Experiences of TRAC-P code at INS/NUPEC

Transcription

1 Exploratory Meeting of Experts on BE Calculations and Uncertainty Analysis in Aix en Provence, May 13-14, 2002 Experiences of TRAC-P code at INS/NUPEC Fumio KASAHARA ( kasahara@nupec or jp) Institute of Nuclear Safety (INS) Nuclear Power Engineering Corporation (NUPEC) 1

2 Contents 1. Preparation of the input data with Fine noding model 2. Large break LOCA analysis by Fine noding model 3. Uncertainty Methods Study by Coarse noding model 2

3 1. Preparation of the input data with Fine noding model (1) Noding We have been prepared the system analysis noding of Japanese 4-loop PWR. It is constituted from 3-D VESSEL and 4 primary coolant loops. And based on the RELAP5 code input data that we use for the demonstration analysis. 3

4 1. Preparation of the input data with Fine noding model Fig.1 4-Loop PWR Fine Noding (52) J49 J47 (81) J50 (50) (80) J26 (82) (53) J51 J53 (55) (51) (11) (21) (54) J48 Loop 1 Loop 2 (2) J52 J3 (12) J2 J1 (10) (13) J5 J28 (14) (15) J6 J4 J30 (85) (70) J31 J27 (17) J29 (16) (1) J7 J12 J25 J8 (20) J33 J11 (25) (24) (23) J35 J10 (71) (86) J32 J36 (27) (26) J34 (22) J9 J55 (83) J58 J62 (84) J59 (58) J57 (57) (56) (31) (59) (41) (61) J61 (60) J56 J15 (32) J16 J14 (33) Loop 3 J13 (30) J17 J38 J18 (34) (35) J40 (87) (72) J37 J41 (37) J39 (36) J24 J19 (89) J63 Loop 4 (Broken Loop) J64 (90) (48) (45) J63 J43 J45 (73) J42 (46) (40) J23 (44) (88) J46 (47) J44 J20 (43) J22 (42) J21 J60 (N):ComponentNo. Jn :Junction No. 4

5 1. Preparation of the input data with Fine noding model (2) Detailed model of invessel structures Include core buffle-barrel region and reactor control rod guide tubes. HTSTRs are modeled to conserve their volume and heat transfer area. Based on the drawings lent from the electric company, so input data are confidential. Fig.2 Reactor Vessel Fine Noding Loop 2 Loop 3 Level17 Level11 Level4 Level1 Loop 1 Loop 4 (Broken Loop) Guide Tube Buffle- Barrel Region 5

6 1. Preparation of the input data with Fine noding model (3) Other reactor coolant system components Pressurizer and accumu-lators are modeled by PIPE components. ECCS low pressure injec-tion flow is modeled by FILL component. (Fig.3) This table is applied to each intact cold legs. m ass flow (kg/s/loop) Fig.3 ECCS Injection Flow Table Input pressure (M Pa) 6

7 1. Preparation of the input data with Fine noding model (4) Containment vessel Fig.4 CV Pressure Table Input Containment vessel free volume is not modeled. Containment vessel pres-sure is modeled by BREAK component. (Fig.4) This table is based on the reactor establishment permit report. Pressure increases linearly to 0.26 MPa by 15 s pressure (M Pa) time (s) 7

8 1. Preparation of the input data with Fine noding model (5) Reactor power Reactor core power is modeled by time table. (Fig.5) It is stepped down to 7% of initial power. pow er ratio Fig.5 Core Power Table Input It includes decay heat of ANS plus 2 sigma time (s) 8

9 2. Large break LOCA analysis by Fine noding model (1) Noding and boundary conditions Fine noding model and boundary conditions are used Figure 1 to 5 show these analysis model Power distribution is defined by Average-power Rod (AV Rod) Hot rod is modeled by Additional Supplemental Rod (AS Rod) 9

10 2. Large break LOCA analysis by Fine noding model (2) Description of Fine noding model Item Number of Components Number of HTSTRs VESSEL cell division (r, t, z) Core sell division (r, t, z) Initial reactor power (MWt) r-direction max power ratio z-direction max power ratio AV Rod max linear power density (kw/m) AS Rod max linear power density (kw/m) Break point Break shape Value *4*17 3*4*8 3, Note ncomp nhtstr ring average of cosine-shape distribution 25 points input based on cosine-shape distribution 3,411/50,952/3.66*1.31*1.45= *1.22=42.3, equivalent to F Q =1.31*1.45*1.22=2.32 loop 4 RCP discharge guillotine 10

11 2. Large break LOCA analysis by Fine noding model (3) Result of transient analysis a. Chronology time (s) event break loss of electrical power to RCPs reactor stop PCT 1 st peak appeared Accumulator injection started Reflood phase started PCT 2 nd peak appeared ECCS injection started PCT 3 rd peak appeared note after 1,000 s Null Transient Assumption, flow coastdown by Semiscale test facility pump characteristics Assumption, reactor power step down to 7% by 0.1 s, followed by decay heat of ANS(1979)+2 sigma 1,120 K set pressure at 4.5 MPa judged by the flow direction at core bottom 1,100 K timer delayed operation 1,074 K 11

12 2. Large break LOCA analysis by Fine noding model (3) Result of transient analysis b. Core pressure Decreases to the CV pressure set as boundary condition about 30s Fig.6 Core Pressure pressure (M Pa) time (s) 12

13 2. Large break LOCA analysis by Fine noding model (3) Result of transient analysis c. Core void fraction Increases to 1.0 from 0.0 instantly after the break initiation, and continues at high value about 0.9 core center void fraction Fig.7 Core Void Fraction time (s) 13

14 2. Large break LOCA analysis by Fine noding model (3) Result of transient analysis d. Maximum hot rod surface temperature Increases to 1,120K from 600K at about 4s. Fig.8 Maximum HotRod Surface Tem perature 1200 Reflood phase begins at about 33s and the second peak of 1,100K appears at 38s. The third peak of 1,074K appears at about 99s. tem perature (K ) time (s) 14

15 3. Uncertainty Methods Study by Coarse noding model (1) Trial of GRS type Ordering Statistics Evaluation parameter is maximum hot rod surface temperature during large break LOCA blowdown phase. Two parameters are selected as cause parameter. (discharge coefficient CD and power peaking factor Q) Based on the Wilk s formula, we can obtain the upper tolerance limit (UTL) at 95% probability on 95% confidence level. 15

16 3. Uncertainty Methods Study by Coarse noding model (2) Noding This noding model is prepared by US NRC for test problem of US 4-loop PWR. There are two hot legs and steam-generators. One of them represents intact loops and is 3-loop-size. 16

17 3. Uncertainty Methods Study by Coarse noding model Fig.9 4-Loop PWR Coarse Noding Loop 2 (3-loop size) (24) (22) (9) (23) (18) (28) (19) (27) (17) (8) (12) (11) (26) (2) (3) (13) (10) (1) (4) (14) (15) (16) (5) (31) (21) (20) (25) (6) Loop 1 (Broken Loop) (7) 17

18 3. Uncertainty Methods Study by Coarse noding model Fig.10 Reactor Vessel Coarse Noding Core region is consist of Three axial noding, Four azimuthal noding, but No radius noding. Loop 2 Loop 1 (Broken Loop) Loop 3 Loop 4 Level7 Level5 Level3 Level1 18

19 3. Uncertainty Methods Study by Coarse noding model (3) Description of Coarse noding model Item Number of Components Number of HTSTRs VESSEL sell division (r, t, z) Core sell division (r, t, z) Initial reactor power (MWt) z-direction max power ratio AV Rod max linear power density (kw/m) AS Rod max linear power density (kw/m) Break point Value *4*7 1*4*3 3, Note ncomp nhtstr 4 points input for distribution 3,250/39,372/3.64 x 1.23= x 1.1=30.7, equivalent to F Q =1.23*1.1=1.35 loop 4 RCP discharge Break shape 150% split 19

20 3. Uncertainty Methods Study by Coarse noding model (4) Cause parameters rank ofq Power peaking factor Q of 124 random samples : normal distribution, myu=1.1, sigma=0.044 Discharge Coefficient CD of 124 random samples : uniform distribution, min=0.8, max= rank ofcd frequency frequency 20

21 3. Uncertainty Methods Study by Coarse noding model (5) Typical maximum rod surface temperature By different Q=1.0, 1.1, 1.2, and fixed CD=1.0 By different CD=0.8, 1.0, 1.2, and fixed Q=1.1 Maximum value of each curve is blowdown peak Q=1.2 Q=1.1 Q= CD=1.2 CD=1.0 tem perature (K ) 700 tem perature (K ) 700 CD= time (s) time (s) 21

22 3. Uncertainty Methods Study by Coarse noding model (6) Discussion of blowdown peaks Blowdown peaks based on random Q and CD Both figures are arranged from the same results of 124 cases. 900 a. Arrangement by Q This distribution shows widely spread and small positive correlation. tem perature (K ) peaking factor Q 22

23 3. Uncertainty Methods Study by Coarse noding model b. Arrangement by CD This distribution shows obvious positive correlation focusing to about 830K with about 50K spread. tem perature (K ) discharge coefficient CD 23

24 3. Uncertainty Methods Study by Coarse noding model c upper tolerance limit based on Wilk s formula These values are equivalent as Upper Tolerance Limit of 95% probability on 95% confidential level based on the Wilk s formula. Number of samples %*95% UTL 1 st max of 59 samples 2 nd max of 93 samples 3 rd max of 124 samples value 863 K 856 K 855 K 24

25 3. Uncertainty Methods Study by Coarse noding model (7) Conclusion and future plan We have made prototype of GRS-type uncertainty evaluation system. We will select several ten pieces of cause parameter, And modify the TRAC-P code to handle those parameters by tracin, And apply this uncertainty evaluation system to evaluate three peaks of large break LOCA PCT using fine noding model. 25