ML20210A605

Addendum a to Vol II of Nusco 140-2, Nusco Thermal Hydraulic Model Qualification

ML20210A605
Person / Time
Site:	Millstone, Haddam Neck, 05000000
Issue date:	09/05/1986
From:	NORTHEAST UTILITIES
To:
Shared Package
ML20210A583	List: B12208, Forwards Addendum a to Vol II of Nusco 140-2, Nusco Thermal Hydraulic Model Qualification ML20210A605
References
NUSCO-140-2-ADD, NUSCO-140-2-ADD-A, NUDOCS 8609170287
Download: ML20210A605 (49)

Text

,

t %

NUSCO I40-2, ADDENLUM A NUSCO TIERMAL HYDRAULIC MODEL QUALIFICATION - VOLUME II (VIPRE)

NORTIEAST UTILITIES SERVICE COMPANY AUGUST 1986 8609170287 860905 PDR ADOCK 05000213 .

A-1 P PDR

ec .

t \

TABLE OF CONTENTS l

fECTION TOPIC PAGE I. Introduction A-1 II. Comparison of 1/4-Core and A-2 1/8-Core VIPRE Models III. FSDA Benchmark - Control Rod A-5 Ejection Incident IV. Conclusion A-9 V. References A-10 Attachment 1 - VIPRE-01 Modeling Sensitivity Studies, October 1985.

(This report was. submitted to the NRC as part of Reference 3.)

A-li

..__a

r- 1 I. INTRODUCTION Topical Report NUSCO 140-2 (NUSCO Thermal Hydraulic Model Qualifica-tion Volume II - VIPRE) was transmitted to the NRC in Reference 1.

The purpose of NUSCO 140-2.is to demonstrate the adequacy of NUSCO plant modeling techniques and to respond to the concerns expressed in NRC Generic. Letter 83-11. It describes the development of a 1/4-core, fuel cycle specific VIPRE model for the Haddam Neck Plant (HNP).

Based on the experience gained in preparing NUSCO 140-2, a new VIPRE model was developed which is suitable for future licensing analysis.

This 1/8-core, 19 channel, fuel cycle independent model was used in NUSCO Report No. 151 (Haddam Neck Plant - Non-LOCA Transient Analysis) which was transmitted in Reference 2.

In this addendum to NU3C0 140-2, the 1/8-core VIPRE model is described and compared to the 1/4-core model. Also, the benchmark to the FDSA Rod Ejection Incident (originally described in NUSCO 140-2) is

~

further evaluated.

A-1

r i II. COMPARIbON~~0F.1/4-COREAND1/8-COREVIPREMODELS The 1/4-core VIPRE model developed in MUSCO 140-2 was fuel cycle specific (i.e. the hot channel location was selected based on physics parameters calculated for Cycle 12). This model was used to perform numerous sensitivity studies on VIPRE modeling and input parameters.

Additionally, several benchmark calculations were performed including comparisons to COBRA-IIIC results, HNP FDSA results, and plant data.

A noding diagram for the 1/4-core model is shown in Attachment 1 (Figures 9A and 9B).

The development of the 1/4-core VIPRE model demonstrated that the VIPRE code is adequate for reactor core thermal hydraulic analysis and that NUSCO's modeling techniques are reasonable anld conservative.

Based on the experience gained in developing the 1/4-core model, a 1/8-core, fuel cycle independent VIPRE model was developed. A limiting thermal-hydraulic environment is created for the hot channel by means of channel noding and a conservative power distri-bution. Therefore, the hot channel location of this " generic" model is not a function of burnup, control rod insertion, or fuel cycle design. The magnitude of the power (i.e. hot channel enthalpy rise and axial power shape) i? still varied based on the nature of the

~

transient and control rod position.

A-2

. t Attachment I contains a report describing the development of the 1/8-core, 19fchannel model which is used for licensing calculations.

This report is part of a transmittal from the Utility Group for Regulatory Applications (UGRA) to the NRC. The transmittal (Reference 3) addressed issues relating to the NRC's generic review of VIPRE.

In the report, transient sensitivity studies'(using forcing functions for a complete loss of flow transient) are described for channel noding and several other modeling options and input parameters. As noted, the 1/8-core, 19 channel model produces a slightly more conservative MDNBR than the 1/4-core model. These two models use similar levels of noding detail, in that at least one row of subchannels surrounds the hot'subchannel and the remainder of the hot assembly is modeled with some detail. Additionally, fuel rod powers are applied consistantly in the two models. However, since the hot subchannel is close to the center of the core in the 1/8-core model,

. flow redistribution is minimized which results in a lower MDNBR.

Also, it is sltbwn that the models with coarser noding (i..e. 8 channel hot assembly model and 8 channel, 1/8-core model) produce less conservative results. This sensitivity to channel noding detail was also seen during the development of the 1/4-core model.

i A-3 l

L

. s In Attachment 1, the results of other sensitivity studies are shown for the 1/8-core, 19 channel model. These studies agree with the sensitivity studies performed for the.1/4-core model (in NUSCO Report 140-2) and show that these' modeling options and input parameters are reasonable and conservative.

., t.

4 4

i i

t l A-4 l

%++p -

w .., - - , - ---- . --m-eww - - -*- -*r- er vr g e --T

III. FDSA BENCHMARK - CONTROL R0D EJECTION INCIDENT In the VIPRE -Topical Report (Reference 1) VIPRE results were bench-marked with FDSA results for the Control Rod Ejectio'n Incident (Reference 4). 'First, a RETRAN analysis of the rod ejection incident j was performed. The predicted core average power and heat flux compared well with the FDSA results. The VIPRE analysis then used the power forcing function from the RETRAN results. Other analysis inputs were assumed or obtained from the FDSA where possible.

Figure 1 (from Reference 1) compares fuel and cladding temperature results for the hot rod from-the FDSA and VIPRE analysis. As noted, the initial conditions of fuel centerline, fuel average, and cladding temperature (T CL' AVG, and CT ' re8Pectively) compare well. Also, during the transient VIPRE prediction of T and T compares CL AVG reasonably well with the FDSA results.

However, the VIPRE results from Reference 1 show no cladding heatup while the FDSA predicts a cladding temperature increase of 1000 F.

(from 600 F to 1600'F). Power generation in the hot pellet is predicted reasonably well (i.e. power forcing function matches the FDSA, hot pellet peaking factor is identical to that reported in the FDSA). .Therefore, this difference in cladding temperature response must be due to differences in the prediction of heat transfer into the cladding (i.e. gap conductance, K ,p) and/or out of the cladding (i.e. convective heat transfer coefficient, h).

l l

A-5 l

l

.O 9 The assumption made for h is not stated in the FDSA, however, several possibilities exist. .These include:

1. The boiling curve is modeled and post-CHF heat transfer

.(film boiling, small h) ~ occurs after about 0.2 second of.

transient time.

2. Specific values of h are input as a function of transient.

time allowing the post-CHF, clad heatup condition ~to occur.

The FDSA states that a-K ,p of 2000 BTU /hr ft*F was used thronghout the transient. However, this value may just have been an initial condition. During the transient, K may have been increased in 8ap order to account for thermal expansion of the fuel pellet. This would allow increased heat transfer to the cladding.

In the Reference 1 VIPRE analysis, Eg ,p = 2000 is assumed for the entire transient. The boiling curve is modeled and a switch to a post-CHF heat transfer regime is not predicted to occur. The switch to post-CHF ueat transfer would occur is MDNBR decreased below 1.0.

As a result, saturated nucleate boiling (large h) occurs throughout the transient and cladding heatup is not predicted.

A-6

In order to more closely match the FDSA results for T ,.tw additional C

cases are performed. These cases ~are summarized below:

Case # gap h-1 increased from 2000 to boiling curve modeled 10000-at time = 0.1 sec. with switch to post-CHF heat transfer- for 8

MDNBR 5 1.0 2 Same as Case 1 boiling curve modeled with switch to post-CHF heat transfer for 8

MDNBR'$ 1.3 Note: a. MDNBR is calculated with the W-3S correlation as detailed in Reference 1.

The results of these cases are shown in Figure 2. As noted, Case 2 compares-fairly well with the FDSA results. This case predicts a maximum clad temperature of_about 1300*F. In this case, K ,p is increased from 2000 to 10000 early in the transient. Also, post-CHF heat transfer is assumed to occur when the MDNBR decreases to 1.3.

A-7

Case 1 is identical to Case 2 except that post-CHF heat transfer is modeled to occur when the MDNBR-decreases to 1.0. As a result,

~

post-CHF heat transfer does not occur until later in the transient and the predicted maximum clad temperature (about 1050*F) is less

-than that in Case 2.

A-8

IV. CONCLUSION The 1/8-core, 19 channel model was shown in the attached report to be slightly more conservative than the 1/4-core model for the analysis of a complete loss of flow transient. Based on other inhouse calculations, this conclusion was also seen to be valid for other transient and steady state calculations. In licensing calcula-tions using the 1/8-core model (such as those reported in Reference 2),

additional conservatism is included in the MDNBR analysis by minimizing the following ratios:

4

1. Peak pin power / Hot assembly average power

2. Hot assembly average power / average power of the assembly neighboring the hot assembly If these ratios are small, then the power gradient in the hot channel region is relatively flat. This results in less cooling at the hot channel due to cross flow and thus, a lower MDNBR.

An additional sensitivity study was provided for the control rod ejection incident which was evaluated in Reference 1. K was gap increased during the transient to account for the fuel pellet I

thermal expansion. Also, post-CHF heat transfer (i.e. film boiling) was modeled to occur when the MDNBR decreased to 1.3. With these assumptions a better comparison with the FDSA results was achieved.

A-9

V. REFERENCES

1. NUSCO 104-2, Thermal' Hydraulic Model Qualification Volume II (VIPRE), dated July 1984.

2. J. F. Opeka letter to C. I. Grimes, dated June 30, 1986,

Subject:

Reanalysis of Non-LOCA Design Basis Accidents and included NUSCO Report No. 151, "Haddam Neck Plant - Non-LOCA Transient Analysis."

3. J. A. Blaisdell (Chairman - UGRA Executive Committee, Northeast Utilities) letter to H. R. Denton (NRC), NE-85-SAB-285, dated October 10, 1985.

l 4. Connecticut Yankee Atomic Power Company, Haddam Neck Plant, Facility Description and Safety Analysis, Docket No. 50-213.

i i

I 3 A-10

- , , ~ - -

n,, - . . - - ,- ., ----...---,---,.,,.---,-.-c- , _ - - , , - . . , , . . . - , ..w,., , _ - . , - . . , -----r. r- --- .. - , - - . - - e r--m-,,,-

l FIGURE 1 (from Ref. 1)

CONTROL ROD EJECTION INCIDENT TEMPERATURE VS. TIME 5000 4

FUEL CENTER i

i C O

.I I

y 3000 -

FUEL AVERAGE w

i I 7 u -

__ _ -- ~

5 r

2000 - '

CLAD FDSA lt M 2

---

VIPRE (Hgp = 2000 BTU /HR-FT ,.7) 0 l , ,

0 1.0 2.0 3.0

) TIME (SECONOS)

I

FIGURE 2 CONTROL R0D EJECTION INCIDENT TEMPERATURE VS. TIME 4500 -

^

, , g Cases 162

^

4000 -

Fuel Centerline 3000-

, _ . - ~~

f -

- x  :: e _

_- m w _n 5 Fuel Average a

m 2000..

c.

E H

/

1000 .

2 e X 'X X X X X  ?

Clad 1

0 - . . . . .

0.0 1.0 2.0 TIME (sec) key: FDSA Case 2 P X X Ref.1- VIPRE O O O Case 1

• E ATTACHMENT 1

. -- - ~ - - . . , . . .- . -

t a

1 I

i COMPLETE LOSS OF FLOW TRANSIENT

{ FOR THE HADDAM NECK PLANT VIPRE-01 MODELING SENSITIVITY STUDIES I

1 i

I NORTHEAST UTILITIES SERVICE COMPANY October 1, 1985 i

f i

?

i f

f

o '.

TABLE OF CONTENTS Page List of Figures -ii  !

List of Tables iii

1. Introduction 1

2. Description of the Haddam Neck Plant 1

3. VIPRE Base Model for the Haddam Neck Plant 2

4. VIPRE Transient Sensitivity Studies 2

a. Introduction 2

b. Radial Noding 3

c. Time Step Size and Solution Method 4

d. Grid Loss Coefficients 4

e. Turbulent Mixing Coefficient 5

f. Crossflow Resistance 6

5. Comparison of Transient and Steady State Results 6

6. References 7 t

(

i

LIST OF FIGURES

1. HNP Reactor Core

2. HNP Fuel Assembly

3. 19 Channel - 1/8 Core _ Base Model

4. a-c Forcing Functions for Loss of Flow Transient

5. 8 Channel - Hot Assembly Model

6. 8 Channel - 1/8 Core Model

7. 14 Channel - 1/8 Core Model 8, 28 Channel - 1/8 Core Model-

9. a,b 34 Channel - 1/4 Core Model

10. MDNBR vs. Time, Radial Noding Sensitivity Study

11. MDNBR vs. Time, Time Step Size Sensitivity Study

12. MDNBR vs. Time, Solution Method Sensitivity Study

13. MDNBR vs. Time, Grid Loss Coefficient

14. HDNBR vs. Time, Turbulent Mixing Coefficient Sensitivity Study

15. MDNBR vs. Time, Crossflow Resistance

16. MDNBR vs. Axial Height - Transient and Steady State Results i

i l

i l

l t

11 l

t

LIST OF TABLES

1. HNP Thermal-Hydraulic Data

2. Summary of Input for VIPRE Base Model

3. Radial Noding Sensitivity Study, Summary of Cases

4. Time Step Size and Solution Method Sensitivity Study, Summary of Cases

5. Grid Loss Coefficient Sensitivity Study, Summary of Cases

6. Turbulent Mixing Coefficient Sensitivity Study, Summary of Cases

7. Crossflow Resistance Sensitivity Study, Summary of Cases iii

=

• _1

~1. INTRODUCTION ,

l:' Proper geometric modeling and input option selection are important~

in order to develop a VIPRE model which produces reasonable and conservative results. This report describes several sensitivity L studies which are performed under loss of flow transient conditions.

. These sensitivity studies investigate the effect on VIPRE's predicted

• MDNBR due to changes in the following items:

1. Flow channel modeling; i.e., radial'noding.

l

2. Solution method.

L 1- 3. Time step size.

4. Grid loss coefficients.

1 5. Turbulent mixing coefficient.

6. Crossflow resistance.

Radial noding detail is required in the hot channel region so that-the thermal-hydraulic conditions, and therefore the MDNBR, will be accurately predicted in the hot channel. Proper representation of grid loss coefficients, turbulent mixing, and gap-to gap crossflow ,

i' resistance are also important factors which effect the channel-to-channel ,

crossflows and thus effect the hot channel conditions. Selection of j the solution method and time step size may effect the stability and

. accuracy of the VIPRE solution.

l j An accurate prediction of the thermal-hydraulic conditions in the .

j hot channel is needed so that the MDNBR will be accurately calculated. '

,i The MDNBR will be calculated using the W-3 CMF correlation.

) Additionally, a quasi-steady state analysis is performed. By this,

] steady state VIPRE calculations are made using the operating conditions i from several points in time during the loss of flow transient. This series of results from steady state VIPRE calculations is compared i

! to the results of the transient VIPRE calculation. Agreement in

} these results shows that the W-3 correlation, developed from a data j base of steady state results, produces acceptable results when j applied to a loss of flow transient calculation.

j! 2. DESCRIPTION OF THE HADDAM NECK PLANT l

The Haddam Neck Plant (HNP) is a 4-loop Westinghouse PWR rated at i 1825 NWt. It began commercial operation in January 1968.

i The HNP reactor core consists of 157 fuel assemblies. The fuel

assemblies are made up of a 15 x 15 array with 204 fuel rods, i j 20 control rod guide thimbles and one instrument tube. The active

! core height is 120.5 in.

i  ;

i j

i

. *- Each fuel assembly has seven grids which are approximately 20 inches apart. The first (at the assembly inlet), second and seventh (at the assembly exit) grids are vaneless spacer grids. The four remaining grids are mixing vane grids.

Figures 1 and 2 show the HNP core layout and fuel assembly cross section, respectively. Thermal hydraulic design information and operating conditions for HNP are outlined in Table 1.

3. VIPRE BASE MODEL FOR THE HADDAM NECK PLANT A 1/8 core, 19 channel VIPRE model is used by Northeast Utilities Service Company (NUSCO) to perform licensing grade core thermal-hydraulic analyses for HNP. This model was developed based on guidance provided in NUSCO's Topical Report describing VIPRE model qualification (Ref. 1). The Topical Report detailed the channel noding and input selection for a 1/4 core, fuel cycle specific model.

The 1/8 core, 19 channel model produces more limiting MDNBR results than the Reference 1 model. Additionally, by means of channel noding and rod power input, a limiting, cycle-independent hot channel location is defined.

Noding of the 1/8 core, 19 channel model is shown in Figure 3. 'As seen, 15 subchannels are modeled in the hot assembly and 16 fuel rods provide heat to these subchannels. Lumped channels and rods are used to model the rest of the 1/8 core. Channels 5 and 6 are the hot subchannel and hot thimble channel, respectively.

Limiting hot channel conditions are produced by having a nearly flat radial power profile in the hot channel region. Also, the hot assembly and its adjacent assembly (Channel 17) are assumed to have equal powers. The effect of this power profile in the hot channel region is to minimize flow redistribution, thereby producing conservative hot channel results.

Some of the input parameters for the HNP base model are summarized in Table 2. An input listing for this model is shown in Appendix A.

.4. VIPRE TRANSIENT SENSITIVITY STUDIES

a. Introduction In this section VIPRE transient calculations are performed assuming a complete loss of flow (LOF) accident.

Sensitivity studies are performed under LOF transient conditions for the following items:

1. Radial noding.

2. Solution method.

3-

3. Time step size.

4. Grid loss coefficients.

5. Turbulent mixing coefficient.

6. Crossflow resistance.

The LOF forcing function input used in VIPRE was calculated for HNP using the RETRAN computer code (Ref. 2). This input is shown for core power, pressurizer pressure, and core inlet flow in Figures 4a, 4b, and 4c, respectively.

For all sensitivity studies, th 19 channel 1/8 core base model for HNP (see Section 3) is used as the starting point. The radial noding rtudy involves major changes to the base model, as noted in Section 4b. The other sensitivity studies involve minor changes to VIPRE input parameters or options. These studies are described in Sections 4c-4f.

b. Radial Noding Six cases are performed for the radial noding sensitivity study. As noted in Table 3, transient calculations are performed with a hot assembly model, 8,14,19, and 28 channel 1/8 core models and a 1/4 core model. These models are shown in Figures 5-7, 3, 8, and 9, respectively.

In each of the six models, the hot channel is modeled individually.

However, the level of noding detail in the remainder of the hot assembly and in the remainder of the core section varies significantly. Rod powers are applied consistently in each model. Other VIPRE options and input parameters are unchanged.

Figure 10 shows MDNBR vs. time for each case during the LOF transient. As noted, the 14, 19, and 28 channel 1/8 core models produce identical results with a MDNBR of 1.507 occurring at 6.0 seconds. The 8 channel-1/8 core and 34 channel-1/4 core models produce slightly less conservative results with HDNBR's of 1.527 and 1.545, respectively. The 8 channel hot assembly model is the least conservative with a MDNBR of 1.592.

These results show that the most conservative results are produced by individually modeling the subchannels which are adjacent to the hot channel. Noding detail in the remainder of the hot assembly and core section has little effect on the KDNBR result.

The 1/8 core models produce more conservative results than the 1/4 core model. This is because the hot assembly location in the 1/8 core models results in more limiting thermal-hydraulic conditions in the hot channel region (specifically due to reduced crossflow and increased hot channel enthalpy rise).

These results show that the 1/8 core nodings produce limiting,

~I

.j fuel cycle independent models. However, fuel cycle specific modeling, typical of the 1/4 core model, may also be used to accurately represent the core.

c. Time Step Size and Solution' Method VIPRE LOF transient results are compared assuming variations in the time step size (At) and solution method. As shown'in Table 4,; cases are performed using the RECIRC solution with time steps varying from 0.05 second to 2 second. Also, cases are performsd with At=0.1 second for each of the solution method options (i.e., RECIRC, Iterative and Direct).

Results of the At sensitivity study are.shown in Figure 11. As noted, each case predicts a MDNBR of about 1.507 occurring at 6.0 seconds. The agreement in those results is deceiving, though.

Since MDNBR occurs at 6 seconds, which is an even multiple of all the time step sizes, each case does well in predicting this point. However, if MDNBR occurred at'5.5 seconds, the results using At=1 second or 2 seconds could vary from the results using At=0.1 second or 0.2 seconds.

The VIPRE RECIRC solution is stable for.the LOF transient for each At studied. However, to insure an accurate prediction of the MDNBR, At=0.1 to 0.2 seconds is recommended. Ultimately, the selection of at is determined by the type of transient (i.e., rate of- change of the forcing function parameters) and the solution method.

Figure 12 shows MDNBR results for each solution method with At=0.1 second. These results are essentially identical. For this transient, the RECIRC solution uses considerably less computer time than the Iterative and Direct solutions. However, as with time step solution, the choice of solution method may vary with the type of transient. The Direct or Iterative solutions may be faster than the RECIRC solution for other transients.

d. Grid Loss Coefficients Five cases are performed for the grid loss coefficient sensitivity study. As noted in Table 5, these cases involve either doubling or halving the grid loss coefficients used in the base model.

The doubled or halved coefficients are applied to all channels or to only the hot channel. The grid loss coefficients used in the base model are listed in Table 2.

The results of this study are shown in Figure 13. As seen, the results of the base model, Case GL1 (coefficients halved in all channels)'and Case GL2 (coefficients doubled in all channels) are identical. Changing the grid loss coefficients uniformly over the entire model effects the overall channel pressure

A #

r-ms drop; however, the hot channel conditions are not significantly

( ,) effected. Each of these case predicts a MDNBR of 1.507 at 6.0 seconds.

In cases GL3 and GL4 the grid loss coefficients are halved and doubled in only the hot channel (Channel 5 - see Figure 3). As seen'in Figure 13, a local change in the loss coefficients has a CignifiCant effect on the hot channel conditions and the MDNBR. By~ halving the loss coefficients in the hot channel, ,

crossflow into the hot channel is increased and MDNBR is increased to 1.557. Doubling the loss coefficients increases the crossflow out of the hot channel, reducing the MDNBR to '

1.440.

In summary, transient HDNBR results are not significantly effected if grid loss coefficients are uniformly halved or

, doubled. If those coefficients are halved or doubled on a local basis, changes in MDNBR of +3.3 percent and -4.5 percent, respectively, are seen.

e. Turbulent Mixing Coefficient i

A turbulent mixing coefficient (p) of 0.019 is used for all gap connections in the base model. The 19 channel base model (see Figure 3) has 28 gaps, 20 of which connect a subchannel to another subchannel and 8 of which connect a subchannel or S lumped channel to a lumped channel. Since turbulent mixing is

(~'l

\~- a subchannel phenomenon, the way in which it is applied in a VIPRE model may significantly effect the MDNBR result. This sensitivity study is performed to determine the effect on VIPRE's calculated MDNBR due to p and the gaps to which turbulent mixing is applied.

Table 6 lists the 8 cases that are performed in this study. In ,

three cases (Base Model, Cases TM1 and TM2), 's of 0.019, 0.0095, and 0.038, respectively, are applied to all gap connections.

In three other cases (Cases TM3 - TMS), the same values for S are used, but turbulent nixing is assumed to occur only in the 20 gaps which connect a subchannel to another subchannel. The remaining two cases assume turbulent mixing only in the 4 gap -

connections from Channel 5 (Case TM6) and no turbulent mixing (Case TM7).

The results of this study are shown in Figure 14. Three paire of cases are each seen to produce almost identical results. ,

These pairs are - Base Model and Case TM3 (MDNBR = 1.507), -

Cases TH1 and TM4 (MDNBR = 1.487) and Cases TM2 and TMS (MDNBR = 1.527). These results show that applying turbulent mixing only to subchannel-to-subchannel gap connections does not produce more limiting results than applying turbulent mixing to all channels. Also, by comparing the Base Model and Cases TM1 and TM2 it is seen that halving or doubling p results

( in a -1.3 percent and +1.3 percent change, respectively, in MDNBR.

t

6-

<-s3 To further show the effect of turbulent mixing, Cases TM6 and

( ) TM7 unrealistically assume turbulent mixing in 4 gaps and no

~'

gaps, respectively. As expected, these cases show decreases in MDNBR over the base case result. Cases TM6 and TM7 predict MDNBRs of 1.471 and 1.447, respectively.

f. Crossflow Resistance As shown in Table 7, the VIPRE base model assumes that the crossflow resistance (Kg) for a gap is proportional to a constant multiplied by the number of rod rows between the centroids of the adjacent channels. In this sensitivity study the constant used in the calculation of Kg is varied from 0.05 to 5.0.

The results of f.his study are shown in Figure 15. As noted, the changes in the constant used to calculate Kg have little effect on the MDNBR. Case CR1, the base model, and Case CR2 predict MDNBRs of 1.509, 1.507, and 1.500, respectively.

5. COMPARISON OF TRANSIENT AND STEADY STATE RESULTS The W-3 CHF correlation was developed based on data taken at steady state conditions. Therefore, transient and steady state VIPRE results are compared in order to show that VIPRE and the W-3 correlation are acceptable for performing DNBR calculations under

(}

V LOF transient conditions.

The transient and steady state calculations both use the 19 channel - 1/8 core base model described in Section 3. The transient calculation uses the LOF forcing function input from Figures 4a-c. The steady state calculations use operating parameters at specific points in time during the LOF transient. As seen in the previous sections, MDNBR occurs at 6 seconds (or 5 seconds from the start of the LOF transient).

Figure 16 compares steady state and transient MDNBR vs. hot channel axial height at time = 6 seconds. As noted, these results are almost identical. Comparison of these results at other times during the transient gave similar agreement.

6. REFERENCES

1. NUSCO Thermal-Hydraulic Model Qualification, Volume II (VIPRE),.

NUSCO Report No. 140-2, Northeast Utilities Service Company, August 1984.

2. RETRAN A Program for Transient Thermal-Hydraulic Analysis of Complex Fluid Flow Systems, EPRI NP-1850-CCM-A, October 1984.

l I

I FIGURE 1 HNP REACTOR. CORE 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 R

t P

x l

N- l X X X X M X X X X X L X X

K X X X X 'X J

x y .

H X X X X X X X i G X X l i F X X X X X

E y y -

D x y x x C

X X X X B

y A

\ ,

x - FUEL ASSEMBLY WITH CONTROL R0D t

p r

.-,c - - ,,, ,, ,, ~,, --..s-,-p ,yr~-,~w

-m ,~~~-------~w-w--*r*'~

FIGURE 2 FNP FUEL ASSEIELY 000000000000000 .

000000000000000 000000000000@O0 0000000@O000000 000000000000000 00@O00000000000 00000.0000000000 000@O0000000000

~

000000000000000 000000000000000 OOOOOOOOOO@OOOO  :

000000000000000 OOOOOOOOOOOOOOO 000000000000000  :

000000000000000 o-~s ua um o - g g g y g g gr g et

F FIGURE 3 19 GWEL /1/8 CORE VIPRE t0 DEL 1

1 2 2 3 3 4 _

6 7 8 7 8 9 9_ 10 11 12 11 12 13 14 3-*

J_3 14 15 16 200000 "

)O O"O O O O x NI X - CHANNEL #

f

_8 \ X - ROD #

18

.l 9_

N 19 22 1

FIGURE 4A NOWALIZED CORE POWR FORCING FUNCTION FOR CCfPLETE LOSS OF FIDi TRANSIENT 1.0-0.95 0.8f I

0.7f N

0.6f I

Z E .

0 0.5-C 0

R E .

0. 4 --

0 W

E R .

0.3-0.2f 0.lf 0 o', , .

0 5 10 15 20 25 T IME (SEC)

. _ . . . __J

l l

FIGURE 4B PRESSURIZER PRESSURE FORCING FUNCTION FOR i COMPLETE LOSS OF FLOW TRANSIENT i

l 2100-2050-P R

E 2000-5 S

U R

I Z

E R

P R 1950-E S j S

U R

E P

S 1900-I R

1 1850-1800 , , , , , ,

0 5 10 15 20 25 TIME (SECl

S'b j( '. '4 :

i f

? FIGURE 4C f'

~ NORMALIZED CORE FLOW FORCING FUNCTION

'., FOR COMPLETE LOSS OF FLOW TRANSIENT 3 -li .(

s' '

y _ _

i i

.t 1.0- ,

l;

/

0.9f

. /

e O.Si N. 0.11

^

-g - ,e.  ;

R - ,

M R

L I 0.62 '

Z ,

E

'O  !

C i 0 0.5' '

l

.R ,

i E ,

i I '

'N '

L 0.42 '

E t

T >

> , F t L

'l"ll rj .

C 0.3 N

r J* '

y 0.2i

. 6 J

• /- e 0.1-0.02

' ' r , , ,

0 .'.1 .5 10 15 20 25 .

' TIME (SEC1

'I

---wew-- _ - . -<-,e..e,_ ,_.

0 $ e FIGURE 5 8 OWfEL / HOT ASSENLY f0 DEL

)1 3 5 6 7 8

hObbb . .

X - CHANNEL #

l.

)

f I

s - - - - -

4 9 FIGURE 6

.8 CHAtt4EL /1/8 CORE IM

-1 2 3 6

20000

) o o o'O O O

\

\ '

\

N X - CHANNEL #

FIGUE 7 14 DiANNEL / 1/8 CORE FDDEL 1

2 3 4 5 6

7 8 9 0 20000

) O O"O O O 2000000 5\

\

x

\

14 X - CHANNEL #

FIGURE 8 28 OMEL /1/8 CORE RJ0EL 2 3 4 5 6

7 8 9 0 11 12 13 14 15 3.

16 17 18 19 20 'l 22 23 20 0 _

000

\2.

x

\

2, x

28 l

X - CEW4NEL #

l I --

FIGURE 9A HOT ASSBELY NODALIZATI0ti FOR 34 CHANNEL /1/4 CORE l'DDEL O OO OO OO OOOO O" d O O'b OO OOOO Oe OOOO O 10 lll0pOs0 1 O Q100 O '*'

OO OO 0000

. OO O@OO Q,0p 1 0p 1 00 0000 0000000 0000000 OO@dOOO 0000000 OCOO@DO<OO@1pCOO 0000000 0000000 00@O0@O 0.000@O0 000d000(0000000 0000000(?O000000 0- n.e 1 aa uoy

9. g7=jg 31 gm;get

FIGURE 9B 34 CHANNEL /1/4 CORE IDDEL

\ 8- 7 6 5 4 3 2 1 _

H I'

G 24 25 26 32 See F Figure 31 23 27

1. 9a l E l 30 29 28 D

l N

g 34 -33 l

B l

A I

L_ _

FIGURE 10 MDNBR VS. TIME 1

RADIAL NODING SENSITIVITY STUDY 2.8-2.5-

{

)

2.2-4 .

i n- ts i g

w /

3 . /

1.9- g /

'?-

y-CN1 /k, N, N /'

s ,-CN2 /

f 1.6-

Q~' N/ -

le Base Model, N 7,)\ '

l i

CN3, CN4 .I 1.3 , , , , , ,

0 2 4 6 8 10 TIME (SEC1 LEGEND: CURVE BASE MODEL - CASE CN1 ------- C ASE CN2

- - CASE CN3 CASE CN4 -

CASE CNS LOF TRANSICNT OCCURS AT 1 SCC.

1

_ _ _ _ _ . )

FIGURE 11 MDNBR VS. TIME TIME STEP SIZE SENSITIVITY STUDY 2.8-

/

/ ,

/ l 2.5-

/

/

/ l

/

2.2-l '

/

a .

~N l 1.9-

\

/'s

! //

l/

1.6- /

/ k 2 3-O' 2 14 6 8 10 TIME (SEC)

LEGEND: CURVE EL -------

LOF TRANSIENT OCCURS AT 1 SEC.

e .

FIGURE .12 MDNBR VS. TIME SOLUTION METHOD SENSITIVITY STUDY 2.8-i 1

2.5-2.2-M D

N B

R H

3 1.9-1.6-1.3 , , , , , ,

0 2 tl 6 8 10 TIME ISEC)

LEGEND: CURVE CASE TS2 CASE TS6 ------- C A S E T S 7 LOF TRANSIENT OCCURS AT 1 SEC.

l L

l 1

1 FIGURE 13 MDNBR VS. TIME GRID LOSS COEFFICIENT SENSITIVITY STUDY 2.8-I

/

jl l 2.5-  :

l ,'

/ /

/ ' ,/

/

2.2-  : / .

/ /

l /

g .

f j

/ ,/ f i.e.

s.

N. <

/ /

/

'\

'vGL1, GL2 Base Model, '

// -

jl N .

./ /

\ r

\ /

\ ' x' / /

N y\ s,,\ '

/ /

1.6-

,'s..J' /!

N's' N

% /

/

I'3~. . , , , ,

0 2 4 6 8 10 TIME (SEC)

LEGEND: CURVE BASE MODEL - CASE GL1 ------- C AS E GL2

- CASE GL3 CASE GL4 LOT TRANSIENT OCCURS AT 1 SEC.

r FIGURE 14 MDNBR VS. TIME TURBULENT MIXING COEFFICIENT SENSITIVITY STUDY 2.8-2.5-  !

/!

////

2.2- //lll//

,l l

M o l' B i

?

R /

---

s\ s , /,,

3 1.9- - -~s' N, N N\ '

'//

///

s\N \4-[TM2,TM5 s g /p

's'N.Ns \' ' .

//

t Base Model,TM3 //

N's-s //

'Q s s;-TM1, TM4 j,/

1.6-

'k'N '

-Q '\ '

TM6 TM7 1~./'

1,3-0 2 4 6 8 10 TIME (SEC) c LEGENO: CURVE EL --- -------

_,_ g

--- CASE TM6 ---- C ASE TM 7 LOF TRANStENT OCCURS AT 1 SEC.

a

a ' e FIGURE 15 MDNBR VS. TIME-CROSSFLOW RESISTANCE SENSITIVITY STUDY 2.8-l 2.5-2.2-M D

N B

R W

3 1.9-1.6-

~~...'

• ~

0 2 Li 6 8 10 TIME (SEC)

LEGEN0: CURVE BASE MODEL -

CRSE CRI ------- C ASE CR2 LOF TRANSIENT OCCURS AT 1 SEC.

?.

I:

f l

FIGURE 16 MDNBR VS.. AXIAL HEIGHT TIME =5 SEC.(FROM START OF LOF TRANSIENT)~

FOR CHANNEL 5 (SEE FIGURE 3) 10- q 8-

\

6-0 2

R 4-l 2-l

~

0 20 40 60 80 100 120 AXIAL HEIGHT (IN.)

LEGEND: CURVE STEADY STATE ------- TR ANS I EN T 4

. . . . . , , _ . . av,\n ,,

o TABLE 1.

HADDAM NECK PLANT, THERMAL-HYDRAULIC DATA Core Power (nominal) 1825 MWth Core Power (maximum steady state) 1861.5 MWth System Pressure (nominal) 2015 psia System Pressure (minimum steady state) 1960 psia Core Flow Rate (minimum) 88.10 Mlbm/hr Core Inlet Temperature (maximum) 544.4*F Number of Assemblies 157 Rod Array 15 x 15 Fuel Rods per Assembly 204 Control Rod Guide Timbles per Assembly 20 Instrument Tubes per Assembly 1 Active Core Height 120.5 in Fuel Assembly Pitch 8.5066 in Fuel Rod Pitch 0.5652 in Fuel Rod Diameter 0.424 in Control Rod Guide Thimble Diameter 0.5456 in Instrument Tube Diameter 0.424 in Number of Grids 7 (3 vaneless spacer grids and 4 mixing vane grida)

Grid Spacing ~20 inches t

L '

• TABLE 2

SUMMARY

OF INPUT FOR 19 CHANNEL VIPRE BASE MODEL Section of Symmetry 1/8 Core Number of Channels 19 (15 subchannels and 4 lumped channels)

Number of Rods 20-(16 single rods and 4 lumped rods)

Axial Noding 5.06 inch nodes with 2.53 inch nodes in MDNBR region Crossflow Resistance coefficient K = 0.5 L/P*

Axial Friction Factor f = 0.32 Re .25 Turbulent Crossflow W' = pS5 Where: mixing coefficient = 0.019 turbulent momentum factor = 0.8 Local Loss Coefficients:

Lower Core Plate and Bottom Nozzle 2.31 Vaneless Spacer Grid 0.99 ea Mixing Vane Grid 1.30 ea Top Nozzle and Upper Core Plate 1.11 Solution Method RECIRC Convergence Limits and Iterations Allowed VIPRE Default Valves CHF Correlation W-3 with L grid factor L grid Mixing Factor 0.042 Grid Spacing Factor 0.06 Axial Power Shape 1.367 top peak Maximum Radial Power Factor (Peak pin) . 1.772 Fraction of Power Generated in Coolant 2.6%.

Other: Local pressure option is used.

Conduction model used for all rods L = Channel-to-Channel Centroid Distance P = Fuel Rod Pitch S = Gap Width 5 = Average Mass Flux for Adjacent Channels i

r-4 at a TABLE 3 RADIAL N0 DING SENSITIVITY STUDY

SUMMARY

OF CASES

1. Channels in

1. Subchannels 6 Lumped Channel Remainder of Total Case # in Hot Assembly in Hot Assembly Model- Channel CN1 - 6 2 NA 8

'CN2 4 3 1 8 CN3 10 1 3- 14 BASE MODEL 15 1 3 19 CN4 21 2 5 28 CN5 9 13 12 34 TABLE 4 TIME STEP / SOLUTION METHOD SENSITIVITY STUDY

SUMMARY

OF CASES Solution Time Step Case # Method Size TS1 RECIRC 0.05 sec.

TS2 RECIRC 0.1 Base Model RECIRC 0.2 TS3 RECIRC 0.5 TS4 RECIRC 1.0 TSS RECIRC 2.0 TS6 ITERATIVE 0.1 TS7 DIRECT 0.1 Note: Time step size is constant for entire transient.

VIPRE default values used for all convergence limits.

m __ _,

{'es 6 TABLE 5 GRID LOSS COEFFICIENTS SENSITIVITY STUDY

SUMMARY

OF CASES Case # Grid Loss Coefficients GLI Loss coefficients halved

• in all channels GL2 Loss coefficients doubled

• in all channels GL3 Loss coefficients halved

• in hot channel only GL4 Loss coefficients doubled

• in hot channel only Base Case See Table 2

• From base case value TABLE 6 TURBULENT HIXING COEFFICIENT SENSITIVITY STUDY Turbulent Gaps Where Case # Mixing Coefficient (S) S Applied Base Model 0.019 1-28 TM1 0.0095 1-28

TM2 0.038 1-28 TM3 0.019 l-17, 19, 21, 23 0.0 18, 20, 22, 24-28 TMs 0.0095 1-17, 19, 21, 23 0.0 18, 20, 22, 24-28 TM5 0.038 1-17, 19, 21, 23

,, 0.0 18, 20, 22, 24-28 TM6 0.019 4 , 15 , 7 , 8 0.0 1-3, 6, 9-28 TM7 0.0 1-28 TABLE 7 CROSSFLOW RESISTANCE SENSITIVITY STUDY

SUMMARY

OF CASES Crossflow Gaps Where Case # Resistance Coefficient (Kg) Kg Applied CR1 0.05 f/p* 1-28 Base Model 0.5 g/p 1-28

CR2 5.0 t/p 1-28 i *f = Channel-to-channel centroid distance p = fuel rod pitch i

A/p2 # rod rows between the centroids of adjacent channels.

1