ML19260D401

Forwards Bench Mark Analysis for Sequential Auxiliary Feedwater Flow to Steam Generators Following Loss of Main Feedwater,In Response to NRC 790821 Ltr.Unit Design Does Not Utilize Auxiliary Feedwater Sys Causing Sequential Feeding

ML19260D401
Person / Time
Site:	Davis Besse
Issue date:	02/05/1980
From:	Crouse R TOLEDO EDISON CO.
To:	Reid R Office of Nuclear Reactor Regulation
References
NUDOCS 8002080700
Download: ML19260D401 (17)

Text

.

TOLEDO

%mm EDISON Docket No. 50-346 I).1$["

w e-License No. NPF-3 4m aar Serial No. 584 February 5, 1980 Director of Nuclear Reactor Regulation Att ention : Mr. Robert N. Reid, Chief Operating Reactors Branch No. 4 Division of Operating Reactors United States Nuclear Regulatory Commission Washington, D. C. 20555

Dear Mr. Reid:

In partial response to your staff's August 21, 1979 letter, question 1A, we are transmitting five copies of a bench mark analysis for sequential auxiliary feedwater flow to the steam generators following a loss of main feedwater.

In the case of Davis-Besse Nuclear Power Station, Unit 1 (DB-1), this response is provided as a generic code verification exercise only. The design of DB-1 does not utilize an auxiliary feedwater (AW) system that will result in undesirable sequential feeding of the steam generators. One AFW pump serves each steam generator independently of the other as illustrated in the DB-1 Final Safety Analysis Report (Figures 10.6 and 10.7).

Therefore, when you review the modeling and analytical technique attached, note that although it is generically applicable to 177 Fuel Assembly Babcock and Wilcox plants, this event scenario is not appropriate for the Davis-Besse design.

Very truly yours, f f ^r_: -

RPC:TKR: cts Attachment -

hwt 1942 063 THE TOLEDO ECISCN CCMPANY EDISCN PLAZA 300 MADISON AVENUE TCLEDO. CH!O 43652 8

M2 080 M

. Docket No. 50-346

. License No. NPF-3 Serial No. 584 February 5, 1980 Response to Question lA Provide a benchmark analysis of sequential auxiliary feedwater flow to the steam generators following a loss of main feedwater. This analysis was provided in a letter from J. Taylor (B&W) to R. Mattson (NRC) dated June 15,1979. However, in this analysis the TRAP-2 code with a 6 node steam generator model was utilized. All small break analyses presented to the NRC have been performed using the CRAFT-2 code with a 3 node steam generator model. We require a benchmark analysis for sequential auxiliary feedwater flow also be performed using CRAFT-2 with a 3 node steam generator representation.

by The Babcock & Wilcox Company Nuclear Power Generation Division o

CONTENTS I IflTR000CTION II SITE EVErlT DESCRIPTION -

III METHODS ,

IV RESULTS V CONCLUSIDMS

, FIGURE 1 CPAFT-2 N0DIt!G DIAGRAM FOR SMALL BREfKS FIGURE 2 STARTUP FEEDWATER FLOW- -

FIGURE 3 STEAMGENERdf0RLIQUIDLEVEL(TEMPERATUREADJUS FIGURE 4 STEAM GENERATOR SECONDARY SIDE PRESSURE FIGURE S PRIMARY A LOOP TEMPERATURE .

FIGURE 6 PRIMARY B LOOP TEMPERATURE j FIGURE 7 PRESSURIZER LEVEL FIGURE 8 REACTOR VESSEL PRESSURE 1942 065 1-4-80

I. INTRODUCTION This report presents an analysis of sequential auxiliary feedwater (AW) flow to the once through steam generators for a loss of main feedwater transie it. The CRAFT 2 codel and the s=all break model described in reference 2 have been used in the study. The calculated results have been compared to a loss of offsite power startup test data obtained from the Florida Power Corporation's Crystal River 3 Unit in which an imbalance in the auxiliary feedwater flows between the two operating loops resulted in an imbalance in the primary loop response. This transient tests several features of the computer simulation, including conditions of asymmetric loop temperatures, an almost dry generator to feed aniliary feed-water into, loss of RC pumps, and establishment of natural circulation. In many cases the absolute validity of the boundary conditions and test data were ques-tionable, and estimates had to be used. However, this analysis does show that the

, data trends can be predicted by a 3 node CRAFT 2 SG representation.

II. SITE EVENT DESCRtPTION The Crystal River 3 Unit is a 2452 MWt, 177-FA B&W reactor with a lowered-loop configuration. On April 23, 1977, a loss of offsite power test was performed.

This test was initiated from approximately 15% full power operation. The secon-dary liquid levels were approx 1=ately 2 feet and was sufficient to remove the power and provide essentially steady-state operation prior to test initiation.

The test was initiated by tripping the reactor, the reactor coolant pump, and feedwater pump power sources. The core power then dropped to the decay heat -

level and, as the pri=ary coolant pumps coasted down, the primary flow decayed to natural circulation level. One diesel generator was started to provide power for the pressurizer heaters, one makeup pump, and other necessary services of secon-dary importance to this analysis.

The main feedwater flow coasted down resulting in both steam generators eventually drying out until the auxiliary feedvater flow became sufficient to start filling the A loop steam generator secondary at about two minutes into the transient.

The B loop steam generator remained dryed out until twelve to fourteen minutes into the transient when the A loop reached normal operating level and the feed-water flow was diverted to the B loop. The imbalance in the feedwater flows, and hence levels, resulted in a corresponding imbalance in the primary system re-sponse including the decay heat removal, the hot and cold leg temperatures and

~

. 1942 Cs6 e

flows between the two loops. The transient results were used to evaluate the ability of the 3 node CRAFT 2 steam generator model used in small break evalua-tions to calculate the effect of the feedwater transient.

III. METHODS A. CRAFT Input Model The input model developed for this calculation was based on the small break model used for licensing.2 The schematic of the flow path nodalization is shown in Figure 1. The initial system conditions were defined based on the available mea-sured data which were required to represent this test. The model was set up to provide a steady-state calculation until two seconds into the transient when the reactor, reactor coolant pumps, and ain feedwater pumps were tripped initiating the transient calculation.

B. Initial Conditions The initial mass flow'kas assumed to be identical to the full power operation value. The measured hot and cold leg temperatures were then used to determine a consistent core power to provide the initial steady-state operating conditions.

This resulted in an initial power of 19% of full power operation versus the 15%

power defined in the summary test report. Hand calculations, using the 15% core power and the measured hot and cold temperatures, resulted in a mass flow con-siderably below that required to balance the pump power. The actual mass flow is believed to have been only 1 or 2% less than full power flow. The pressure dis-tribution around the system was revised, because of the new hot and cold leg temperatures, to maintain the loss coefficients defined by the referenced model.

The liquid levels in the pressurizer and steam generator secondary were changed to reficct the measured data.

C. Boundary Conditions The makeup pump flow was modeled by defining the pressure flow characteristic curve for normal operation with the recirculation line open. The makeup pump was actuated when the pressurizer level dropped to 30",below the initial liquid level value. The makeup pump flow was equally distributed between the two cold leg pump discharge modes as shown in Figure 1. The feedwater flows were defined by the test data and are given in Figure 2. An auxiliary feedwater enthalpy of 58 btu /lba, which is the nominal enthalpy of the system,was used.

1942 067 I

The safety relief valves were set to 1030 psia to model the effect of the turbine bypass valves, which are fully open at 1030 psia. The safety relief flow is the only ellowance made in the model for steam flow.

The heat transfer to the secondary was assumed to be to the mixture in the lower partic.t of the steam generator and the fraction which may have been deposited in the steam region was assumed to be negligible. A preliminary short-term transient evaluation demonstrated the need to define the heat transfer multiplier based on the steam generator secondary levels. Consequently, the final model contained a heat transfer multiplier as a function of time based on the measured secondary levels.

IV. RESULTS This section presents a comparison of the CRAFT 2 analysis to the data taken for the first 20 mintites of the CR-3 lo9s of offsite power test. As will be shown, some of the data utilized in the evaluation is questionable and greatly influence the transient response.' However, even with the uncertainties in the measured

~

data, the CRAFT 2 code is shown to adequately calculate the RCS behavior.

A. Secondary Response Figure 3 shows the secondary side SG 1evels during the test. The test data shows that, following the loss of main feedwater, the initial level in both steam gen-erators decreases. At approximately 1 minute into the transient, the auxiliary feedwater system initiates, as shown in Figure 2, and preferentially feeds the A loop steam generator. Thus, the liquid level in SG A increases. At 12 minutes, the liquid level in SG A stabilizes because it has reached its control point. At that time, the feedwater flow is diverted to SG B and its level increases.

The CRAFT 2 code calculated results shows reasonable agreement with the SG A level during the first 12 minutes. After this time, however, the CRAFT 2 calculation continues to increase the SG level while the data shows a level stabilization after this time. This difference is probably due to an overestimation of the auxiliary feedwater flow to SC A after this time. The auxiliary feedwater flow, as indicated in Figure 2, is very stable and at a relatively high flowrate af ter 12 minutes. Examining other data, such as the A loop hot and cold leg tempera-tures, does not support a high auxiliary feedwater flowrate. In light of the ability of the CRAFT 2 code to reasonably predict the SG response up to 12 min-utes and the inferences obtained from other data, the flowrate given in Figure 2 after 12 minutes is believed to be in error.

}hk2 bb

TheSGBliqjidlevelresponseisgenerallyoverpredictedbytheCRAFTcalcula-tion. This again is believed to be caused by an overestimation of the auxiliary feedwater flowrate to SC B, especially between 3 and 9 minutes. Figure 2 shows the auxiliary feedwater flow to be very low over this time period and very stable.

This may be due to an initial instrumentation offset and no feedwater may have been delivered to the steam generator in this period. Once a sustained auxiliary feedwater flow is established to the SG, the CRAFT calculated level increases are in reasonable agreement with the data.

Figure 4 shows the SG secondary side pressure response duridg the transient.

CRAFT 2 predicts the pressure response for the A loop SG reasonably. Between 4 and 6 minute;, the calculated SG pressure increases above the data. Over this time period, it is believed that the measured auxiliary feedwater flows are low.

This conclusion is consistent with the level comparison shown in Figure 3. For .

the remainder of the transient, the prediction is higher than the measured SG

, pressure.

The secondary side pressure for SG B was generally underestimated throughout the transient. This is caused by con'ensation d of the steam within the SG due to the excess auxiliary feedwater flow utilized in the calculation.

B. Primary System Response Figure 5 shows the A loop te=perature response during the test. The hot leg tem-perature compares well with the transient data until 13 minutes. After this time, the CRAFT 2 calculation ~ continues to show a decrease in the i.ot leg temperature .

due to the continued feeding of the A loop SG. The data shows a flattening of the hot leg temperature due to the control of the SG level. This supports the belief that the auxiliary feedwater flows after 12 minutes is lower than the values indicated by Figure 2.

The calculated A loop cold leg temperature response is consistent with the data

. trend, but generally overpredicts the data after 4 minutes. This is caused by the overprediction of the SG A secondary pressure discussed previously.

The B loop temperature response is shown in Figure 6. Due to the overprediction in the B loop SG level and underprediction in the SG pressure, the hot leg tem-peraturas are underpredicted.

1942 069

Figures 7 and 8 show the pressurizer level and system pressure comparison. Hand calculations which were performed indicate that these parameters are not consis-tent. Examining these figures, it is seen that the calculated pressurizer level response is in good agreement out to approximately 12 minutes. After 12 minutes, the continued overcooling of the A loop, due to the overestimation of feedwater flow, results in an underestimation of the pressurizer level.

The pressure response shown in Figure 8 shows that the CRAFT 2 calculation under-predicts the data. However, as mentioned previously, this is not unexpected as the system pressure and pressurizer level are not consistent.

V. CONCLUSION A sequential auxiliary feedwater flow transient has been benchmarked in this analysis using tha CRAFT 2 code with the 3 node SG model used in small break evaluations. The site data trends were reasonably reproduced by the code. In me "r cases the validity of test boundary conditions were questionable and esti-mat:s of the test data were used. However, the results provide assurance that the CRAFT 2 code is capable of reasonably predicting the primary system behavior irl'.cated by the test if the boundary conditions were well defined. Thus, this it.udy has demonstrated that, in spite of the simp,licity of the CRAFT 2 steam generator model, the CRAFT 2 code can estimate, with reasonable accuracy, a tran-sient highly dependent on the steam generator. Thus, the ability of the small break model to calculate the effect of steam generator heat removal during a small break transient is reasonably assured.

e_

l942 070

. REFERENCES 1

.R.A. Hedrick, J.J. Cudlin, and R.C. Foltz, " CRAFT 2 Tortran Program for Digital Simulation of a Multinode Reactor Plant During Loss-of-Coolant,"

BAW-10092, Rev. 2, Babcock & Wilcox, April 1975.

2 Letter J.H. Taylor (B&W) to S.A. Varga (NRC), July 18, 1978.

e 4

0 9

k e

e WgI 5

9 e

9 9

O

~

1942 071

Fl4W11 Ctaf'2 sWihG gl8Geas 704 IsaLL gatsas G G 2

. . C, f he ,,

o 2, e

, S z

- is is F

4 o re '

e ,,

h 3 N0fE5 438tflGn&L Cafa 23f ! sage Cs flattas atlas il 20CE

{ 15 C0nfalhatuf eCSE t t_0_

Pats it Ltas Pats it3s catal TO Canttimetti LPi G ,

2 Fifu 15 REfUen List P4tt fBCu CGhfAinstut fg gains t@t

, P4tu stretsinf5 C64faltstti $Psat $75785 c *)

wrt =0 int eritiCa t ics Pafn na intefif sCarlos t 8094CCEt4 1.2

• CCRI 2 LCett PLinct 3.4.10.18 a0f Ltt PI'inC 3 C38t 5.!S.40.42 #0f LtC. UPPER 4 to m0f LIG Pirias s.28 55 futts 5 13 $$ 4 UPPER ntag 7.22 SG LOtti utA8 6.18 17tas singtaf at fetts e CCPt STPst!

7.17 5tC naast 15 f.13 j e COLS Lts PIPleG 8,10 15 LCett stAO 19.14,23 PUt?$

9.11.19 COL 8 LIG PIPimC 11.12.11.18,28.!? COLS f tG PIPING 10.12.28 COLO LES PIPih4 17.31 fotocastl

,3 UPPts 4:suCCatt 23 LP, 21 Pat!!Utilti 28.29 BPPER KXCatt

!! Ccufstartnf 38 Pet!!UtlZit 23 UPett Pttabt 22 Vtaf fatst 24.29 $$ UPPER h(AS 23.34 LEAR 8 RifAE PAIN 39.34 inPI 37 Ccarsinstaf setafs eI.43 18 ilPPIS set 18

• t i4

.}

s o$s- .

1942 072 e

4

FIGURE 2. STARTUP FEEDWATER FLOW  !

i A LOOP TEST DATA

_ _ 8 LOOP TEST DATA

/

^.

l 5 .. ,,-

/

/

d /

. /

>/

4 . . r * *** ' '

l . ,

^

I w l 2 I x I 3 ..

• I

/ g i / I 2 .. j 5 g

i f -

t .

It /

l1 A / -

l --

I \ l\

I \l \

/ \ /

<f,!

I

,) --- /

L- , .

~ .Y 0 ,

0 4 8 12 16 20 Time, Min 1942 073 f.

a

FIGURE 3 STEAM GENERATOR LIQUID LEVEL (TEMPERATURE ADJUSTED) .

A LOOP TEST DATA

, A LOOP CALC. DATA

---

B LOOP TEST D ATA

. . . B LOOP CALC. DATA

^.

25

/

/ .

20 - -

/, -) .

l

/ .

/

m~

E 15 .

j' =

/

/ /

/ c

=

i .

e /

/ r o .

/

/

i

[

/

10 - -

,/ j, u ,

t / /

. m .

/ /

5 "

E / - /

/ .

. i i.

/ ./ /. 5 g /

/

1942 074 4,e m /

~~~~p 0  :

0 4 8 12 16 20 Time after loss of power, Win

~~

FIGURE 4 S.G. SECONDARY SIDE PRESSURE

.' A LOOP TEST DATA A LOOP CALC. DATA

.---- 00P TNT DATA 1100 - - ,

- B LOOP CALC. DATA 1000 -

p\ s ,'s r

N g

\ o -

. g no g 3

900 - -

.\s s Ns /

\

\

.7 h

e

\\

h. \~N \

A s

800 - -

\ s

\

/ -

s'N'N,N\

\

/ \

\

. \ ~

700 - -

c,,,, \. \

L s ~ .\a.~ \

600 - - \

\

. . \

\

\

\

500  : .  :  :

0 4 8 12 16 20 Time, min 1942 075

. I

.;.

s FIGURE 5 PRIMARY A LOOP TEMPERATURE ,

= I -

LOOP A HOT TEST DATA l LOOP A HOT CALC, DATA

---

LOOP A COL 3 TEST DATA

- -- LOOP A COLD CALC. DATA l

\u m 580 - N

\ N (s.

N x - -

N 570 -

N ..,

t\ .N g rg Ng

= 560 - -

1\

5 i\ N

& 1 -

\

=

\ r

\ \.

550 - -

i\ \ '

\\ '\

\\ '\. \

540 - -

\

\ -

s

\

s

\

\ -

s

\ '

\ \ . \

. \ '

N. x 530 -- N

\ N. .~ ~

\ \. .

s \

N -

y '

520 . . .

0 4 8 12 16 20 g Time, min 1942 076

FIGURE 6. PRIMARY B LOOP TEMPERATURE s .

. LOOP B HOT TEST DATA LOOP B HOT CALC. DATA

---

LOOP 0 COLD TEST DATA

-.-. LOOP B COLD CALC. DATA N

' d 580 .

kN N -

m,kn .

h .

N;N .

570 - . \ 'N,- \ *

\ N \ -

.V N.N N ds s 'N \

560 - .

e 3

2 sx\x\ \. \,

N g -

.g 3 \-

e- c ;o . .

\. \ ,

\\/

540 - -

\ >

{

a

\ \'

k 530 . .

\

520 . .

0 4 8 12 16 20 Time, Hin ]942 077 i 1-4 Pf1

j - __

,,,-h-FIGURE 7.

• j PRESSU21ZER LEVEL

' e e

TEST DATA e

~ ~ ~ ~ - ~ ~ ~ -

e

, CALCULATED DATA 9

. a

, e 1

• = ,

e 250 .- .

e P

e N

• O .

.c u

c 200 i e

\

. \

o b, U

a s

\

t 150 . .

N N

$ g%

3 W

W C.  %

6

\\

100 -

s ,

N N

\

\

~

\

\

M ,

\

\

0 si

• E2 I6 gn Time, Min e

e 1942 078

, FIGURE 8. REACTOR VESSEL PRESSURE ,

TEST DATA

. ----- __ CALCULATED DATA

=O I

2200 . . 4 i t e

' \

-I i

\ .

t ' '

2l00 -I -

1 wN N

N -

m \s

l 2000 . N g ,

N O N'

=  %

M NN 8 g

' \

o 1900 -- -

= N .

N N

\

\

\

1800 .

\g N

N O 4 8 12 16 20 Time, Min e g O

e 1942 079

.o

. 1-4-80