ML18025B822

BWR Transient Analysis Model Utilizing Retran Program, Second Part

ML18025B822
Person / Time
Site:	Browns Ferry
Issue date:	07/23/1982
From:	TENNESSEE VALLEY AUTHORITY
To:
Shared Package
ML18025B821	List: ML18025B819 ML18025B822
References
TVA-TR81-01, TVA-TR81-01-01, TVA-TR81-1, TVA-TR81-1-1, TVA-TR81-2, NUDOCS 8207300243
Download: ML18025B822 (102)

Text

A system transient model for the Browns Ferry Nuclear Pl ant based on the RETRAN program is described.

The model Is appl icable to a wide range of transients but Is prlmarl ly Intended for analysis of the I lmltlng pressurization transients-considered for reload core I icenslng.The model is quallf led by comparisons to a range of startup test transients and to special turbine trip transients performed on a boll Ing water reactor of essentially Identical design as the Browns Ferry units.The results of a special NRC test problem with comparisons to other codes'alculations are also presented.

A representative appllcatlon of the model for licensing basis calculations of the limiting pressurization transients (based on Browns Ferry unit 3 at projected end of cycle 5 conditions) ls presented.

Results of extensive sensitivity studies are presented for the licensing basis calculations.

Two procedures for determining conservative critical power ratio limits from the model results are developed and their use in updating plant technical speclf lcatlons demonstrated.

0 S207300243 820723 PDR ADOCK 05000259 ,~P PDR k

The authors wish to acknowledge the assitance provided by many other Individuals In several TVA organizations during the course of development of this model.The assistance of J.Naser and L.Agee of EPRI and J.McFadden of Energy Incorporated Is also gratefully acknowledged.

Specfal thanks are due to Vlvlan Davis for her diligent efforts ln typing the manuscript.

The assistance of John.Strange ln performing some of the analyses in chapter 7 and Sheree Hutcherson In preparing many of the f lgures is also appreciated.

11

TABLE OF CONTENTS ABSTRACT~~~~~~~~~~~~~~~~~~~~~~~~~,~~ACKNOWLEDGEMENT.

~~~~~~~~~~~~~~~~~~~~~~o.....II 1~INTRODUCTION

~~~~~~~~>>~~~~~~~~,~~~~~~~~~~~.1 Purpose.1 1.2 Appl icable Events.1.3 The RETRAN Code.1.4 Thermal Limits Evaluation

..................4 1.5 The BFNP Model........................4 2.DESCRIPTION OF MODEL.~~~~~~~~~~~~~~~~~~~~~~~2.1 Introduction.

2.2 Model Geometry.2.2.1 Recirculation Loops.2.2.2 Feedwater and Steam Lines.2.2.3 Vessel Internals..2.2.4 Core Region.~~~~~~~~~~~~~~~~~~~~~~~~~~~~'1 17 18 19*2.3 System Components

.20 2.3.1 2.3.2 2.3.3 2.3;4 2.3.5 Reclrcul ation Pumps.Jet Pumps.Steam Separators

.Safety/Relief Valves.Core Hydraulics.

~~~~~~~~~=~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~,~~~~~~~~~~~~~~~~~~25 25 28 29 31 2.4 Trips and Control Models.32 2.4.1'.4.2 Trips.Controls~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~32 36 2.4.2.1 2.4.2.2 2.4.2.3 2.4.2.4 2.4.2.5 Sensed Inputs and Miscellaneous Cal Reactor Water Level Modeling.Feedwater Control System.Electro-Hydraulic Control System.Recirculation Control System.cu I at I ons.~~~~~~36 42 49 51 57 TABLE OF CONTENTS (Continued)

ZhGE 2.5 Condition Specific Inputs..................

63 2.5.1 One-Dlmenslonal Kinetics Data.............

63~2.5.1.1 2.5.1.2 2.5.1.3 Procedure for Generating Kl Verification of Procedures Kinetics Data.Generic Kinetics Inputs.netlcs Data....64 for Generating 68 72=2.5.2 Initial izatlon.~~~~77 3.COMPARISONS TO BFNP STARTUP TESTS~~~~.~...........

82 3.1 Feedwater Transients...

-..................

83 3.1.1 Level Change.3.1.2 Feedwater Turbine Trip.83 86 3.2 Reclrcul ation System Transients...............

93 3.2.1 Two-L'oop Trips.3.2.2 One M-G Trip.97 110 3.3 Pressurization Transient.

..................119 3.3.1 Sequence of Events 3.3.2 Model Comparisons to Measurements.

126 127 4.'OMPARISONS TO PEACH BOTTOM TURBINE TRIP TESTS.~~~~~~~~137 4.1 Peach Bottom Turbine Trip Tests...............'137 4.1.1 4.1.2 4.1.3 Test Objectives.

Test Description

.Reactor Description.

~~~~~~~~~~~~~~~~~~~~1-37 138 139 4.2 4.2.1 4.2.2 4.'2.3 4.3.4 RETRAN Pe'ach Bottom Model Description.

Modl f lcatlons to BFNP Model.Inli i al Conditions

.Model Inputs.~~~~~~~~~~~~~~~~~~~~140 142~142 145~~~~~~~~~~~~1 40 4.3 RETRAN Analyses of Peach Bottom Tests............147 4.3.1 4.3.2 4'.3 RETRAN Calculations.

Measured Data.Data Comparisons

.147 148 151 TABLE OF CONTENTS (Continued) 4.3.3.1 Pressure and Flow Comparisons

.4.3.3.2 Power and Reactlvlty Comparisons.

151 167 4.4 CPR Calculations.

~~~~~1 91 4.4.1 Calculations

.4.4.2 Measured (Inferred)

Delta-CPR.

4.4.3 Delta-CPR Comparisons.

~~~~~~~~~~~~~~~~~~~~~~~~~~~191 192 193 5.NRC TEST PROBLEM.~~~~~1 96 5.1 Description of Test Problem.196 5.2 Model Inputs.5.3 Results for Test Problem.~~~~~~~~~197~~~~~~~~~~~199 6.REPRESENTATIVE LICENSING BASIS ANALYSES~....~........217 6.1 L lcenslng Basis Inputs...................., 217 6.1.1 6.1.2 6.1.3 6.1.4 6.1.5'6.1.6 6.1.7 Core Exposure.Inltlal State Point.Scram React 1v I ty.Fuel Rod Gap Conductance Separator Inert l a.Equi pment Speci f l cat l ons Hot Channel Model ing..~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0~~~~~~~~~~~~~~~~~~~~~~~~~~-~~~~~~~~~~~~~~~~~~~~217 21 9 220.222.222.223.223 r 6.2 Generator Load.Rejection Without Bypass...........

225 6.2.1 Sequence of Events.6.2.2 Results of RETRAN Analysis.~~~~~~~~~~~~~~~~~~225 226 6.3 Feedwater Controller Failure.................238 6.3.1 Sequence of Events.6.3.2 Results of RETRAN Analysis.~~~~~~~~~~~~~~~~~~239 240 6.4 Main Steam Isolation Valve Closure with Flux Scram......

247 6.4.1 Sequence of Events..6.4.2 Results of RETRAN Analysis.~'~~~~~~~~~~~~~~~~.252 253 TABLE OF CONTENTS (Continued) 6.5 Summary of Transient Results.7.MODEL SENSITIVITY STUDIES.261 267 7.1 GLRWCB Sensitivity Studies..................268 7.1.1 Nuclear Model.....................268 7.1.1.1 7.1.1.2 7.1.1.3 7.1.1.4 Void Reactivity

.Scram Reactivity.

Doppler Reactlvlty.

Prompt Moderator Heating.~~268 273 273 274 7.1.2 Core Thermal-Hydraul lc Model lng............

274 7.1.2.1 7.1.2.2 7.1.2.3 7.1.2.4 Fuel Pin Model ing.Core Pressure Drop.Core and Core Bypass Modell Void Models..ng~~~~~~~~~~~~~275 276 276 278 7.1.3 Rec I rcu I at I on System Mode i..............

279 0, 7.2 7.1.3.1 Reclrcu I at 1 on Loop.7.1.3.2 Jet Pump Model.7.1.3.3 Steam Separator Model 7.1.4 Vessel and Steam Line Nodes.7.1.4.1 Vessel Nodes.7.1.4.2 Steam Line Model..7.1.5 Ml seel I aneous Sensitivity Results.FWCF Sensitivity Studies.~~~~~~~~280 280 281 282 282 283 285 286 7.2.1 7.2.2 7.2.3 7.2.4 7.2.5 Nuc I ear Mode I 0 I~~~~~~~~~Core Thermal-Hydraul lcs Model lng Reclrcul ation System Model Steam Line Model lng.Miscellaneous.

286 286 288 288 289 7.3 MSIVC Sensitivity Studies.......'..........'290 TABLE OF CONTENTS (Continued) 8.ALLOWANCES FOR MODEL UNCERTAINTIES.

.............,..294 8.1 Option A Operating Limit MCPR....,...........294 8.2'Option B Operating Limit MCPR................

296 8.2.1 Statistical-Process for Margin Eval uatl 8.2.2 Model Response Surface.8.2.3 Statistical Adjustment Factors.one~~~~296 297 303 8.3 Determination of Actual Operating Limit MCPR.........305

217 6e REPRESENTATIVE L I GENS ING BAS IS ANALYSES The three I iml ting pressurization trans I ents for reload I l censing analyses were identified ln chapter 1 as: the generator load rejection with failure of the turbine bypass system (GLRWOB), feedwater controller failure to maximum demand (FWCF), and closure of all main steam Isolation valves w Ith Indirect scram on high neutron flux (MSIVC).The basis for selecting these three events as Ilmltlng was given ln chapter I and discussed In reference 6-1.This chapter w lll describe the model Inputs and Initial conditions for I Icenslng basis analyses and Indicate how these Inputs compare to expected values.Representative results for each of the three events will be shown using a hypothetical Ilcenslng basis analysis for Browns Ferry unit 3 at conditions proJected for the end of its fifth operating cycle.6.1 The basic model util Ized for I lcenslng, analyses was described In chapter 2;,There are some conservative Inputs for the licensing basis analyses and conservative Initial conditions are employed.'able 6-1 shows the relationship of Ilcenslng basis model inputs and Initial conditions to the expected values.The"expected" values and conditions are meant only to show potential conservatlsms ln the I lcenslng basis modeling and not to define a practical"best estimate" model.6.1.1 The Ilcenslng analyses are performed at the maximum cycle exposure in the interval for which the analysis applies (e.g., BOC to EOC-2 GWD/MT, BOC Tab l e 6-1" Transient Model Inputs 8, initial Conditions Compared to Expected Valves Cycle Exposure Power/Exposure Distribution Initial Power ($NBR)Initial Steam Flow ($NBR)Initial Core Flow ($NBR)Initial Dome Pressure (psla)Feedwater Temperature Vessel to Rel ief Vlv Pressure Drop (psi)Vessel to Steam Header Pressure Drop (psl)Control Rod Initial Insertion Control Rod Motion CRD Scram Time (seconds to 20$insertion)

Scram Setpolnts..

Protect"lon

'System Logic Delay-(msec)Number of Rel lef Vlv's Rel lef Vlv Capacity Rel lef Vlv Setpolnt Relief Vlv Response (msec delay/msec stroke)Turb.Stop/Control Vlv Stroke Time (msec)Turb.Bypass Vlv Response (msec to 80$open)Reclrcu I ation Pump Trip Delay (msec)Recirculation Pump Coastdown Constant (sec)Flow Control Mode Controller Settings Separator Inertia Fuel Rod Gap Conductance Inside interval Nominal<100.<100.<Limiting value 1020<Max.value Nominal (<15)<42.Nominal pattern Rods at d I f f erent speeds Nominal (approximately 0.71)More conservative than tech.spec.Nominal (30)13 Nominal Nominal Nominal (300/100)Nominal (150/250)Nominal (200)Nominal (135)Nominal (4.0)Manual Nominal Spl lt between Inlet 8, exit Junctions Nominal, vary lng axially 4 during transient Max.value for interval being analyzed Conservati,ve target 104.5 105.0 Limiting value 1035 Max.value Max.(15)46.Minimum scram worth configuration All rods at same speed (conservative)

Tech.spec.upper I lmlt (0.90)Tech.spec.limiting value Max.(50)12 (one Inoperable)

With 0.9 ASME derate-Nomlnal+1$S I owest spec.(400/150)Fastest spec.(100/150)S I owest spec.(300)Maximum spec.(175)Conservatively slow (4.5)Manual Nominal Al I on inlet Junction Conservatively low, uniform axially and constant during transient 219 to EOC, etc.).As cycle exposure increases, the inventory of partially Inserted control rods ls reduced and this In turn decreases the rate of scram reactivity Insertion.

The reduction In scram reactivity Insertion rate Is the domln'ant phenomenon for pressurization transients so that the most severe results occur at the maximum cycle exposure..Near the end of an operating cycle when essentially all control rods are fully withdrawn from the core, the axial power dlstrlbutlon is controlled by the accumulated axial exposure dlstrlbutlon.

Since the axial power dlsirlbutlon affects the Initial rate of scram reactlvlty Insertion during a transient, the scram reactivity insertion rate ls influenced by ihe exposure dlstrlbutlon used ln the analysis.An exposure distribution ls utilized in the analyses which produces a conservative scram reactlvlty insertion rate relative to that of the expected exposure dlstributlon.

The target exposure distribution used ln the analyses ls normally that produced by the power-exposure Iteration (reference 6-2)or the so-called"Haling principle" dlstrlbutlon (reference 6-3).However, lf the plant operational strategy Is expected to result ln an exposure dlstrlbutlon more Ilmltlng than the Haling distribution, another target distribution conservative relative to the expected operation ls used ln the analyses.6.1.2 The initial power'" ln the model ls set consistent with the maximum steam, flow.capabl I lty of 105-percent NBR.A high value of Initial steam flow results ln a.more rapid pressurlzatlon and higher maximum pressures.'

maximum value of f'eedwater temperature is util Ized along with a nominal 0.2 percent steam carryunder from the separators.

The lnltlal reactor dome 220 pressure I s set at 1055 ps I a Whl ch I s conservat I vel y hi gh re I at I ve to normal plant operation.

The core flow Is Initial Ized at the maximum value expected to be utilized by the operating unit.This is normally the rated capacity of 102.5 mlb/hr.However, the recirculation system of the Browns Ferry units has the physical capability to produce core flows ln excess of the rated capacity at rated'power (up to approximately 105'ercent of rated flow).The use of the increased flow capability has substantial benefits In simplifying plant operations.

For cycles In which use will be made of the increased core flow capability the analyses will be performed for the limiting core flow value.6.1.5 The dominant conservatism in the I lcenslng basis modeling Is ln the representation of the rate of scram reactivity Insertion.

The Initial control rod conflguratlon Is selected to minimize The rate of scram reactivity insertion (1.e., the minimum use is made of partially inserted control rods consistent with maintaining the power dlstrlbutlon wlthln appl lcable operating I Imlts).An additional conservatism is inherent In ihe assumption that all control rods move at the same speed following scram..Use of a uniform speed for all control rods yields a slower Inltlal scram reactivity Insertion rate than achieved by a dlstrlbution of control rod speeds with the same average motion.The licensing analysis util lzes the control rod movement versus time following scram solenoid deenerglzatlon I Isted as the upper limit" conformance speclf lcatlon on average rod motion In the unit technical specif lcatlons (reference 6>>4).Table 6-2 shows the assumed rod motion 221 H Table 6-2 Technical Speclf lcatlon Upper Limit on Average Control Rod'otion After Deenerglzatlon of Pilot Valve Scram Solenolds Insertion 20 50 90 Time 0.375 0.900 2.000 3.500 222 fol lowing scram and has a large degree of conservatism relative to actual measured rod motion data.'6.1.4 The I Icenslng basis core model utilizes a conservatively low fuel rod gap conductance that Is uniform axIally and constant during the transient.

The actual gap conductance tends to be higher In the central areas of the core and the axial power shape tends to shift upwards In the core during pressurization transients increasing the importance of high gap conductance areas.The actual gap conductance ls also expected to increase during the transient due to fuel pellet expansion resulting ln a further conservatism In the licensing basis model.6.1.5 The ef f ect I ve f I ul d I nert I a of the separator Is determIned from manufacturer's data (reference 6-5)as a function of the separator Initial Inlet qual lty.For, best-estimate calculations the separator Inertia Is divided between the separator inlet Junction (125)and liquid exit Junction (141).The ca I cu I at I ons per f ormed f or compar I son to measured data presented ln chapters 3 and 4 used this best-estimate model lng.However, sens I t I v I ty stud I es I ndl cated that the peak trans I ent power and heat flux were insensitive to the, Inertia of Junction 141 but quite sensitive to the separator Inlet Junction (125)Inertia.Since the peak power and heat flux were increased ,for higher Junction 125 lnertlas, the licensing basis modeling places all of the separator inertia on Junction 125 to provide an additional margin of conservatism ln the calculation.

223 6.1.6 The model inputs for equipment performance (e.g., valves, protective system, control system, etc.)are chosen from a combination of expected performance data, conservative equipment design speci f lcatlons and plant technl ca I spec I f i cat Ion I iml ts.The p I ant control I er sett I ngs do not slgnlf lcantly affect the I lcenslng basis analyses of the I lmltlng pressur Ization events, therefore nominal plant values are employed.Conservative Inputs are employed for rel lef valve opening response and for closure rates for stop, control, and main steam Isolation valves.Reactor protection system setpolnts and delays are also conservatively set.6.1.7 The hot channel model described ln chapter 2 is employed to compute the varlatlon ln thermal-hydraulic conditions ln the limiting fuel bundle.The transient thermal-hydraulic data ls used ln evaluating the change ln critical power ratio (CPR)via the GEXL correlation (reference 6-6).A standard 1.4 peak design axial power distribution ls utilized ln the hot channel calculation.

The transient variatlon ln normalized bundle power is taken from the.RETRAN system model run w 1th 98 percent of the power deposited In the fuel rods and 2 percent deposited directly in the coolant.The initial hot bundle power and flow are determined as a function of initial CPR by using a steady-state thermal-hydraulics program with a i mul tip le, paral I el channel representation of the reactor core, hot fuel bundles, and core bypass paths (reference 6-7).The initial hot bundle power and flow are selected corresponding to an initial CPR (ICPR)which will result ln a minimum CPR during the transient within+0.02 of the safety-limit CPR (1.07).Table 6-3 shows the lnltlal conditions for the hot bundle calculation for Browns Ferry unit 3 at EOC 5.The limiting bundle utilized Is a GE pressurized 8 x 8 DR design bundle.

224 Tabl e 6-3 Hot-Channel Analyses Inltlal Condltlons E~m:tee Radial peaking factor 1.48 Axial peakfng factor R-factor Bundle power (mWt)Bundle flow (klb/hr)Upper plenum pressure (psla)Inlet enthalpy (Btu/Ib)Initial MCPR 1.40 1.051 6.259 107.6'1045.2 523.3 1.29 225 6.2 6.2.1 lt A loss of generator el ectr I ca I load from h I gh power cond l tl ons produces the approximate sequence of events shown below for the portion of the event Important for determining lf applicable fuel damage limits have been violated: a.Electric load ls lost and turbine-generator begins to accelerate (0.000 sec).b.The loss of load ls sensed by the power-load unbalance (PLU)device which lnltlates a turbine control valve fast closure to protect the turbine-generator from overspeedlng.

The Imbalance between power and load also generates a signal to open turbine bypass valves but failure to open Is assumed (approximately 0.005 sec)., c.Turbine control valve fast closure ls sensed by the reactor protection system which lnltlates a scram for power levels above 30-percent NBR (approximately 0.035 sec);'I d.Sensed fast control valve closure Initiates opening of breakers between reclrculatlon M-G sets and pump motors beglnnlng pump coastdown (approximately 0.180 sec).e..Pressure rises to the relief valve setpoints causing them to open and discharge Into suppression pool.Flow through the relief valve terminates the pressure increase and begins pressure reduction to the relief valve reclosure setpolnt (approximately, 1.4 to 7.0 sec).For the conservative assumptions utilized in the licensing basis analyses, the posltlve reactivity created by void collapse during the Initial-reactor vessel pressure rise Is sufflclent to overcome the negative 226 reactlvlty caused by scram for a short period of time resulting In an Increase ln reactor power.6.2.2 The analysis of the GLRW(8'or Browns Ferry unit 3 at proJected end of cycle 5 condlilons was performed with the RETRAN model described ln chapter 2 and the licensing basis Input as Identified In section 6.1.The fast closure of the control valve Is simulated by linearly decreasing the flow at fill Junction 340 (representing steam flow to the turbine)to zero at 0.075 seconds.This causes a rapid Increase in the pressure in the steam line near the turbine as shown In f lgure 6-1.The pressure disturbance propagates at the speed of sound back to the reactor vessel causing the large osclllations In vessel steam flow shown in flgu're 6-2.The large negative (I.e., back Into the vessel)portion of the vessel steam flow oscillation causes the very rapid pressurization of the reactor dome shown In figure 6-3.The short flat portions of the vessel pressure rise occur when the steam flow oscillation Is allowing large positive (I.e., out of vessel)flow rates.The delay"In the vessel pressure rise following control valve closure Is approximately 0.20 seconds and ls determined by the length of the steam lines.The pressures of the core inlet (vessel lower plenum)and core exit (upper plenum)are closely matched and follow the reactor dome pressure.Beyond approximately 0.35 seconds.the pressurization rate of the reactor core ls causing, a net Insertion of positive reactivity since the void reactivity Is suff lclent to'overcome the Initially very low scram reactivity insertion rate.As shown ln figure 6-5 Bt=3 COC5 LI CCNS INO BRSI S OLRWOB FIGURE 6-1 0.8 ZA.3 12 4.0 T I f16 (SEC)4.S BP'3 COC5 LI CFJlS IND BASE S OLRhJOB FIGURE 6-2 0.8 3,7 T I ME (SEC)4e0 4.8 BF3 COC5 l I CENS I NO'BRS I S OLRMOB FIGURE 6-3 lQ~Q o CV 4J (Q Q-o QJ~03 4J~O Q LLJ C)0.9 3.2 TIFF[SEC)4.0 4,Q o 04 BF3 CQC5 LlCENSING BRSIS GLRhlOB FIGURE 6-4 Ct w o~o CE o~o mQ LLI~D a (Q~n~~ld'Q~o~C4 X 4J o (0~N~QJ 0 I o bj~z UJ~o o bJ oG CKl~o 0.8 T I.HE i SE.C l 4~0 4.8 5.6 o eA Bt 3 EOC5 L I CEPlSI No BRSI S DLRhlOB FIGURE 6-5 L0.0.3.6 232 the net reaciiv I ty reaches a maximum of approximately

$0.72 at 0.615 seconds then begins to decrease as negative scram reactlv 1ty insertion rapidly increases.

The transient varlatlon ln, reactor power ls shown ln figure 6-6.The reactor power rises rapidly to a peak value of 393-percent NBR at 0.63 seconds then rapidly decreases as the scram reactivity terminates the excursion.

The behavior of the core average clad surface heat flux during the GLRWOB Is shown In figure 6-7.The initial pressure rise In the core causes a reduction In clad-to-coolant heat transfer due to the rise In saturation temperature of the liquid phase.The core average heat flux quickly turns around and begins to rise due to the'ncreased power generation and reaches a peak heat flux of 120.3 percent of the rated steady-state power value at 0.85 seconds then begins to decrease at a rate determined by the reduction In power and the fuel rod time constant.The core inlet and exit flow rates fn figure 6-8 show the compression and expansion oscillatlons excited by the steam line pressure wave.The magnitudes of the Initial core Inlet flow Increase and core exit flow decrease are Influenced by the lnertlas of the Jet pumps and steam separators ln addition to the size of the steam line pressure wave.The feedwater flow and narrow range (NR)sensed level behavior during the GLRWOB are shown In figures 6-9 and 6-10, respectively.

The feedwater flow Is initially reduced due to the reduced output of the feedwater turbines as the pressure increases.

The feedwater flow begins to Increase later in the transient as the pressure decreases and the controller demand Increases.

The NR level transient Is relatively mild with a reduotion of only 12 inches, leaving a large margin to the MSIV closure setpolnt.l' BF3 COC5 LTCCNSI'NO BRSI'S OLRMOB FIGURE 6-6 o.8 2.4 3 2 T'I NE (SEC)4.0 4 8 BF3 EOCS LI'CCNSI'NO BRSI'5 t LRWOB FIGURE 6-7 0.8 3%2 TINE f SEC)4.0 4 8 FIGURE 6-8 4 8 Bf 3 EOC5 L I CEPS I Nt BRS I'8 GLRWOB FIGURE 6-9 o.8 1~4 2.4 342 T'If1E (SfC)4 0 4.8 aF>COCA Lrccvsrvs sasrs oLe~oa FIGURE 6-10 0.8 3.2-Trne.[sec)4.0 4.g 238'he GLRWOB system transient'un.was used to prov I de the time dependent relative power plus, thermal-hydraul lc boundary conditions for the upper plenum, lower plenum, and core bypass volumes for the hot channel model Inftlallzed as discussed ln section 6.1.7.The hot-channel run produces time dependent thermal-hydraulic data which Is used with the GEXL correlation to compute the change In critical power rate's during the event.The minimum CPR calculated was 1.07 and since the Initial CPR was 1.29, a value of 0.22 for the hCPR for the GLRWOB was obtained.6.3 This event Is postulated on the basis of a single failure of a control device, speclf lcally one which can directly cause an Increase ln coolant Inventory by Increasing the feedwater flow.The most severe applicable event is a FWCF to a maximum demanded flow output.The peak pressure, power, and heat flux values are largest when the event Is initiated from maximum power and steam flow.However, the relative Increase ln power and heat flux may be larger at the lower end of the f I ow control range s I nce this generates a large Increase In feedwater flow and a correspondingly greater reduction ln Inlet subcool lng.The Improvement ln the Initial scram reactivity Insertion rate due to either the axial power shape shift towards the bottom of the core (for decreases of power along a flow control I inc)or due to initially inserted control.,rods (for operations below the maximum power flow lines)Is generally suff lclent to cause the reduced initial power FWCF operating I lmlt CPR to be bounded by that obtained for maximum power conditions.

Even neglecting the Improvement In scram reactlvl'ty insertion rate, the consequences of the FWCF at reduced flow operation ls conservatIvely bounded by the maximum p'ower results when corrected by'the applicable Kf curve (reference 6-4)for core flow bel"ow i".75 percent of rated.

239 I The change In hCPR for the FWCF event ls sl lghtly more severe for feedwater enthalples less than the maximum value as shown in chapter 7.To account for the potential si lghi nonconservatlsms In the I lcensing basis conditions for the FWCF a penalty of 0.03 ls added to the RETRAN model~CPR results.This penalty ls significantly larger than the potential changes due to uncerta I nt I es I n f eedwater temperature or due to I nl ti a I power I evel.6.3.1 The ana lysi s of the FWCF event Is based on the assumptions and sequence of events I lsted below.a.With reactor operating ln manual flow control mode (which results In most severe transient), feedwater controller Is assumed to fall to a maximum demanded output (0.0 sec).b.Feedwater turbines accelerate at maximum rate to maximum runout capability (approximately 3.0 sec).c.Excess in feedwater flow results ln an Increase ln core Inlet subcooling which In turn causes a rise ln core power (approximately 9.0 sec).d.Feedwater flow Increase creates a mismatch with steam flow which eventually Increases vessel water level to high water level turbine trip-setpolnt (15.5 sec).e.High water level causes tripping of feedwater pumps and turbine trip (15.5 sec).f.Turbine trip lnltlates reactor scram and closure of stop valves go begins pressure Increase (approximately 15.53 sec).Turbine trip signal Inltlates'pening of RPT breakers beginning pump coastdown (15.675 sec).

240 h.Increase ln steam1 Ine pressure causes turbine bypass valves to open (approximately 15.80 sec).I.Pressure rises to setpoint of relief valves which open, termlnatlng ihe pressure Increase and begins pressure reduction to relief valve reclosure pressure (approximately 17.15 to 20.0 sec).As can be seen from the above sequence of events, the FWCF evolves into a turbine trip with bypass event from a higher than initial power level and lower inlet temperature.

6.5.2 The FWCF event was.analyzed for the lnltlal conditions previously described and was initiated by setting the output of the feedwater controller to Its maximum output.The resulting feedwater flow Is shown ln figure 6-11.The Increase In feedwater flow decreases.

the average temperature In the mixing downcomer and after the transport time through the lower downcomer (approxlmateiy two-thirds of flow)and recirculation loops (approximately one-third of flow)causes an Increase ln the core Inlet subcool lng as In figure 6-12.The excess feedwater flow also causes the reactor water level to Increase as shown ln figure 6-15.The NR sensed water level reaches the high level turbine trip setpolnt at approximately 15.5 seconds causing a turbine trip.Closure of the stop valves causes the pressure to Increase.Part of the steam flow Is rel Ieved by opening of the turbine bypass valves (figure 6-14)but for high initial power levels the bypass capacity Is not suff Iclent to prevent further reactor pressure increases as shown In figure 6-15.For the conservative assumption of the I lcenslng basis analysis, the positive reactivity from core pressurization following the turbine trip Is Initially sufficient to overcome the negative scram reactivity Insertion (figure 6-16)and a rapid Increase In power BF3 EOCS LI GENS I NO BFIS I S F HCF FIGURE 6-11 Z.S S.O 7.S 10-0 T JOE (SEC l 12.5 l5.0 BF3 COCA LI CENSING BRS'IS FWCF FIGURE 6-12 2.5 5-0 V.S CO.0~Z.S T I f1E{SE'C J 15-0 BF3 EOCS LJCENSIHO BRSJS FhlCF 2-5 5.0 7.5 10-0 TI VE[SEC)12.5 15.0 17.5 za.o Bt 3 EOCS LICEHSINO BRSIS FOCI FIGURE 6-14 2.5 5.0 7.5 i0-0 T2.5 TIrlE (SECJ i5.0 BF'3 DOCS L!CENSLNO BHS!S FACY FIGURE 6-15 7.S 10-0 TI HE (SEC)tK.S 15-0 BF3 COCS LI CEHS I HO BFIS I S t 4ICF FIGURE 6-16 7.5 10.0 12,5 T J f1E'S E C J 15.0 17 5 247 occurs (figure 6-17).The increase ln reactor power causes an increase ln\core average heat flux as depleted ln figure 6-18.The pressure continues to Increase until the relief valves open and additional steam flow Is relieved to the torus (f.lgure 6-19).The closure of the stop valves generates an oscillation ln the vessel steam flow also shown In figure 6-19.The pressure wave excites the oscillation In core inlet and exit flows shown In f Igure 6-20.The overall reduction ln core flow near the end of the simulation ls due to the coastdown of the=recirculation pumps follow Ing the opening of the RPT breakers.The power excursion Is eventually terminated by the scrammed control rods and the pressure rise ls reversed by the turbine bypass and relief valves.Over the longer term (portion of the event not sfmulated) the reactor level will be reduced since the feedwaier pumps have been tripped and eventually the level will be mafntalned by the HPCI/RCIC systems.I r The hot-channel model was utilized with boundary conditions from the FWCF system run (as previously described for the GLRWCB)to determine the transient varlatlon ln critical power ratio.For the initial hot-channel conditions I lsted In table 6-3, a hCPR of 0.14 was obtained for the FWCF event and adjusted to 0.17 as described In section 6.3.6.4 The simultaneous closure of all main steam isolation valves with indirect scram on high power or flux (direct scram on MSIV position disabled)event was selected by the Browns Ferry Nuclear.Steam Supply System vendor as a conservative basis for analyz Ing compliance with ASME Boiler and Pressure Vessel Code for"upset" conditions.

The boiler and pressur'e vessel code defines four categories of conditions for overpressure protection system design: (1)normal, (2)upset, (3)'mergency, and BF3 EOCG L I CENS J Nt'RS I S EHCF FIGURE 6-17 2.5 5.0 75'100125.

TIVE (SECJ 15.0 Bf 3 EOCS L.I CENS I NO.5RS I S F'HCF FIGURE 6-18 2.5 S-0 7.5 t0-0 i2.5 T I HE (SEC J 1S.0 BF3 EOCS LICCNSIH&BRSIS F4CF'lGURE 6-19 CK Q3-0>g7~e C3 cr 4 4 4J O)>4 4J 4J D 2-S 5.0 7.5 10.0 TINE[SEC)12 5 t5.0 17.5 at=a cocG t rCENsIVo ORsLs f=WCP FlGURE 6-20 Z.S 7.5 10.0 12.5 TI5E{SEC)1S.O 252 I (4)faulted.The'ompl lance criteria for upset condltlons ls that the maximum vessel pressure not exceed 110 percent of the design pressure (for Browns Ferry 1.1 x 1250=1375 pslg).Based on the probabll lty of occurrence the MSIVC flux scram event could reasonably be placed ln the"emergency" condftlon category and thus provides a conservative basis for testing of compliance with upset condltlon I lmlts.The maximum pressures for emergency and faulted condltlons are 1500 and 1875 pslg, respectively and analyses by the NSSS vendor have previously established these tlmlts to be far less restrlctlve than the analysis of MSIVC flux scram event under upset condltlons.

The MSIVC with Indirect scram has a probabll lty of occurrence far below that consfdered for abnormal operational transients and thus ls not considered In determlnlng the operating I Imlt CPR.The MSIVC event with direct scram on valve posltlon has consequences bounded by the GLRWCB and thus analys.ls for each reload Is not required.6.4.1 The main steam Isolation valves on all four main steam I Ines are assumed to close s lmul taneous I y at the fastest rate allowed by plant technical speclf Icatlons (3 sec)and a conservative nonlinear valve closure characterlstlc ls assumed.WIth the direct scram on MSIV position.disabled the approxfmate sequence of events shown below occurs.a.Isolation tr,lp lnltlates closure of.MSIVs (0.0 secs).b.Sensed APRM signal reaches 120 percent of Inltlal value and lnltlates reactor scram (1.75 sec).c.Control rod motion begins and slows rate of Increase of power (2.04 sec).

253 d.Worth of scram reactivliy becomes larger than positive reactivity from void collapse and power Increase Is terminated (2.2 sec).e.Pressure reaches lowest setpolnt of relief valves and 3 of 4 In the group open (1 assumed failed).The remaining relief valve groups open as pressure reaches their setpolnts (2.82 sec).f.MSIVs are fully closed (3.0 sec).g.High pressure causes tripping of M-G s'ets and coastdown of M-Gs and pumps begins (3.28 sec).h.Maximum pressure ls reached In reactor vessel and pressure begins decreasing (approximately 3.9 sec).The times for many of the items in the above sequence of events apply to the RETRAN analysis presented In the next section and ihe times are dependent upon reload speclf lc klnetlcs data and setpolnts.

6.4.2 The steam flow rate through the closing MSIVs Is shown lnfigure 6-21 along with the rel lef valve flow.The highly nonl inear closure characterlstlc assumed for the MSIVs results ln the MSIV flow being largely shut off by 1.7 seconds.The rapid reduction ln MSIV flow causes a corresponding rise ln the steam I inc pressure near the MSIVs as shown ln figure 6-22.The steam flow at the reactor vessel and pressure rise In the vessel steam dome are shown In figures 6-23 and 6-24, respectively, The net (void+Doppler+scram)and scram reactivity components are shown ln figure 6-25.The maximum positive value of net reactlvliy was$0.7 I and occurred at 2.03 seconds.The power level varlatlon during the event ls I shown In figure 6-26 with the peak power of 476-percent NBR occurring at 2.22 seconds.The maximum val ue of core average heat f I ux was 135.5-percent NBR at 2.59 seconds as shown ln figure 6-27.

BF3 EOCS LICENSING ARSIS HSI VC FIGURE 6-21 o-8 TI NE (SEC J 4.8 CD tA C4 BF3 EQCS LL CENSING BRS!8 Ks!'Iti.FIGURE 6-22 CK~cn (Q w T 4J (Q~(Q 4J 0~CD~t I (Q CD Oo8 3 bi T!NE i SEC)4.0 BF3 COC5 LICENSING DRSIS NSIVC FIGURE 6-23 o.a 2,4 3.2 TI HE[SEC)4.0 4.$

" BF3 EOCS LICE'NS ING BRSIS HS I VC FIGURE 6-24 Oo8 2,%3.2 TI5E (SEC]4,0 BF3 DOCS LICENSING BRSJS MSIYC FIGURE 6-25 2.4 3,L TILDE.(SEC)4,O 4,Q BF3 EOCS LL CEHS I NO BRS l 8 AS I VC FIGURE 6.-.26 o.a Z,$3.2-T I APE ('SEC)4,0 BF3 DOCS.LICENSING BHSIS t1SIVC FIGURE 6-27 Oa8 3.2.4.0 T I f1E (SEC 3 4,Q 261 Because the cutoff of steam line flow for the MSIVC event Is not as rapid as for fast closure of the turbine control valves, the pressure wave'C that fs excited Is not as severe and results ln smaller core Inlet and exit flow oscillatlons (figure 6-28)than occurred for the GLRWCB.Because of the much slower coastdown of the pumps for a M-G trip ln comparison to opening of the RPT breakers, the overall reduction In core flow rate ls not as readily evident for the portion of the event shown ln figure 6-28.Figure 6-29 shows the behavior of the calculated feedwater flow during the MSIVC event.The reduction fn feedwater flow is caused by the reduced feedwaier pump output (at approximately constant speed)as the reactor pressure increases.

Later In the transient the feedwater flow Increases due to both ihe action of the feedwater controller to Increase the pump speed and to ihe reduction In reactor pressure.The behavior calculated for the narrow range sensed water level is shown ln f lgure 6-30.The Initial reduction ln water level Is primarily due to the collapse of voids Inside the core shroud increasing the mass of water ln that region and decreasing the level In the vessel downcomer (sensed by NR level Instru-ment)., The decrease ln level later ln the transient is due to the reduction In feedwater flow below the steam flow rate and thus decreasing vessel Inventory.

Shortly beyond the time scale of f lgure.6-30, the h feedwater flow w I I I increase to a rate higher than the steam flow and recovery of the level wll I begin.Over a longer period, the feedwater flow w I I I terminate due to loss of extract Ion steam to dr I ve the'eedwater turbines and level wll I be maintained by the HPCI and RCIC systems.6.5 The key transient simulation results for the three limiting pres-surization events are summarized In table 6-4.The value of aCPR ls given for the limiting P8 x 8R bundle for the GLRWCB and FWCF.

BF3 EOCG LICENSING BRSJS HSIY.C FIGURE 6-28 0-8 R.4 T I t1E (SEC)48 BI.3 EOCS L.ICE'CASINO BRSI S HS!VC FlGURE 6-29 O.S ZA TINE (SEC)4,a 4.8 BF3 EOCS L I CENSING BRS I S AS I.YC FIGURE 6-30 0.8 rrvc (sec)4.0 4-8 265 Table 6-4 Summary of Pressurlzatlon Transient Results Power (5 NBR)Core avg.heat flux ($NBR)Steam line presure (psla)Vessel pressure (psla)393.120.3 1212.1234.0.22 224.114;6 1176.121 4.0.17+476.135.3 1237.1276.n/a>>Includes 0.03 adder to account for potential ly nonconservatlve Inltlal conditions.

\

2666-1"TVA Reload Core Design and Analysis Methodology for the Browns Ferry Nuclear Plant," TVA-EG-047, January 1982.6-2 R.L.Crowther,"Burnup Analysis of Large Boiling Water Reactors," Proceedings of a Panel ln Vienna, Aprl I 1967 on Fuel Burnup Predictions ln Thermal Reactors, IAEA, Vienna (1968).6-3 R.K.Mallng,"Operating Strategy for Maintaining An Optimum Power Distribution Throughout Life," TID-7672 (1963).6-4 Browns Ferry Nuclear Plant Technical Speclf lcatlons, Unit 3, Tennessee Valley Authority.

6-5" R.B.Llnford,"Analytical Methods of Plant Transient Evaluations for the General Electric Boiling Water Reactor," NED0-10802, page 2-30, February 1973.6-6"General Electric Thermal Analysis Basis Data, Correlation, and Design Application," NED0-10958A, January 1977.6-7 A.F.Ansari, R.R.Gay, and B.J.Gltnick,"FIBWR-A Steady-State Core Flow Distribution Code for Boiling Water Reactors," EPRI NP-1923, July 1981.

267 7.MODEL SENSITIVITY STUDIES This chapter will present ihe results of a wide range of sensitivity (model perturbatlon) studies.The sensltlvlty studies fall into the general categories below: a.Those performed to assess the effect of modeling options and to verify reasonable functioning of models.b.Those performed to quantify the effect of model Inputs for which the value ls uncertain.

c.Those performed to quantify the effect of uncertainties In actual conditions In the operating plant.d.Those performed to quantify conservatlsms ln licensing basis model lng.e.Those performed to identify Ilmlting lnltlal condltlons for analyses.The base cases for all sensitivity studies are the licensing basis analyses for Browns Ferry unit 3 at projected end of cycle 5 conditions as presented ln chapter 6 and are typical of expected future operating cycles of all three Browns Ferry units..Sensitivity studies were performed for each of the three limiting'ressurization transients but the most extensive set of studies was performed for the GLRWOB event since It ls normally most limiting for crifical power ratio.Some of the sensitivity studies performed for the GLRWOB were repeated for the FWCF and MSIVC events to verify the appl I-cabillty of conclusions based on the GLRWOB for these transients.

In addition special sensltlvlty studies were made for the FWCF and MSIVC events for model options or Inputs exercised by these transients but not used by the GLRWCB event.

268 7.1 A summary of the sensitivity studies performed for the GLRWOB event ls presented In table 7-1.As discussed earl ler, this table includes analyses performed for several'different reasons and not al I of the perturbatfons In table 7-1 reflect uncertainties In the llcenslng basis RETRAN model for Browns Ferry.The table presents the change from the base case in maximum transient reactor power level (AQ), maximum core average fuel rod heat flux (~q), peak vessel steam dome pressure (A PySD), and the change I n the ratio of transient ACPR over initial CPR (b,RCPR).'7.1.1 The major uncertainty components related to the reactor core nuclear model are the three reactivity components (void, scram, and Doppler.)and the uncertainty In the prompt moderator heating.The uncertainty ln each of these components ls discussed and the model sensitivity described In the sections below.7.1.1.1 The uncertainty In the void reactivity coeff tclent inherent In the I-D kinetics model Is composed of four components:

(1)uncertainty ln the 3-D simulator void model;(2)uncertainty In the dependence of reactivity on water density In.the basic lattice physics code;(3)uncertainty,'In the transformation between 3-D and 1-D'water densities and;(4)uncertainty due to Inexact fitting of the collapsed 1-0 cross section to the polynomlnal forms used by RETRAN.The,f Irst two components reflect the uncertainty In the 3-D simulator calculation (reference 7-1)and the second two components represent the additional uncertainty In the 1-D representation.

Ie 7-1-Key Results of Sensitivity Studies for GLRWCB Transient 0<PVSD Void coefficient 13$more negative Scram reactivity reduced 104 Doppler coefficient reduced 10$Prompt moderator heating reduced 25$+149.9+23.2+6.6+10.0 5.53 1.33 0.40 1.29 4.9 3.3 0.8-1.5+0.030+0.012+0.003+0.007 Fuel pin radial nodes Increased 50$Fuel rod gap conductance increased 25$Fuel conductivity reduced 5$and heat capacity Increased 5$Increase core pressure'drop by 1.5.psf Redistribute 5$of core inlet pressure loss to core exit Increase active core nodes to 24 Reduce initial core bypass flow by 20$El lminate core bypass flow junction 109 Detailed nodlng of fuel channel conductor Use HEM ln thermal-hydraulics solution Reduce subcooled voids by 304 0.5 11.0 9.4 4.2 1.5 1.7 1.1 63.5 24.5 115.3 66.6 0.22 1.55 0.42 0.39 0.13 0.08 0.13 3.01 1.61 0.30 2.24 0.3 1.5 0.4 0.8 0.3 0.1 0.1 2.7 2.0 4'2.0+0.002" 0.015+0.009+0.003++0.002"+0.004++0.002+-0.015-0.013+0.009++0.012+Reduce reclrc pump head 10$Double recirc loop fluid Inertia Double jet pump fluid Inertia Jet pump M ratio Increased 7 C Jet pump N ratio Increased 10$.Jet pump head increased 10$No carryunder from separators Initial separator I lquld inventory reduced 25$Double-Inertia on separator junction 141 Best estimate separator Inertia modeling Reduced separator pressure drop by 0.5 psl Equilibrium separator model 5.8 9.8 60.9 48.0 0.5.6.8 3.9 0.7 5.8 95.4 2.8 15.7 0.31 0.68 2.10 2.30 0.03+0.35 0.16 0.03 0.24 2.48 0.27 0.66 0.3 0.9 0.6 2.3 0.0 0.3 0.4 0.2 0.2 0.3 0.4 1.3+0.002++0.005++0.005++0.011 0.000+0.002+0.000 0.000+0.001" 0.002-0.001-0.004 Table 7-1 (Continued)

Key Results of Sensitivity Studies for GLRWCB Transient wQ&%K~PVSD ABBE~hKE3 Increase Inertia of volumes 100, 180, 8, 190 by 20$Decrease steam dome volume by 5$Decrease upper downcomer volume by 5$Reduce steam I inc volume by 5$.Reduce steam line flow area by 5$Increase steam I Ine inertia by 7$Reduce steam I inc pressure drop by 10$Nominal rel ief valve model ing (capacity, setpoints, 8 delay)Increase steam I ine nodes to 11 1.7+21.9+'.6+.10.8+2.8+14;8+19.3 0.0 4.2 0.05+0.97+0.15+0.20+0.12+1.11+0.82 0.00 0.39+0.1+3.2+0.6+0.5 0.0+0.8+.0.5" 36.9 0.7+0.001+0.007++0.001+0.002"+0.001+0.009++0.005"+0.017" 0.003 Nominal scram solenoid delay Nominal fast TCV closure time Nominal RPT delay Nominal pump coastdown constant Initial Ized for 100$-tSR steam flow Ncmlnal measured scram speed Final feedwater enthalpy reduced 20 Btu/Ib Initiated on load line at reduced flow (approx.714 rated)20.8 50.9 17.0 6.9 48.7 99.7 14.0>>209.6 0.78 2.05 0.94 0.46 7.15 5.28+1.93-30.78 1.0 2.3 1.0+0.3-11.9>>14.6 0.0-32.1-0.005-0.012-0.005-0.003-0.006-0.055-0.012-0.006 aQ=perturbed case peak power (5 NBR)minus base case value hq=perturbed case peak core average heat flux ($NBR)minus base case value hPySD=perturbed case peak vessel dome pressure (psia)minus base case value hRCPR=perturbed case hCPR/ICPR minus base case value~Indicates Items Included ln determination of uncertainty In model RCPR 271 Comparison of the 3-D simulator to measured plant data as presented fn ref erence 7-2 I s usef ul In assur I ng that no gross b I as I n the 3-D slmul ator's void reactlv lty ca 1cul ation exists.However, such comparisons do not readily allow accurate quantification of the uncertainty.

Adequate measured data ls noi available to allow rigorous determination of the 3-D code's uncertainty therefore an estimate of the posstbl e uncertainty was devel oped, by examlnlng the analytic model s.The voids fn the 3-D simulator are calculated using the empirical CISE slip correlation (reference 7-3).The empirical parameters ln the CISE correl at I on were developed to min lmlze the dl f f erences In model calcul ations and experimental data for a wide range of conditions.

The standard deviation of the differences between measured void fractions and model calculations was II percent.To determine the uncertainty In the CISE model for void coeff lclents, revised empirical parameters

'were developed which maximized the void change for Increases ln pressure while maintaining a standard deviation of less than 11 percent ln void fraction from the orlglnal model for the range of data used ln developing the correlation.

Utilizing the Browns Ferry unit 3 EOC5 core and transient model.Initial conditions, the reactivity change associated with a pressure increase of 75 psi was evaluated with the 3-D simulator for the original CISE correlation parameters and those for maximum pressure coeff Iclent.These analyses Indicated a.difference of 5.3 percent In void reactivity which ls a measure of the uncertainty ln the 3-D simulator void model.The varlatlon ln nodal k for Instantaneous changes in void ln the 3-D simulator is based on tables computed by the TVA LATTICE program (reference 7-4).To estimate the uncertainty In void reactivity due to uncertainties in the calculation of the change ln k w Ith void, Lk(v)/k(40)3, computed by LATTICE, calculations were performed with the KENO Monte Carlo neutron 272 transport program (reference 7-5).The varlatlon In nodal reactivity with void changes was computed with KENO and the dl f f erence In k(v)/k(40) between LATTICE and KENO was developed for three fuel bundles.The effect of.these dl f f erences on void react 1v I ty was eval uated by apply ing the correction to each node ln the 3-0 simulator and evaluating the effect on reactivity for a 75 psl pressure increase.These analy'ses IndIcated an uncertainty of approximately 8.2 percent In void reactivity based on the differences between LATTICE and KENO for I nflnlte lattice"physics calcu-lations.The total 3-0 simulation voId reactivity uncertainty (CISE model and lattice physics data)was estimated to be 9.8 percent.The uncertainty ln the water density transformation between the 3-D and I-D codes is basically due to the uncertainty In the manner the water density perturbatlon from transient initial conditions Is distributed ln the radial plane.Table 2-7 yields an estimate of 6-percent uncertainty fn reactivity due to uncertainties In the radial dlstrlbutlon of water density perturbatlons.

The uncertainties due to errors ln fitting 1-D cross sections cannot be completely separated from the uncertainty ln 3-D to I-D water density transformation since these also result ln fitting errors.Based on a range of cross section flies developed for use with RETRAN, the combined reactivity uncertainty due to transformation and f Ittlng errors Is estimated to be less than 8.5 percent.Combining the I-D uncertainty (8.5 percent)with the uncertainty In the 3-0 simulator void reactivity (9.8 percent)results In'a total of 13-percent uncertainty In the void reactivity ln the I-D mode,l.The base case RETRAN cross section f I I e was modl f led to obta In approximately a 13-percent Increase (more negative)in void coeff iclent of reactivity.

Util izing the modified cross sections the GLRW(S transient was 273 rerun to obtain the sensitivity results for void reactivity shown In table 7-1.The peak power and heat flux are greatly Increased for the increased void coeff lclent case due to the relative closeness to prompt criticality.

A moderate Increase of 4.9 psl ln peak vessel dome pressure occurred.The 13-percent void coefficient change resulted In a 0.030 increase in RCPR which ls the largest component ln the model uncertainty.

Comparisons were made between LATTICE and KENO of the Inflnlte lattice control strength (change In kfor control rod Inserted to control rod withdrawn configurations).

The comparisons were made for several bundle designs and Inchannel void fractions.

The differences In control strength ranged between 0.5 and 4.8 percent.These comparisons confirm that an uncertainty In scram reactivity of 10 percent due to lattice physics uncertainties Is a conservative estimate.As noted ln section 6.1.3 there are several conservative assumptions employed In the I-D representation of scram reactlvlty which tend to offset any potential nonconservatlsms in the lattice physics data.The scram speeds were adjusted In the RETRAN model to achieve a 10 percent reduction ln scram reactivity during the pertinent part of the GLRW(B transient (I.e., before peak heat flux ls reached).The results of the RETRAN analysis.Indicate a moderate increase ln peak power, heat flux, dome pressure, and RCPR for the 10-percent decrease ln scram reactlvlty.

7.1.1.3 The Dopp I er react I v I ty ca I cu I at I ons by TVA'LATT I CE program were compared to He I I strand'exper lmenta I resonance I ntegra I data f or U-238 (references 7-6 and 7-7).A normal ization factor of 1.12 (appl led ln the 3-D simulator) was found to give excel lent agreement between LATTICE and 274 the Hei I strand data." The uncertainty In the Dopp I er react I v I ty was determined by examining the quoted uncertainties in the Hei I strand corre I at I on parameters (I o and 8)whi ch are est imai ed to contr I bute 9-percent uncertainty to the Doppler reactivity.

The uncertainty In the calculation of the average Increase ln fuel pin temperature during a pressurization transient was estimated to be less than 4 percent yielding a combined Doppler reactivity uncertainty of 10 percent.The base case RETRAN cross section file was modified to obtain a 10-percent reduction In the Doppler coefficient of reactivity.

The modlf led cross section file was then utilized to perform the GLRWOB analysis.As shown by the results ln table 7-1, a 10-percent reduction ln Doppler coeff lclent produces a slightly more severe transient but its effect Is small relative to the void and scram reactivity uncertainties.

7.1.1.4 In the RETRAN model for Browns Ferry, the fraction of power deposited directly In the moderator decreases approxlmateiy I Inearly with water density.For the water density dlstrlbutlon initially present for the base case the core average fraction of power deposited promptly ln the moderator was 0.019.Based on reference 7-8, the uncertainty In the prompt heating was assumed to be 25 percent and this reduction was made uniformly as a function'of water density.The RETRAN calculations with a 25-percent reduction ln prompt heating y lelded increases of 10-percent NBR ln peak power, 1.3-percent NSR in peakheat flux, 1.5 psl ln peak dome pressure, and 0.007 In RCPR.7.1.2 A range of core thermal-hydraul lc model lng sensltl v ltl es was Investigated Including nodal izatlon, irreversible pressure loss magnitude 275 and distrlbutlon, core bypass flow model lng and magnitude, effect of sl ip and subcooled voids, fuel channel conductor model lng, and fuel pin model ing and properties.

7.1.2.1 A 50-,percent increase ln radial nodal Ization (from 10 to 15 nodes ln pellet and from 4 to 6 in clad)was made In the fuel rod conductors to verify that the base model nodlng is adequate.Only.very minor differences from the base model were observed for the Increased nodlng indicating that the base case nodlng ls adequate.The core-average fuel.rod gap conductance used in the llcenslng basis model ls set conservatively low and ls speclf led as uniform axially and'onstant during the transient which further increases the conservatism.

To assess the amount of conservatism ln the gap conductance model Ing, a run was made with gap conductance Increased.

Ii was estimated that a 25-percent increase would be approximately equivalent to a best-estimate value and the effects of the expected axial and'ime variation ln gap conductance.

Best>>estimate gap conductance modeling was found to decrease the peak power by 11-percent NBR, increase peak core average heat flux by 1.55-percent NBR, decrease peak dome pressure by 1.5 psl and reduce RCPR by 0.015.Thus the licensing basis model Ing,of gap conductance yields a significant conservative blas in thermal limits.The uncertainties in U02 and Z Ircaloy properties (conductivity and speclf Ic heat)were estimated to be approximately 5 percent.The properties as a function of temperature were changed 5 percent each in the direction required to Increase the fuel rod time constant (reduced condu'ctivlty and Increased heat capacity).

As shown In table 7-1, the change in transient results with the modlf led fuel properties Is less than 0 276 the estimated conservative blas ln fuel rod gap conductance model lng.Therefore the overal I I lcenslng basis fuel rod model has a conservative bias and no addltlonal uncertainty penalty ls appropriate.

7.1.2.2 The uncerta I nty In the core pressure drop I s estimated to be less than 1.5 psi at design conditions.

Increasing the core pressure drop by 1.5 psl resulted In small increases (slightly more severe)ln all quantities.

It should be noted that changes In core pressure drop cannot be made without a corresponding change ln the driving head.In this study the core pressure drop decrease was balanced by reducing the frictional pressure losses In the Jet pump dlffuser.Alternate approaches would be to modify the head produced by the recirculation system.However, the uncertainties ln pump head and operating point are considered separately in this study.The distribution of pressure losses between core inlet, Internal, and exit areas for the Browns Ferry RETRAN model was developed to provide agreement with a program which performs detailed core thermal-hydraulic calculations based on empirical models verlfled against measured data..To assess ihe effect of uncertainties in the distribution of the pressure losses, the core Inlet pressure loss was reduced 5 percent with a corresponding increase In the core exit loss.The redistribution of pressure losses between core inlet and exit produced a slightly more severe result for the GLRWCB event (RCPR Increased 0.002).7.1.2.3 The adequacy of using twelve active core volumes and fuel rod conductors to provide water density and fuel temperature feedback to the 277 nuclear model was assessed by subdlvldlng the nodlng to obtain 24 active sections.The change ln GLRWCB transient key results were small for the Increased nodlng lndlcatlng that the base model core nodlng ls adequate.The uncertainty In the Initial core bypass flow fraction was conservatively estimated to be less than 20 percent.The Inltlal bypass flow was reduced by 20'percent which resulted In a sl lghtly more severe transient but the overal I effect was smal I.The base model utilizes two bypass paths.One path (junction 101)ls between the vessel lower plenum and core bypass volume such that the flow Is proportional to the presure difference across the core support plate.The second path (Junction 109)ls between ihe unheated core Inlet section and the bypass volume such that the flow ls proportional to the fuel channel wall pressure dlfferentlal.

The effect.of the bypass Junction 109 on the system response was evaluated by Inltlallz lng a deck with Junction 109 removed and the flow at Junction 101 Increased to maintain the Inltlal bypass-flow fraction.Utilizing this deck for the GLRWCB transient showed that removal of Junction 109 significantly reduced the severity of the event.The reason for the sensitivity was traced to differences in the active core inlet flow (junction 1).The flow through Junction 109 rapidly decreases during the initial pressurization forcing ihe active core Inlet flow higher than occurs lf Junction 109 is removed.The higher active core Inlet'flow augments the void coll apse caused by the pressure Increase further increasing The positive void reactivity insertion and producing a more severe transient than occurs without Junction 109.A portion of the bypass flow In the operating reactor Is dependent upon the fuel channel wall pressure differential; however, the amount varies slgnlf lcantly from bundle to bundle depending upon the bundle power.=An average power channel pressure differential as used ln the RETRAN model overestimates the 278 reduction In bypass flow fraction for a pressure increase relative to that which would be obtained from a multi-channel model with a distribution of bundle powers.Therefore, part of the difference observed when Junction 109 Is removed represents a conservative blas ln,the RETRAN model.A single lumped conductor (912)represents ihe fuel channels of all 764 bundles ln the core preserving the total surface area, channel volume, and thickness.

Conductor 912 ls bounded by the lumped bypass volume on one side and a mid-core volume (12)on the other.During a pressure Increase the thermal equllbrium assumption In RETRAN-02 causes the coolant temperature to increase resulting ln heat being transferred into conductor 912 and stored.Since this heat Is not available to produce voids the severity of the power rise Is Increased.

To examine the effect.of the simplified fuel channel conductor modeling, a deck was created with ihe core bypass volume subdivided Into an axial'stack of 12 volumes with 12 channel wall conductors each associated with a bypass volume on one side and the corresponding active core volume on the other side.The results of the GLRWOB transient with the more detailed fuel.channel conductor.

modeling were slgnlf Icantly less severe than the base model results.In fact the detailed conductor model results were comparable to those obtained by eliminating conductor 912 from the base model.Thus the slmpl lf led fuel channel conductor modeling Introduces a conservative blas.Into the~~I icens I ng bas I s model.'The RETRAN"Algebraic Sl lp" option Is employed In the TVA model.This option ls a drift flux model developed by EPRI (reference 7-9).To assess the effects of uncerta'Intles ln drift flux parameters on transient results an analysis was performed without sl Ip between the liquid and vapor phases 279 us I ng the RETRAN homogeneous equi I ibr I um mode I (HEM)as a bound I ng assumption.

The HEM assumption was also used ln making the transformation between 3-D.and i-D model water dens l ties for producing the I'-D cross section flic.The results of the GLRWCB event util izing the HEM assumption resulted In a much lower (115-percent NBR)peak power but no slgnlf leant change In peak heat flux.The peak dome pressure Increased 4.4 psi and RCPR was Increased by 0.009, representing one of the larger*model uncertainties.

Since RETRAN-02 assumes thermal equilibrium between the vapor and I lqul d phases (except In speci al separated volumes utl I lz lng a nonequl I lbrium model), subcooled voids are not directly treated.However, a profile fit subcooied void model developed by EPRI (reference 7-10)Is used to determine the water dens I tl es f or eval uat I ng the 1-D cross sections.The effects of uncertalntles In the subcooled void model were evaluated by performing an analysis without the profile fit model (I.e., A densli les were taken directly from RETRAN's thermal-hydraul lc solution for evalutlng the cross sections).

Since the uncertainty ln subcooled voids is estimated to be less than 30 percent the resulting changes from the base case were decreased by multiply lng by a 0.3 factor.The 30-percent reduction ln subcooled voids slgnlf'lcantiy increases all the key quantities I I lsted In table 7-1.I 7.1.3 I ncerta I nil es I n operat I ng cond 1 tl ons assoc I ated w 1 th the recirculation pump, loop piping, Jet pumps, and steam separators were estimated.

The effects of the uncertalntles and some modeling assumptions on the GLRNCB event were evaluated with sensitivity studies.

280 7.1.3.1 The pressure head produced by the recirculation pump operating at licensing basis conditions was reduced by 10 percent.A compensating reduction in the recirculation loop frictional pressure loss was made so that the Jet pump operating point was not changed.The reduction In recirculation pump head caused each of the four key quantities In table 7-1 to be slightly more severe.To account for the uncertainty In the effective fluid inertia ln the recirculation loop the inertia for volumes 200, 220, and 240 was doubled and the associated Junction lnertlas determined.

Doubling the recirculation loop fluid lnertlas produced a moderate Increase In transient severity.The sens I t I v I ty of model results to uncerta I nt I es I n the ef f ect I ve fluid Inertia associated with the Jet pumps was establ lshed by doubl lng the base case value of Jet pump inertia.The increased Inertia caused a 60.9-percent NBR decrease In peak power, 2.1-percent NBR decrease ln peak core average heat flux, a 0.9 psl increase ln peak dome pressure, and an increase of 0.005 ln RCPR.The reduced peak power and heat flux are caused by a reduction in the maximum core inlet flow during the initial pressurization.

The decreased inlet flow also offsets the decreased heat flux and Increases the value of RGPR.The effects of uncertalntles ln the initial operating point of the jet pumps were evaluated by Increasing (separately) the initial M ratio by 7 percent, the Initial N ratio by 30 percent, and the Jet pump head by 10 percent.These changes were accomplished by changing the irreversible pressure loss coeff Icients at Junctions 181, 240, and 260.The Increase In 281M ratio produced moderately more severe results for all of the, key quantities.

Increasing the Jet pump head produced only sl lghtly more severe results while the N ratio change had no appreciable effect.7.1.3.3 The effect of the steam carryunder fraction from the separators (0.2 percent ln base model)was examined by decreasing the core inlet enthalpy to allow Initialization for zero carryunder.

Initialization for zero steam carryunder was a minor perturbation and did not significantly affect any key result.A slmllfarly insignificant change occurred when the Initial liquid inventory ln the steam separators was decreased by 25 percent.In the I lcenslng basis model the fluid Inertia for the I lquld exit path (Junction 141)was changed" to the value obtained by dividing the separator height by the cross sectional flow area.When the inertia of Junction 141 ls doubled only slight Increases ln peak power and RCPR are , observed.As described ln"section 6.1.5 the effective fluid inertia of the steam separators from vendor test data Is applied at the separator inlet Junction (125)In the licensing basis model.In the"best estimate" modeling used for comparison to measured transient data In chapters 3 and'4 the test data separator Inertia was divided between Junctions 125 and 141.The conservatism in the licensing basis modeling was assessed by comparing the base case to the results with the inertia divided equally between the-separator inlet and exit., The best estimate separator Inertia modeling produced large reductions In peak power (95-percent NBR)and heat flux (2.5-percent NBR)but only small reductions In peak pressure (0.3 psi)and RCPR (0.002).

282 The pressure drop across The steam separator ln The TVA RETRAN model is Initial Ized to a value determined by the Inlet quality and flow rate using an equation developed from manufacturer's teststand results and shown to y leld conservatively high values.The sensitivity of GLRHOB transient results to the separator pressure drop was assessed by decreasing the value ln the base model by 0.5 psl.The 0.5 psl reduction In separator pressure drop slightly reduced the severity of the event.The Browns Ferry, model utilizes the RETRAN separator model with the state property solution which does not assume thermal equlllbrlum between the vapor and liquid phases ("nonequlllbrlum separator" model).RETRAN-02 also has available an equilibrium separator model which does assume thermal equilibrium ln the state property solution.The magnitude of the nonequll ibrlum effects ln the separator was tested by running a GLRl/OB trans lent ut1 I lz lng the equi I lbr I um separator model.Use of the equll Ibrlum separator reduced the severity of the transient as expected.7.1.4 The sections below pr'esent the sensitivity studies performed on the vessel nodes (primarily In the, dome and downcomer) and main steam line representations.

Most of the sensltlvity studies relate to uncertainties ln geometric data (volumes, areas, and lnertlas).

Since"as built" drawings were employed In geometric data calculations the uncertainties are small.One contributor to the geometric data uncertainty ls due to the need to base areas on the"stream tube" area ln one-dimensional thermal-hydraulic codes such as RETRAN.7.1.4.1 The fluid Inertia for vessel downcomer and lower plenum volumes (100, 180, and 190)ls low due to the large flow area ln relation to flow length.

283 The inertia of these volumes Is not expected to significantly affect the trans lent results and this was conf lrmed by Increasing the inertia by 20 percent for these volumes.The uncertainty In the base model volume for the vessel steam dome (170)and upper downcomer (160)was estimated to be 5 percent.Decreasing these volumes by 5 percent produced more severe transient results for the GLRWOB as expected.The effect of the reduction for volume'160 was slight but a 5-percent reduction for volume 170 produced significant Increases ln peak power and RCPR.7.1.4.2 Uncertainties ln steam I inc geometric data were Investigated by uniformly reducing the available fiuid volume and flow area In the steam I Inc by 5 percent.Each of the reductions caused slightly more severe results for the GLRWOB.A uniform 7-percent Increase ln the steam line fluid Inertia caused modest Increases In peak power, heat flux, and pressure but resulted fn a substantial (0.009)increase ln RCPR.The Increased steam line Inertia causes a longer period and higher amplitude pressUre wave ln the steam line.This has the effect of delay lng the core pressurization but making lt faster and more severe.The steam I Ine form loss coefficients ln the Browns Ferry RETRAN model were developed to provide a pressure drop between the vessel steam dome (170)and last steam line volume (340)which provides good agreement with measured data.The comparisons to pump trip transients presented ln chapter 3 indicate that excellent agreement ls obtained and Indicate a uncertainty of less than 10 percent.The steam I inc loss coeff lclents were changed uniformly to lower the pressure drop by 10 percent and the GLRWOB transient was reanalyzed.

The lower steam I Inc pressure drop results ln a 284 slightly greater pressurization rate ln the core and produces more severe transient results as shown In table 7-1.In the licensing basis model, the relief valve opening delay and stroke time are speclf led at maximum speclf lcatlon values (slower than expected).

Also the capacity Is set ln compliance with the ASME rating which Is less than the expected values, and the setpolnts are Increased by 1 percent over their nominal values to account for calibration uncertainty and drift.The licensing basis modeling of relief valves produces a conservatively high estimate of peak vessel pressure.However, when the RETRAN model was utilized with nominal modeling of the relief valves the calculated value of RCPR Increased for ihe GLRWCB event.The primary cause of the increase ln RCPR was traced to the earlier opening of the relief valves.When the relief valve initially opens there Is a momentary Increase ln the local flow rate near the exit of the core.This acceleration lasts for only a few tenths of a second but tends to decrease the CPR value.The nominal relief valve modeling causes this temporary decrease in CPR to reinforce the minimum CPR calculated due to longer term heat flux and core flow trends.Because of the brevity of the CPR decrease caused by the initial relief valve opening, It would not be associated with any fuel damage even lf incipient boiling transition Is calculated with a steady-state correlation.

Since several conservative biases In the licensing basis model have been Identlf led which combined are of substantially greater magnitude than the relief valve opening effect, no penalty or additional uncertainty to the licensing basis results Is warranted.

The main steam lines for the Browns Fer'ry units are approximately 260 f.feet long between the vessel and stop/control val ves.In the model the 285 steam I ines are divided into six volumes.In order to determine if the base model nodlng of the steam I lne Is adequate, a model with, 11 approxl-'ately equal length volumes was developed.

The results of the GLRWOB transient w'1th the 11-node steam I Ine were sl lghtl y I ess severe than the base model but no slgnif leant differences were observed.7.1.5 Severa I sens I t I v I ty runs were per f ormed to assess the degree of conservatism ln I lcenslng basis Inputs for scram solenoid delay, turbine control, valve closure time, recirculation pump trip delay, and recirculation pump coastdown rate.The conservatism due to the rapid closure'of the control valve relative to the expected rate is substantial and the combined conservatism In RCPR of these quantities alone ls approximately 0.014.The amount of conservatism ln the use of 105-percent NBR steam flow instead of the nominal 100 percent was evaluated In addition to the'difference between using nominal measured scram speeds and technical speci f Ication conformance I lmlt speeds.The I lcenslng basis deck Is lnltlal ized for the maximum capability final feedwater enthalpy.To demonstrate the conservatism In utilizing the maximum feedwater enthalpy a deck was Inltlallzed for 105-percent NBR steam flow but with the feedwaier enthalpy reduced.The sensitivity results In table 7-1 are for a reduction In enthalpy of 20 Btu/Ib which Is suff lclent to account for uncertainties ln the feedwater enthaipy and the'effect of'peration with a feedwater heater steam extraction I inc valved out.As exp'ected the reduced feedwater enthaipy results ln a milder transient demonstrating the conservatism ln using the maximum feedwater enthalpy for the GLRWCB transient.

286 The effect of operation at reduced core power and flow on the GLRWOB transient was investigated by'lnltiatlng the event from Initial conditions determined by reducing the recirculation pumps speed to obtain a load Ilne-point at approximately 71-percent core flow.All key GLRWOB transient results were less severe for the load line reduced power and flow case as indicated ln table 7-1.7.2 A summary of the sensltlvlty studies performed for the FNCF to maximum demand event ls presented In table 7-2.The 1 1st of studies presented for the FWCF Is not as extensive as for the GLRWOB event since the effect of most of the perturbatlons for the GLRWOB can be conservatively applied for the FWCF.A representative set of perturbed case results ls presented for the FNCF transient and the perturbatlons were made In the same manner as described for the GLRWOB event.Additional studies were performed on the FWCF transient for models and Input which uniquely affect the FWCF.7.2.1 As for the GLRWOB transient, the major uncertainty contributions result from the assumed 13-percent uncertainty In void coefficient and 10-percent uncertainty In scram reactivity.

Due to the less severe pressuri-zation for the FWCF event, the sensitivity of the key results ls approxi-mately one-half as large as'btained for the GLRWOB.The FWCF transient results were found to be insensitive to the 25-percent reduction ln prompt moderator heating.7.2.2 The lower power Increase for the FWCF event lessened the sensltlvlty to fuel rod gap conductance relative to the GLRNOB transient.

In fact, Tabl e 7-2 Key Results of Sensitivity-Studies for FWCF Transient E dq~me~PVSD M~GEB Void coefflclent 13$more negative Scram reactivity reduced 10$Prompt moderator heating reduced 25$+44.1+8.8 0.1+2.1+0.79+0.58 0.8+0.015+1.6+0.006+0.3 0.000 Fuel rod gap conductance increased 25$Reduce Initial core bypass flow by 20$El lmlnate core bypass flow Junction 109 Remove all passive conductors Use HEM in thermal-hydraulic solution Reduce subcooled voids by 304 3~7 4.8 18.8 14.1 40.1+25.8+0.33 0.28 1.00 0.99 1.93+1.15 0.5 0.5 1.0 1.2 2.3 0.2-0.004" 0.002" 0.008-0.006-0.009+0.008 Jet pump M ratio Increased 7$Jet pump head reduced 10$+7.6 1.0+0.54 0.06+0.4 0.000 0.1+0.001 Increase steam I inc inertia by 7$Reduce steam line pressure drop by 10$Nominal rel lef valve opening delay In'crease rated turbine bypass capacity 10$Nominal bypass'ervo time constants+3.0+11.3 0.0 9.1 37.2+0.16+0.51 0.00 0.53 2.03 0.1'.8-10.4'0.1 0.3" 0.001+0.005 0.000-0.003-0.023 Nominal RPT del ay Decrease maximum FW runout by 5$%R Inltlalized for 100$NBR steam flow Nominal measured scram speed Ffnal feedwaier enthalpy reduced 20 Btu/Ib Initiated on load line at reduced flow (approx.71$6.9+3.7 19.1 45.7+11.2 rated)-86.5 0.44 0.28 5.58 2.65+3.38" 28.7 0.5 0.1 5.8 6.4+1.3-26.7" 0.001-0.001+0.002-0.014+0.004--0.004 zQ~PVSD ZRCPR=perturbed case peak power (5 NBR)minus base case value=perturbed case peak core average heat flux (5%R)minus base case value=perturbed case peak vessel dome pressure (psla)minus base case value=perturbed case CPR/ICPR minus base case value 288 lower sensitivities to all perturbatlons to core thermal-hydraulic models were exhibited for the FWCF event.In general, the direction of change In a result caused by a given model perturbatlon was the same for FWCF and GLRWCB;however, due to differences ln timing of various phenomena, the reduction of core bypass flow by 20 percent and the use of ihe HEM thermal-hydraulic solution resulted ln less severe values of RCPR for the FWCF while both caused more severe results for the GLRWCB transient.

The major uncertainty component from core thermal-hydraulics model of the FWCF event (as for the GLRWCB)arose from the 30-percent reduction In subcooled voids.7.2.3 The 7 percent increase ln Jet pump M ratio slightly increased the peak values of power, heat flux, and vessel pressure but by slgnlf lcantly smaller amounts than for the GLRW(B transient.

The value'of RCPR was not appreciably affected by.the M ratio change.A 10-percent reduction ln Jet pump head caused very slightly less severe results for all key results except RCPR which increased by an Insignificant amount (0.001).7.2.4 Increasing the steam line inertia by 7 percent had an lnslgnlf leant effect on computed transient results for the FWCF.The Importance of the tlmlng.of various phenomena ls demonstrated by the fact that two key results were slightly more sever'e and two less severe for the FWCF but all four were more severe for the GLRWOB transient w 1th Increased steam line inertia.The change ln results caused by decreasing the steam I Ine pressure drop w'as of comparable magnitude to that shown by the GLRWCB with RCPR 289 increasing by 0.005.The use of a best-estimate rel lef valve opening delay I reduced the maximum pressure by 10 psl but did not affect power, heat flux, or RCPR.This ls In contrast to the nominal rel lef valve model lng results for the GLRWCB transient where the tlmlng of the valve opening was such that RCPR was Influenced for a short time.Additional sensitivity studies were performed for the FWCF event to assess sens I tl v I ty to turb I ne bypass mode I I ng.A 10-percent I ncrease (2.62-percent NBR steam flow)In rated turbine bypass flow capability produced a modest decrease ln all key results.The use of nominal turbine bypass servo delays and time constants greatly reduced the value of RCPR relative to the base case using licensing basis (upper I lmlt)time constants.

7.2.5 Only minor changes ln key results occurred for the FWCF event when a nominal recirculation pump trip delay was util Ized or when the maximum runout capability of.the feedwater pumps was decreased by 5-percent NBR to the nominal,value; I'he use of nominal measured scram speeds results ln a large decrease In severity of the FWCF event, as expected.The.reduction ln final feedwater enthalpy produces a slightly more severe transient than the Iicensln'g basis result Initiated'from maximum feedwater enthalpy because the'ame Increase In feedwater f,low following the controller failure results in a greater Increase ln core Inlet subcool lng.The Initiation of the FWCF event from reduced power may also result ln a slightly more severe transient since the amount-of Increase'n feedwater flow to maximum runout Is larger.The larger reduction ln core inlet subcoollng for a greater 290 f eedwater f I ow increase may be suf f I c I ent to overcome the I ess severe pressurization rate from reduced power y leldlng a slightly larger value of RCPR.For the base case used ln these studies (Browns Ferry unit 3 EOC5);1 however, this effect was not sufficient to yield a net Increase ln RCPR.For core flows less than 75 percent of rated the operating I lmlt CPR Is always Increased by multiplying by the Kf factor ln the unit technical specifications.

The Kf muitlpl ler Is computed to provide protection for a s I ow pump runout trans I ent and The I ncrease I n operat I ng I lml t CPR required by the Kf muitlpi ler ls slgnfflcantly larger than the FWCF event would-necessitate.

To account for any potential nonconservatlsm In The licensing basis FWCF analysis from 105-percent steam flow and maximum feedwater enthalpy, a 0.03 adder Is applied to the RETRAN results for the FWCF event as Indicated ln table 6-4.The 0.03 adder ls larger than any potential nonconservatlsms arising from reduced feedwater enthalpy or reduced power operations not covered by the Kf multlpiler.

7.3 Since main steam Isolation valve closure with Indirect scram on high power Is not a transient expected to occur during the life of a plant It Is not analyzed for meeting the safety-limit CPR and no RCPR sensitivity results will be presented.

The.primary purpose of the MSIVC event ls to demonstrate compliance with the 1375 pslg I lmlt on maximum vessel pressure, therefore the primary sensitivity result ls the change ln peak vessel lower plenum pressure.A summary of a representative sample of the sensitivity studies performed for the MSIVC event ls shown ln table 7-3.The manner ln which Table 7 Key Results of Sensitivity Studies for MSIYC (Flux Scram)Event hQ bq~me~PYL Lmll Vol d coe f f I c I ent 13$more negat I ve Scram reactivity reduced 10$.Doppler coefficient reduced 10$Prompt moderator heating reduced 25$+22.0+17.7+13.6+31.7+1'.05+1.14+0.75+2.64+2.4+3.4+1.3+1.9 Fuel rod gap conductance Increased 25$Reduce.Initial core bypass flow by 20$Eliminate core bypass flow Junction 109 Remove all passive conductors Use HEM ln thermal-hydraulic solution Reduce-subcooled voids by-30$-48.7+5.4-14.1-45.7+55.2+35.6+0.18+0.59-0.57-2.43+4.85+1.24 2.0+0.6 0.6+2.0+10.3+1.5.Double reclrc loop fluid Inertia Jet pump M ratio increased 7$Inreased M-G Inertlas by'5$202+29.2+0.1-0.06+1.56+0.01+0.2+2.0+0.2 Reduce steam I ine volume 5$.Increase steam line Inertia by 7$Nominal relief valve capacity+5.6 4.3 0.0+0.21+0.04 0.00+0.1~-, 0.6-10.8 Reduce FW flow pressure correction by 33$Initialized for 100$-NBR steam flow Nominal measure scram speed+1.4"+0.12-45.8-7.53-64.8.-4.31+2.0-14.5-13.7 hQ-=perturbed case peak power ($NBR)minus base case value hq=perturbed case peak core average heat flux (5 NBR)minus base case value APYLP=perturbed case peak vessel lower plenum pressure (psla)minus base case value 292 The model perturbatlons were introduced ls the same as was described for the GLRWCB transient.

The major effects on peak pressure were found to be due to the scram reactivity and slip (HEM)uncertainties.

However, the combined uncertainty of the nonconservative components was only 12 psl which Is of comparable magnitude to the conservatism ln the relief valve capacity used ln the licensing basis modeling and less than the conservatism due to the use of upper I lmit technical specification scram speeds.An additional large conservatism (not quantlf led in this study)ls due to the assumed MSIV closure characteristics which was very nonlinear allowing the flow area to be reduced by 99 percent after 1.7 seconds and the valves to be ful ly closed In the mlnlmum technical speclf ication time of 3 seconds.

293 7-1.S.L.Forkner, G.H.Merlwether, and T.D.Beu,"Three-Dimensional LWR Core Simulation Methods," TVA-TR78-03A, 1978.7-2 TVA-TR79-01A,"Verification of TVA Steady-State BWR Physics Methods," January 1979.7-3 A.Premol I, et al.,"An Empirical Correlation for Evaluating Two-Phase Mlxtur'e Density Under Adiabatic Conditions,".

An Informal Contribution to the Private Meeting, Heat Transfer Laboratory, Grenoble, France, 1970..7-4 B.L.Darneli, T.D.Beu, and G.W.Perry,"Methods for the Lattice Physics Analysis of LWRs," TVA-TR78-02A, 1978.7-5 L.M.Petrle and N.F.Cross,"KENO IV-An Improved Monte Carlo Criticality Program," ORNL-4938, 1975.7-6 E.Hellstrand,"Measurements of the Effective Resonance Integral ln Uranium Metal and Oxide ln Different Geometries," 1493-1509 (1957).~~~~n u 7-7'E.Hellstrand, P.Blomberg, and S., Horner,"The Temperature Coeff lcient of the Resonance integral for Uranium Metal-and Oxide," 497 (1960).7-8"Qualification of the One-Dimensional Core Transient Model for.Boiling Water Reactors," NED0-24154, Volume 1, pages Q5-1 to Q5-3, October 1978.7-9 G.S.'ellouche and B.A.Zolotar,"A Mechanistic Model for Predicting Two-Phase Void for Water ln Vertical Tubes, Channels, and Rod Bundles," EPRI (to be published).

7-10 B.A.Zolotar,"An Approximate Form of the EPRI-Void Formulation Model," EPRI (to be published).

294 8.ALLOWANCES FOR MODEL UNCERTAINTIES The procedure to be employed by TVA In determining updates to unit technical specifications relating to the operating I lmlt CPR for pressurization transients w lll be consistent w Ith the approach util Ized ln current technical speclf lcatlons'or the Browns Ferry units.The"deter-mlnlstlc" value of hCPR (or RCPR)from the RETRAN model for the GLRWCB and FWCF events with licensing basis Inputs (as described ln chapter 6)will be corrected to values which y leld 95-percent probability with 95-percent confidence (95/95)that the safety-I lmlt CPR (1.07)wll I not be violated by the event if Initiated at or above the operating I lmlt CPR determined by adding the adjusted hCPR to the safety limit.Two separate methods are util ized for determining the adjustment to the determlnlstlc dCPRs.These methods (referred to as"option A" and"option B")will be described In the sections to follow and the manner ln which they will be employed In updating the unit technical speclflcatlons described; 8.1 The option A approach takes no credit for the large conservatlsms In the I lcenslng basis models and Inputs which were demonstrated In chapter 7.The uncertainty In transient RCPR as computed by the model fs determined based on the sensltlvlty studies.The sensitivity study components In table 7-1 Indicated by an asterisk (~)ln the last column are considered applicable In setting the model uncertainty.

Table 8-1 shows the combined uncertainty for each major model component (In terms of RCPR)and the overall model uncertainty.

Since upper limit component and equipment 295 uncertainties were utilized ln the sensltlvlty studies it is reasonable to equate the 0.04I RCPR uncertainty as being an upper bound or 95/95 level.The approach utl I ized in apply ing the option A uncertainty ls shown by equation 8-1.Option A OLMCPR=1.041>>(SLCPR+hCPR)(8-1)Where SLCPR ls the safety-limit CPR value of 1.07 and ACPR fs the deter-minlstic value calculated by the RETRAN model for either the FWCF or GLRWCB transient.

Application of equation 8-1 to the RETRAN model deterministic ACPR values from table 6-4 results in operating limit MCPR values of 1.34 and 1.29 for the GLRWOB and FWCF events, respectively.

Tab l e 8-1 Components ln Browns Ferry RETRAN Model Uncertainty Uncertainty Nucl ear Model Core Thermal-Hydraul lcs Model ing Reel rcul ation System Model Vessel and Steam Line Nodes Combined uncertainty 0.033 0.016 0.013 0.041 296 8.2 The option A operating I lmlt MCPR Is very conservative In that no credit ls taken for conservatlsms Inherent In the I lcenslng basis analyses Including the s igni f leant,conservatlsms ln the use of technical speclflcatlon upper I lmlts on average control rod motion fol lowing scram and ihe use of 105-percent NBR Initial steam flow.The conservatlsms are compounded by the use of a model uncertainty penalty to the operating I imlt CPR.The option B method's an approach to reduce the unwarranted conservatism Introduced by compounding the uncertainties.

The option B approach utlllzes the conservatism Inherent ln the statlstlcal varlatlon of expected operating condltl.ons (for initial steam flow and scram speed)from the limiting conditions assumed ln Ilcenslng basis calculations to compensate for potential nonconservatlsms resulting from uncertainty In model predictions.

Statlstlcal convolution of Initial steam flow and scram speed uncertainties w 1th the model uncertainty was employed to determine statistical adjustment factors (SAFs)to the deterministic licensing basis RCPR value which malntaln a 95-percent probability (at 95-percent confidence level)that the safety-I Imit CPR wll I not be violated for the I lml tl ng pressur I zat I on trans I ents.The stat I st I ca I ad J ustment factor determination wli I be b~sed on the Browns Ferry unit 3 projected cycle 5 conditions but this cycle Is representative of expected operation for al I three Browns Ferry units and the SAFs are generically appi lcable to future operating cycles at a I I three Browns Ferry units.8.2.1 The ob Ject I ve of the stat I st I ca I eva I uat I on I s deve I opment of the probabli lty dlstrlbutlon for RCPR given the statlstlcai distribution of the 297 key transient Input variables.

The probabl I ity distribution of RCPR ls then utilized to obtain the value of RCPR which has 95-percent probablllty at 95-percent confidence of not being exceeded by the operating.

plant if the I Imitlng pressurization event occurs.The direct approach to developing the RCPR probability dlstrlbutlon would be to run trials with the RETRAN model with the key Inputs selected randomly from their uncertainty distribution.

However, the Monte Carlo approach requires a large number of trials to develop a precise probability dlstrlbutlon so direct simulation of each trial with the RETRAN model is impractical.

Instead a response surface ls constructed which predicts the RETRAN model calculated value of RCPR as a function of the value of the key inputs.The response surface Is developed by fitting model results to a polynomial with the key transient Inputs as Independent variables.

The advantage of I'he response surface is" that far fewer model calculations are required to'I develop an accurate response surface than to directly probab I I liy distr 1butlon on RCPR.deve I op the'I 8.2.2 The response surfaces used ln the analyses to be presented have the form shown ln equation 8-2.RCPR=(Ao+Ai"SF A2"SF Ag"SS A4"SF"SS A5"SS URS)"URM where, SF=random value of lnltlal steam flow (5 M3R)minus the nominal value SS='random value of time (seconds)to 20$scram insertion minus th'nominal value (8-2)URS=random response surface fitting error URM=random fractional uncertainty in RETRAN model predictions and the Al are fitting coefficients unique to each response surface.

298 In order to develop the f lttlng coefficients (Al)for the response surfaces 17 RETRAN simulations were performed for each one.The procedure used was an augmented variation of the factorial design process for three parameter levels (reference 8-1).Five values of each of the key input variables (steam flow and scram speed)over the approximate range of+4.5 standard deviations were utilized.A matrix of possible combinations of key inputs for model calculations with these values Is shown ln table 8-2 with combinations actually used denoted by an"X." The value of the'Input variable index ls used to denote the relative deviation of the variable value from Its expected or mean value.The 17 cases def Ined by table 8-2 were run for the GLRWOB and FWCF events ai end of cycle (all control rods withdrawn) and for an earlier point In the cycle for which some control rods were still lnltlally present ln the core.Tables 8-3 and 8-4 show the comparison of fitted response surface RCPR values to the RETRAN model calculation for the GLRWOB and FWCF events, respectively.

A standard least squares fitting technique was 1 used and the I ow order (six constant)equation resulted In very smal I fitting errors.The 95-percent confidence level.(upper bound)estimate of the standard deviation of the fitting errors was determined by use of chl-squared statistics.

The reliability of the response surface was also tested by developing ihe fitting constants util Izing only a portion of the data and then comparing to the error obtaining using all 17 points.The fitting coefficients were not significantly affected and the standard deviation of the f lttlng error was of a similar size as obtained when all 1.data po I nts were used I n the f It.The 95-percent conf I dence standard deviation (with a zero mean)for the flttlng error was employed to generate 299 Table 8-2 Matrix of Response Surface'uns Initial Steam Flow Index 0 CV'0 I Cl Cl Q W cO I 8 CI$4 O V X X X 4J 0 N 4+X X indicates RETRAN model calculations performed 300 Tabl e 8-3 Accuracy of Response Surface for GLRWOB ai End of Cycle Observation 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Model JKEB 0.11350 0.12510 0.11750 0.10830 0.09790 0.16990 0.14500 0.08340 0.05590 0.18450 0.06430 0.15490 0.04960 0.15410 0.09140 0.13800 0.07690 Fit JKEB 0.11397 0.12485 0.11974 0.10755 0.10048 0.17192 0.14288 0.08525 0.05666 0.18640~0.06395 0.15483 0.04678 0.14954 0.09012 0.13556 0.07973-0.00047 0.00025-0.00224 0.00075-0.00258-0.00202 0.00212-0.00185-0.00076-0.00190 0.00035 0.00007 , 0.00282 0.00456 0.00128 0.00244-0.00283 Average Difference

=0.00000 Standard Deviation=0.00214 95$Confidence S.D.=0.00303 301 Table 8-4 Accuracy of Response Surface for FWCF at End of Cycle Observat I on 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Model JKEB 0.09510 0.08810 0.09530 0.09530 0.09770 0.10940 0.10350 0.08420 0.07060 0.10460 0.06140 0.11050 0.07620 0;10320 0.08480 0.10200 0.08600 Flt JKE8 0.09549 0.08929 0.09293 0.09697 0.09737, 0.10934 0.10373 0.08465 0.07117 0.10538 0.06273 0.10898 0.07528 0.10173 0.08153 0.10465 0.08668 Average Difference Standard Devi,altlon 95$Confidence S.D.-0.00039-0.00119 0.00237"0.00167 0.00033 0.00006-0.00023-0.00045-0.00057>>0.00078-0.00133 0.00152 0.00092 0.00147 0.00327-0.00265-0.00068 0.00000 0.00151 0.00214 302 the values of the response surface uncertainty (variable URS in equation 8-2)by selecting randomly from a normal dlstrlbutlon for each trial evaluation of the response surface.Application of the N-test.(reference 8-2)to the fitting errors showed no basis to reject the assumption of x normal I ty.The trial values of lnltlal steam flow (variable SF ln equation 8-2)and time to 20-percent scram insertion (variable SS)were also assumed to be normally dlstrlbuted.

The mean-Initial steam flow was 100-percent NBR with a 2-percent NBR standard deviation (reference 8-3).'he mean time to 20-percent control rod Insertion fol I ow lng scram sol eno id deenerg1 zat1 on was assumed to be 0.71 seconds with a standard deviation of 0.053 seconds.These values are conservative relative to measured data for Browns Ferry and consistent w Ith option B scram time conformance testing ln current Browns Ferry Technical Speclflcatlons

<reference 8-4).Two measures of the uncerta I nty assoc i ated w I th pred I ctl ons of transient RCPR w 1th the Browns Ferry RETRAN model are aval I abl e.Neglecting any conservative biases the sensitivity studies In chapter 7 were employed to arrive at an estimated uncertainty In RCPR of 0.04I In section 8.1.Since the base value of RCPR for the GLRNG3 transient ls 0.17 the estimated model uncertainty (2 standard deviations) ls 24 percent.A second measure of the model uncertainty was obtained by'omparing the differences In RCPR between the normal Ized model calculations and those inferred from measured data for the Peach Bottom turbine trip test presented In table 4-12.Based on this comparison the model predictions conservatively overestimated RCPR by an average of 6.6 percent with a standard deviation of 2.6 percent.Using the chl-squared test these three data points yield a 95-percent confidence model uncertainty (2)of 23 percent, neglecting the conservative blas.

303 Based on the comparisons to Peach Bottom turbine trip test data and the sensi tlv I ty=study resul ts, the model uncertainty (URM var I abl e In equation 8-2)was conservatively assumed to be normal ly distributed with a mean of 1.0 and standard deviation of 0.125 (model uncertainty of 25 percent).8.2.3 The fitted response surface equation (8-2)along with the uncertainty distribution for each of the four input variables was evaluated for several hundred thousand trials.Each trial selects a random value for each of the four variables ln accordance with their assumed uncertainty distribution,.

and equation 8-2 Is evaluated to obtain the corresponding RCPR value.The maximum range of RCPR values was divided Into approximately 100,000 intervals to obtain a resolution of RCPR better than-10".A count was kept of the number of trials which resulted in an RCPR value In each interval, thus generating a=probability density function (PDF)for RCPR.To obtain the RCPR value which is greater than 95 percent of the'rials the PDF Is Integrated from the lowest interval up to the value at which 95 percent of'the.trials have been accumulated.

In general.to obtain the RCPR value for a prescribed probability (P)and one-sided confidence Interval (C)after N trials have been performed, the number of trials which must be accumulated (n)by the Integration ls: n=PN+g(C)>>CNP(l-P)l~(8-3) 304 where g(C)ls 1.645 for C equal to'5 percent.Equation 8-3 Is based on the normal distribution approximation of a blnomlnai dl strlbutlon (referen'ce 8-5)and has been shown to be very accurate for large values of N and C.After Integrating the PDF up to a value of RCPR such that n trials have been accumulated, RCPR (95/95), the I lcensing basis value of RCPR ls subtracted from RCPR (95/95)to obtain the statlstlcal adJustment factor.Table 8-5 shows the resulting SAFs for the GLRWOB and FWCF events of EOC and for a point with inltlai control rod insertion (MOC).Table 8-5 Yalues of Statlstlcal Adjustment Factors for Browns Ferry RETRAN Model GLRWOB at EOC GLRWCB at MOC FWCF at EOC FWCF at MOC-0.025-0.022+0.007" 0.002 The SAFs are used to adjust the deterministic licensing basis RCPR values by equation 8-4.l-CRCPR'+SAFE (8"4)The resulting option B operating I Imlt MCPR value based on the determin-istic values from table 6-4 Is 1.25 for both The GLRWOB and FWCF events at end of cycle 5 for Browns Ferry unit 3.

305 8.3 The overall operating limit for MCPR specified as a limiting condition for operation In a unit's technical specif lcations Is speclf led as a function of scram time (with adjustments based on core flow)based on the envel ope of maximum OLMCPR va I ues resu I t I ng from al I saf ety analyses.Since the primary difference between the option A and B OLMCPRs is due to the assumed scram speeds, the app I i cab I e I imlt I s determined by Interpolation between these I lmlts based on the actual average scram speed measured during the operating cycle.The Browns Ferry technical specifications have surveillance require-ments that all control rods be scram tested after each refuel Ing outage and 10 percent of the control rods at I6-week Intervals.

The surveillance testing data ls utilized'to compute the average scram time to 20-percent insertion from the fully withdraw position (Tavg): n<avg=Z<I/n 1=1 (8-5)where<I ls the 20-percent insertion time of rod I and n ls the total number of survell lance rod tests performed to date in the cycle Including the N active rods measured at beg Inning of cycl e.The interpolation between option A and B OLMCPRs ls based on the'ractional difference (v)of the average measured scram time (~avg)between the option A scram time (~A=0.90 seconds to 20-percent insertion which ls ihe upper conformance limit on average scram time)and the option B adJusted scram time.

306 T av-B T Tmaximum of 0 and A-B (8-6)w 1 th: T=P+i+65 I,n]G N where p Is the average time to 20-percent scram Insertion (0.71 seconds)utilized In the option B analyses and o Is the corresponding scram time standard deviation (0.053 seconds).The OLMCPR required by the GLRWCB and FWCF events by TVA analyses is shown In figure 8-1 as a function of T.Also shown on figure 8-1 are the OLMCPR values required by the nonpressurlzatlon transient safety analyses, in particular, the 100'F loss of feedwater heating (LOFWK)event, the control rod withdrawal error (RWE)at high power, and the rotated bundle error (RBE)analyses.The overall operating I Imlt MCPR (before correction for core flow)Is obtained as an upper envelope of all safety analysis results and Is Indicated by the solid line ln figure 8-1.

307 Figure 8-1*Hypothetical Browns Ferry Unit 3, Cycle 5, Operating Limit MCPR for GE P8 x 8R Fuel Bundles Based on TYA Methods I 1.36 I I I I I 1.34 I I I GLRWOB I I 1.32 I I I I'1.30 I I I I FWCF I 1.28 I I I RBE I I RUE 1.26 r I 1.24 I.I I I I 1.22 I I I I I 1.20 LOFWH 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 3088-1 Prentlss-Hall, Inc., 1977, pages 375-418.Irwin Miller and John E.Freund, 8"2 Gerald J.Hahn and Samuel S.Shapiro, John Wiley 8, Sons, Inc., 1967, pages 294-302.8-3 NRC Safety Evaluation for the General Electric Topical Report, Qual If Ication of the One-Dimensional Core Transient Model for Boiling Water Reactors, NEDO-24154 and NEDE-24154-P, Volumes I, II, and lll, June 1980.8-4-"Browns Ferry Nuclear Plant Unit 3 Technical Specifications," Tennessee Valley Authority.

8-5 Gerald J.Hahn and Samuel S.Shapiro, John Wiley 8, Sons, Inc., 1967, pages 144-171.