ") s ' -8 '00 YANKEE ATOMIC ELECTRIC COMPANY TWX 710-380-7619 me
~j' e'
', 1671 Worcester Road Framingham, Massachusetts 01701 2 c.2.1 J_' ;ya,g FYR 85-121
".bss K G-/li
'f November 1, 1985 l
. United States Nuclear Regulatory Commission Washington, C. C. 20555 Attention: Mr. John A. Zwolinski, Chief Operating Reactors Branch No. 5 Division of Licensing
References:
(a) License No. DPR-3 (Docket No._50-29)
(b) Licenso No. DPR-28 (Docket No. 50-271)
(c) License No. DPR-36 (Docket No. 50-309)_
(d) USNRC Letter to YAEC and MYAPCO, dated May 11, 1984 (e) Topical Report: "RELAP5YA - A Computer Progrant for Light Water Reactor System Thermal-Hydraulic Analysis,"
Yankee Atcmic Electric Company, YAEC-1300P, October 1982 L (f) MYAPCO Letter to USNRC, dated January 31, 1985 (g) YAEC Letter to USNRC , dated August 15, 1985
Subject:
Response to NRC Question on the RELAPSYA Computer Program
Dear Sir:
Reference (d) requested answers to 197 questions on YAEC's Topical Report on RELAP5YA [ Reference (e)]. Our schedule called for the submittal of responses to the last 50 RELAP5YA questions by November 1, 1985.
Responses to 7 of these questions 7 have already been submitted to you in Reference (g). Responses to t6e remaining 43 questions are provided as nn attachment to this letter.
This submittal _ completes our response to all the 197 questions on RELAP5YA. Yankee has invested a substantial amount of manpower and computer resources in the development of RELAP5YA and in responding to all the NRC questions. We are very anxious to apply this code in the licensing analyses of our plants which will start upon the receipt of the SER on
_RELAPSYA. Mr. Ed Throm of your staff has been involved with the review of RELAP5YA since our first presentation in August 1982. During his involvement, he has gained considerable expertise in the code. We understand that Mr. Ed Throm will be transferred to a different group as part.of the recent NRC/NRR reorganization. We are concerned that this transfer may lead to the assignment of RELAP5YA to another member of your staff who is unfamiliar with this. computer program, and may cause undue delay 0\
,l0 8511110037 851101 i 8 io.
E
r
.
- 0, a
United States Nuclear Regulatory Commission l' November 1, 1985 Page 2 in the issuance of an SER on RELAP5YA. Therefore, we request that Mr.
Throm be retained as the reviewer of RELAP5YA, and an effort be made to expedite the. issuance of an SER on the RELAP5YA computer program.
., . We trust.this information is satisfactory. However, if you have any questions,'please contact us.
Very truly yours, YANKEE ATOMIC ELECTRIC COMPANY George apanic Senior Project Engineer Licensing GP:kem i.
r- .
. .os RESPONSE TO 43 QUESTIONS ON RELAP5YA NOVEMBER 1, 1985 e
i
l I. ONDENSATICE HEAT TRAN87ER AND NONCONDENSABLE GASFS l
LLJ, Clarify the concern expressed in the following from Page 22 of Reference 13 where R5 stands for RELAPS: "In R5 the mass transfer I
is modeled with a simple relaxation formula making proportional to.(I-I,). Two different proportionality functions are used; one for condensation and one for flashing. Both of these relaxation i rates have been adjusted to fit experimental data obtained from f pressure change-dominated experiments (vs. wall heat-dominated tests), mostly depressurisation tests. More experimental comparisons are needed to test the condensation rates. Because the l i
present model has been adjusted to fit the pressure change-related mass transfer rates, it is not suitable for wall heat flux-induced mass transfer (i.e., mass transfer caused by subcooled boiling or any other wall heat transfer machanism has not been adequately addressed in RS). Because the mass transfer rates due to pressure change are larger than the corresponding rates due to wall heat flux, R5 at present calculates most heat flux-generated mass transfer. problems as nearly equilibrium situations. In developing the modeling needed to address mass transfer due to wall heat flux mechanisms, it appears that some of the detailed partitioning of i
energy that was avoided in the five equation formulation will have ,
to be examined and included in the p model." !
8 !
A.I.1 The interphase mass transfer models in RELAP5YA are similar to RELAPS/MODl. Hence, the concern expressed in the question is valid, but mainly for realistic (best-estimate) LOCA analyses. In licensing analyses of reactor systems, the code assessment work at Yankee Atomic Electric Company (YAEC) has shown that the impact of the models is in the conservative direction. This has been discussed in Section 2.1.3.1 of Reference I.1-1.
Reference (I.1-1) Fernandez, R. T., et al., "RELAP5YA - A Computer Program i for Light-Water Reactor (LWR) System Thermal-Hydraulic j
Analysis, Volume I," YAEC-1300P, January 1983.
i i
4 i
, - - + - - - - - - - + - - . - - - - - - , e a-- - . , . - - - -a- , -- ,,-a e,rr- - - ,,--,--g - , - - ,~-----w .e---s , - , , , v g , ---,
LLS, The condensation model is described as follows on Page 35 of Reference 10:
Q=-K(1-I+1)(I,-X),where: c K = 1.0 x 10 (Eg/m -sec), empirical constant
~
I,= 1.0 x 10 , empirical constant.
"From the code assessment, we have inferred that this model tends to overpredict the condensation rate when subcooled ECC is injected into a steam environment. This causes the system pressure to be low, and results in degraded heat transfer in the fuel bundle region as seen in the Two-Loop Test Apparatus (TLTA) and Loss of Fluid Test.(LOFT) code assessment cases discussed in Section 5.0 of Volume III. An improved model in this area is desirable for best-estimate analyses. For licensing analyses, the degraded heat transfer effect l' and the conservative assumptions imposed by Appendix K yield conservative results."
' Clarify why this equation is conservative for licensing analyses, including its effects on ECC injection, pressurizer refill and condensing any vapor in high points of the system.
A.I.9 The conditions of p::essurizer refill and condensation of vapor in high points of the system are not important from the point of view of small break LOCA analysis since these occur beyond the point of core recovery and are, therefore, beyond the scope of the analysis.
1 For the case of ECC injection, the above equation may not be conservative. However, the input assumptions used along with this
! condensation model will be such as to ensure conservative ECC l injection flow rates. These ICC modeling guidelines are described in A.II.5 to A.II.8.
0.I.31 Clarify for the LOFT L3-1 accumulator comparison in Section 2.4.2 of Reference 12, which curves are the data and which are the RELAPSYA calculated results.
, , _ - .-- .y.,
A.I.31 As described in A.I.23 (July 1985 submittal), the accumulator model described in Chapter 3 of Reference I.31-1 has been replaced by the RELAPS/ HOD 1 Cycle 18 model. Therefore, the LOFT L3-1 accumulator response has been recalculated using the revised model as part of the system test prediction. Comparisons of the RELAP5YA results to the LOFT L3-1 data are presented in Appendix V.1-1.
Figures V.1-6 and V.1-7 present comparisons of the calculated and measured accumulator level and pressure. As a result of a later calculated accumulator injection time and the assumption of a higher ECCS water temperature, it is hard to assess the accuracy of the accumulator model from these plots. For this reason, the curves were reconstructed to show accumulator level versus accumulator pressure. Figure I.31-1 compares the RELAPSYA calculation to the LOFT L3-1 data. The comparison shows that RELAPSYA accurately predicts the accumulator expansion.
Reference (I.31-1) Fernandez, R. T., R. K. Sundaram, J. Ghaus,'A. Husain, J.
M. Loomis, L. Schor, R. C. Harvey and R. Habert, "RELAPSYA
- A Computer Program for Light-Water Reactor System Thermal-Hydraulic Analysis, Volume I: Code Description,"
Yankee Atomic Electric Company. Report YAEC-1300P, Volume I, October 1982 (Proprietary).
4
-,-.v, ,- ---- -,--w.--- --- -- , -w-
RELRP6YR COMPARISON TO LOFT LS-1 DATR RCCUMULATOR LEVEL.VS ACCUMULRTOR PRES 8URE 60 ===== ' 8 - = = - - - - - - ' - - - - - - - - - n - - - - - - - - -
. I I I I .
i
- 50- _ _ ;
i .
40- _ _ ;
n . .
i >_ -
a u.
~
3 0; _ _ ;
_s . .
k,J . .
us
.J 20: -
m--e RELAP5YR LOFT L3-1 T
1 0; _ _ ;
~
0.0- '
70C t . b ' ' ' ' ' ' ' 6'OC I . b ' ' ' ' ' ' ' 6'0C t . b ' ' ' ' ' ' ' 4'0C l . h ' ' ' ' ' ' ' $'0(1.b ' ' ' ' ' ' ' 2'OC ' . 0 PRES 8URE CP8IR)
Figure I.31-1: Comparison of Accumulator Level vs. Pressure in LOFT L3-1 to RELAP5YA Calculation
0.I.32 "As part of model verification studies perforined using test data from the LOFT facility, the sensitivity of accumulator model response to the size and location of the node attached to the accusaslator should be assessed" (page VIII-59 of Beference 4).
Clarify how this has been done in the submittal because the Maine Yankee and LOFT L3-1 analyses modeled only the accumulators with pressure boundary conditions.
i
+ A.I.32 'As mentioned in A.I.31, the LOFT L3-1 accumulator response was reanalyzed as part of the integral test prediction. In addition, a node size sensitivity was performed for this test, the results of 4 which are presented in A.II.S. No assessment of the sensitivity to the injection location has been made since only the actual location has been used in model assessment and will be used for plant
{
applications (see A.II.3),
f 4
1 0.I.32 "As part of model verification studies performed using test data from the LOFT facility, the sensitivity of accumulator model response to the size and location of the node attached to the accumulator should be assessed" (Page VIII-59 of Reference 4).
Clarify how this has been done in the subalttal because the Maine Yankee and LOFT L3-1 analyses modeled only the accumulators with pressure boundary conditions.
A.I.32 As mentioned in A.I.31, the LOFT L3-1 accumulator response was reanalyzed as part of the integral test prediction. In addition, a node size sensitivity was performed for this test, the results of which are presented in A.II.S. No assessment of the sensitivity to the injection location has been made since only the actual location has been used in model assessment and will be used for plant applications (see A.II.3).
II. 50NEQUILIBRIUN EFFECTS 0.II.1 If a volume contains subcooled liquid with the quality less than 0.5, clarify what happens to the liquid energy if enough liquid is rencved from the volume to make the quality greater than 0.5 so that l
- the liquid temperature is then jumped up to the saturation temperature.
A.II.1 RELAPSYA will generally predict significant nonequilibrium effects only under conditions of rapid pressure transients. Under all other conditions, RELAP5YA will predict very little nonequilibrium between the phases. Hence, if a volume initially contains subcooled-liquid, at a quality of less than 0.5 (but greater than 0.0), the amount of
.. subcooling calculated by RELAP5YA will be very small. If this system is perturbed in such a way to cause the quality to exceed 0.5
.and the liquid to be saturated, the effect of the perturbation on the liquid energy is also expected to be small.
0.II.2 Because of the minimum-phase-at-saturation assumption in the code, any subcooled water injected into a node containing some steam would condense instantaneously all the steam needed to bring it to saturation as long as the liquid remained the minimum phase. This can cause a rapid drop in pressure that can cause the ECC injection rates to be overestimated because the delivery rate increases as the system pressure decreases. Also, the amount of steam available and the resulting depressurization is dependent on the node size.
Clarify how overestimation of ECC injection rates is avoided.
!. A.II.2 Based on the RELAPSYA calculation of LorT tests L3-1 and L3-6, it was'found that raising the ECC water temperature to 200 F mitigated the artificial rapid pressure decrease in the injection mode. Thus, in PWR applications, raising the ECC temperature to i
near saturation (corresponding to the containment pressure) provides a method for avoiding overestimation of ECC injection rates. A
' further discussion of the guidelines used for modeling ECC injection is provided in the answers to questions II.5 through II.8.
0.II.3 If alternate injection locations are used. clarify that the heat transfer in the core is not overpredicted because of higher pressure than would have occurred with cold les ECC injection.
A.II.3 Alternate injection locations will not be used. A realistic representation of the injection location will be used in plant j applications.
0.II.5 Clarify how the condensation rate is affected by the size of the volume and also the sizes of the adjacent volumes.
A.II.5 To assess the sensitivity of the condensation rate and the ECC injection flow rate to node size, a node size sensitivity was performed for LOFT Test L3-1. The base case calculation, described in Appendix Y.1-1 and A.II.6, used an ECC water temperature of 200 F. The sensitivity case consisted of reducing the node size of the cold leg injection node and the adjacent volume by one half as shown in Figure II.5-1. This sensitivity calculation also used an ECC water temperature of 200 F.
The primary system pressure, the accumulator pressure and the accumulator level are presented in Figures II.5-2 through II.5-4.
For each plot, the node size sensitivity case is compared to both the base RELAPSYA case, as well as, the test data. The results show no significant difference in the two RELAP5YA calculations. This implies that the node size has no significant impact on the RELAPSYA calculated condensation rate for these conditions.
i O.II.6 Justify that the injection model will result in conservative values for the ECC flow rates.
A.II.6 In modeling the accumulator injection for LOFT Test L3-1, RELAP5YA calculated a sharp drop in primary pressure, unlike the data, along with an overprediction of the decrease in the accumulator level during some periods of injection. To prevent the overprediction of the accumulator flow rate, the temperature of the ECC water was artificially raised to limit the condensation rate (see Appendix l - . - - . . - -- , .__- -.. - .-- -- . .- - - - - - - -
l l
143 151 ; 152 605 ECC Downcomer Base Case Cold Leg Nodalization 143 : 151 - + 154 : 155 - + 152
--e 605 ECC Downcomer Cold Leg Nodalization Used for Sensitivity Calculation Figure II.5-1: RELAP5YA Cold Leg Nodalizations Used in LOFT L3-1 Prediction
- _ _ . ~ . .
- , RELAP5YA C0tFARISON TO LOFT L3-1 DATA l , , , , , , , , , , , , , , , ,
j ii o
)
i
! o j h~ l
- 2
=
1 -
E l "l2=
i - I-- L0rT ts-1
- .- RsrA ease a M R5YA NODE SS i
e .
i Q-
- ~
@o s-b- ~
S' I 6-e s- -
O
~
- ~
' Z _
l F W W W y y g y y y y W W W W W 3 y 0.0 100.0 3 DOC 0 300.0 m.0 500.0 800.0 700.0 000.0 900.0 1000.01100.0130.01300.01400.01900.0155.01700.01800.0 TIME (SEC)
Figure II.5-2: Comparison of Primary System Pressure in LOFT L3-1 to RELAPSYA Base and Node Sensitivity Calculations h
o.
- - . - - - l . o o
' s _
a M i o.
s da
. o. -
o a i s
a o t
o.
a i a 1 s s - n
- a 3 o Li
_ o. t o T a a i F l e O u s
L c
- o. na l
a i a i C s
R a ey T rt A o. ui D a i o sv a
s si 1 et
- ri 3 o. P s L o n a da re T s oS t
F' o. ) ae O o Cl d L a i E uo O
c s S( mN u
T o. cd E cn N a i o MA c a O s I S Tf e I os R o. a o nB A a i
s o f
l sA i Y o
C
- o. rS a io aP pA A
n mL Y oE 5 o. CR Fa io U s s 3
E -
R S o. 5 a
S s
'os I
_ 1EE I SD o. e 3AO o r a
LBN it u g
TAA
' i FYY o. F a O55 io T LRR.e :. z W
- io o.
a v a :..
m v o.
a io e
s y
T o.
o
=
- g
- g Og N' o.g 9g og aN "ceI~
oC W gm1.:gha yOI] E3 Y 1 o8 lj !
i i
l I
l RELRP5YR COMPRRISDN TO LOFT L3-1 ORTR g.0 , . . . n . ...n ....n ....n . . ..n . . . . n . . .a n . . . . n . . . . n . . . . i l i i i i i i i i
i.
1 i : :
l 5.0 _ - ;
(
4 0: _
g . .
w . .
w, i -
~
- 30 _ -
a: .
o g
m a : :
m . .
E LOFT L3-1 :
8 2 0:: -
m -eR5YR BRSE
-~
m
- ---* R5YR NODE SS :.
i 1 0; _ _ _
00- ' -
ct
'2Ct0' ' ' '4C O' ' ' '8C O' ' ' '8do' ' ' io ld ' ' i2 )d ' ' i4 )d ' ' iB )d ' ' i8 )d ' ' f0 30 TIME (SEC1 Figure II_5-4: Comparison of Accumulator Level in LOFT L3-1 to RELAP5YA Base and Node Sensitivity Calculations i
V.1-1). 'An ECC water temperature of 200*F was selected since it is closer to.the saturation temperature corresponding to containment pressure yet low enough to avoid the injection of superheated liquid. j l
Figure V.1-6 compares the accumulator level calculated by RELAP5YA, l
-using a temperature of 200 F for the ICC water, to the LOFT Test L3-1 data. Since there is no measurement of the accumulator flow rate, the flow rate will have to be inferred by the rate of change in the level. The calculated rate of level decrease is lower than that in the data implying a lower volumetric flow rate. In addition, raising the temperature reduces the density of the ECC water and reduces the mass flow rate (for the same volumetric flow).
For plant applications, the ECC water temperature will also be raised to 200 F to allow for conservative ECC flow rates.
0.II.7 Clarify what guidelines will be used for injection node size in SBLOCA analyses.
A.II.7 The only guideline used in RELAP5YA nodalization of Reactor Systems is to provide sufficient detail to capture physical phenomena. This general guideline is also used in determining node sizes around the ECC injection locations. The node size sensitivity study discussed in A.II.5 has shown that this guideline provided an acceptable nodalization for LOFT Test L3-1. Similar nodalization schemes will be used in plant applications.
0.II.8 Clarify how the effects of injection node size and location have l been assessed against data and how the range of data compares with l
expected SBLOCA conditions.
I A.II.8 Our test' facility assessment models and LWR System models contain relatively detailed nodalizations that use actual ECC injection locations. These nodalizations allow the pressure and flow to
- connunicate readily between surrounding volumes. Thus we expect the i
l depressurization rate to be sensitive to the ECC condensation rate, but not sensitive to the ECC injection node size. This has been l
confirmed by the injection node size sensitivity case described in Answer A.II.5 above. No sensitivity studies have been performed with alternate RCC injection locations since we have used actual injection locations in our assessment and LWR System models. 1 Two integral tests, LOFT L3-1 and L3-6/L8-1, have been used to assess RELAP5YA for PWR SBLOCA applications. These tests were scaled to represent a four-inch equivalent cold leg break in an LWE. This break size is close to the limiting break size expected for our PWR SBLOCAs. Table II.8-1 compares the nominal accumulator' l conditions for LOFT Test L3-1 to those~for the Yankee and Maine Yankee PWRs. This comparison shows that the L3-1 accumulator flow contains more subcooling than either plant due to the higher initial pressure and lower fluid temperature. Therefore, with respect to -
condensation rates, the LOFT accumulator bounds the plant "pplication.
a Three TLTA tests, consisting of two large break and one small experiments, have been used to assess RELAPSYA for the BWR full ,
break spectrum. -These are described in Reference II.9-1. We believe these tests span the range of conditions for BWR LOCA applications.
TABLE II.8-1 Nominal PWR Accumulator Conditions Pressure Temperatura Subcooling Plant (psia) (OF) (OF)
Maine Yankee 220 110.0 280.0 Yankee Rowe 488 100.0 364.5 LOFT L3-1 634 88.5 403.7
. . .~ _ _ _ _ _ -
0.II.9 Page VIII-29 of Referesace 4 says that submittals need ".. .to confirm that HPI and accumulator flows during small breaks will not result .
l in the formation of water slugs, or if they do, to show that the l
'- structural design bases of the primary system include loads due to: 1
- (a) water slug inertial motion, (b) water slug impact and (c) i pressure oscillations due to steam condensation." Clarify how RELAPSYA models water slugs if they occur and how sensitive this modeling is to noding.
A.II.9 We assume that the formation of water slugJ is defined as the f
phenomenon of periodically filling a fluid volume completely with water. In the RELAPSYA analyses carried out so far at YAEC, there have only been a few cases where water slugs have been calculated to occur. However, in these cases there was also a buildup of mass error, and when corrective action was taken to minimize mass error (see response to Question I.1), water slug formation was not observed. All the RELAPSYA calculations described in Reference
'II.9-1 and the additional calculations described in response to Questions V.1, V.2., V.3 and VI 9 have acceptable mass errors and have not shown occurrences of. water slugs. The same principles used in developing the nodalization for these calculations will be applied in plant calculations.
Reference (II.9-1) Fernandez, R. T., et al., "RELAP5YA - A Computer Program for LWR System Thermal-Hydraulic Analysis, Volume III,"
YAEC-1300 P, January 1983 0.II.10 Page 13 of Reference 8 states: "The thermal energy source term associated with the virtual mass acceleration term is not necessarily positive or zero. However, consideration of the second law of thermodynamics dictates that such must be the case. This contradiction is not resolved at this time, and since the energy dissipation associated with virtual mass acceleration force is small, this energy source term is presently omitted." Clarify approximately what the er.ergy error is, based on this assumption.
t
--A.II.10 This broad hypothetical question is difficult to answer precisely for all conditions that might be encountered in PWR SBLOCAs and the
~
BWR full break spectrum. However, the RELAPSYA code has been assessed against numerous separate effects and integral system tests that are discussed in Reference II.10-1. These cases span a very wide range of conditions that might be encountered in PWR SBLOCAs or the BWR full break spectrum. In all assessment cases, the last term on the right-hand side of Equation 15 in the cited reference was neglected. The favorable RELAP5YA code assessment results indicate
- this appears to be a reasonable assumption. Therefore, we believe I this term has a negligible effect on the thermal energy equation.
Reference (II.10-1) Fernandez, R. T., R. K. Sundaram, J. Chaus, A. Husain, J. N. Loomis, L. Schor, R. C. Harvey and R. Habert, "RELAP5YA - A Computer Program for LWR System Thermal-Hydraulic Analysis, Volume III: Code Assessment," YAEC-1300P, Volume III (October 1982),
(Proprietary).
l l
I l
IV. STEAN GENERATOR HEAT TRANSFER 0.IV.4 Page 301 of Reference 12 states: "The RELAPSYA calculation generally provided a reasonable analysis of the LOFT system response for Test L3-6. The only major discrepancy between the calculation and the data seems to have been in the underprediction of pressure in the Secondary Coolant System. This is believed to be due to overprediction of condensation rates. The calculated mass error in the primary system was, at most, 20 lba, which is well within the data uncertainty. The Secondary System mass error was considerably higher. This may have resulted from the overprediction of steam condensation." Clarify why this overprediction of steam generator secondary condensation rates and large secondary mass errors will not produce nonconservative results for SBLOCAs.
A.IV.4 To address the impact of underpredicting the secondary coolant 1
system pressure for small break LOCAs, the RELAPSYA calculation of LOFT Test L3-6 was reanalyzed. In this reanalytis, the steam generator secondary side pressure was set to match the data and the
-impact on the primary system conditions were investigated.
Figure IV.4-1 compares the criginal and revised RELAPSYA secondary system pressures to the LOFT L3-6 data. At about 1800 seconds into the transient, the original and revised RELAPSYA secondary pressures differ by as much as 300 psi. The primary system pressures for the two RELAPSYA calculations are compared to the LOFT data in Figure IV.4-2. Despite the large differences in the calculated secondary pressures, the difference in the primary pressures is no more than I
30 psi. This small impact on the primary system is also illustrated in the comparison of the primary system inventories presented in Figure IV.4-3. The revised RELAPSYA system inventory is only L 200 lba lower than the original.
I The overall effect of these differences in primary system conditions t
on small break LOCAs is best determined by comparing the cladding temperature responses for LOFT Test L8-1. The starting point of L8-1 is the final system conditions of L3-6. Therefore, the t
l differences in primary system conditions will be carried through to i . Test L8-1. Figure IV.4-4 presents the peak clad temperature responses for the original and revised RELAP5YA calculations compared to the L8-1 test data. The two RELAP5YA cladding temperature responses are almost identical with the only difference occurring during the recovery period. The reason for this difference is a result of accumulator modeling differences described in A.V.9.
l Since there is virtually no difference in the two calculated peak cladding temperatures, underpredicting the secondary system pressure is not expected to produce nonconservative results for small break LOCAs.
0.IV.8 Page 15 of Reference 16 says: "The results of a FLECHT SEASET steam generator separate effects test analysis using RELAPS/ MOD 1... suggests that the nonequilibrium behavior on the
. secondary side of a steam generator in reverse heat transfer cannot be correctly calculated with MOD 1. The presence of such a subcooleo layer on the tube sheet is closely associated with the propagation of a ' quench front' up the insides of the U-tubes. This quench front cannot be calculated without a reflood model containing a
.nonequilibrium heat transfer correlation package together with a moving fine-mesh temperature grid..." Clarify how secondary-to-primary heat transfer is treated in RELAPSYA.
A.IV.8 No special treatment is given to the secondary-to-primary heat transfer. The standard RELAPSYA heat transfer models described in Sections 2.1.3.6 and 2.1.3.4, Appendix A of Referenes IV.8-1, are used.- The nonequilibrium behavior described above was seen in LOFT Test L3-6 and not predicted in our RELAP5YA calculation.
Reference (IV.8-1) Fernandez, R. T., et al., "RELAPSYA - A Computer Program for LWR System Thermal-Hydraulit Analysis, Volume I: Code Description," YAEC-1300P, October 1982 (Proprietary)
d
. RELAPSYR COWARISON TO LOFT L3-6 DATR g
O h- P__
t
- o r -
l l- -
I o l N' .
0 O
~
h~
i as i
. g-J -
C g- .
I j
i E
lg-9
' 9
. r w l- -
l
- 8a
- l g S. LOFT L3-6 _
- s *-4 R5YR REVISED ga" F-*R5YR ORIGINAL 5 =
No
~ ~
- $5 =
g- .
! a I == -
f = = =
i i U y g 1ko U U U
- 0.0 2 10. 0 400.0 500.0 000.0 15 5.0 1210.0 1NII.0 1NII.0 31[5.0 2NII.0 St[II.0 2800.0 i TIME (SEC) l Figure IV.4-1
- Comparison of Secondary System Pressure in LOFT L3-6 to Original and Revised RELAPSYA Calculations I
i - - - - - - -
a RDAP5YA CON'ARISON TO LOFT L3-6 DRfA g , , , , . . . . . . . .
Hli i a
- a i
- g. -
! a j g. -
1 -
Si LOFT L3-8
- m - ~
a_
- -*R5YA REVISED
- MR5YR ORIGINFL ma i I
'e Eo
- e Q. ,
G taJ
! J i
3 H"0- _
a l %o.
e-e y 1
a
~
! o a
I 0.0 Ei0.0, 4d0.0 0d0.0 8d0.0 1050.0 1NO.0 1450.0 1NO.0 1850.0 2150.0 2NO.0 30El.0 2000.0 1
TIME (SEC) i Figure IV.4-2: Comparison of Prfmary System Pressure in LOFT L3-6 to i Original and Revised RELAPSYA Calculations i
i
]
. . . . 0
)
EEiEEE
!E[:E.EEi=E:E_:El=2lEEi:El
~ _ _ _ _ _ _E!.g:i:::2
_ _ 2 d
- 6
'I ,7 i b O
_. d
_7
- L '
6 DA _r 6 -
EN b 3
nI SI _- e L 6IG s VI _- i T
- 3ER ' F s LRO _-
O n L o
- TAA _-
6 nt i
FYY g b i a A nI O55 y1 _ i s l T -
LRR i yu R M- ^ : rc O E
' ol r '
t a nC 8 T-S-
^ : -
6 e vA 3 Y 5 i b nY L S'- 4 I S T Y-
.]CI i
' mA eL P
F R - ' t E O A- sR L IM- 6 y Sd O R' 8 i b2 )C ys e
T P- (S(E ri av N E-H- ' me D i R ST- 6 E r I
N8 I e b IM Pd n
R I O T f a R -
i o P S-
' l M na O S- ' on C A si i g M- 'O ri R
iI i d ar YL-R b pO m
5 T- '
oo P O- -
' Ct R
L T -
E 'O 3 R iI i d
- b 4 V
- I
'O e e r iI d u
- '4 g i
F I1
'O i8 i d
- h 5EjEEi iil:5Ei22 2.:E 3 r2 ] E E i. :Ei=Elg - E5 o 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 2 1 0 9 8 7 6 5 4 3 2 1 1 1 1 1 n rmJ.w32(uE a_$OH
/
8woi
0
. . - ~ - - . " - 0
' 2 DA L "
^ 1 EN -
SI 8 1IG 0 L
. VI T 8ER 35 F
- LRO 1 D
THA FYY I
n O55 i
LRR _- 0
- e e
. _- r s
- 5W1 u n t o 2m
,_ 0 a
r it e a pl m u e c A .
0 T l T W 4 a A 1 gC D n i
1 Ili d Y A
- d 5 S # , l P L
0 CaAL 50 )
E T
F 2
1 C6 E2 R O S0 L ( - d 8 e O EG s i T 0 MS v I
N . I Te e O
WME d R S 1 o d I N n R f a A
f ol f n a 0 .
0 o in C s g 3W i A r ir Y a pO 5
P m o o A
2 0 C t E .
R 3W :
4 4
_. 1 V I
0 0
e 4 r u
- g i
F
^_
r 0
- - 5 2
^
0 0
.g o ol al o
sa air gd ag
= k' Oh aic
_b t - ge8m i u 4 l
,ii!iI i ! !
4 0.IV.9 Clarify how the condensation heat transfer that could occur in the J primary side of the steam generator U-tubes during SBLOCAs has been l assessed.
A.IV.9 The RELAPSYA calculation of condensation heat transfer in vertical flow is based on two commonly used correlations:
For laminar flow the Nussel theory of film condensation as documented.by Collier in Reference IV.9-1 is used.
i For turbulent flow the correlation of Carpenter and Colburn documented by Collier (Reference IV.9-1) is used in the primary side of the steam annerator U-tubes. The condensation heat transfer was assessed in benchmarking the code against integral tests.
Two SBLOCA tests performed in the LOFT facility, Test L3-1 and L3-6 were assessed using RELAPSYA. The system behavior in these tests which depended in part on condensation heat transfer in the steam generator U-tubes, was adequately predicted by RELAP5YA.
Reference (IV.9-1) J. G. Collier, " Convective Boiling and Condensation,"
McGraw-Hill Book Company, London (1972).
V. SHIEEE_HRInGAT1.qNmuRPT_Ak_v__rRInGATl_of II.K.3.30 Page 3-177 of Reference 1 states'that: "The revision should account for comparisons with experimental data, including data from the
~
i ~ LOFT Test and Semiscale Test facilities." "In addition to the modeling
- concerns identified, the task force also concluded that, in light of the TNI-2 accident, additional systems verification of the small-break LOCA model
- as required by II.4 of Appendix K to 10CFR50 was needed. This included
^
providing predictions of Semiscale Test S-07-108, LOFT Test L3-1, and providing experimental verification of the various modes of single-phase and two-phase natural circulation predicted to occur in each vendor's reactor during small-break LOCAs."
Page 3-178 of Reference 1 states that: "Other separate effects tests (e.g., ORNL core uncovery tests) and future tests, ae' appropriate, should also
- be factored into this assessment."
- Three types of integral system tests are discussed in Reference 12.
These include tests at the Thermal-Hydraulic _ Test Facility (TNTP) at ORNL, tests at the TLTA at General Electric, and a Ningl's LOFT test. The TNTF tests were not modeled as integral tests, but rather only the test section was modeled using boundary conditions from the experiments, so questions on the TNTF tests are presented in Section VII, along with other questions related to core heat transfer. The TLTA is for BWRs, so it was not reviewed for PWR SBLOCA applications. Therefore, LOFT Test L3-6/L8,-1 was the only integral PWR calculation included in the submittal. '
9M Clarify Why additional LOFT tests and also semiscale tests were not simulated as recoamended in II.K.3.30 (Reference 1). It would i appear the semiscale MOD-2A Tests S-NC-5 and S-NC-6 that involve
, injection of a noncondensable gas for both tests and natural circulation for Test S-NC-5 and a reflux cooling for Test S-NC-6 would be especially useful.
A.V.1 Additional assessments of RELAP5YA have been made. A prediction of LOFT small break Test L3-1 has been performed and is contained in ,
Appendix V.1-1. To assess the various modes of natural' circulation.
~
t- g- ~ , .-, , , - - , , - - , , . . - - , . - - - - . m - ,-.. , , , - ..,--..--.,.,m,, . ..m..._,,,..-,..,,,,,,u.-,,-,-e.-... , - , - . . . , - . - . - - -
semiscale test S-NC-2 has been analyzed. This test contains a range of power levels and primary system inventories. The results are contained in Appendix VI.9. This test was chosen instead of S-NC-5 l or S-NC-6 since noncondensable gas has been determined to be insignificant for our SBLOCAs (see A.I.3 and A.III.1 to A.III.5, July 1985).
f o
I
Appendix V.I-1 l
l Introduction To provide auditional assessment of integral PWR raall break phenomena, a RELAP5YA calculation of LOFT Test L3-1 was performed. The main difference between this test and L3-6, which was previously analyzed in Reference V.I-1, cre as follows:
a) The reactor coolant pumps were tripped at the start of the accident, b) The break was located in the inactive broken loop, and c) Accumulator injection was allowed during system depressurization.
~
LOFT Facility A description of the LOFT facility is contained in Section 5.3.1 of Reference V.I-1.
RELAP5YA Model The RELAPSYA input used to model LOFT Test L3-1 was based on the input previously used to analyze Test L3-6 (see Section 5.3.2 of Reference V.I-1).
Therefore, only the changes made to the model will be discussed and are listed below:
o The cold leg break location was changed from the intact loop for L3-6 to the broken loop for L3-1.
o The break junction was modeled with a discharge coefficient of 1.0 for both subcooled and two-phase critical flow.
o The ECCS injection was directed into the intact loop cold leg as per the test configuration.
l
o The accumulator and associated piping were modeled to inject into the cold leg.
o The temperature of the ECCS water and the auxiliary feedwater was artificially raised to 200 F to Italt the steam condensation rate in the cold les injection volume, o The reactor coolant pump flywheel system with its variable inertia was modeled through a code modification.
o Initial conditions were entered to match the measured initial conditions of the L3-1 test.
Small Break Test L3-1 l
LOFT Test L3-1 was designed to simulate a 2.5% small break (4-inch l equivalent) in the cold les of a large PWR. The actual break diameter was ;
0.6374 inch and was located in the broken loop cold leg. Safety injection was provided by means of the HPIS and the acccmulator directed into the intact loop cold leg. The reactor coolant pumps were tripped at break initiation.
The experimental procedure for LOFT L3-1 was as follows. The reactor was operated for 92 hours0.00106 days <br />0.0256 hours <br />1.521164e-4 weeks <br />3.5006e-5 months <br /> prior to the initiation of the experiment. The reactor was then scrammed 2.15 seconds before the break was opened.
Electrical power to the reactor coolant pumps was terminated at blowdown initiation and the pumps began to coastdown under the influence of a flywheel system. Below a speed of 750 rpm, the flywheel system was disengaged.
Coastdown of the reactor coolant pumps was completed at about 19 seconds.
During the blowdown, ECC injection was directed into the intact loop cold leg. The HPIS flow started automatically at 4.6 seconds while the accumulator injection started at 634 seconds when the system pressure dropped below 633.8 psia. At 1,741 seconds, the accumulator and surge line emptied I
! and the nitrogen cover gas entered the system. The auxiliary feedwater pump l
l started at about 75 seconds and operated for 30 minutes. The fuel rods remained well cooled throughout this text. A more complete description of Test L3-1 is provided in the Experiment Data Report (Reference V.I-2).
' The results from the RELAP5YA prediction of LOFT Test L3-1 are presented in Figures V.1-1 through V.1-7. The calculation was ended slightly beyond the measured accumulator depletion time since this generally l corresponds to core recovery in PWR SBLOCAs. Figure V.I-1 compares the predicted and measured primary system pressure. The comparison shows very good agreement for the first 350 seconds. Between 350 and 500 seconds, the calculated decrease in pressure is less than the data. It is suspected that this slower depressurization is a result of a lower break flow quality and the subsequent lower energy removal rate. Beyond 500 seconds, the calculated and measured depressurization rates are similar, although the calculated pressure remains about 40 to 100 psia above the LOFT data.
The secondary system pressures are compared in Figure V.I-2. For the first 725 seconds, the predicted secondary pressure agrees with the measured value. From 725 seconds on, the code underpredicts the data. This underprediction is suspected to result from an overestimate of the mixing of the auxiliary feedwater and a resulting overprediction of the steam condensation rate. This overcondensation occurs in spite of the fact that the auxiliary feedwater temperature was raised to 200 F. However, it has been determined by a sensitivity study for LOFT L3-6 that the steam generator secondary side conditions have little impact on the primary system response (see A.IV.4).
i Figure V.I.-3 shows the RELAPSYA calculated break flow rate compared to the LOFT L3-1 results. The comparison shows that RELAPSYA predicts the overall trends of the data. However, between about 350 to 1,000 seconds, the calculated break flow rate is slightly higher than the data.
The calculated intact loop and broken loop cold leg densities are compared to the LOFT L3-1 data in Figures V.I-4 and V.I-5. For the first 630 seconds, the LOFT cold leg data show a gradual decrease in density followed by an increase due to accumulator injection. The data also suggests that steam voids were uniformly distributed between the intact and broken cold legs throughout the test as,sh,own a ,
by comparing the center beam data from each loop. The RELAPSYA calculation shows a more abrupt density decrease at about 150 seconds in the broken loop and 250 seconds in the intact loop. This suggests somewhat more vertical stratification was predicted for the primary
_ ~ . _ . . _ . _ . . _ . _ _ _ . _ _ . . , . - . _ , _ . - . . . _ , _ - - ._.,,.-..__,m.m_ ~ , ,_ ,_-. ,., _ , _ , _ _ . _ . , . , , . . _ _ _ _
.. = . _ .-. - - - -- . - - ..
t cystem early in the test-than indicated in the data. Beyond 630 seconds, RELAP5YA calculates a lower density in both cold legs than indicated by the data. We believe these differences are due to the following:
- a. The predicted break flow rate from 350 to 1,000 seconds is somewhat higher than the data.
- b. The predicted accumulator flow rate is lower than the data and initiates about 100 seconds later. This is discussed further below.
- c. These two effects imply that we calculate a lower system mass inventory than probably occurred in the test, and would explain the !
lower cold les densities that were calculated. However, the test report does not contain the mass inventory history to confirm this observation.
- Figure V.I-6 presents the calculated and measured accumulator. level.
Since there was no measurement of the accumulator flow rate, the accumulator level provides an indirect means to determine volumetric flow. First, the comparison shows that the calculated time of injection is about 100 seconds 1 later than the data as a result of a higher calculated primary system !
- j. pressure. Second, the RELAP5YA calculated level decreases at a slower rate than the data, indicating a lower volumetric flow rate. The main cause for this lower calculated flow rate is the assomption of a higher ECC water temperature, that is 200 F used in the calculation versus 88.5 F used in 3
the test. This higher temperature reduces the condensation rate and thereby i reduces the pressure difference between the accuanalator and the primary system. The reason the ECC temperature.was raised was to: 1) prevent an
- unphysical rapid drop in primary system pressure, and 2) ensure conservative ECC injection flow rates. Figure V.I-7 compares the calculated and measured i accumulator pressure. The results are similar to the level comparison presented in Figure V.I-6.
t i.
i
Conclusion The RELAPSYA calculation generally provided a reasonable analysis of the LOFT System response for Test L3-1. For those areas where the RELAP5YA calculation deviated from the data (i.e., densities and accumulator level),
the code results tended to be in the conservative direction. In particular, we believe we underpredicted the primary system inventory.
References (V.I-1) R. T. Fernandez, R. K. Sundaram, J. Ghaus A. Husain, J. N.
Loomis, L. Schor. R. C. Harvey and R. Habert, "RELAP5YA - A Computer Program for Light-Water Reactot System Thermal-Hydraulic Analysis, Volume III: Code Assessment,"
Yankee Atomic Electric Company Report YAEC-1300P, Volume III (October 1982) (Proprietary).
(V.I-2) P. D. Bayless, J. B. Marlow and R. H. Averill, " Experiment Data Report for LOFT Nuclear Small Break Experiment L3-1,"
NUREG/CR-1145, January 1980.
(V.I-3) LOFT L3-1 Data Tapes.
l 1
a
.; g
'" O
~
y ' '
I. ~h
=
g !
e 1 3 .l 1
- S i
. .g
- e 6 '
. 5 o-e a
a
.. .g g
,, - m-r * *%
- g. .g gg T l.- ; #3
.g #6
. G. .
~
d6 e *-
82 g- -l- t $
o h ." Gu"8 oo mm
- g. g le.e 8-a 36 to 1
- -g H t3 o 8.C g
-l uS 7
.I "
'N $
9 i
. .g o
. .g
. .g
.g
- ci o ooot o oos o nos om o coe a autz o coat o cost o oost owt o ooet (WISd) 380SS38d 03110H dOO11091NI
.. 0
- ~ - . ~ ~ - . I 0 0
0 1
0 0
N
- 1 1 - -
A- 1 Y3 _
0 5L 0 F
fT
- 5 8
- 1
- 0 n e 0 i
- 5 4 e
- 1 r 0 u s
0 s
- . 5 e A
3 1 rn Po T 0 i A 0 mt D . 5 2
ea tl 1 1 su yc
- 0 3 0 Sl a
- L . . 5 1 yC 1 r T 0 aA P dY O 0 nS L .
5)0 oP 1C cA O eL T -
0.
0(
SE SE R f
N .
d9 oot O E S M n
_ I 0. I o1 R
d0T s-
. A .
8 i3 P rL M 0 a
O pT mF C .
d 0
oO
- R 7 CL Y
. 5 O.
o :
P . d 2 A. e 1 1
E 0 .
. R 0 V
d5 e r
0 u g
0 i
. d4 F 0
o
. d 3
0
- o
. d 2
0 o
. di 0
1 0
- - 1 ~ - _ .
~
.g. . o g- o og *8 a-l #g o R E.nhe o
t a t O. s$>.n (
e d z I3' j I I :,
1 I I I l
~ El ,
cl
', o.
[ BM C
I d
-o R
S l
to O % c5 A '
< > o. ou U
_ f w' _8 N
U$
o - %. 8 d 3 om $$
o do%
H 8$
j$a
- t3 z o 2.-
o O _ g _d oE $b s CD H g ..
C
- H2 L i E _ _o .
O O $
o a m
W _@o a_
T
__] U.
Ecl _
_o E O o.
I O 1 I I I 0'09 0*0S 0*0F 0~0E 0'05 0'01 0'O (03S/W97) 3198 M0lJ MU389
RELAP5YA CalfARISON TO LOFT L3-1 DATA y , ,
3
\ -
RELAPSYA m-eDC-PC-001A a
f j *-4 DE-PC-001B x- DE-PC-0010
(
q RELAP5YA l r A WWM ;
, ll 5
Q a g
f g, AW M *NY h
o h {
So j d n'"
h 1"'
l l l g^ & ah l 4 i
o.o acIo.o aqIs.o acio.o 4cIo.o soIo.o soIo.o FoIo.o soIo.o soh.o lo66.o'1160.0 La60.oilio.c 1460.o'15dodMb '
%.o161.0 TIME (SEC)
Figure V.1-4: Comparison of Intact Loop Cold Leg Density in LOFT L3-1 to RELAPSYA Calculation 4
i o
g j ~
.l
.8 -i, m
s
- N
-g E
- a f -i 58 l- N C
-n 4
sii et o, 3 T'
9- g
.g ja
. cN
-g Eh 8 8g ?S $oo
-go
- E 8 -
1
.gC aA l
l
' i- -
M 8
3, -f o b
$[S t .
9 **
~h ,!.
F f a Q
i
-8 O*ts eIk ON 0*bt 0*N 0*EE 0*bt 0*El 0*bt Ok O*C (EIJ/W87) 111SN30 S31 0100 d001 N3N088
. . . , _ _ . _ . . _ ~
RELRPSYR COMPRRISON TO LOFT L3-1 ORTR 80 a = = ='===a ' = = * '== ==='===a '
= = 8 = = = = ' - = =
. I I I I I I I I I .
i . .
l 50- .
4
'- 4 . 0.- - _ .
- h. . .
- w . .
& w : :
ui J .
1 i i 30
- e . .
i o .
i 6-i e; - -
1 i a
, m . .
i 8 o
20- _ m -eRELAPSYA _ ;
m :
LOPT L3-1 :
i . .
~
! 1 05 .
0 0- ' ' '
1 '
Cl '2( O' ' ' '4( O' ' ' '6( O' ' ' '8(10' ' ' i0 ld ' ' (2 3d ' ' i4 )d ' ' i6 )d ' ' i8 )d ' ' f 0 30 TIME (SEC) '
l Figure V.1-6: Comparison of Accumulator Level in IDFT L3-1 to RELAPSYA Calculation
1 _ _w - - - - - . - - - - - - -
e i
' ' i a _
a
. (
s, 83
- 'N~
E e9 g
'N 5h 7 - a<
- c. ' Y lM
~
3
- 6. <.
% 8
. *Q w* 3 R~
~
iM
. t3!-
n n e
~
g .,
~
i .
7 jj "
9 .g o
i l
i l .
, I 1 F OM 0* lib 0*M 0*($5 ON O*O 0*M 0*tkE 0* TEE 0000iOSIE d (WISd) 380SS38d 80Itn0Wn00W 1
-- - -~~ . . - - - - - - , _
W Clarify other tests or actual plant transients that might be available to verify other areas or parts of areas that require verification.
W Please see the response to Q.V.1.
W Provide justification for not performing analyses for the other tests and plant transients that could provide additional verification.
W Please see the response to Q.V.1.
0.V.9 Figure 5.3-14 of Reference 12 shows that RELAPSYA overpredicts the primary pressure decrease caused by condensation following accumulator injection. I,,
Clarify how this condensation will be modeled for SBLOCA calculations to prevent too much condensation occurring with the resulting overprediction of ECC injection rates.
A.V.9 The ECC modeling guidelines described in A.II.5 to II.8 reconunends artificially raising the ECC water temperature used in the analysis close to the saturation temperature corresponding to the containment pressure. The recommended temperature is about 200 F to avoid the possible injection of superheated water.
Figure 5.3-14 described in the above question refers to the RELAPSYA prediction of LOFT Test L8-1. To check the proposed ECC modeling guidelines, the L8-1 test prediction was reanalyzed with the following changes:
a) The accumulator water temperature was raised from an original value of 88.5 F to 200 F.
b) The accumulator model was replaced by the RELAPS/ MOD 1 Cycle 18 model (see A.I.23 July 1985).
c) The initial conditions for Test L8-1 were taken from a L3-6 run in which the secondary system pressure was set to match the test data. This was done to consolidate computer runs.
Figure V.9-1 presents the original and revised primary system pressure compared to the L8-1 test data. The revised RELAPSYA primary pressure is slightly above the original calculation as a result of a higher secondary system pressure. However, both calculations are somewhat below the measured data. As the accumulator injection occurs, the original RELAP5YA calculation tends to underpredict the system pressure While the revised calculation follows the data reasonsbly well. Figure V.9-2 compares the peak node cladding temperatures for the two RELAP5YA calculations and the test data. The revised calculation shows a slightly later quench time than either the original calculation or the test data.
In LOFT Test L8-1 the accumulator injection was delayed. This created a higher initial pressure difference between the accumulator and the primary system that resulted in higher accumulator flow rates. This atypical test procedure causes more rapid condensation than expected for PWRs. For this reason, raising the ECC water temperature to 200 F should limit the condensation and injection rates for PWR SBLOCAs.
o.
- . - . - - - o o
~
- m
^
o.
' s
. i s
a
- o. 1
_ o -
. i s 8 a L
_ Ts
- o. F n o Oo
. Li d t
_ a na il
__ u
__ o. ec o rl
__ i s ua
__ a sC R s T __ eA A __ o. rY D o PS
. i s P 1 a mA eL
- __ t E 8 o. sR y
L o
. Sd T
F da )CE y se O S ri L av o.
o
(
me O .
da EM Pd iR r
T I W T f an
( o. o S o l I . i s na
. R a on f si i g f o. ri I
( .
o ar pO C da m l
f oo Y Ct 5
r.
f
- da o.
o 1
L -
E - L 9
R o.
DA EN o V
. i SI s e 1IG a r
_ VI u 8ER g
., o. i
. LRO o F TRR i
FYY m
-,- O55 LRR o.
r ^ o ae d a
- - M 0
. m. 5 0
3 3
^
o s
Wi ~. - - - - -
s a
o40 Ol e og 9an aR 9I[ oB ad E ua_
n wEQ. c.wJ g s8szw Iw i
,':: ,, .l! ?i : j :,1l} i,1i,..j;il' l , ,(l}l:11 i
i I o RELRP5YR CoffRRISON TO LOFT LA-1 DRfA g
l t
=
B' i
! =
] g- -
i
" LOFT L8-1 i
-l-b #
a-*R5YR RE: VISED M R5YR ORIGI M ,
i 1 kl o 1 C 4
i o
a eI f: -
\
kg m
^^
_ :^ - - -
1 o - ~
l o n s ,_
s% ~
oM l I i Ho i
i ln-o g- .
o a
0.e ai.e 45.0 si.e si.e sein.e gio.e lio.e leis.e se's.e 30.0 TIME (SEC)
Figure V.9-2: Comparison of Node SG8-026 Cladding Temperature in LOFT L8-1 l to Original and Revised RELAPSYA Calculations
VI FLOW REGINgS 0.VI.1 Areas Where horizontal countercurrent and/or stratified flow of steam and water may occur are in the. hot legs, cold legs and pressuriser surge line. Countercurrent flow could occur in the hot leg with steen going from the vessel to the U-tube steam generator, t t condensing and then flowing back along the hot les to the vessel (reflux cooling). Countercurrent flow might also occur with potential slugs of liquid introduced by cold 3CC water injected into a largely vapor-filled cold leg. It might also occur in the pressuriser surge line. Clarify When countercurrent and/or stratified flow will occur and how the transition from nonstratified to stratified flow is modeled.
A.VI.1 Countercurrent vapor-liquid flow is calculated in RELAP5YA as a consequence of the solution of the basic governing equations.
Stratified flow is assumed only in horizontal components and only when the stratification criterion is met. This is described in the i
i horizontal flow regime map shown on Page 28 of Appendix A of Reference VI.1-1. A transition region between stratified and nonstratified flow is also shown in the flow regime map. Within this region, the interphase drag is calculated by interpolating on mass flux between the stratified and nonstratified regime boundaries. This interpolation provides continuity for the transition to either regime.
Reference i
i (VI.1-1) Fernandez, R. T., et al., "RELAP5YA - A Computer Program for LWR System Theriaal-Hydraulic Analysis, Volume I,"
l YAgC-1300P, January 1983 0.VI.2 Page 14 of Reference 8 states: "The additional force term which arises for stratified flow geometry in horizontal pipe is added to the basic equation when the flow is established to be stratified from flow regime considerations." Clarify how momentum surges are avoided when the stratified flow model is turned on and off.
l
A.VI.2 In s' typical BRLAP5YA plant model, there are only a few horizontal components which generally represent horizontal runs of reactor coolant pipes. It is not obvious that if the flow goes from nonstratified to stratified in one of these components, a momentum surge will ensue that will have significant impact on overall system behavior. The inclusion of the extra pressure gradient term in stratified flow is an attempt to improve the modeling of stratified flow phenomena (refer to Page 14, Appendix A. Reference VI.2-1).
This ters only appears in the difference momentum equation Which primarily determines the relative velocity between the phases. This
-equation contains other tsras (for example, the interphase drag term) which are smoothly behaved between nonstratified and stratified flows. Hence, we do not believe that a transition from nonstratified flow to stratified flow will result in a nonphysical momentum surge.
It should be pointed out that momentum surges can occur due to a variety of reasons (for example, propagation of density waves). In general, no attempt is made to avoid them if they are not accompanied by nonphysical numerical behavior and buildup of mass error.
0.VI.3 Page 32 of Reference 8 states that: for the virtual mass coefficient, C, "It mey be appropriate to assume that C=0 should be used for separated or stratified flows. At present, the value of C defined by Equations (109) or (110) is used without regard to the flow regime." Clarify how much different the separated or stratified flows would be if C=0 was used.
A.VI.3 A value of C=0 might be appropriate for completely separated or stratified flows only when the two phases do not significantly interact. However, this will only occur when (a) the phasic velocities are small, (b) the temporal derivative of the relative velocity is small and (c) the spatial gradients of the phasic velocities are small. For these conditions, the dynamic drag terms in the momentum difference and the thermal energy equations that contain C will be small (see Equation 107 of the cited reference).
w--.-._.------- - - . - ---__
l
! Therefore, we expect small phasic flow rates for completely separated or stratified flows regardless of the value for the virtual mass coefficient, C.
0.VI.9 Clarify how the various modes of single-phase and two-phase natural circulation have been verified analytically and experimentally.
A.VI.9 To verify RELAP5YA's ability to predict the various modes of natural circulation, an assessment of semiscale Natural circulation Test (S-NC-2) has been performed and is described in Appendix A.VI.9.
Appendix A.VI.9 Introduction An assessment of Semiscale Natural Circulation Test 2 (S-NC-2) has been performed to demonstrate that RELAP5YA predicts the single-phase, two-phase and reflux modes of natural circulation.
Semiscale Facility The Semiscale MOD-2A test facility at the Idaho Nati 41 Engineering Laboratory was used to investigate the thermal-hydraulic phenomena cccompanying various hypothesized loss-of-coolant accidents in a PWR System.
The Standard Semiscale MOD-2A System (see Figure VI.9-1) is approximately a 1/1700 scale model of a four-loop PWR. It consists of two primary coolant loops connected to a pressure vessel which has an external downconer. The intact loop is scaled to represent three loops while the broken loop represents a single loop. The pressure vessel contains electrical heaters to simulate the reactor core.
Several separate effect tests were performed to examine the important system parameters during natural circulation. The Semiscale Natural Circulation Test 2 (S-NC-2) was performed using a subsystem of the MOD-2A facility (see Figure VI.9-2). This subsystem had the broken loop, the upper head and the intact loop pump removed. Removing the broken loop makes the facility a single-loop system. Removing the upper head reduced the primary system volume, but did not affect the natural circulation modes since the bottom of the upper head was located above the hot legs. The intact loop pump was replaced by a spool piece to eliminate leakage from the pump. Hydraulic resistances were simulated by an orifice in the line.
8-NC-2 Test Description The S-NC-2 experiments were performed in the MOD-2A facility to examine various modes of natural circulation [ Reference VI.9-1]. Three cases at different power levels (30, 60 and 100 kW) were examined. At each power level, the system started at 100% primary mass inventory with the pressurizer establishing subcooled fluid conditions. The pressuriser was then valved out of the system and mass was drained from the primary system in discrete
_ amounts. The primary pressure was allowed to float while core power and steam generator secondary conditions were held constant. Test data were recorded when steady-state conditions were established at each mass inventory.
l At 100% inventory, single-phase natural circulation existed with subcooled liquid circulating through the system. As liquid was drained from l the primary system, the liquid level in the vessel dropped. Once the level dropped below the hot les elevation, bubbles appeared in the hot les and i
two-phase natural circulation was established. As the mass inventory was further reduced, the flow rate increased until bubbles were observed in the steam generator outlet. With continued reduction of mass inventory, reflux flows were observed in both sides of the steam generator.
i
- All modes of natural circulation were capable of removing core heat.
In addition, with the exception of the reflux mode, no dryout or heater rod temperature excursion was observed.
1 RgLAP5YA Model I
The RELAP5YA input used to model the Semiscale Natural Circulation Test 2 (S-NC-2) is based on the RELAPS/ MOD 1 base model described in Reference VI.9-2. The RELAPS nodalization for S-WC-2 is shown in
- Figure VI.9-3. The following changes were made to the base model to obtain the RELAP5YA model
(a) The pressurizer component was replaced by a time-dependent volume for the 100% mass inventory cases. This change is not significant l
since the time dependent volume, like the pressurizer, establishes the desired primary pressure when the system is at 100% mass inventory. For primary mass inventories less than 100%, the pressurizer in the experiment was valved out and the corresponding time-dependent volume in the RELAP5YA model was eliminated.
(b) All the* passive heat structures were removed from SNLA's RELAPS model except the heat structures that model the steam generator U-tubes. This change was made to simplify the model and should be insignificant since S-NC-2 consisted of steady-state tests. In addition, the tests were conducted with strip heaters designed to offset the environmental heat losses. Therefore, the assumption of an adiabatic boundary in RELAP5YA adequately represents these hast structures.
Cases Analyzed The 3-NC-2 experiments were performed at three power levels (30, 60 and 100 kW). We analyzed the 30 kW and 100 kW cases at various primary system inventories. -The results for these two cases are presented below.
)
1 Results Figures VI.9-4 through VI.9-6 compare the measured and calculated hot les temperature, primary system pressure and mass flow rate as a-function of primary mass inventory for the 30 kW case. The calculated hot les temperature and primary pressure compare well with the measurements. As in the data, the primary pressure decreases because the hot leg is at saturation conditions.
The maximum flow in the loop was predicted when the inlet flow to the steam generator U-tubes was two-phase and the outlet flow was single-phase fluid.
The peak flow rate was overpredicted by about 25% and also occurred at a slightly higher inventory than observed in the experiment. As the system mass inventory was further reduced, the calculated mass flow rate also decreased, closely following the experimental trend. The single-phase, two-phase and reflux modes of natural circulation were predicted reasonably well by RELAP5YA. Reflux was calculated and experimentally seen at the lowest inventory (approximately 60%).
At some system inventories, oscillations in mass flow rate were calculated. The range of oscillations at these inventories is indicated by vertical bars in Figure VI.9-6. The experiment shows flow oscillations at some system inventories as well. The atact magnitude and period of these oscillations is difficult to determine due to poor resolution of presented experimental data [ Reference VI.9-1].
Figures VI.9-7 through VI.9-9 show the measured and calculated hot leg temperature, primary system pressure and mass flow rate as a function of cystem inventory for the 100 kW case. As in the 30 kW case, the calculated temperatures and pressures compare well with the measurements. The maximum flow in the loop was predicted when the inlet flow to the steam generator U-tubes was two-phase and the outlet flow was single-phase fluid. The maximum flow in the loop was overpredicted by about 25% and occurred at a slightly lower inventory than the experimental data. As the system inventory was further reduced, the calculated mass flow rates also decreased. The experiment identified reflux cooling to occur for the lowest inventory (56.9%)
chown in Figure VI.9-9. This mode was not calculated to occur at this inventory. However, at a lower inventory (55%), periodic countercurrent flow was calculated to occur in the hot legs accompanied by periodic core heatup.
This is indicative of the beginning of the reflux cooling mode. As in the 30 kW case, mass flow oscillations were calculated at some inventories. The cause of these oscillations are also believed to be due to an unsteady calculation of void distribution.
Conclusions The results of our natural circulation calculations show that RELAPSYA qualitatively describes all modes of natural circulation correctly. The calculated mass flow rates were generally higher than the data. The impact of these higher flow rates should not be significant from the point of view of LOCA analysis, since adequate core cooling is maintained for the single-phase and two-phase modes of natural circulation.
l References (VI.9-1) O'Connel, Thomas M. , NUREG/CR-2454, EGG-2141, " Experimental Data Report for Semiscale MOD-2A Natural Circulation Tests S-NC-2B, S-NC-3 and S-NC-4B," December 1981.
(VI.9-2) McClaun, J. M., and L. M. Kmetyk, NUREG/CR-3258, SAND 83-0833, "RELAp5 Assessment: Semiscale Natural Circulation Tests S-NC-2 and S-NC-7," May 1983.
0.VII.3 Pages 18-20 of Reference 12 show the RELAP5YA results compared with the test data for a GE level swell test. Again, the predictions are under the data for low void fractions. Clarify why at 10 s, 100 s, 150 s the RELAP5YA void fractions decrease significantly in the upward direction in the volume just before the steep increase in void fraction occurs.
A.VII.3 RELAP5YA does not contain models to predict and track two-phase levels within a volume. In the calculation of the interphase drag term. Fgy, at junction locations, a volume-weighted void fraction is used. This would result in some smearing of liquid above the location of a two-phase level and, hence, some inaccuracy in the prediction of interphase drag near regions of steep void gradients.
Also, RELAPSYA uses a donoring scheme to determine void fractions and fluid properties at junction locations. This would also have
'some impact near regions of sharp void gradients because of the occurrence of countercurrent flows. It is believed that the above characteristics combine to yield the void profiles referred to in the question. However, the dip in the void prediction below the level does not propagate to other regions and did not result in a buildup of mass error. In the calculation, the location of the level is assumed at the location Where the void fraction reaches a value near unity. In general, the calculated level is always below that observed in the data. 1This is in the conservative direction and believed to be acceptable for licensing analyses. .
O.VII.4 Clarify why the void fraction at the 10 and 12 ft elevations takes so much longer to reach 1.0 for the calculations than the data.
A.VII.4 There is an error in Figure 2.1-14. The data are the solid lines and the RELAP5YA calculations are the dashed lines (the reverse is indicated by the figure). The calculated void fraction at a given location reaches unity faster than indicated by the data. The corrected figure is attached.
I I# . . . . . . . . y r r - .
J
.S . ,o .
.8 .
.7 .
f o
,, DATA s j : N/ 1
.4 . ,' .
o
= #
A . $ .
l
.2 .
\
.1 .
g,g . . > > . i n .
8.0 10 2.0 30 48 SJ S.8 10 98 00 18 8 11 8 18 0 13.o 14 0 E 1AL ELETM IGR. (FT)
Figure 2.1-13: Axial Void Distribution at 150 Seconds for.
GE Level Swell Test 1*88 r , ,-. -r 1 1 1 .
u ,v-1 1 i,1 I g
.=
l l wn .
, a
--- usa m
.so j j 10 ge, .
.M .
8 12ft./ 0 5 .so . l ,' .
e , ,
l m. ,
- o l .
i, e a g .4e . s, .
I i
.= .
l .
a
.ao .,so'. ,*'%% ,P
< 2 ft.
~ K ~~~ ~** ~~* **~~~~~' ^- * *
.to 1 ,e ~~ .
- 9 e.co. > > > > a e ia e i e n = = i i i a to to so de se so w pg og sto tes tso too too too 17o too Figure 2.1-14: Nodal Void Fraction Versus Time
'O.VII.6 Page 78 of Reference 10 states: "RELAP5YA uses a large finite value (10 N-S/m ) for F7 in ne void fraction e m e 0.0 M g<
- to ensure homogeneous flow behavior. To provide continuity with this scheme, a narrow-range of void fraction is defined as 0.001<af <0.01. In this void fraction range, F h cahdated by 7
exponen ial interpolation between 1020 ,,3j ,4 and the value calculated by the undistorted bubble model ats( = 0.01." Clarify if the F is used directly or if a relaxation method is applied.
7
+
A VII.6 Please see A.VII.8.
0.VII.7 Clarify if this large change in F with 7 small changes in void fraction can lead to instabilities.
A.VII.7 Please see A.VII.8.
0.VII.8 Page 101 of Reference 10 states:- "After the interphase drag term F is calculated for a particular junction in subroutine FIDRAG, 7
it is averaged over space and time." Clarify how this averaging is done and how long it takes for a very large F to change to the 7
much lower values for void fractions greater than 0.01.
A.VII.8 The interphase dras coefficient F 7 is first calculated according to the methods described in Section 3.1 of Reference VII.8-1 for each hydrodynamic junction using junction properties. This value, designated as Fg* , is further adjusted as follows:
Given F
- for all junctions:
g
- a. Volume-based F are calculated using inlet and outlet 7
junctions:
Fg = 1/2 R gy y IN OUT a
- b. Junction-based F are recalculated using volumetric weighting 7
of volume-based F values of adjacent volumes:
s F
IJ
= IVK
- K+ I V
. L VK+ L
- c. At each juncticn, the F7is time-averaged as an arithmetic average between old time value and the latest value to give the new time value:
yn
,n+1 ,
,IJ IJ IJ 2
Thus, in response to Q.VII 6, the F7 calculated by the models in Section 3.1 of Reference V'II.8-1 (designated above as Fg* ) are not used directly, but some under relaxation is provided as described above. In response to Q.VII.7, the above scheme will bias the final value of F g towards the 1seger value between Fg **
and F" . In general, larger values of F 7y will provide stronger coupling between the phases and, hence, less o'scillatory behavior. Thus the above averaging scheme is believed to help in mitigating instabilities.
In response to Q.VII.8, the impact of averaging F7 as described above will depend on the node and time stop sizes, as well as the type of problem being analyzed. The motivation behind this averaging procedure has been discussed in Section 3.1.3 of Reference VII.8-1. Its largest impact will be when a junction undergoes a change between near-single-phase conditions (voids below 0.001 or above 0.999 Fg = 10 N-S/m ) to two-phase conditions (voids between 0.01 and 0.99, F7 = IM to 1[
N-S/m ). The time-averaging schene will allow this transition over about 50 time steps. For small break LOCA analyses, the time step sizes are typically 0.01 to 0.05 seconds. Thus, the impact on the transient will be a: delay of 0.5 to 2.5 seconds in the transition between near-single-phase and two-phase conditions. This appears reasonable for small break LOCA transients which generally occur over a time scale of 10 to 10 seconds.
. _ . . . - . .- - _ _ . _ _ . - _ - ~ _ _ _ _ , _ - _
Reference (VII.8-1) Fernandez, R. T., et al., "RELAP5YA - A Computer Program for LWR Thermal-Hydraulic Analyses, Volume I,"
YAEC-1300P, January 1983.
O.VII.23 Page 2-74 of Reference 19 describes the Yankee Atomic earlier experience in modeling a Lehigh University low flow film boiling test with RELAP5: "In this case, the code underpredicted the wall temperature profile because the heat transfer coefficient (based on T,g) was overpredicted by approximately a factor of two.
Also, the code does not predict the thermal nonequilibrium (approximately 400 F of vapor superheat) measured in the test.
This illustrates the need for improvements in modeling low flow film boiling heat transfer and nonequilibrium vapor generation rates."
Clarify if RELAP5YA has been used to model this same test because the test provided such a severe test for RELAPS.
9 A.VII.23 The reference cited in the question deals with YAEC's initial assessment of RELAPS/ MOD 1 for predicting post-CHF heat transfer phenomena. Subsequent review of the Lehigh test indicated that there were many uncertainties in modeling the Lehigh test facility, especially with regard to the " hot patch" at the test section inlet designed to generate post-CHF conditions in the test section.
Hence, alternate tests have been used for RELAPSYA assessment, described in Reference VII.23-1.
Reference (VII.23-1) Fernander, R. T., et al., "RELAP5YA - A Computer Program for LWR System Thermal-Hydraulic Analysis, Volume III,"
YAEC-1300P, January 1983.
0.VII.25 Clarify how detailed the noding needs to be to get proper superheating rather than forcing the steam to stay at saturation l because it is the minimum phase in a large volume partially filled with liquid.
1
. . _ . - - . -. - - = _ - - _ - . _ _ _ _ . _ _ _ - _. - . - _ _ _
A.VII,25 In general, RELAPSYA will not predict vapor superheating to occur in the presence of the liquid phase. Hence, to calculate superheating of steam, the void fraction has to reach 1.0. This can happen if (a) the noding is detailed enough to show the existence of a two-phase level, if it can occur; and (b) the heat supplied to the volumes above the level is high enough to completely evaporate entrained droplets. This can be seen in the RELAP5YA assessment i against the TNTF boiloff teste 3.09.101, described in section 5.1.5 ;
of Reference'VII.25-1, and further illustrated in Figures 5.1-15 and
- l. 5.1-16 of Reference VII.25-1.
Reference i
et al., "RgLAPSYA - A Computer Program 4
(VII.25-1) Fernandez, R. T.,
for LWR System Thermal-Hydraulic Analysis, Volume III,"
YAgC-1300P, January 1983.
4 O.VII.26 Clarify if forcing entrained' liquid drops to be at saturation reduces the superheating of the vapor.
l A.VII.26 Entrained liquid drops, whether at saturation or not, will reduce superheating of the vapor. Further, in PWR LOCA analyses, it is difficult to envision situations of significance where entrained liquid droplets will not be at, or very close to, the saturation temperature. The vapor temperature is, however, mere influenced by
< the interphase mass transfer coefficient than by the liquid
temperature. RgLAP5YA generally uses a very high mass transfer !
coefficient and, hence, will predict very little vapor superheat in the presence of entrained liquid droplets. This results in higher vapor generation rates. However, this. leads to conservative results for LWR LOCA analyses as discussed in Section 2.1.3.1 of Reference VII.8-1.
0.VII.27 Clarify and justify that the range of CHF data and post-CHF data covered by the experiments includes the range of conditions expected for SBLOCA calculations.
i
._ _, _ _ _ . . . . , _ , _ _ _ _ . - _ . _ . . - . , . , . , _ , . _ . _ . _ _ , _ . _ _ , , . . , _ . . - _ _ _ , , . . , . . .~,
A.VII.27 Reference VII.27-1 presents the assessment of the CHF and post-CHF models implemented in RELAP5YA.
CHF Assessment l
The criteria considered in selecting the steady-state CHF tests j include the range of thermal-hydraulic conditions expected for asLOCA calculations. Twelve tests were selected to assess the CHF options. The tests were performed in three different facilities as follows:
- 1. Five tests in the medium pressure heat transfer flow loop at ;
the Chemical Engineering Research Laboratory of Columbia 1 University (References VII.27-2 and 3).
.2. Four tests in the Nine Rod Test Section at General Electric (Reference VII.27-4). ,
- 3. Three tests in the Thermal-Hydraulic Test Facility at Oak Ridge National Laboratory (Reference VII.27-5).
The range of conditions for all the steady-state tests is summarized in Table VII.27-1. Further ass.essment of the CHF models implemented in RELAP5YA has been performed for the system tests described in Section 5.0 of Reference VII.27-1.
Post-CHF Assessment 4
The most relevant tests for the post-CHF assessment are:
- 1. The steady-state flim boiling tests, 3.07.9B, 9K and 9I, r
conducted at Oak Ridge National Laboratory in the Thomal-Hydraulic Test Facility. These tests are described in Table VII.27-2. The tests provide steady-state film boiling heat transfer data in rod bundle geometry. These data were ,
used in assessing film boiling heat transfer correlations. The heated pin diameter and rod pitch are typical of later
1 generation PWRs with 17x17 fuel bundles. The conditions in the tests K and I are representative of conditions expected during SSLOCA transients.
- 2. .The quasi-steady-state bolloff test, 3.09.10I, conducted at Oak ,
Ridge National Laboratory in the TNTF (Reference VII.27-6).
The objective of the bolloff test series was to study the heat transfer and mixture level swell under SBIACA conditions in Pressurized Water Reactors (FWRs). .
.The conditions for Test 3.09.10I are shown below:
System pressure 650 psi Mass flux 2.19 x 10 lbm/hr-ft Linear power / rod 0.68 kW/ft
- 3. The system bolloff experiment Test 6441/6, conducted at General Riectric, San Jose in the TLTA facility (Reference VII.27-7).
1 The boiloff tests attempted to simulate system conditions which
Emergency Core Cooling Systems, as well as the ADS, were available.
In these tests, the recirculation loops were blocked and the liquid inventory was slowly boiled off at a constant pressure and constant bundle power. The power level was representative of decay heat in a BWR. The main objective of these tests was to evaluate heat transfer in a partially covered bundle at decay power levels and low flows.
The phenomena and the range of conditions encountered in this test might be encountered during SBLOCAs in PWRs before the accumulator actuation if the flow from the high pressure safety injection pumps bypasses the core.
I
The initial conditions for Test 6441/6 are: ,
System pressure 395 110 psia Bundle power 250 i2 kW Initial two-phase level Bundle top LOFT Tests L3-6/L8-1, used in the sasessment of RELAP5YA, simulate a small break LOCA with core uncovery (Reference VII.27-8). Test L3-6, which was a SBLOCA with the pumps running, was extended into a more severe transient.
Test L8-1, which produced core uncovery and heatup. These experiments were configured to simulata a small break equivalent to a 4-inch diameter rupture in the cold leg of a
- large (approximately 1,000 MWe) commercial pressurized water reactor. In Tast L3-6, the primary coolant pumps were operating until the hot leg depressurized to 311.83 psia. The I coolant pumps and High Pressure Injection System '(HPSI) flow l were then terminated ar.d the break lef t open. Experiment L8-1 started when the primary coolant pumps were tripped. When the maximum fuel cladding temperature reached 600.53 F, core reflood was initiated.
Experiment L3-6/L8-1 was initiated from primary coolant system conditions of:
Hot leg temperature 579.1 1 3.24 F Cold les temperature 544.5 3.24 F Hot les pressure 2,156.7 i 20.3 psia Intact loop flow rate 1,065.48 5.72 lb/sec Power level 50 1 1 MW Maximum linear heat generation rate 16.06 i 1.12 kW/ft A detailed account of the integral test conditions is presented in Reference VII.27-1.
References VII.27-1 Fernandez, R. T., R. K. Sundaram, J. Ghaus, A. Husain, J. N. Loomis, L. Schor. R. C. Harvey and R. Habert, "RELAP5YA - A Computer Program for LWR System Thermal-Hydraulic Analysis, Volume III: Code 1
Assessment," YAEC-1300P, Volume III (October 1982) (Proprietary). l l
l VII.27-2 " Critical Heat Flux Correlation for CE Fuel Assemblies with Standard Spacer Grids. Parts 1, 2. Non-Uniform Axial Power Distribution,"
Combustion Engineering Topical Report CEEPD-207, June 1976 VII.27-3 Electric Power Research Institute Report, RPRI-RP-813-1 (to be published).
VII.27-4 Janssen, E., "Two-Phase Flow and Heat Transfer in Multirod .
Geometries, Final Report," General Electric Company Report GEAP-13347, March 1971 VII.27-5 Yoder, G. L., et al., " Dispersed Flow Film Boiling in Rod Bundle Geometry - Steady-State Heat Transfer Data and Correlation Comparisons," ORNL/5822 (to be published).
4 VII.27-6 Anklam, T. M. .et al., " Experimental Investigations of Uncovered Bundle Heat Transfer and Two-Phase Mixture Level Swell Under High Precsure, Low Heat Flux Conditions" (Final Report for THTF Tests 3.09.10I-N and 3.09.10AA-FF-DRAFT), Oak Ridge National Laboratory, September 1981 VII.27-7 Seedy, D. S., and R. Muralidharan, "BWR Low-Flow and Bundle Uncovery l Test and Analysis," EPRI Report No. NP-1781, June 1982 VII.27-8 Bayless, P. D., and J. M. Carpenter, " Experiment Data Report for LOFT Nuclear Small Break Experiment L3-6 and Severe Core Transient Experiment L8-1," NUREG/CR-1868, January 1981.
i l
Table VII.27-1 Steady-State CHF Test Conditions Average Average Heated Rod Heated Pressure Mass Flux Heat Flux Diameter Length Rxperiment (psi) (Mib/hr-ft 2) (NBtu/hr-ft2) (in.) (in.)
Columbia 1,500 - 2,005 1.968 - 2.008 0.282 - 0.436 0.382 150.0 GE Nine Rod 997 - 1,005 0.249 - 1.248 0.289 - 0.522 0.570 72.0 OREL TNTF 635 - 1,849 0.166 - 0.525 0.14 - 0.29 0.374 144.0 Table VII.27-2 THTF Steady-State Film Boiling Data Ranges Pressure Mass Flux Heat Flux Test No. (psia) (1bm/hr-ft 1 2 (Btu /hr-ft21 3.07.9B 1,849 5.25 x 10 2.9 x 10 3.07.9K 635 1.66 x 10 1.4 x 10 3.07.91 872 2.50 x 10' 1.9 x 10 l
O.VII.38 Page 215 of Reference 10 describes the calculation of internal gas pressure for the fuel behavior model. Clarify what values of the user-specified parameter, A TFP, are recomunended for the range of j SBLOCA transients expected to be modeled.
A.VII.38 The RELAP5YA fuel behavior internal pressure model has been improved to achieve greater accuracy through more detailed modeling. The former fuel behavior model assumed that the fuel rod internal pressure was proportional to the absolute coolant temperature adjacent to the fuel rod plenum plus a user-specified offset. The l
accuracy of the forimer model was limited by the specified temperature offset and the neglect of geometric changes. The new internal pressure model addresses both of these limitations. The new model is based on the fuel behavior models in RELAP4/ MOD 3 (Reference 1) and T00DEE-2-EM (Reference 2). The only difference is that RELAP5YA neglects fuel pellet cracks. The internal pressure j
calculation in RELAP5YA lis now based on the volume and temperature in the fuel-clad gap and plenum. The plenum gas temperature is assumed to be the temperature of the adjacent coolant plus an offset. Yankee uses a 2 F offset in our current licensing calculations for large break LOCAs. The same offset value of 2 F will be used for small break LOCAs as well.
The fuel rod internal gas pressure is calculated via a two-region model using the ideal gas law. The model considers a plenum region and a fuel-cladding gap region. The internal pressure, Pg , is calculated by the ideal gas law as:
R . M T
P int " -+
P El T p Eg,y
_Tgy l
.- . . _ _ - . . . .- -~ .. .
where:
i M,= total gram-moles of gas in the fuel rod R = universal gas constant 3 Yp = plenum region volume (m )
Tp = plenum region gas temperature (deg-K)
(assumed to be adjacent coolant temperature plus a user-specific offset) n- = total number of axial fuel segments
! V g= fuel clad gap volume of axial segment i T = fuel clad gap temperature of axial segment i i The plenum volume is calculated using the hot dimensions of the fuel and cladding such that:
Y p = Yp , - 1T ( gr , + & gr ) A Lg +1(ed *A #c) ALc where:
V = initial free volume in the plenum region (m )
Po r initial cladding inner radius (m) d-
&r = change in cladding inner radius taken as that of the top axial node (m)
., r g,= initial fuel outer radius (m)
Ar g = change in fuel outer radius taken as that of the top axial node 6L = total length change of the cladding due to thermal expansion (m)-
AL g= total length change of the fusi column stack due to thermal effects (sum of axial thermal expansion at the pellet dish node (m) f i
i 2
I
References
-(VII.38-1) "WREM: Water Reactor Evaluation Method (Revision 1),"
NUREG-75/056, Division of Technical Review, U.S. Nuclear Regulatory Constission (USNRC), Washington, D.C. , May 1975.
(VII.38-2) Lauben, G. N., "T00DEZ2-EM, A Two-Dimensional Time Dependent Fuel Element Thermal Analysis Prograa,"
Division of Technical Review, USNRC, May 1975.
1 II. BRgAK FLOW 0.II.1 Investigations and experiments "...have demonstrated a wide variation in mass flux as a function of break geometry. Mass flux was shown to be influenced by the degree of curvature at the break inlet, flow passage diameter, flow passage length and the ratio of the break diameter to the vessel diameter. Correlations incorporating all these factors are not available at the present time. Moreover, small break geometries postulated for reactor systems could range from splits in pipes to double-ended breaks restrained by pipe supports, and could include full ruptures in small diameter pipes" (Page VIII-40 of Reference 2).
One approach the submittals might use to account for uncertainties in break flow would be to model a spectrum of break sizes with
'enough size range to account for the uncertainties in break flow modsis. However, note that the uncertainty for liquid break flow is different than for two-phase break flow. Clarify how the uncertainties described above will be treated in SBLOCA analyses.
1 A.II.1 YAEC will provide a break spectrum study in our plant-specific LOCA I analysis submittals as required by Paragraph I.C.1 of Appendix K to 10CFR50.46. Each break spectrum study is aimed at achieving-the following goals:
. a. Identify the maximum' cladding temperature for each accident category to assure that 10CFR50.46 criteria are met.
- b. Account for break flow modeling uncertainties.
' Minimize the number of cases to be analyzed in order to achieve c.
the above.
Each break spectrum study will use the RELAP5YA subcooled choking model for liquid discharge and the Moody critical flow model for the two-phase and steam discharge in our licensing submittals.
Substantial evidence shows that the uncertainties for each model,
. , - - - + - - ,- . . - - - - - . - - - . - + . . - .. . - , , - - , , - - .- ---.----,.v -,
derived by comparing predicted to measured break flow rates, are essentially the same. Therefore, a separate variation of the discharge coefficients for each break size is not necessary since the variation of break size within the spectrum will account for the uncertainty of the two break flow models. This is further explained below.
Subcooled Discharme Coefficient Variation h Comparison of predictions by the RELAP5YA subcooled choking model to subcooled discharge data from LOFT and TLTA yields variable discharge coefficients during subcooled blowdown that generally lie in the range of:
0.6gC D-SUB k wh.re C
D-SUB "- EST YA-SUB*
This is shown in Figures 5.2-5, 5.2-26 and 5.3-5 (attached) from Reference II.1-1. These tests used relatively small nozzles.as shown in Table II.1-1.
This same trend is shown in Abdollahian's comparison of predictions from the Burnell subcooled choking model to Marviken subcooled discharge data. This is shown in Figures 5.1 through 5.4 (attached) from Reference II.1-2. Please note that the RELAPSYA subcooled choking model is very similar to the Burnell model l (References 11.1-3 and II.1-4). These Marviken tests were conducted with relatively large nozzles as shown in Table II.1-1.
Moody Discherme Coefficient Variation Durinz Blowdown comparison of predictions from the Moody Model to two-phase discharge data from Marviken also yields variable discharge coefficients that generally lie in the range of 0.64C y (1.0 4
where C
D-20 " EST DY*
This is shown by Abdollahian in Figures 5.1 through 5.4 Where he has used-the upstream thermodynamic state at each point in time.
BREAK FLOW IBODELING UNCERTAINTIES i The preponderance of these data show that the discharge coefficients for each model applied over a range of conditions are essentially comparable and lie in the range from about 0.6 to 1.0. These discharge coefficients are directly related to the uncertainty of-each break flow model as follows:
Uncertainty,'% = DEL ~ EST EST
=
( CD - U x 10 M Since the discharge coefficients for both models span a similar range, then we infer that these models contain similar uncertainties that can be treated as equal. This implies that parametric studies could be performed using the same discharge coefficient for each model to bound the uncertainties. However, this same result is
- obtained when we vary the break size within the break spectrum study. Therefore, we conclude it is not necessary to separately vary the discharge coefficients for each break size.
We further observe that if a break size is dominated by either subcooled or two-phase discharge, then variation of the break size alone is sufficient to bound the uncertainty of the dominant model a
since that variation is equivalent to a variation of the discharge coefficient. Finally, we note that errors in break flow modeling
can be somewhat self-correcting. This stems from the fact that both the subcooled and two-phase discharge flow rates decrease with pressure. Thus, if a model predicts too large a flow, the predicted pressure will drop more rapidly and reduce the flow. The converse also occurs. This may explain why computer code predictions with l fixed discharge coefficients often yield results that compare well to test data, provided that the fixed discharge coefficient is reasonably close to the mean variation of actual values.
Based upon this background, YAEC proposes to use the following method to account for break flow modeling uncertainties within the break spectrum study:
- 1. Define the break size spectrum for each accident. category based upon plant hardware features, system capabilities and single failure assumption.
- 2. Analyze selected break sizes that span this range. Use single and two-phase discharge coefficients of unity, and the Moody two-phase critical flow model. Determine the break size within this spectrum that yields the maximum cladding temperature.
References (I1.1-1) Fernandez, R. T., R. K. Sundaram, J. Ghaus: A. Husain, J. N. Loomis. L. Schor, R. C. Harvey and R. Habert, "RELAPSYA - A Computer Program for LWR System Thermal-Hydraulic Analysis, Volume III: _ Code Assessment,"
YAEC-1300P, Volume III, October 1982 (Proprietary).
(I1.1-2) Abdollahian, D., et al., " Critical Flow Data Review and Analysis," EPRI NP-2192, Electric Power Research Institute, Palo Alto, CA, January 1982 (II.1-3) Tong, L. S., " Boiling Heat Transfer and Two-Phase Flow,"
John Wiley & Sons, Inc., New York, 1965 l
a . _ - - _ _ _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ ___ . , - - _ , - _ _ .
(II.1-4) Ransom,' V. H., R. J. Wagner, J. A. Trapp, K. E. Carlson, D. M. Kiser, H. H. Kuo, H. Chow, R. A. Nelson and S. W. James, "RELAPS/ MOD 1 Code Manual, Volume 1: System Models and Numerical Methods," NUREG/CR-1826. EG&G Idaho, Inc., March 1982, Pages 54 and 55
l Table II.1-1 Critical' Flow Test Nozzle Geometry Nozzle Diameter Aspect Ratio Test (mm) (L/D)
TLTA 6425 and 6426 18.9 9.42 LOFT L3-6 16.2 3.34 Marviken Test 4 509 3.1 Marviken Test 24 ,
500 0.33 Marviken Test 18 300 3.7 Marviken Test 6 300 1.0 ll l i l l l l 0
1 3
5,
(
C )1 E .
S (
e
/
M B
)
3 N A 2 T L ( N L T
O = n L -
i F 7 M 0 n s
s " ,
y 3
2 wo oi N i g l t O G g A F la 2I Y S k u T
C E P ac A el U , L r a N S , E B C U
I R
eE R s. d B 1 ' o i S YA C
nS oP iA S \ ) t L 0 c cE 2 e. 5 e uR 4 t 1 S S
- c. ( o G 3 f t .
/ e. E o 2 T
3 m /s 0*
A M I n/
E bl m h a T o5 s2 I
M i 4 2
3 3 U" A._ r6 a
pt ms
= 2 9 oe
- CT g
e
= 0 i s :
- s. s 5 A
u = _ 2 y s 0 .
$ - B 7 5 R t I t e c S r O
e - u g
S r D i 0
G O F
'{
_ 0 1
0 ,
4
} h-lf f i
0 1
3 3
5, o
=
O.
o A X )
1 T
L T
W (5 n w
i
, 0 3 w t A 2 o s y '
l n n s g . F o
't i kt
- G aa el 5 ru N
U a , B c l
R o ea s nC i
LA o Y C )
c nS oP e iA 6 S tL
- ( cE
- 6 'm
/ .
0 5
1 E
M I
T S
uR f t o
o T
S m n n/
1 E
T k /
s N A Y .
o6 s2 i4 8
5 r6 3 6 1 & P A
L a
pt ms
= i E oe 9 R CT 2
s 0 e
s
=
s =
y 6 2
- /
a o g 0 2 y s u 7 s 5 s~~
a s A _
r T A
e r
G s t
D D u 9 7 C i g
o G - F y
e lI 0
- 1 0
0 O 2
' 2 -
qe* e wh E'i ;
a?
s e
e i -- ..- ..< k '.
. s
. s a
s
. a s
e e n
- s. s no ii
. . t ea tl
. s e au i R c
. t l wa oC 8
l 8 5
. FA M
5
. e Y
. e s
k5
. o 0 t aP eA
) rL
=
c BE e
d R
S
= c .
e( f s oot n
. . t e
- - e e / t E n
' s 4( M s o6
.=
. s / .
I s -
/ i N . T i3
= rL
. - hi m
4 5
. s e
a 4 1 . pT
. 1 0- s. s e mF 38 5 s o . oO 2 l CL tM .
. A Y
. = 5 :
. = P e 5
. A s
. _, s s o
L E
. 1 3 e o R o s s 5 s - -
e c
m a .
s r
. o f e s
u g
. c 4 .
e A 3 L .
i F
/
e T F
O .
s
. L e
. s
)
. .f. Lg j ( - el
- Y '.
- g. n g I rE W.. o.. , es st . a, E" oeeNc#J.
n
,s oJg- wasE*
n e
5o1
TEST 4, D = 509 m. L/D = 3,1 -
3.0
\ L Data HEH '
m
-- - Moody Slip
$ _b
\ I g
Burnell
~
A w 1.0 -
d ". '
~ -
l 2.0
}\ ,
A 6 }\* n
'o j, 'o u L 3 ~ w -~--~- x a
%'% m 3 % E 9
y 5 - -
~ '
5 N g f' '
g m5 ~ ~ _ ,.,_ _ % M t k% .
E 4 j
'. 0.5 - o* w %
4 %
O
- i
- 1,0 j ,
AA AAAAAAAA A AAAA
~
.E
_ l E
4 =
- V308Co* Lib => u 9AroRelach 7 0 l l l 0
~[ l ,
r 0 10 20 30 40 50 - -
Time (S)
{ Figure 5.1 Mass Flow History for Test 4
---~
O
^
1.5 TEST 24. D = 500 era. L/D = 0.33
- A Data Hm "
\ 3.0 .
- -- Moody Slip s
,_\' - - -- Burnell
\
\
2 2.5 io -
[ At
- _ 2.0 T
~
b e
i\ 's x As =
b ' -- - - % . s - -
1.5 a -
3
=
',p~ s 1 .} , =
E s- e E" 0.5 '
h {'=a y,, i E
[*g ,M ,__,Ih' - 1.0
< a sg R. : 0 bb g n, .
I
- .l
- 0.5 1 1
- -Snemtb -*
l" ureurn :
\
i 0 I l j l l l
/o+ s 0
0 10 2d 30 40 50 60 Time (S)
Figure 5.2 Mass Flow History for Test 24
W:
w 0.5
=
TEST 18. D = 300 nun, L/D - 3.7 A,9 g
=
(s a Data g,\ \ HEM
~
1.0
=
t9 1 --- Moody Slip
~ '
4 h -
O.4 -
$ \ ~ ~ ~ Bu m il
_"\
\
%'\4%
( ,
0.3
~
\ s h , E ~ ~ ~- _ , _
, s -
\ .
I g ) 'o
- ' x s
3 g,,,,,<---~__,,--,
0.5
{3 gD M l $
b _
a+
i 0.1 -
7 E F
~ g
+ Suscoat.cb : + Mrungreb -n ,
s' v
E I.0 ' j b
0 ! I I I I ! !
0 0 10 20 30 40 50 60 70 80 90 =
Time (S)
- Figure 5.3 Mass Flow History for Test 18 - =
i r
r 1
0.5 _
TEST 6, D = 300 m, L/D = 1.0 O s A Data _
HEM s - - - Moody Slip 6 k ~
0.4 -
-- Burnell a,
h6 _
AT@-
io b "
'o
- 0.3 - I x N(
{
G E
..._.__- E
~
'l o C ". M a 3
w
. d j % ,, , b 0.5 g 1 6
,a -
-, r 9
0.2 - 3' oA 2 N 9
7
<fM .
4 T
t 4
~
0.1 4
- Suscooteb : $ .n--- SntvM TO --p D
u
= ,
4 h
I I I I I 1 1 0 0 0 10 20 30 40 50 60 70 80 Time (S)
Figure 5.4 Mass Flow History for Test 6 0.11.2. 3 and 4 Page 2-7 of Reference 3 states: "The staff finds that the predicted flow through the Power Operated Relief Valve (PORV) has a large uncertainty when the flow is two-phase in composition. Section 2.1.2 of NUREG-0578 requires that PORV and safety valves be qualified to perform under conditions of both solid water and two-phase flow."
0.II.2 Clarify the uncertainty in the RELAPSYA models for single-phase and two-phase flow through PORV and safety valves.
0.II.3 Clarify the effect of liquid entrainment and the need for detailed modeling below the PORV or safety valves.
O.II.4 Clarify how the flow through valves has been assessed.
A.II.2. 3 and 4 The answers to Q.IX.2, Q.II.3 and Q.IX.4 are provided below for each plant model.
Maine Yankee (NY)
MY has two PORVs and three Safety Valves (SVs) located on the top of the pressurizer. The two PORVs are set to open at 2,385 psia. The three SVs are set to open at 2,485, 2,510 and 2,535 psia, respectively. None of these valves actuate during SBLOCAs due to their high pressure setpoints and the generally decreasing primary system pressure from 2,250 psia. Therefore, they are not modeled for SBLOCAs.
A stuck-open PORV is analyzed by adding a TRIP VALVE (latched open) i to model this component. The valve passes steam because the pressurizer spray is terminated by the decreasing primary system pressure. The valve throat area and a discharge coefficient of one I
are used to maximize the steam flow rate through this path. A more realistic value would lie in the range of 0.6 to 0.9 based upon our l
assessment of the critical flow model for steam.
L-
[ MY has six Secondary Safety Valves (SSVs) located on each of the I three main steam lines that exit from the steam generators. The 5 group on each steam line is set to open as follows: 1 at 1,000 psis, 1 at 1,020 psia, 2 at 1,035 psia and 2 at 1,050 psia.
2 These valves pass steam when actuated during SBLOCAs. Each group of SSVs is modeled using a TMDPJUN component Where the steam mass flow k rate is specified as a function of the upstream pressure. This I incorporates a discharge coefficient to match the Technical Specification ratings for these valves.
Yankee Rowe (YR)
E_
YR has one PORY (2,400 psig setpoint) and two SVs (2,485 and 2,560 psig setpoints) located on top of the pressurizer. None of these valves actuate during SBLOCAs due to their high pressure h
setpoints and ths generally decreasing primary system pressure from 2,000 psig. Therefore, they are not modeled for SBLOCAs.
A stuck open PORV is analyzed in a similar manner to that described for MY above. _.
m YR has three SSVs located on each of the four main steam lines from the steam generators. Each group is modeled using a TMDPJUN component in a similar manner as that described for MY above.
=.
_ VY has one SV each on two of the four main steam lines. The SVs are - -
nominally set at 1,225 i 15 psis. These valves do not operate during any LOCAs (small to large) due to their high pressure setpoints and the generally decreasing primary system pressure from 1,020 psia. . ', g c y k L-t'. . -;
- p ';
VY has one Safety / Relief Valve (S/RV) on each of the four main steam F: ;5[
y lines. The S/RV setpoints are as follows: 1 at 1,080 psig, 2 at . t; t.
['., .
I 1,090 psig and 1 at 1,100 psig. For small break LOCAs, usually one 6. 4'f.~,5
. q.f valve is sufficient to provide prossure relief prior to actuation of O^.. e - -
- avQ
'j- g
[n .:.. ;.
- gb
'. _
- w1ey "
- f. ,
L e
the Automatic Depressurization System (ADS). During this period, wet to dry steam (static quality > 0.8) passes through the valve.
After ADS is initiated, the static quality in the upstream steam line may drop as low as 0.5 and then increase toward unity.
The S/RVs are modeled as mott.,e valves with a short opening and closing time based on test data. The valve throat areas and a
(
discharge coefficient of unity are used for the junctions in order L to maximize the loss of fluid. Comparison of predicted to rated
[ flow rates shows that a more realistic value for the discharge f
r-coefficient is 0.84.
I 0.II.6 Page 3-17 of Reference 17 states: "The flow from assumed small breaks is strongly affected by stratification. The mass discharge b rate is dependent upon whether or not the break is covered by the liquid-vapor interface. Vapor pull-through and liquid entrainment occur and the resulting flow is highly complex. Semi-empirical E models are required for these effects." Clarify how the break flow effects of stratification have been assessed and what the resulting uncertainties are for break flows in cold legs where stratified flow i exists.
A.II.6 As pointed out in the question, stratification effects can lead to highly complex two-phase flows at the break. Reliable models for k calcalating the phenomena of vapor pull-through and liquid entrainment are currently under development. RELAPSYA contains an approximate method to account for the effect of stratification on b break flow. This is discussed in the response to Question VI.7. An
? assessment of this method is provided in the response to Question g V.1 which discusses the RELAP5YA calculation of the LOFT L3-1 E experiment. Further assessment and improved modeling of the phenomena would indeed be desirable for realistic analysis of SBLOCAs. However, for licensing analyses, a spectrum of break sizes I will be investigated. It is believed that the break spectrum analysis would cover the uncertainties in modeling stratification
~
effects on break flow.
E
.f
{
0.II.13 Page 261 of Reference 10 states: " Analyses with discharge coefficients ranging from 0.6 to 1.0 will be performed for large breaks in boiling water reactors. However, small break analyses for PWRs, as well as for BWRs, will be performed with a discharge coefficient of 1.0." Justify why only one value of discharge coefficient needs to be considered for SBLOCAs.
A.II.13 See the answer to Question 11.1. ;
+
q I.. ADDITIONAL OUESTIONS THAT ARE CONCERNED WITH SH ERAL AREAS Q.I.1 Page 19 of Reference 21 states that: "Under some conditions, RELAP5 does a poor job of conserving mass 'and/or energy. These errors can dominate the transient history. The problem seems to lie in the basic formulation of the finite-difference equations and donor cell determination." Large mass errors were also observed in a-calculation reported in Reference 22. Clarify how mass errors are either avoided or affect the results presented in Reference 12 and What will be done to eliminate or avoid them in 85LOCA calculations.
A.I.1 If the mass error becomes large, the nueerical solution becomes questionable. The results shown in Reference I.1-1 and the additional calculations furnished in response to Questions V.1, V.2, V.3 and VI.9, have low mass error (less than about 101.).
Techniques that have been successful in minimizing mass error are (a) reducing time' step size and (b) appropriate renodalization without sacrificing physical modeling. These techniques will be used to keep mass errors to a minimum in plant calculations. It should be noted that mass errors (or energy errors) are a basic feature of most finite-difference solution schemes used for two-phase flow analysis and, hence, they cannot be eliminated-completely without introducing other numerical compromises.
Reference
\
(I.1-1) Fernandez, R. T., et al., "RELAP5YA - A Computer Program for LWR System Thermal-Hydraulic Analysis, Volume III,"
YAEC-1300P, January 1983 Q.I.5 A better appreciation for the verification of SBLOCA models provided by the previous comparisons is needed. Clarify the areas verified by the comparisons; for example, by an itemized list of area and aspects of the area verified for each comparison.
W Table 1.5-1 lists the area and aspects of area verified by the tests presented in Reference 1.5-1. In addition, two new tests were added as part of the RELAPSYA assessment, the semiscale test S-NC-2 and the LOFT Test L3-1.
' TABLE X.5-1 Core Thorinal-Hydraulics Steam Wall ECCA Injection Generator Core Levels Post-CHF Condensation & Interfacial Natural Thermal- Break Void Fraction Heat Reuet 4 T '~ t Heat Transfer Condensation Circulation Hydraulics Flow Distribution CHF Transfer Quench FRIGG Loop I GE Level X
.Swe11' Test Co. 1004-3 Mirviken X Tszt 10
'C31umbia X X CHF Tests Gentral Electric X X CHF Tasts THTF Steady- X X X Stcts Film Boiling Test 3.07.g K and X THTF Quasi- X X X Stscdy-State Boilsff Test 3.C9.10 1
TABLE K.5-1 (cont'd)
Core Thermal-Hydraulics Steam Wall ECCA Injection Generator Core Levels Post-CHF Condensation & Interfacial Natural Thermal- Break Void Fraction Heat Rewet T~st Heat Transfer Condensation Circulation Hydraulics Flow Distribution CHF Transfer Quench THTF Reflood K K K K T= t 3.09.100 end 3.09.10.Q TLTA Bolloff X X X X X Expiriment T:st 6441/6 LOFT Test K X X X X X X L3-6/L8-1 Semiscale K K K T;2t S-NC-2 LOFT Test L3-1 K K K K K TLTA-SBLOCA K K K K T;zt 6432/1
i W The staff's experience with advanced thermal-hydraulic computer programs has shown an important sensitivity to modeling of the steam generators when analyzing small break loss-of-coolant accidents. In specific, the modeling of liquid entrainment, condensation and hydraulic resistances (e.g., flow regime maps) could significantly l depress the mixture level in the core. This phenomenon has been observed in Semiscale Ruperiment S-UT-8. Recognizieg semiscale's ;
l atypicality, the staff nevertheless believes this phenomenon'to be l real and therefore, plausibiu in a full-scale reactor. It is for this reason that we request integral experimental validation of your computer program to predict this phenomenon, should it occur in a full-scale reactor. Validation with the ?-lecale experiment would-be acceptable. Use of other integral experiments for validations requires that these experiments simulate this hydraulic behavior.
W Introduction The staff requested that YAEC validate the RELAP5YA computer program against the S-UT-08 experiment. The request was made because this test uncovered certain phenomena that.were not prevf.ously observed to this degree in other UT non-dHI tests. Specifically, it appeared that total core uncovery occurred prior to clearing of the loop seals. This uncovery was attributed to complex thermal-hydraulic phenomena in the steam generators which depressed the core coolant by developing significant resistance to steam venting. RELAPSYA has ;
been used for a SBLOCA analysis of the Maine Yankee PWR. This calculation showed that RELAP5YA can predict the phenomena observed in S-UT-08. Thus, we believe that a simulation of Test S-UT-08 is not necessary to further demonstrate this code capability. The S-UT-08 test results and the RELAP5YA results for the Maine Yankee SBLOCA case are described below.
Description of Phenomena Observed in Test S-UT-08 The phenomena observed in S-UT-08 are described in detail in Reference I.6.1. The important aspects of the test behavior are briefly described below.
semiscale Test s-UT-08 simulated a SBLOCA in a Westinghouse PWR resulting from a 5% communicative break in one of the cold legs.
The break was located between the broken-loop primary coolant pump and the inlet to the downconer. The experiment included early pump trip, high and low pressure ECC injection and accumulator injection into the cold less of the broken and intact loops.
After the break was initiated, the primary pressure rapidly decreased and approached the secondary side pressure due to SCRAM and pump trip. The loop flows coasted down and natural circulation was quickly attained. As the pressure continued to decrease slowly, the loop void fractions inemased and an asymmetry developed in the liquid level between'the upside and downside of both the U-tube steam generators. Figure I.6-1 shows the collapsed levels in the upside and downsida of the broken-loop steam generator. At about 85 seconds, the upside level became larger than the downside level due to CCFL at the steem generator entrance. This caused steam flow to be blocked and resulted in a depression of the core level seen in Figure I.6-2. The core level continued to decrease until the loop seals cleared at about 210 seconds. This caused the liquid held up in the steam generators to be swept up and over the top of the U-tubes. The ECC water was then able to recover the core level.
Subsequently the core underwent a slow boiloff followed by recovery due to accumulator injection starting at about 500 seconds. There were two core heatup periods. The first heatup, at around 200 seconds, was caused by the core level depression associated with the liquid holdup in the steam generators. The second heatup was caused by the core boiloff. The core thermal behavior can be seen in Figures- X.6-3,1.6-4 and 1.6-5 which show cladding temperature responses of the hot rod at three axial elevations.
Observation of S-UT-08 Type Behavior in Maine Yankee Calculation RELAPSYA was used to assess the Maine Yankee plant behavior for various postulated break sizes as part of the Maine Yankee Pump Trip Study (Reference 1). The worst break size, 0.05 ft , was used to perform a pump trip time sensitivity. Two times were chosen to trip the pumps, two and ten minutes following the HPSI injection. Both these transients exhibit similar behavior as the Test S-UT-08. We will concentrate our discussion on the transient with the RC pumps tripped two minutes following HPSI injection. The major assumptions utilized in this the analysis are presented below:
- 1. Off-site Power Available - The assumptions of continued operation of RCPs during a small break LOCA event requires the availability of off-site power.
- 2. 102% Steady-State Power Operation.
- 3. Only one HPSI and one LPSI Pump Avsilable - Of the two high pressure pumps which are energized automatically on safety injection actuation signal, it was assumed that only one was available. Due to the assumption of off-site power availability, the delay in starting the pump is 0.9 second.
- 4. One auxiliary feedwater pump available.
- 5. ANS decay heat + 20% uncertainty.
- 6. No Steam Dump and Bypass is Available - The steam relief is through the safety relief valves only.
- 7. Metal Water Reaction - The Baker-Just model was used for metal water reaction.
- 8. Appendix K (Reference 7) Lockout Options - These flags force a degraded heat transfer calculation during the blowdown phase even in circumstances where calculated local conditions indicate that rowetting occurs.
- 9. Break Location - The break was located in the cold leg pump discharge pipe.
The Maine Yankee nodalization is given in Figure 1.6-6. The trends ;
predicted by BgLAP5YA for some of the Maine Yankee system parameters are compared with the corresponding trends observed in S-UT-08.
Similar to S-UT-08, after break initiation, the primary pressure :
rapidly decreased to near the secondary pressure and then continued to decrease at a slower rate. The loop flows coasted down, the loop void fractions increased and two-phase natural circulation was quickly established. The steam generator liquid holdup phenomena
- - are remarkably similar to those observed in S-UT-08. Figure I.6-7 shows the collapsed levels in the upside and downside of the broken loop steam generator. It can be seen that at about 400 seconds, the upside level begins to be higher than the downside level, indicating liquid holdup and a consequent blockage of the steam venting path.
Figure I.6-8 shows the same behavior for the intact loop steam
- generator. -Consequently, there was a rapid core level depression beginning at about 400 seconds shown in Figure 1.6-9. The level continued to drop _below the bottom of the core. At about 700 seconds, the loop seals cleared and the core level was recovered.
Figure I.6-10 shows the void fractions in the loop seals. The j clearing of the loop seals is indicated by the void fractions increasing rapidly from near zero to near unity at about 700 seconds.
l Beyond this time, the core underwent a steady bolloff and the core level decreased steadily. At about 2,000 seconds, accumulator ,
injection began and the core level began to recover (Figure 1.6-9).
However, as seen in Figure I.6-10, the loop seal in the broken loop refilled at this time thus eliminating one path for steam venting.
Hence core recovery was slower than that observed in S-UT-08.
Similar to S-UT-08, there were two core heatup periods. Figure 1 I.6-11 shows the cladding temperature response of the hot rod. The first heatup is associated with the core level depression caused by liquid holdup in the steam generators and is seen to occur between 400 and 700 seconds. The second heatup is associated with the slow core boiloff and is seen beyond about 1,000 seconds. The refilling of the loop seal in the broken loop and the ensuing blockage of
=
.-.m... - - , . .,_. _ _ _ _ _ . , _ _ . _ _ _ _ . . . . . . _ _ . . , _ . , . . . _ _ , , . _ _ _ _ _ _ _ _ . _ _ _ _ _ . _ _ _ . _ _ _ _ , . _ _ _ _ _ _ _ _ _ _
steam venting resulted in an intermittent quenching of the core seen in Figure I 6-11 beyond about'2,100 seconds.
Conclusion Test S-UT-08 exhibited certain phenomena which were not encountered to this degree in other tests. Phenomena very siellar to that in Test S-UT-08 were also observed in a BRLAP5YA calculation for a
. Maine Yankee SBLOCA. This calculation demonstrates that RELAP5YA l has the capability of predicting the phenomena observed in S-UT-08. l Therefore, we propose that further validation of RELAP5YA against S-UT-08 test is not necessary.
-Reference (I 6.1) Robert Fujita, " TRAC-PF1/ MOD 1 POST-TEST Analysis of Semiscale Small Break Test S-UT-08," Proceedings of Third
. International Topical Meeting on Reactor Thermal-Hydraulics, Newport, Rhode Island, October 1985.
1200 .
1 1
1000 -
7 3 ano .
d Upside N '"
a 600 -
O s 400 -
Downside _
a a ~
E 200 -
5 a
8 o _I - -
i , t
-200 400 600 800 0 200 Figure X.6.1 Intact Loop Steam-Generator Collapsed Liquid Level (S-UT-08) 400 - -
1 l '
~
300 - -
e . .
a S 200- -
B 3 . .
8
" 100 - -
a - -
8 0 - -
l
-50 ' l - I ' I l 0 200 400 600 600 l 1
TIME (s)
Figure X. 6.2 Core Collapsed Liquid Level (S-UT-08)
I i a 800 - .
Q-H 800 - -
8 at 400 ' I ' I I '
O 200 400 800 800 TNE (s)
Figure X.6-3: Heater Rod Clad Temperature at the 137-cm Elevation (S-UT-08) 1000 i , . , ,
2 800 - -
M ac I ** ,L L
400 ' '
O 200 400 600 800-TIME (s)
Figure X.6.4: Heater Rod Clad Temperature at the 208-cm Elevation (S-UT-08)
!- l l
100C .
i i
. 900 -
~
$ 800 -
M
=
.a: : 700 -
B N
~
SOO -
8 -
a:
600 -
400 ^
400 600 800 0 200 Figure X.6-5: Heater-rod Temperature at the 253-cm Elevation (S-UT-08) l l
l
$ . ,. _ , . - . . r - , - , . , ..%.- ,, - _. - - ,
. . , . _ - , . - ,,.,p-, _ . - . -a e-.- -
m 8
ZL$G SL 56:
Mfs7 l -1987 956 g 9 Passsuaster F 4' '
I [ts I ..
I _ 9 l' q4,5 gg
-~*
950
, 3
.g .:
-b &
4 4 3 e ist i ~;i
.i B6.:s* .i en ~.i
".i i 9st ir ~",
- i i I vEss t t 4
G i kr
- ne 8
-r- z Y 25o I is l 20o INTACT .,l.2:-3% 8RO"EN g l gg l d -l!-1l-4
- 8e l- Loop too P a _ . ca.rrAmW@d a
. sr sso n 15 I sa. . .. , *
-. :-a Mss. 4-tr : l-a:- FQ -a...
-+ Jt I des ,
I n *
" n , -s~ een Acc m
~
y acca n ggs 's,.s --
l u l .-
O'E ,,.h W y Mz y
- u. =
n.g s_' LF1_
-+ in =
N, ,, ->
4
~F_ :
.a .
I T --m.
-s -'"*L .
-6 l 28 l T T l an l T
I 34 I Figure X.6-6: RELAPSYA Nodalization of Maine Yankee for SBLOCA Analysis i
l
e t t t t i
o.
n
-l
- c. .
- 0 N 9 9
-o .
O Q^
- -g PIE 3 g 0
- m
~%
. u
=g1 e
-g ao
- 59 o- e M h
- @' u O o>m m
o
- O
- -NE-p o .t .c
$ 5 d
s ~e 9
i o.
ee N N o o .I - e M M i w M ~'
- o 3
- l. s.
)
~@
o
- R
- ...s is N( .v._--- ............. -
o
_g
"? . o
,,.. -g o
i i a i s O'S; .0*0 0*SZ 0 *0C 0*S1 0*01 0*S 0*0
(.LI) 12A31 cIn5Is casarI1oo 3
t i e t i i 2
-$ n O
~
n O.
m O W
- 0 M 'O O. W N
O ke
" ~
i
~ g e [
(O
~
O b C*
~
~
]
.M O. H 4.
g U
- -8 ~3 a E a O M O.
U c a
- : i o
M u
o i
9 O e,
- u) H '
-g g i s' o l g . O u
~
~
w 4
,... ==*,,,....===***.......- O.
. .g
,f
~, .
..- -g O
6 i i i s i G*SE 0 *0 0*SC 0*05 0*St 0*01 0*S 0*0 (IJ) 12A2'I GIr15I1 casavTIoa l
I -
^
3-3-
R l
l lI-g.
D -
b .
p s
\
3-I I
l
'i s$ d ais e mis asis adeTIE d 15C1 s misTIEs d d ais mis mis seis as Figure X.6-9: Reactor Vessel Collapsed Liquid Level (MY)
I ipr 9 6 ll Intact l
Loop l
l l l
" Broken Loop E
3-I U-i f 0.0
) FOIO.O 14E0.0 TIPE (SCC)
']iN &
3150.0 3e50.0 3500.0 Figure X.6-10: Void Fraction in Pump Suction Leg Loop Seal (MY)
d' o ; !.!. .A. ^'!.,a.,
.... HOTROD 099005 '
- HOTR00 099004 4 l l,
- HOTR00 099001 ,/
i ..<8 f.Y, hb ifli,l 'g I
o k
fA ' ='?
- d. ,**
wa Jg.,,,
x **
t *
- 1. !
l- l' f '.. * '.
h7 4 , = ,
'. [ ,
. . u.v- __-
I .
o.o 700.o i400.0 2:00.o asco.o asco.c T!rE ISCC)
Figure X.6-11: Hot-rod Temperature (MY) 6