ML20010B743

Evaluation of Natural Circulation Tests in Fftf, Informal Rept

ML20010B743
Person / Time
Site:	Farley
Issue date:	03/31/1981
From:	Bari R, Chen L, Perkins K BROOKHAVEN NATIONAL LABORATORY
To:
Shared Package
ML20010B738	List: ML20010B743
References
BNL-NUREG-29348, NUDOCS 8108170494
Download: ML20010B743 (61)

Text

. . _ - _ . - _ _ _ - _ _. _ _ _ _ _

, qd8f. a,

! L.NUREG -29348 /

lui

/

ff b.

0flJLJp,fh N

i ORMAL REPORT '

j t.IMITED DIsTRInuTIcn f-4 6

'lUGl 31981 h l k' "%yw p(2

s O ,

S P l 2 ui  :

I, AN EVALUATION OF NATURAL CIRCULATION TESTS IN FFTF i

1

c

^

i e

i l K. R. PERKINS, R. A. BARI, AND L. C. CHEN i

i i

l 1

l l

DATE PUBLISHED - MARCH 1981 l

3 4

i

{ DEPARTMENT OF NUCLEAR ENERGY BROOKHAVEN NATIONAL LABORATORY

] UPTON, NEW YORK 11973 i

i i

bi l Tj 3., Fj Prepared for the U.S. Nuclear Regulatory Commission l 71 O ,,; 3N (;f f OfEco of Nudear Reactor Regulation j 3,;j jj b;; y [ Contract No. DE-ACO2-76CH00016

. /r ?? m p i

ilu 4n (!n.!?!a pI!w 4

l 8108170494 810810 PDR ADOCK 05000348 j P PDR

3, d

.s C

i I

l 9

)

NOTIC 2 This reimrt was prepared as an aranint of work siensored by the United Stairs Gmernment. Neither the United States nor the United States Nuclear Regulatory Commiwinn, nor any of their emph >ces, nor any of their i ontrai tors, sula nntractors, or their ernploien. makes any warranty, express or implied, or acumes any Ireal liabiht) er res[=msibility for the as rura(y,< ornplcirness or usefulness of any informa-tion, apparatus, produc t or prot ew diwbred, or reprevnts that its use would not infrince prn ately ow ned rights l

l I

BNL-NUREG- 29348 INFORMAL REPORT LIMITED DISTRIBUTION AN EVALUATION OF NATURAL CIRCULATION TESTS IN FFTF L

K. R. Perkins, R. A. Bari, and L. C. Chen Department of Nuclear Energy BROOKHAVEN NATIONAL LABORATORY Upton, New York 11973 March 1981 -

Prepared for U.S. Nuclear Reg..latory Commission Washington, D. C. 20555 Under Interagency Agreement DE-AC02-76CH00016 I

l b

ABSTRACT Preliminary results of the FFTF natural circulation test program have been analyzed and predictions for the remaining tests have bee made using DEM0-F, IANUS, and FLODISC. The post-test analysis of the secondary tests with IANUS indicate that with some adjustments the code is capable of repre-senting the data. However, the comparisons indicate the need for some model-ing refinements (more detail in the DHX, changes in the pump coastdown model, and more detail in the cold-leg modeling) in order to provide accurate pre-dictions of secondary system transients. The secondary loop natural circula-tion performance (flow versus temperature gradient) compares favorably with noninal modeling, but the margin between the nominal and worst case modeis is found to challenge the accuracy of the experiment. The present neminal predictions for natural circulation performance in the primary loop are in good agreement with the HEDL predictions, but flow redistribution predictions appear to have important differences which are attributable to the different core representations.

O

-111-

TABLE OF CONTENTS l

Page ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii LIST OF FIGURES ..................... ...... v LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii s

1.0 INTRODUCTION

........................... 1 2.0 PREVIOUS WORK . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 l 3.0 TEST DESCRIPTION ......................... 3 3.1 SECONDAaY TESTS ....................... 6 3.:.1 Steady-State Tests .................. 6 3.1.2 Transient Tests ................... 6 3.1.3 Comparison to Analyses ................ 9 3.2 PLANNED PRIMARY TESTS .................... 21 3.2.1 Nuclear Steady-State Tests .............. 21 3.2.2 Nuclear Transient Tests . . . . . . . . . . . . . . . . 27 3.2.3 Uncertainty Analyses ................. 47 4.0

SUMMARY

AND CONCLUSION S . . . . . . . . . . . . . . . . . . . . . . 50 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 O

-iv-

LIST OF FIGURES Ficure Title Pace 1 Steady-state natural circulation flow in the secondary loop, j ralculated by IANUS and DEMO-F, Compared to HEDL Data (18) and Pretest Predictions (3), 7 2 Comparison of HEDL Heat Loss (19) Data (18) to Previous LMEC Data and the Acceptance Limit Suggested (3) by HEDL. 8 3 The nominal IAMUS calculation of the DHX outlet temperature conpared to the data (6) and HEDL pretest predictions (3) for the LOP from refueling conditions. 10 i

4 The nominal IANUS calculation of sodita flow through the DHX compared to the data (6) for the LOP from refueling conditions. 14 5 The nominal IANUS calculation of the secondary cold leg tem-perature at the TC location compared to the data (6) for the LOP from refueling conditions. 15 6 Schematic of the FFTF DHX operation during an LOP from refuel-ing conditions compared to 7 node modeling in IANUS. 17 7 HEDL data for the LOP from refueling conditions compared to IANUS calculations of the secondary cold leg temperature at the TC 1ocation using the adjusted (6-pass) model of the DHX. 19 8 HEDL data for the LOP from refueling conditions compared to IANUS calculations of the secondary flow rate using the ad-justed (6-pass) model of the DHX. 20 9 IANUS calculation of the secondary flow rate during pump coast-down, compared to the data and the HEDL calculations. 22 l

l. 10 IANUS and DEMO-F predictions for the primary flow versus loop aT compared to the HEDL prediction. 24 11 FLODISC prediction of F0TA AT's during steady-state testing l compared to a calculation assuming no flow redistribution. 26

-v-

1 LIST OF FIGURES (Cont.)

Figure Title Page '

12 FLODISC nominal and worst case prediction of row 2 F0TA AT versus loop AT for steady-state testing compared to hypotheti- -

cal (unmodeled) transition behavior. 28

~

13 Nominal IANUS prediction of core temperatures for the 5% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 5% power. 31 14 Nominal IANUS prediction of core temperatures for the 35% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 35% power. 32 15 Nominal IANUS prediction of core temperature for the 75% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 75% power. 33 16 Nominal and worst case IANUS prediction of core exit tempera-tures for the 100% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 100% power. 34 17 Nominal DEM0-F prediction of the core exit temperature for the average channel for the 5% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 5% power. 35 18 Nominal DEM0-F prediction of the core exit temperature for the average channel for the 35% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 35% power. 36 19 Nominal DEM0-F prediction of the core exit temperature for the average channel for the 75% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 75% power. 37 20 Nominal and worst case DEM0-F prediction of the core exit tem-perature for the average channel for the 100% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 100% power. 38 21 Nominal IANUS prediction of the core exit temperature for the ,

100% LOP test compared to the identical prediction using limit-ing gap conductance. 41

-vi-

LIST OF FIGURES (Cont.)

Figure Title Page 22 Primary and secondary flow rates for the 35% unbalanced LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 35% power. 45 23 Primary and secondary IHX outlet temperatures for the P-loop (representing 3 loops with pony motors off) during the 35% un-balanced LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 35% power. 48 24 Primary and secondary IHX outlet temperatures for the I-loop (representing the 1 loop with a secondary pony motor running) during the 35% unbalanced LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 35% power. 49 i

l 9

-vii-

LIST OF TABLES Table Title Page 1 Sur. nary of Differences for the FFTF Sfstem Models, used in the Lill Analyses, Compared to the Default Modeling in IANUS. 4 .

2 Summary of Flow Redistribution Modeling used in the BNL Analyses Compared to the 12-Channel FLODISC Model . 5 -

3 Modifications of I ANUS to Simulate the Loss of Power from Refuel-ing Conditions Experiment in FFTF. 12 4 Assumed Conditions for the Simulations of the FFTF Primary Loop, Loss-of-Power Transient Tests. 30 5 Comparison of Nominal and Worst Case Predictions of Maximum Tem-peratures for the FFTF 100% LOP Test af ter 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of Full Power Operation. 40 6 Comparison Between HEDL and BNL Nominal Predictions for the 5 Per-cent Loss-of-Power Test in FFTF after One Hour of Prior Operation at 5 Percent Power. 44 7 Summary of BNL Test Predictions of the Maximum Temperatures for the 35% Unbalanced Loop Loss-of-Power Test in FFTF. 46 O

-viii-

h

1.0 INTRODUCTION

As part of the Nuclear Regulatory Commission's (NRC) safety review of the Fast Flux Test Facility (FFTF) the Safety Evaluation Group at Brookhaven National Laboratory (BNL) has used pertinent computer codes to aid in eval-uating some of the safety analyses presented in the FFTF Final Safety Analysis Report (1) (FSAR). The NRC's Safety Evaluation Report (2) has "... condition-ally accepted the premise that there is reasonable assurance that natural cir-culation will be demonstrated as a viable method of removing decay heat". As part of the Natural Circulation Test Program (3) the NRC has requested (2)

... substantive verification of the mathematical models of the IANUS(4) and FLODISC(5) codes" .

The purpose of the present investigation is to review the available data from the secondary system tests (6) and to provide independent predictions for the nuclear test series. The present app- ach has been to retain consistent modeling between the test predictions and previous safety analyses (1,7-9) and to specifically identify those modifications which are used to generate the nominal (expected) behavior.

Although the primary system tests will be the most relevant to the safety I

analysis, the emphasis of the present investigation has been to understand the unanticipated behavior demonstrated by the secondary system test data. The 5%

transient test was performed on November 20th,1980 but the results are not available as yet.

2.0 PREVIOUS WORK There have been nunerous investigations (7-11) into the natural circula-tion capability of fast breeder reactors. The Clinch River Breeder Reactor Project (CRBRP) used DEM0(12) to analyze the system response to LOP tran- ~

l sients(10) and concluded that there would be adequate natural circulation to prevent bciling from occurring for the postulated LOP event.(10) However, a

~

subsecuent BNL investigation (ll) indicated that the results were quite sensi-tive to unmodeled effects, the most significant of which was flow redistribu-tion away from the breeding blankets. The CRBR design was modified (13) to

( provide back-up D.C. power for one heat transport train to ensure at least some I forced circulation for the LOP event The FFTF design does not include radial blankets and the longer primary pump coastdown is predicted (7) to give substantial margin to boiling for the l postulated LOP event. However, some unmodeled effects (predominantly strat-l ification and flow redistribution) as well as some faulted conditions (8) l appear (9) to have the capability to greatly diminish the margin to boiling.

Analyses at Argonne National Laboratory have indicated that severe stratifica-tion can occur in the upper plenum (14) as well as in the piping (15) at nat-

! ural circulation conditions. An on! Sing experimental study (16) is attempt-ing to define the significance of three-dimensional effects in the heat ex-changers.

While the limited measurement capability at FFTF precludes the likelihood of precisely defining the significance of such unmodeled effects, it is impor-tant to recognize their presence and their possible contribution to unantici-pated resul ts.

e 3.0 TEST DESCRIPTION l

The stated objective of the natural circulation tests (3) is "to demon-i strate the natural circulation capability of the plant and to show that perfor-j, mance is acceptable for all Emergency Loss-of-Power Events". A second and broader objective of the test program is to " provide necessary tools and empir-

', ical data to pertinent future breeder reactor designs to rely on sodium natural circulation as a back-up means of ensuring decay heat removal (7)". Since the test program cannot cover all hypothetical conditions, achievement of both ob-jectives will necessitate the evaluation of computer codes which can be used to extend (interpolate and/or extrapolate) the experimental results to other conditions. We note that, although this approach appears to be necessary, i t might not be sufficient with the present analytical tools and for a broad range of natural circulation conditions.

As a preliminary form of this code evaluation, test predictions have been made, using nominal and worst case modeling, to indicate the range of expected behavior. For the case of the non-nuclear tests of natural circulation in the secondary loop, the available test data (6,18) tend to agree reasonably well with the nominal modeling (as expected), but the estimated uncertainty in the data covers a large part of the difference between the worst case and nominal I resul ts. This general agreement with the nominal case cannot be construed as a rejection of the " conservative" modeling in the worst case. Rather, any mod-ifications to the modeling must be based on statistical analysis of the data uncertainty and modeling sensitivity. This is particularly true when the mod-els are to be used to analyze conditions out of the testing range (e.g., to ana-lyze the behavior for a hypothetical accident). A sumary of the nominal and worst case models as used for the BNL analyses are provided in Tables 1 and 2, respectively. The worst case models are intended to be conservative and are essentially those models which were used by the Project staff (7,8) and the present authors (9) to analyze loss-of-power transients in FFTF. The nominal models' incorporate best-estimate modeling of key parameters which have been tested at the Hanford Engineering and Development Laboratory (HEDL) and the Liquid Metal Engineering Center (LMEC) .

L TABLE 1 Sunnary of Differences for the FFTF System Models,

1. Used in the BNL Analyses, Compared to the Default Modeling in IANUS l

IANUS IANUS DEMO-F DEM0-F Parameter (Worst Case) (Nominal) (Worst Case) (Nominal)

. Decay Power

(% of Nominal) 125 100 125 100 -

Core AP (psi, kPa) 134.4 (926.5) 112(772) 134.4 (926.5) 112(772)

! Core Flow Ex onent, n (aP = kVn = 1.82(a) = 1.82(a) 1.75 1.82 Pump Stopped Rotor AP

(% of Nominal) 115 65 115 65 lChack Valve AP

(% of Nomina 1) 125 100 125 100 l Upper Plenum l Stratification None None Immediate Stratification Stratification Delayed by 60 Seconds lAssemblyLength(m) 2.43 2.43 1.54 2.43 Reactor Days at Power 1.0(b) 1.0(b) 1.0(b) 1.0(b)

Pump Trip Delay (Sec. Af ter Scram) 1.0 1.0 0.01 0.01 Scram Reactivity (Dollars) 18 18 18 18 Operating Powt:r TestSpec(b) Test Spec (b) Test Spec (b) TestSpec(b) l Operating ? low TestSpec(b) TestSpec(b) TestSpec(b) TestSpec(b)

= 1.8(c) = 1.8(c)

Piping Flow Exponent 1.75 1.8 -

~

Notes: (a) IANUS uses a polynomial fit of data which is approximately equivalent to an l exponent of 1.82 at 3% flow.

(b) Post-test analyses will incorporate the actual days of operation, pre-test power, rod and scram delay.

(c) The piping flow exponent in IANUS changes with relative roughness of the pipe, c/D.

TABLE 2 Summary of Flow Redistribution Modeling Used in i tiie BNL Analyses Compared to the 12-Channel FLODISC Model(5) l 1

l Parameter FLODISC - Nominal FLODISC - Worst Case Flow Exponent, n (aP = kVn) 1.83* 1.75*

Upper Assembly length (m) 2.43 1.54 Core AP (psi) 112 134.4 Decay Heat (Fraction of ANS) 1.0 1.25 Power / Flow (Normalized) Variable Variable Transient or S teady-S ta te Transient Transient Number of Channels 12 12

• Note:

The flow exponent is represented approx {mately by a single value but actually has three components, ki V , k 2Vl 917 and k 3Vs1.75,

\

3.1 SECONDARY TESTS The first set of tests on the secondary systen have been performed.(6)

_The reactor core had not yet been loaded so the heat input was provided by pump work and trace heaters. The main objectives of these tests ere(6).

(1) To detemine the natural circulation flow rate in the secondary loop -

as a function of the temperature difference between the hot and cold leg.

~ '

(2) To verify the ability to isolate DHX modules without causing severe thermal transients.

(3) Demonstrate a secondary loop natural circulation flow start-up at refueling conditions.

(4) Measure DHX heat losses with the air path closed.

These objectives appear to have been met with only one major surprise (a much better margin to freezing than predicted for the loss-of-power from refuel-ing conditions).

3.1.1 Steady-State Tests The two steady-state natural circulation test points are shown in Figure 1, along with the nominal and worst case calculations (post-test) using DEM0-F and IANUS. Note that the nominal calculations from both codes agree quite cicaely with the two data points, but when the estimated data uncertainty is considered (shown as error bars on Figure 1), there appears to be little basis for rejecting the worst case models for being too conservative. In fact, there is considerable overlap at the low-flow data point.

The steady-state dump heat exchanger (DHX) heat loss characteristics are shown in Figure 2. The HEDL(18) heat loss data are about 20% below both the previous LMEC(19) data and the acceptance limit established by the FFTF Project staff. However, the uncertainty in this data is about 20% to 40%.

3.1.2 Transient Tests Fortunately, unlike tm LOP tests from operating conditions, the 1.0P ,

tests'from refueling conditions are very close to demonstration tests (in that pertinent parameters are identical to those hypothesized for the accident).

Thus, the tests, themselves, demonstrate that freezing will not occur for the 1 l 1 1 I i

~~

-HEDL PRETEST O I AN US -NOMIN AL a U

el ANUS-WORST CASE 0 3.0 - 0 DEMOF-NOMINAL

.5 -

4 DEMOF-WORST CASE Q DATA WITH UNCERTAINTY ESTIMATE /

+  !

Y 3 ..

O

_J

u. 2.0 -

6m -

O. .

O ..

O

.J O

Z O

O

$ l.O - -

0.0 O.O 10 20 30 40 50 60 70 SECONDARY LOOP AT (K)

Figure 1: Steady-state natural circulation flow in the secondar loop, Calculated by IAtlUS and DEM0-F, Compared to HEDL Data 18)and Pretest Predictions (3),

i 600 i , ,

ACCEPTANCE LIMIT I3) t

--- HEDL DATA (I7) 500 - O LMEC DATA (le) _

/

- /

0 /

$ /

i 400 -

k /

/

3 x /

~ /

/

m 300 -

/

m

/

S

$ /

W /

I 200 -

f X /

I /

O 100 - -

I i i O 10 0 200 300 400 ~

Tav - Ta mbie nt Wi Figure 2: Comparison of HEDL Heat LossU } Cata Data and the Acceptanca Limit Suggested )tobyPrevious HEDL. LFEC

. _ . _ . _ _ _ _ _ . _ _ _ _ . _ _.. h

given(20) hypothetical conditions (although the measurement location does not

! correspond precisely to the coldest point in the system).

The DHX outlet temperature measurements are shown in Figure 3, along with the nominal IANUS calculations. The minimum temperature is about 20K higher o than originally predicted (3) and adds additional assurance that premature freezing will not occur for the hypothetical e sent.

However, it is important to note that, even with this added margin, freez-ing will occur eventually as the decay heat decreases and operator intervention is required to prevent all of the loops from freezing simultaneously (freezing would block all heat removal paths to the DHX).

3.1.3 Comparison to Analyses While the tests of the primary system will be the most pertinent to plant safety, the data from the secondary tests will provide important verification in formation. The present approach has emphasized the use of the codes and mod-els used in the previous safety analyses (1,7,8,9,19), specifically dei 10-F ,

IANUS(4) and FLODISC(5), While these codes are not intended to simulate these types of experiments, and they use relatively large amounts of computer time (due to the large time constants at natural circulation conditions), the experience gained with the verification process appears to be well worth the effort. The nominal IANUS prediction agrees surprisingly well with the two steady-state data points in Figure 1, while the worst case IANUS prediction tends to be at the lower bound of the estimated experimental uncertainty range (shown as error bars on the data points). The fact that the worst case IANUS calculation more or less bounds the data is not surprising (in fact it is rather reassuring), since the worst case model incorporates upper bound esti-mates of the pressure drop characteristics of the secondary system. It is more surprising to note that the original HEDL predictions (3) (also shown in Fig-ure 1) are weil below both the nominal and the worst case predictions. The large margin between the HEDL prediction and the data seems to imply that the natural circulation flow rate is much better than expected. However, most of the margin appears to be due to differences between the IANUS and the HEDL model. The data are only marginally above (the margin is approximately equal to the experimental uncertainty) tne worst case prediction and very close to the nominal prediction. Similar calculations have been performed using

I I I I I I I I I O DATA, AVERAGE O HEDL, PRETEST EVALUATION MODEL V IANUS-NOMINAL DEFAULT DHX OUT 640 620 -

E w -

w a: -

D600 -

ka:

w _

Q.

E W -

- 580 -

560 -

l 540 -

. I I I I I I I I I O !O 20 30 40 50 -

TIME (minutes) i Figure 3: The nominal IANUS calgylation of the DHX outlet tempergture compared to the datalb> and HEDL pretest predictions (3) for

+ the LOP from refueling conditions.

DEMO-FI9) and are also shown in Figure 1. Again, the nominal results agree quite well with the data and the worst case results tend to fall at the lower range of the estimated experimental uncertainty interval .

The behavior for the LOP from refueling conditions (6) is even more in-teresting in that there was a large disparity in both the shape and magnitude of the DHX outlet temperature between the nominal HEDL predictions (3) and the

. experiment as shown in Figure 3. Much of this disparity is said(6) to be due to neglecting the thermal capacity effects in the DHX tubes (an effect which is accconted for in IANUS). However, the post-test calculations (6) accounting fo? tube thermal capacity (shown in Figure 5) still demonstrated a wide dis-parity in shape and magnitude when compared to the data.

The present approach utilized the phenomenological modeling incorporated i in IANUS to analyze the transient. DEM0-F was not used becaus) it uses a sim-i plified " perfect" DHX and the modifications to get DEMO-F to simulate the tran-sient would be fairly extensive. The modifications to get IANUS to run the secondary LOP transient were minor and most of them could be accomplished by input specification. These modifications are summarized in Table 3. The first two modifications are simply manipulations of the IANUS variables to obtain the unusual boundary conditions (they are unusual only in that they are not associ-ated with normal operating conditions and/or expected transients) associated with the secondary LOP test. The third modification can still be accomplished by a manipulation of the input, but it involves the implicit assumption that all " heat losses" are caused by leakage through the closed dampers. This leak-age assumption is justified by examining the geometry of the DHX tube bundles.

The tube-to-tube temperature uniformity (6) (with the exception of edge tubes) j along with limited radiation view factors for interior tubes appears to be suf-ficient infomation to rule out radiation as a dominant heat transfer mecha-nism. There may, of course, be recirculation loops (i.e., internal natural convection), but the temperature unifomity precludes downflow within the tube bundle and the edge baffles should minimize downflow at the periphery. In any

- case, the assumed heat loss phenomenon (by leakage flow) results in reasonable agreement with the experiment and appears to identify the source of' peculiari-ties in the temperature data.

1 l

l TABLE 3 l

l Modifications to IANUS to Simulate the .

Loss-of-Power from Refueling Conditions Experiment in FFTF  !

l l

l I Model Change from IANUS Default Configuration ,

Pump Work Energy input is modeled as initial fis-sion energy input (3.6 MW nominal) with step change to 15% of initial to hold -

primary system temperature constant.

Secondary Pump Trip To get detailed hydraulics calculations all pumps are tripped with pony motor speed set to 100% of initial full flow in the primary system and 0.0% in the secondary system (this circumvents the normal trip of all pumps).

DHX Heat Loss Fans are shut off and air flow is held constant to obtain equilibrium heat loss (1.6 MW per loop nominal at 370*C).

DHX Thermal Capacity The modeling of the DHX in IANUS is equivalent to a 7 pass heat exchanger l with most of the heat transfer taking place at the bottom pass. For realis-tic heat capacity and temperature dis-tributions the air temperature at the bottom two nodes was held constant.

l l

I The IANUS nominal case DHX outlet temperature and the secondary mass flow rate are shown in Figures 3 and 4, respectively. Note that the tube bundle exit temperature (also shown in Figure 3) drops much more rapidly than the data, but this is expected since there are roughly 4 meters of pipe between the thermocouple (TC) location and the tube bundle exit. The calculated tem-perature at the TC location can be obtained (approximately) from the detailed printouts and this compares quite favorably in shape, but the response time is still more rapid than the data indicate, as shown in Figure 5. There are four features of the nominal IANUS calculation which merit some elaboration.

(1) The IANUS calculation yields the same shape as the data show for the first 12 minutes.

(2) The calculated temperature at 40 minutes tends to be somewhat high rather than low (as calculated by the Project staff).

(3) The IANUS calculation of the exit temperature shows a flattening off at about 15 minutes (as does the data somewhat later), but the calculation does not show the continued reduction several minutes later.

(4) Due to the imbalance in flow rates, most of the heat transfer takes place at the bottom node (of seven nodes) in the DHX. Above this node the air is too hot to provide effective cooling of the sodium.

I The generally good agreement observed in (1) in shape and magnitude tends to indicate the acceptability of the convective heat loss modeling in IANUS.

However, the tendency for the nominal model to overpredict the flow rate re-sul ts in a relatively high outlet temperature (the product of m and AT equals q

the heat loss at steady-state conditions) .

The overprediction of the flow rate tends to indicate that the nominal case underestimates the flow resistance at these low flow rates (as was pre-viously observed for the steady-state data) . However, it should be recognized that the calculated results are extremely sensitive to the heat loss. The val-ue of 400 KW per module was taken from the heat loss data (18), but these data are expected to have a large uncertainty (not yet provided b,' the Project).

The sudden flattening off of the DHX outlet temperature tas seen in both the data and the IANUS calculation at 15 to 20 minutes) is sanewhat surprising

{

I 6 i i i i i

- O F FT F DATA 5 -- -

A l ANUS-N ORM AL W O V

<4 m -- -

3 A O

I

u. 3 -- -

e 4

o2 z 9 O

O w

m1 - -

O O 10 .20 30 40 50 60 70 TIME (minutes)

Figure 4: The nominal IANUS ca]cylation of sodium flow through the DHX -

compared to the data (61 for the LOP from refueling conditions.

640 i i i i i i

. o FFTF DATA WITH SCATTER -

O HEDL NOMINAL MODEL o I ANUS NOMINAL MODEL E 620 w

a: -

D H

, $600 w

i 1

c.

1 _ -

w i-o 580 -

w

_J g _ _

_J O

o 560 - -

540 O 10 20 30 40 50 60 70 TIME (minutes)

Figure 5: The nominal IANUS calculation of the secondary cold lea temperature at the TC location compared to the data (6) for the LOP from refueling conditions.

s

! 15-

in view of the Project's calculations. However, the fourth observation 'made from detailed printouts of IANUS) helps to explain the mechanism behind the dramatic change in slope of both the experiment and the calculation. With the very low air flow rate, the top six sodium nodes are essentially at the inlet temperature while most of the heat transfer takes place in the bottom node. ,

Immediately after coastdown, the sodin flow is almost stagnant (a flow rate of about 0.%), but the heat transfer continues to dominate in the bottom node.

(Phenomenologically, the bottom node still has sufficient heat capacity to heat the air to nearly the sodium temperature.) This leads to a dramatic nonlinear-ity in the sodim temperature distribution. This nonlinearity is apparently not, included in the HEDL model and would not be seen at the exit except that the thermal transient causes an increased buoyant head and resultant increase in mass flow (by a factor of about four) . This increased flow eventually

, drives the hot sodim through the last node resulting in the sudden change in slope of the temperature curve. The reason for the difference in timing be-tween the IANUS prediction and the experiment is simply that the seven node model in IANUS, in effect, models a seven-pass, counterflow heat exchanger, whereas the FFTF DHX is actually a four-pass, cross-flow heat exchanger. For the actual DHX, shown schematically in Figure 6, only the bottom pass has rel-atively cool inlet air flowing over it thus limiting the effective zone to 1/4 of the entire heat exchanger. Whereas, in the IANUS modeling, also illustrated in Figure 6, the bottom pass is only 1/7 of the total heat exchanger. Thus, the effective portion of the DHX mo<iel has about 1/2 (actually 4/7) of the theimal capacity and transport time of the effective portion of the FFTF DHX.

While ths esent authors have not attempted to do so, it is fairly obvious that the ag, ment between the code and the experiment can be greatly improved by changing th, modeling to a cross-flow representation with noding capability

in multiples of s In an attempt demonstrate the importance of the nonlinear temperature ,

distribution on the e tuing transient, the air-side temperature calculations were overridden to hold the air temperature to a constant value (175'C) for the .

bottom two nodes. Thus, the length of the effective heat exchanger is doubled to 11.1 m (as opposed to the actual length of the first pass of 9.7). This modeling would be expected to result in a more consistent response time between the model and the experiment.

)

l TNa=640 K

l. s T

, air =64OK s

. b

! l '

! C

) )

C 1

Noir =600K) l

g 6 4Tair =3OO K T Nq 2600K

]

f IANUS MODEL i

l TNa=640K jpTair=640K

! )

C l /}Tair = 600K ]

,CF f Tair =3OO K

, TN a=600K 1-1 I FFTF DHX TUBE i

Figure 6: Schematic of the FFTF DHX operation during an LOP from refueling conditions compared to 7 node modeling in IAtlVS.

4

Several calculations were perfomed with this adjusted (6 pass) model .

The chronology of these calculations is not particularly relevant, but it is I interesting to point out the good agreement which can be achieved with this adjusted model . The preliminary calculations indicated that about 1/3 of the heat losses occur above the bottom pass of the DHX. While the precise distri- ,

bution of this remaining heat loss does not appear to be important for the l present calculations, it was assumed to be radiation from the top pass of the

~

! heat exchanger. With this heat loss distribution and the experimentally deter-

! mined heat loss, the worst case (maximum flow resistance) model gives good agreement with the data as indicated in Figures 7 and 8. The tendency for the l calculations to rise back to equilibrium more quickly (at about 60 minutes) than the data indicate is probably due to one or more factors, including:

numerical diffusion of heat in the cold leg; thermal stratification; errors in geometry including lack of pmp tank; or the coarse nodalization of the DHX.

(Each of the seven nodes in the DHX model represents about 5.6 m of tubing i while the outlet temperature is an average of about 15 m of pipe.) Note that the IHX utilizes an optional neber of nodes and that 21 appears to be a mini-mum number to follow rapid themal transients.

The nominal IANUS calculation (not shown) with the modified (6 pass) DHX l consistently overpredicts the flow rate and underpredicts the temperature drop.

l However, by increasing the assumed heat loss to the acceptance limit, the cal-culated exit temperatures show improved agreement with the data (as shown in Figure 7). However, this good agreement in temperature is obtained at the ex-pense of the flow rate. Thus, the nominal flow rate calculations are consider-l ably above the data shown in Figure 8. The tendency for both the nominal and worst case IANUS models to predict high flow rates after about 25 minutes (in spite of following the DHX thermal transient to about 35 minutes) leads us to the conclusion that the themal transport or geometry in the cold leg is in-correctly modeled. It is also noteworthy that the adjusted HEDL model does ,

show consistent transient response with both the flow and temperature data.

I This tends to indicate that there are differences in the geometry and/or ther-mal transport for the cold leg which do not show up during steady-state (when the cold leg is at uniform temperature). Informal discussions with HEDL per-sonnel indicate that the only known difference in the modeling of the cold leg is the inclusion of the pump tank in the HEDL model .

640 i i i i E --- F E TF DATA w o I ANUS-WORST CASE,4.8 MW -

O l ANUS-NOMINAL CASE,3.6 MW

' ~

H A l ANUS-NOMINAL CASE,6.0 MW

$ 620 w

CL 2 -

w l }-

e 600 -

w

! J h -

7

o

U 580 -

s s -

/ -

T

< /

s' g _ 's - _

O

w 560

• O 10 20 30 40 50 60 70 TIME (minutes)

Figure 7: HEDL data for the LOP from refueling conditions compared to l IAflUS calculations of the secondary cold leg temoerature at j the TC location using the adjusted (6-pass) model of the DHX.

1

. _ - _ - ~ _ __ _ - . - . - _

i 1

4 6 i i i i i i 1

4 O FFTF DATA 3

• 5 -

O I ANUS- NOMINAL,6.0 MW _

Q I ANUS-NOMINAL,3.6 MW

• O a I ANUS-WORST CASE,4.8 MW

<4 x -- -

3 A o

.J t

u. 3 - -

i >- a x

i

<2 o

z O O

O W

e I - -

4 0 ' ' ' ' ' '

) O 10 20 30 40 50 60 70 TIME ( minutes) 4 Figure 8: HEDL data for the LOP from refuelino conditions compared to IANUS calculations of the secondary ficw rate using the -

adjusted (6-pass) model of the DHX.

t

During the initial coastdown, (shown in Figure 9) both the worst case and nominal case IANUS underpredict the loop flow rate. The DEM0-F calculations (not shown) were fairly consistent with a tendency to be slightly lower than IANUS. It should be recognized that both the " nominal" and " worst case" models are somewhat conservative predictors of the mass flow rate during pump coast-down (a., they are intended to be). This slight conservatism in flow rate can l be eliminated by adding an inertial term to account for the sodium in the pump

('

l impeller (as was apparently done for the HEDL calculations).

3.2 PLANNED PRIMARY TESTS t

As previously mentioned, the emphasis of the present report is on the evaluation of the available secondary test data. However, sevual illustrative predictions of the primary tests are included to emphasize a key point, spe-cifically: <

There is a very limited margin between the worst case calculations and the nominal calculations. It is felt that the verification program should acknowledge uncertainties in the data and attempt to attach some confidence level to the conservative model, or equivalently, attempt to realistically quantify the uncertainty band (not simply data scatter) around modified models.

It should be recognized that many details of the natural circulation tests from nuclear power may change significant% lrra9 the planned conditions (delay before pump trip, decay power, contrM (* o iv i ty, environmental temperature, etc.). In fact, the limited DHX cito a- dicates that the 100% power test cannot be run unless the inlet air u.mperature is considerably below the design temperature of 32*C.

In any case, no attempt has been made to " fine tune" the analyses to imprecisely defined test conditions. The present predictic.n are presented ta illustrate existing differences between the Project's models and the present

. models as well as to indicate the magn'tude of the limited margin between the predictions using the nominal modeling, and predictions using the worst case modeling.

3.2.1 Nuclear Steady-State Tests The steady-state primary tests are scheduled to be completed early in 1981. As with the secondary tests, the primary steady-state tests will provice

I00 i i i i i

~

HEDL 90 i --- I ANUS-NOMIN AL -

1 0 DATA

\

80 g 1

- k 8

~

70 -1 -

\

w i h  !

y 60 i

• \

O J 50 - \ -

u_ \ '

p \

\

O 40 -

g z \

O \

o 30 -

\

W NO M \

NO 20 -

s -

NO

'sO' 10 -

's s -

'~ -

~,~~ ,

O ~

O 10 20 30 40 50 60 TIME (seconds)

Figure 9: IANUS calculation of the secondary flow rate during pump coastdown, compared to the data and the HEDL calculations.

important verification of the integral pressure loss characteristics of the primary loop. Since the core AP tends to dominate the pressure losses even at low flow rates, these measurements will provide important (albeit indirect) confinnation of the estimated pressure loss characteristics at low flow rates.

. The nominal and worst case DEM0-F and IANUS predictions f the mass flow versus loop aT in the primary system along with the HEDL prediction are shown

, in Figure 10. Somewhat surprisingly (in light of the disagreement on the sec- j ondary side), the nominal DEM0-F and IANUS predictions are in substantial agree-ment with the HEDL calculations. This agreement does not necessarily mean that all the flow resistance models are identical for each of the predictions, it simply refleu.s the predominance of the precisely defined core pressure drop (i .e. , the nominal core pressure drop of 770 kPa is defined by extensive exper-imentation and has been used identically in each calculation) in the overall pressure loss estimate. The most important aspect of the comparison is that the worst case predictions are only about 10% lower than the nominal predic-tions. Given the inherent difficulties in making meuurements in a nuclear environment (with thermal and irradiation induced drift) it is expected that it will be extremely difficult to discriminate between the nominal and worst case models. For example, a flow measurement which is accurate to within 0.2%

of full scale gives rise to an uncertainty of about 10% at natural circulation conditions.

The steady-state natural circulation tests will al so provide impor-tant data to verify flow redistribution predictions. As pointed out previ-ously(21) the flow redistribution model is one of the most important aspects of the verification program since it provides the basic mechanism to extrapo-late the computer code prediction of the average temperature rise to the sta-tistically worst assembly ( the " hot" assembly hwing a power-to-flow ratio 55%

above the average assembly). The row 2 fuel s-open-test assembly (F0TA) will

- provide an intermediate point (with a power-to-flow ratio 27% above the aver-age) for verification, but the model must remain on sound phenomenological

, ground in order to perform the extrapolation with confidence. While IANUS and DEM0-F do not account for flow redistribution explicitly, the beneficial ef-fects (the effects are beneficial in that the hot channel factor is reduced) of flow redistribution are accounted for in the flow-dependent "maldistribution"

4 i i HEDL PREDICTION o I ANUS - NOMINAL e I ANUS - WORST CASE A D E MO-F-NOMIN A L A DEMO-F-WORST C ASE -

l 3 - -

g 0 a

o d2 -

4 l

g _ _

Z Q.

l O

O 50 10 0 15 0 PRIM ARY LOOP AT (K)

Figure 10: IANUS and DEliO-F predictions for the primary flow versus loop AT compared to the HEDL prediction.

factor. Buoyancy-induced redistribution is a real phenomenon, but the effects of flow transitions in the orificing may negate the beneficial effects.

As with the secondary tests, only careful application of uncertainty analysis can be expected to reduce the margin between the nominal and worst

- case models. In fact, for the steady-state tests, the conservative flow re-distribution model (as used in IANUS) and the nominal model are essentially

. equivalent in that the main conservatism built into the IANUS model is to limit the power-to-flow ratio to be at most equal to 1.0. Since the power-to-flow ratio is expected to be less than 1.0 for all the steady-state tests, the mar-gin between the conservative model and the cominal model is zero. Obviously ,

the data cannot distinguish between identical models and verification of the conservatism in the IANUS flow distribution model will require transient data wi th a high power-to-fl ow ratio. l As an indication of the difficulties inherent in verifying flow re-distribution models, the prediction of the nominal AT across the row 6 F0TA versus the row 2 F0TA AT is shown in Figure 11. While these two assemblies have significantly different power levels (6830 KW for the row 2 F0TA versus 5080 for the row 6 F0TA), the flow orificing is designed to keep the AT across the two assemblies nearly equal at full power conditions. The resul t is that the buoyancy-induced flow redistribution is predicted to be nearly the same for both assemblies and the row 2 AT is predicted to be only 10% higher than the row 6 AT as shown. If there were no flow redistribution, the row 2 AT would be 9% higher than the row 6 AT as shown in Figure 11. Thus, the margin between the expected flow redistribution and no flow redistribution translates into a 1.0% difference in AT. Obviously, when one considers the measurement uncer-tainties (power level, aT and AP) it will be difficult to use this comparison as a verification tool.

It appears much more productive to compare F0TA AT to the total loop AT.

~

Since flow distribution to the loop remains constant (at 100%), any significant (larger than measurement uncertainties) change in the AT ratio can be ascribed to flow redistribution effects (although it still may be difficult to determine which effects are due to buoyancy-induced redistribution and which effects are due to friction-induced redistribution) . For this case the nominal FLODISC model predicts that the row 2 AT will be about 20% higher than the loop AT as

200 i i FLODISC-NOMINAL NO FLOW REDISTRIBUTION -

~

l50 -

/

n /

5 /

/

h /

/

< /

l-- / 4 o100 -

/

/

N l

/

3 /

o O'

50 - -

I I O -

0 50 100 15 C-ROW 6 FOTA AT (K)

Figure 11: FLODISC prediction of F0TA T's during steady-state testing compared to a calculation assuming no flow redistribution.

indicated in Figure 12. This is about 25% better than the case of no flow re-distribution. It is interesting to note, however, that while the 25% margin is likely to be larger than the experimental uncertainty, the adverse orificing effects at low flow (the orificing is designed for full flow) have the poten-tial to negate this margin. For example, the worst case model suggested by Perkins and Bari(20) (see Table 2) predicts that in the limiting case there could be nearly no flow redistribution. This bounding estimate does not imply that any adverse flow redistribution will show up as a uniform departure from i the nominal predictions. Rather, such adverse behavior is expected to show up over a limited range (representative of turbulent-to-laminar transition region) as indicated for a hypothetical experiment by the dashed line in Figure 12.

Thus, it will be important to evaluate the data over the full range of the ex-periment in order to ve-ify the flow redistribution models. The tendency to reject data as spurious if it does not fit the model must be avoided. This is particularly true if one is to attempt to apply the data obtained from a few test assemblies (the most reliable flow redistribution data is expected to come from the three F0TA's) to all 76 assemblies in the reactor. While there are only three orifice zones in FFTF, the " orifices" themselves are simply holes drilled in reflector blocks and they cannot be expected to have the reproduc-ible characteristics of a standard, sharp-edged orifice.

3.2.2 Nuclear Transient Tests The transient natural circulation tests will provide important data f or verifying the computer codes at conditions approaching the cr;nditions of the postulated loss-of-power event. While the tests are programmed to gradually lead up to the full power tests with increasing levels of con-fidence in the predictive tools, it is clear that even the full power test is not a demonstration test for the postulated events (i.e., the worst case LOP events (7,8,20)). The greatest utility of the tests will be for verifi-cation of the codes used in safety analyses. The codes can then be used to extend results (aith adequate allowance for experimental uncertainties) to

. the specific conditions of postulated accidents. In the particular instance of the worst case LOP event (7) , the essential variables which require extra-polation beyond the data base are the power level at transition to natural cir-culation (from about 2%, depending on power history, to 3.5%) and the power-to-flow ratio (from about 1.27 for the F0TA to 1.55 for the hypothetical acci-den t) .

1 I HYPOTHETIC AL REDISTRIBUTION IN TRANSITION REGION o FLODISC - N OMIN AL a FLODISC-WORST CASE /

/ .

150 -

/ -

/

- /

M /

F- <

/

<3 5 /

O 10 0 -

/

LL N

3 o

M 50 - -

O -

0 50 100 150 PRIMARY LOOP AT (K)

Figure 12: FLODISC nominal and worst case prediction of row 2 F0TA AT versus loop AT for steady-state testing compared to hypothetical (unmodeled) transition behavior.

)

The nominal IANUS model (see Table 1) has been used to predict the core outlet temperature for the 5%, 35%, 75% and 100% tests as shown in Figures 13 through 16. The conditions for each prediction are uescribed in Table 4. The intent of these predictions is not to show details of the predicted behavior

. since the initial conditions (particularly operating history) may change enough to make the details irrelevant. Rather, the predictions are included to

(1) show the anticipated range of behavior (related to the requirements for experimental accuracy);

(2) benchnark the behavior of the existing models in IANUS; and (3) establish whether significant differences exist between the FFTF Project's predictions and the present predictions using modeling that is theoretically identical .

The most important aspect of these predictions is th margin between the worst case and the nominal predictions. Even for the 100% power case the peak worst case core outlet temperature is only 35 K higher than the nominal predic-tion as shown in Figure 16. As with the other phases of the test program, this limited margin is expected to challenge the accuracy of the experimental re-sults.

Predictions, for the identical cases using the nominal and worst case models for DEM0-F, have been made and the results are presented in Figures 17 through 20. While it is not obvious from the plots (due to the slightly dif-ferent graphics routines which generated the plots), the nominal cases for the DEM0-F and IANUS predictions are essentially identical .

The worst case OEM0-F prediction at 100% power (also shown in Figure 20) is significantly worse (by about 20K) than the worst case IANUS prediction due to the presence of upper-plenun stratification and the t 'nding core AP used in the DEM0-F worst case model . At this time neither worst case model can be recommended over the other, except to say that the DEMO-F model is more conser-vative. The utility of the different models is expected to be seen in the post-test evaluations in that detailed data comparisons will aid in establishing the significance of the different modeling approaches. For instance, comparisons of core outlet and vessel outlet temperatures with DEM0-F calculations can be expected to establish the importance of stratification in the upper plenun (which is not modeled in IANUS) .

-2 9-

l TABLE 4 Assumed Conditions for the Simulations of the FFTF Primary Loop, Loss-of-Power .

l Transient Tests Hours of Operation Core Inlet Initial Power at Initial Initial Flow Temperature Test (% of Design) Power (% of Design) (K) 5% 5 25 75 583.5 35% 35 25 75 583.5 35% 35 25 75 583.5 (Unbalanced *)

75% 75 25 75 634.3 100% 100 25 100 634.3 i

i I

(

• Informal discussions indicate that the 35% test will actually be run in an unbalanced configuration with one secondary pony motor remaining active.

l I

NRT CIRC ON PRIMARY LOOP 5% POWER 7S'i FLOW

> l CASE NUMBER 4041 F:::

LEGEND 620-

/ O- CORE INLET O- HVG.NO OUT 61s

. g/

/ g  %

O- HC NA OUT A- HVG. CLAD I V- HC CLAD IN g 610 m

l f f

e e

Y / 7 e  :-

5 e /

l/

5,5_

/' -

/

,,0 L//

//

585- f

[ H3-0 0 0 580 0 20 10 60 80 100  !?O 110 160 180 200 220 240 260 TIME (SECONDS)

Figure 13: Nominal IANUS prediction of core temperatures for the 5% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 5% power.

.. , l si i . - - - - - .

< DETRILED CORE TEMPERATURES (3 X 5 CORE USED)

NOT CIRC ON PRI!1RRY LOOP 35% POWER 75% FLOW

" CRSE NUMBER 3040 LEGEN0 720 -- ^

O- CORE INLET O - AVG.NR 00T O- HC NA OUT 700-

/ A- AVG.CLfiO 1 V- HC CLAD IN l

2 D 680- j

/

d M e'

E e- x

/ l s

E x #

640 --

7 te 620 ' -

/ .

3 ,

600 Wx g

[] O O C S 60 80 100 120 110 160 180 O 20 10 TIME ISECOND3)

Figure 14: Nominal IANUS prediction of core temperatures for the 35% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 3S% power.

l' 1

DETRILED CORE TEMPERATURES (3 X 5 CORE USED) .

757, LOP TEST AT 25 HOURS l~ t CASE NUMBER 3039

~  ! M 8: ES" bus O- MC NA OUT

' CLR0 IM a

5 C" ,

5 #

8 l

f

[

ir =  ;

7 i

i,  ;

f

/ r h \ /

~

t / /

t s l

]"

l, l l'

,, ,- i g d d. I -! l -

l l c to ra io ao so so n so so non iso iao tit'C tSECONOSI Figure 15: Nominal IANUS prediction of core temperature for the 75% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 75% power.

1 j

1000 i i i i i i i i i i i O NOMINAL CASE O CORE INLET a WORST CASE

900 - -

i x

~

i

] w _ _

l @ ~

4 l-

, 4 800 - -

i- m w

o. '- -

2

! w

,. 6-

! 700 - -

i I

C C C C D 600 O 20 40 60 80 10 0 120 TIME (seconds)

~

i Figure 16: Nominal and worst case IANUS prediction of core exit tempera- -

tures for the 100% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 100% power.

4 4

l 8

e.

o Q<

8 e

8 m . 1 S<

8 rg.

E E8 Oj.

8 3, .

o 8

m 8

b.00 40.00 80.00 120.00 160.00 700.00 240.C0 280.00 3?o.00 TISECl

(

Figure 17: Nominal DEM0-F prediction of the core exit temperature for :he average channel for the SS LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 5'; power.

k b .

k R.

r~

k e.

r-5-

k r{.

i a c8 N.h e

k R.

e k

e.

e

\ .

k 8 _. 4 9 .00 20 00 40 00 60.00 80 00 100 00 I?0 00 140 00 163 00 T I SEC 1 Figure 18: Nominal DEM0-F prediction of the core exit temperature for the average channel for the 35% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 35% power.

l

k g..

\.

H-4

g. ,

S o

I' 4

xp.

8 E4

- g.

4 E-

?

H-

?

e i .00 2U.00 ed.00 60.00 80.00 100 00 120.00 140.00 160 00 T(SEC1 Figure 19: flominal DEMO-F prediction of the core exit temperature for the average channel for the 75% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operai. ion at 75% power.

- 3 7-

4 2-8 DEM0F - Worst Case

~

2 g DEM0F - Nominal Case C

go f

ge 8

x:g. .

W gg '

~f- ,

f '

8 '

6 '

i r2 o -

o -

/

8 ,

p ..- _

"b.co 20.00 40 00 60.00 80.00 tm.co 120 m 140 N l'.1 N

, TISECl .

Figure 20: flominal and worst case DEtt0-F prediction of the core exit temperature for the average channel for the 100% LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 100% power.

As previously indicated, the margin between the nominal FLODISC predic-tion and that used in IANUS is nonexistent for steady-state conditions. There-fore, the verification of the conservatism inherent in the FLODISC calculation will rely heavily on the transient natural circulation tests. The predictions

. for the 100% test using the various models are sunmarized in Table 5. Note that the average F0TA AT is used in preference to the maximum F0TA AT because FLODISC does not realistically model intra-subassembly flow redistribution. In regions where there is a steep power gradient (e.g., the row 6 F0TA), FLODISC is expected to be very conservative (since it does not model crossflow or con-duction between channels), but this is almost irrelevant for verification of the statistical hot-channel where the power gradient is assumed to be flat.

The predicted hot-channel AT has been included in Tabie 5 for the purpose of benchmarking differences between the Project's calculations and the present calculations. It is honed that this will aid in establishing continuity be-tween the analyses used to support the FSAR and the ongoing verification pro-gram.

There are several features of the predictions summarized in Table 5 which bear some elaboration. First, it may be noticed that there is good agreement between HEDL's IANUS predictions and the present nominal IANUS predictions.

This does not mean that all features of the modeling are identical . It only neans that most of the differences (if there are any), " wash out" by the time the peak in temperature occurs. The few degrees difference in the results are probably due to the present use of nominal pipe and pump stopped rotor pressure drop (instead of the conservative default values). Some details of the models (pump trip time, reactivity insertion, etc.) will have important consequence on the short-term behavior but will have all but disappeared by the time transi-tion to natural circulation occurs. Specifically, the gap conductance of fresh fuel is likely to be much lower (particularly at low power levels) than the de-

. fault value which is characteristic of highly irradiated fuel. A low gap con-ductance increases stored energy in the fuel rods and affects the initial cool-down as seen in Figure 21, but after about 20 seconds there is no appreciable difference in the calculation.

It is also worthy of note that the first three columns of Table 5 are given for purposes of illustration and cannot be directly verified. The

Table 5 Comparison of Nominal and Worst Case Predictions of Maximum Temperatures for the FFTF 100% LOP Test After 25 Hours of Full Power.0peration 4

Average Assembly Core AT Average Average -

Outlet Hot Channel (Without Row 2 Row 6 Temperature aT Bypass) F0TA AT F0TA AT (K) (K) (K) (K) (K)

IANUS 796 246 162 - -

Nominal IANUS 835 305 200 - -

, Worst Case DEM0-F 793 240 159 - -

Nominal DEM0-F 845 255 210 - -

Worst Case FLODISC(1) 764 235 130 140 125 Nominal FLODISC(1) 767 260 133 165 160 Worst Case FLODISC(1) 785 246 151 159 145

] Steady-State 4

HEDL 798 232 164 N/A(2) N/A(2)

FLODISC HEDL 800 252 165 N/A(2) N/A(2)

IANUS 4

1 (1) Based on nominal core flow predictions and accounting for redistribution '

from bypass to core.

. (2) N/A - Not presently available.

4

._ _ _ - _ , ,.-..._____._.,z. . , - _ _ . _ . .

_e_-_.___. ._ _

_._.v.--- ____-,_.-_y ,

1 l

1 1000 i i i i i i i i i i i i o I ANUS NOMIN AL A I ANUS REDUCED GAP CONDUCTANCE 900 _

CORE INLET _

D 800 -

700 - -

c c c c o.

600 0 20 40 60 80 10 0 120 TIME (seconds)

Figure 21: Nominal IANUS prediction of the core exit temperature for the 100". LOP test compared to the identical prediction using limiting gap conductance.

core aT and resulting outlet temperature are effectively flow-weighted averages of all assemblies and will be difficult to interpret during the transient. The hot-channel AT is a worst case combination of variables and cannot be measured directly. Thus, verification of transient core temperature predictions will rely on the FLODISC predictions of F0TA AT's and/or FOTA flow rates. The F0TA ,

AT's have been used for illustration since they are expected to be more accur-ate, but this will depend on the final resolution of measurement uncertainties. -

The most significant feature of Table 5 is the radical difference between the FLODISC calculations. This difference shows up most dramatically in the core AT predictions. The BNL p"edictions of core AT range from 130 to 151 K while the FFTF Project predicts 164 K. This inconsistency in the predictions can be attributed to the difference in core models. The FFTF Project used a nine channel model of the core which did not include the bypass region. This treatment precludes redistribution from the bypass regions, a factor which greatly affects the overall prediction. The present FLODISC predictions used essentially the same 12 channel model which had been submitted in support of the flow dependent hot-channel ! actors with siight modifications to the flow and power of channels 2 and 7 to represent the F0TAs. Thus, redistribution from the unheated bypass regions causes the predicted core AT to be substan-tially lower than the IANUS calculation (with a fixed flow fraction). It should be made clear that while IANUS assumes a constant flow split (result-ing in a relatively high core AT), the hot-channel factor is based on flow re-distribution calculation and in that respect is inconsistent with the core AT calculation.

It is clear from Table 5, that there is a great diversity of predictions to be verified by the test program. The present nominal test predictions are much less conservative than the Project's test predictions, but the pres-ent calculations appear to be more consistent with the previous safety analy-ses.(1,7,8,9,20) This consistency seems to be necessary if the verification -

pr , gram is to have any relevance to the previous safety analyses.

HEDL has recently provided(22) more detailed test predictions for the 5% ~

transient test. In this case, HEDL has used CORA(23) instecd of FLODISC(5) to make individual subassembly calculations. Comparisons between the present predictions

• and the HEDL predictions are summarized in Table 6. Note that in this case the agreement between HEDL and BNL calculations is quite good but it is not clear whether this is due to a change in the HEDL model (from FLODISC to CORA) or due to the lack of beneficial flow redistribution at these extremely

. l ow fl ow rates. (For the 5% test the minirr m Reynolds number in the test as-semblies is predicted to be 500 and the tendency toward beneficial flow redis-tribution is negated by the relatively high frictional resistance of the fueled assemblies.) Without this beneficial redistribution, the active core AT pre-dicted by DEM0 and IANUS becomes nearly equal to the average core AT predicted by FLODISC (57 K).

I A recent change (transmitted informally) to the 35% power test will make it a valuable tool in verifying unbalanced loop performance. Present plans call for leaving one secondary pony motor on while the others are shut down.

This changes the thermal center of the affected IHX and causes the natural circulation in the affected primary loop to be substantially higher than that in the remaining two loops. Such flow imbalance is characteristic of the " con-trived" earthquake and tornado events (8) which lead to boiling in the core.

While the predicted flow imbalance (shown in Figure 22) is not as substantial as the imbalance predicted for some of the " contrived" cases (where negative loop flow can occur), the predicted flow in the affected loop is 30% greater than the unaffected loops. This difference in loop flow may be seyere enough to demonstrate whether unbalanced loop flow can produce asymmetric flow in the core (a possibility which is not accounted for by either FLODISC or CORA). In any case, this unbalanced test should provide valuable data for validating the transient IHX perfonnance.

The BNL predictions for the 35% test are summarized in Table 7. The cor-responding HEDL predictions are not yet available. Note that the unbalanced loop configuration has very little effect on the core temperatures in that the predicted temperatures would only be 4 to 6 K higher for a balanced loss-of-power test (not shown). The FOTA temperature data for the 35% test can be

• As of the report date, the 5% test has been run but the term " predictions" is still used since the results are not yet available.

_~

l TABLE 6 Comparison Between HEDL and BNL Nominal Predictions for the 5 Percent loss-of-Power Test in FFTF After One Hour of Prior Operation at 5 Percent Power HEDL CALCULATION BNL CALCULATION USING CORA USING FLODISC Peak Row 2 F0TA Coolant 616.8 617.6 Temperatures at Top of Fuel (K).

> Peak Row 6 F0TA Coolant 609.5 611.5 Temperatures at Top of i Fuel (K).

Minimum Flow Rate for 0.52 0.41 Row 2 F0TA (% of initial).

I' Minimum Flow Rate for 0.50 0.44 Row 6 FOTA (% of initial).

I I.

4 l

PLANT FLOW RATES NAT CIRC ON PRIMARY LOOP 35% POWER 75% FLOW CASE NUMBER 3040 i=tb 95 - LEGEND 9

O- PRIM FLOW O- PUMP-VES I e5 -- '- '

O- PUMP-VES P A- I SEC. FLOW 80 V- P SEC. FLOW 75 70 65-60 e 55 d .["

s

• =

6 cL 50 u Y' 45 10 1'

35 k

\\

30

" A\

X(

C w

lu -~

x N m ___ n .,

0 0 20 40 60 80 100 120 140 160 180 TIME (SECONOS)

Figure 22: Primary and secondary flow rates for the 35% unbalanced LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 35% power.

TABLE 7 Summary of BNL Test Predictions of the Maximum Temperatures for the 35% Unbalanced Loop Loss-of-Power Test in FFTF AVERAGE ASSEMBLY AVERAGE CORE AT R0W 2 R0W 6 0UTLET TEMP. (WITHOUT BYPASS) F0TA AT F0TA AT (K) (K) (K) (K)

IANUS-Nominal after 648 64 - -

1 hr Operation IANUS-Nominal after 675 91 - -

1 25 hrs Operation DEM0-Nominal af ter 677 92 - -

25 hrs Operation FLODISC(l) 640 57 64 55 Nominal FLODISC(l) 642 59 76 66 Worst Case i

j (1) Based on nominal core flow predictions for one hour operation and accounting for redistribution from bypass to core.

=

=

1 1

h i

1 expected to provide important validation of inherent natural circulation capa-bility before progressing to tests at higher power levels. Relative differ-ences in loop behavior will also help to characterize the IHX under off design conditions. The flow rates for the I-loop (representing the two unaffected loops) and the P-loop (representing the one loop with the secondary pony motor still on) are shown in Figure 22. The predicted IHX outlet temperatures for the primary loops are shown in Figure 23 and for the secondary loops are shown in Figure 24.

3.2.3 Uncertainty Analyses A thorough analysis of measurement uncertainty appears to be needed if the verification program is to succeed in identifying the capabilities and deficien-cies in the codes. Nominal and worst case models of the plant have been used to anticipate the range of behavior. The computer codes used for these predic-tions attempt to model the important phenomena affecting performance during the LOP event. However, it should be recognized that several unmodeled effects have been identified which may have significant effects for some of the postu-lated events. Special tests and measurements have been identified which will aid in establishing the significance of these unmodeled effects, but as with the rest of the test program, it will be important to compare the measurement accuracy to the range of anticipated behavior.

I l

IHX TEMPERATURES P LOOP NAT CIRC ON PRIMARY LOOP 35% POWER 757. FLOW 6so CASE NUMBER 3040 LEGEND sto -- O- Plitxso

'\ 0- P IHX PR 0

[ f \ ^

630 N

2 ,,

N -

O sis-O 610 b'

s" l2 sm N '

g sw g

s s90 -

ses sao-s7s-

L 570- 'x o 20 40 so ao im im tai iso iso Tit 1C (SECONOSI Figure 23: Primary and secondary IHX outlet temperatures for the P-loop (representing 3 loops with pony motors off) during the 35%

unbalanced LOP test in FFTF assumina 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation

, , at 35% power. . .

' ' l IHX TEMPERATURES I LOOP NRT CIRC ON PRitiflRY LOOP 35% POWER 75% FLOW

'" CASE NUMBER 3040

" U 615 LEGEND 6' O- I Dix S 0 6E T O- 1 IHX PR O 630 625 )

620 0

M 615 8

9 6:0 h

ge.-

it' 5 59s-

>=

590 585- .

Seu 575

\ t ~.

N 57U V_ n l l us ,

I40 160 100 0 20 10 60 80 100 120 TIME ISECONOSI Figure 24:

Primary and secondary (representing the 1 loop IHX with outlet temperatures a secondary ior the pony motor I-loop) running during the 35% unbalanced LOP test in FFTF assuming 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> of prior operation at 35% power.

4.0

SUMMARY

AND CONCLUSIONS DEM0-F, INIUS, and FLODISC have been used at Brookhaven to analyze the preliminary results of the FFTF natural circulation tests and to make predic-tions of the primary test series scheduled after the core is loaded. The com- ,

parisons of the BNL analysis with the data and previous HEDL calculations in-dicated that: .

(1) There is generally good agreement with nominal modeling for the steady-state tests.

(2) The BNL INIUS transient calculations followed the data trends, but were inaccurate in several respects, as discussed below.

(3) The pump coastdown model tended to underpredict secondary flow rate, but it has very little effect on the thennal transient for the secondary tests.

(4) BNL and HEnl calculations show substantial differences although they employ superficially identical modeling.

(5) The calculated results are very sensitive to the measured heat losses which appear to be dominated by air infiltration through leakage paths.

Some assessment needs to be made of the accuracy af these measurements and of whether the heat losses can change with time (i.e., whether dampers can become misaligned with usage).

The generally good agreement of the data with tne steady-state tests is reassuring with respect to predictive capability, but careful analysis of ex-perimental uncertainties needs to be perfonned in order to evaluate the uncer-tainty interval and/or the confidence level associated with the conservative (worst case) models. This is particularly true at low flow rates where the i data tends toward agreenent with the worst case models.

The main reason for inaccaracies in the IANUS transient thermal calcula- .

tions appears to be due to the fixed numberI7) of nodes in the DHX alcag with modeling more suitable to a counterflow heat exchanger. The FFTF actnally has ,

a four-pass, cross-flow dump heat exchanger. The errors in the transient pre-diction can be substantially reduced by changing to a more appropriate nodali-

ation. Even with improved modeling in the DHX (e.g., using air temperatures rept esentative of a four-pass cross flow heat exchanger), the flow response to the themal transient in the cold leg appears to be somewhat more rapid than the data indicate. These errors in response time are probably due to a com-bination of effects including: coarse nodalization in the DHX and cold leg; neglecting thermal capacity of the outlet manifold, pump, and pump tank vol- I umes; differences in pipe lengths from the various modules; stratification in various cold leg segments; and fictitious energy transport via numerical mixing (NMIX).

O While it is clea- that more detailed data comparisons (e.g., temperature at the pump and IHX inlet) are needed to deternine which of these effects domi-nate the transient behavior, the severity of the thennal transient is a direct result of the mismatch in thermal capacity between the air and sodium sides of the DHX. This mismatch is not nearly as severe for the primary system (with a sodium-to-sodium heat exchanger). Although the LOP from refueling conditions j is interesting for code validation (since it provides a good test of modeling capability), the predictive capability (or lack thereof) has limited safety implications. That is, the LOP from refueling conditions is essentially a dem-onstration test (for this event) and limited success in modeling the details of the DHX performance has very little impact on the initial response of the reac-tor core for the hypothetical LOP #com full power event (due to the large trans-port times involved).

The di fferences between the Brookhaven and HEDL calculations for both the primary and secondary tests point out the need for complete documentation of the models to be verified and their relation to the models used in previous safety analyses. The differences in modeling detail are highlighted by the comparisons to the data for the LOP from refueling conditions. The default IANUS model of the DHX demonstrated good qualitative agreement with the data and required only a. more realistic air temperature distribution (characteristic of the actual four-pass, cross-flow heat exchanger rather than the seven-pass

- model employed) to obtain reasonably good agreament with the non-linear per-formance characteristic of the unbalanced DHX.

REFERENCES

1. "FFTF Final Safety Analysis Report", HEDL-TI-75001, (December 1975).

2. U.S. Nuclear Regulatory Commission, "FFTF Safety Evaluation Report",

NUREG-0358, ( August 1978). .

3. HEDL, " Interim Summary of FFTF Natural Circulation Test Plans", sub-mitted to DPM/NRC for review, (September 1977). -

4. S.L. Additon, T.B. McCall and C.F. Wolfe, " Simulation of the Overall FFTF Plant Performance", HEDL TC-556, (March 1976).

5. J. Muraoka et al ., "FLODISC - A Dynamic Core Flow Distribution Code:

Evaluation of the Total Loss of Electrical Power Event", HEDL-TC-874, (May 1977).

6. T.R. Beaver, " Evaluation of Secondary HTS Transient Natural Circulation Test", interim report submitted to NRC for review, (June 1979).

7. S.L. Additon, A.E. Strait and C.F. Wolfe, "FFTF Natural Circulation Evaluation: Transition to Natural Circulation", HEDL-TC-557, ( April 1976).

8. S.L. Additon, E.A. Parziale and C.F. Wolfe, "FFTF Decay Heat Removal Analysis: Effects of Natural Phenomena", HEDL-TI-75222, (October 1976).

9. K.R. Perkins, R. A. Bari, L.C. Chen and D.C. Albright, " Analyses of FFTF System Transients", Informal Report, BNL-NUREG-25560, (January 1979).

10. R.R. Lowrie and W.J. Severson, "A Preliminary Evaluation of the CRBRP Natural Circulation Decay Heat Removal Capability", WARD-D-0132, (March 1976).

11. K.R. Perkins, R. A. Bari and D.C. Albright, " Uncertainties in the Cal-culated Response of the Clinch River Breeder Reactor During Natural ,

Circulation Decay Heat Removal", Informal Report, BNL-NUREG-22715,

( April 1977). .

12. Westinghouse, ARD, "LMFBR Demo Plant Simulation Model (DEM0)", Rev. 4, WARD-D-0D05, (January ,1976).

i l

l

13. Amendment 32 of the "CRBRP Preliminary Safety Analysis Report", Project Management Corporation, (docketed June 1975).

14. H. Domanus, R.C. Schmitt and W.T. Shaw, " Numerical Results Obtained from the Three-Dimensional, Transient, Single-Phase Version of the COMMIX Code", NUREG-0355, (October 1977).

15. K.E. Kasza, R.D. Schmitt and W.T. Sha, " Thermal Buoyancy Phenonena in a Horizontal Pipe During a Flow Coastdown Thennal-Hydraulic Transient",

ANS-CT-77-31, (September 1977).

16. K.E. Kasza, M.M. Chen and M.J. Binder, " Initial Considerations on the Influence of Thermal Buoyancy on Heat Exchanger Performance", Technical Memorandum, ANL-CT-47, ( August 1978).

17. S.L. Additon, " Presentation to the Advisory Committee on Reactor Safety",

(November 1977).

18. D.M. Turner, " Pretest Predictions for HTS Secondary Transient Natural Circulation Test", submitted to NRC/RSB for review, (May 1979).

19. A.Z. Frangos, "FFTF HTS /DHX Prototype ibdule Testing in SCTI, final Report", LMEC-76-4, (October 1976).

20. S.L. Additon, P. A. Edwards and W.L. Knecht, "FFTF Long Term Natural Circulation Decay Heat Removal Analysis", HEDL-TC-598, ( April 1976).

21. K.R. Perkins and R. A. Bari, "Interassembly Flow Redistribution at Natural Circulation Conditions in the Fast Flux Test Facility", Trans.

Am. Nucl . Soc. , 30, 413 (November 1978).

22. H.G. Johnson, " Pretest Predictions of the Thermal and Hydraulic Responses of the Fueled Open Test Assemblies to the 5 Percent Power Natural Circula-tion FFTF Plant Startup Test", HEDL-TC-1778, ( August 1980).

. 23. H.G. Johnson, "CORA - A Computer Code for Thermal and Hydraulic Coupling of Reactor Core Assemblies", HEDL-TC-1505, (September 1979).

9 O

l l,

I t

i I

e e l' l

l l

e e

l I

l l

l l \

l DISTRIBUTION LIST Dr. Raymond Alcouffe (1)

Los Alamos Scientific Laboratory

. Mail Stop 269 P. O. Box 1663 Los Alamos, New Mexico 87545 Mr. Harry Alter, Chief (1)

NE 563 RRT Division Department of Energy Washington, D. C. 20545 Dr. Robert Avery, Director (2)

Reactor Analysis & Safety Division Argonne National Laboratory 9700 South Cass Avenue Argonne, IL 60439 Mr. Marshall Berman (1)

Sandia Laboratories P. O. Box 5800 Albuquerque, New Mexico 87115 Mr. Robert M. Bernero, Director (1)

Division of Systems & Reliability Research Office of Nuclear Regulatory Research V. S. Nuclear Regulatory Commission Washington, D. C. 20555 Mr. Walter R. Butler, Chief (1)

Containment Systems Branch Division of Systems Integration Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Dr. L. W. Caffey, Director (1)

CRBR Plant Project Office Department of Energy P. O. Box U Oak Ridge, TN 37830 Dr. Ivan Catton (1)

University of California at Los Angeles Los Angeles, California 90024

Central Mail & Files (2)

U. S. Nuclear Regulatory Commission Public Document Room (LMFBR)

Washington, D. C. 20555 .

Mr. Paul S. Check, Assistant Director (1) for Plant Systems ,

Division of Systems Integration Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Dr. Robert T. Curtis, Chief (1)

Analytical Advanced Safety Technology Branch Division of Reactor Safety Research Office of Nuclear Regulatory Research U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Dr. William Davey (1)

Q Division Leader Mail Stop 561 Los Alamos Scientific Laboratory l P. O. Box 1663 I

Los Alamos, Nt1 87545 Mr. Harold R. Denton, Director (1)

Office of Nuclear Reactor Regulation l

U. S. Nuclear Regulatory Commission Washington. D. C. 20555 Director (1)

Division of Reactor Safety Research

Of fice of Nuclear Regulatory Research U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Director (1)

Fast Flux Test Facility Project Office Department of Energy -

P. O. Box 550 Richland, WA 99352 Dr. Raymond DiSalvo, Chief (1) -

Operational Safety Research Branch Division of Reactor Safety Research Office of Nuclear Regulatory Research U. S. Nuclear Regulatory Commission Washington, D. C. 20555

Mr. Darrell G. Eisenhut, Director (1)

Division of Licensing Office of Nuclear Reactor Regulation

. U. S. Nuclear Regulatory Commission Washington, D. C. 20555

, Mr. Malcolm L. Ernst, Assistant Director (1) for Technology Division of Safety Technology Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Mr. Frank Gavigan (1)

NE 563 RRT Division Department of Energy Washington, D. C. 20545 Mr. Stephen S. Hanauer, Director (1)

Division of Human Factors Safety Office of Nuclear Reactor Regulation  ;

U. S. Nuclear Regulatory Commission Washington, D. C. 20555 fir. Robert W. Houston, Chief (1)

Accident Evaluation Branch Division of Systems Integration Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission l Washington, D. C. 20555 )

Dr. Harry Humrel (1)

Applied Physics Division Argonne National Laboratory Building 208 9700 South Cass Avenue Argonne, IL 60439 Mr. William V. Johnston, Chief (1)

Core Performance Branch Division of Systems Integration Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555

I Dr. William Kastenberg (1)

Department of Chemical, Nuclear and Thermal Engineering University of California ,

at Los Angeles Los Angeles, California o0024 Dr. Charles H. Kelher, Assistant Director (1) for Advanced Reactor Safety Research Division of Reactor Safety Research Office of Nuclear Regulatory Research U. S. Nuclear Regulatory Conmission Washington, D. C. 20555 Mr. George W. Knighton, Chief (1)

Research A Standards Coordination Branch Division of Safety Technology Office of Huclear Reactor Regulation U. S. Nuclear Regulatory Conmission Washington, D. C. 20555 Mr. William E. Kreger, Assistant Director (1) for Radiation Protection Division of Systems Integration Office of Huclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Manager, Safety Engineering (1)

Hanford Engineering Develonment Labora tory P. O. Box 1470 Richland, WA 49352 Dr. James F. f' eyer (15)

Reactor Systens Branch Division of Systems Integration Office of Huclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Mr. Warren flinners, Technical Assistant (1)

Division of Safety Technoloay Office of Huclear Reactor Regulation .

U. S. Nuclear Regulatory Comnission Washington, n. C. 20555 Or. Robert flinoque, Director (1)

Office of Huclear Regulatory Research U. S. Nuclear Regulatory Commission Washington, D. C. 20555

l l Mr. Thomas Murley, Director (1)

Division of Safety Technology Office of Nuclear Reactor Regulation

, U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Mr. Thomas M. Novak, Assistant Director (1) for Operating Reactors Division of Licensing Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Dr. David Okrent (1)

Department of Chemical, Nuclear and Thennal Engineering University of California at Los Angeles

' os Angeles, California 90024 Mr. Denwood F. Ross, Director (1) i Division of Systems Integration Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Mr. Francis H. Rowsome, Acting Chief (1)

Systens Analysis Branch Division of Systems & Reliability Research Of fice of Nuclear Regulatory Research V. S. Nuclear Regulatory Commission Washington, D. C. 20555 Mr. Lester S. Rubenstein, Assistant Director (1) for Reactor Systems Division of Systems Integration Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Mr. Frank Schroeder, Assistant Director (1) for Generic Projects Division of Safety Technology 9ffice of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Secretary, Advisory Committee on (5)

Reactor Safeguards U. S. Nuclear Regulatory Commission l Washington, D. C. 20555 4

fir. Melvin Silberberg, Chief (1)

Experimental Safety Technology Branch Division of Reactor Safety Research Office of Nuclear Regulatory Research

• U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Dr. Themis P. Speis, Chief (1)

Reactor Systems Branch Division of Systems Integration Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Mr. Victor Stello, Director (1)

Office of Inspection & Enforce.ent U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Dr. Michael Stevenson (2)

Los Alamos Scientific Laboratory P. O. Box 1663 Los Alamos, NM 87545 fir. John F. Stolz, Chief (1)

Systens Interaction Branch Division of Systems Integration Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20565 Dr. David G. Swanson (1)

Applied Science Associates, Inc.

P. O. Box 214

~

Hawthorne, California 90250 Technical Information Center (2)

U. S. Nuclear Regulatory Commission P. O. Box 62 Oak Ridge, TN 37830 fir. Ashok Thadani, Chief (1)

Reliability & Risk Assessment Branch Division of Safety Technology .

Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Dr. Theo G. Theofanous (1) 132 Pathway Lane Lafayette, Indiana 47906

Dr. Long Sun Tong, Chief Scientist (1)

Assistant Director for Water Reactor Safety Research

, Division of Reactor Safety Research U. S. Nuclear Regulatory Commission Washington, P. C. 20555 Mr. Steven A. Varga, Chief (1)

Operating Reactors Branch No. 1 Division of Licensing Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Mr. Richard H. Vollmer, Director (1)

Division of Engineering

, Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D. C. 20555 Dr. J. V. Walker, Dept. Manager (1)

Reactor Research and Development 4

Sandia Laboratories P. O. Box 5800 Albuquerque, NM 87185 i

)

l BNL Distribution DNE Chairman (1)

DNE Deputy Chairman (1)

RSP Associate Chairmen (3) l Safety Evaluation Group (14) l Nuclear Safety Library (2)

S I

- - . _ , - - - . , _ -