ML20151K992
| ML20151K992 | |
| Person / Time | |
|---|---|
| Site: | Haddam Neck File:Connecticut Yankee Atomic Power Co icon.png |
| Issue date: | 07/26/1988 |
| From: | Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML20151K977 | List: |
| References | |
| NUDOCS 8808030329 | |
| Download: ML20151K992 (24) | |
Text
,
e' * ,
.put auq'o UNITED STATES c
8 ' 'g'o NUCLEAR REGULATORY COMMISSION WASHINGTON, D. C. 20555 3 ,
\....+/
EfiCLOSURE 1 SAFETY EVAll!ATION BY THE OFFICE OF NUCLEAR REACT 0P _ REGULATION RELATING TO THE USE OF RETRAN COMPUTER CODE FOR SYSTEFS TRANSIENT ANALYSIS CONNECTICUT YAPTEE AT0t!!C POWER PLANT NOPTHEAST lfTILITIES DOCKET f:0. 50-213 1.0 INTPODUCTION The Northeast litilities (NUSCOT Topical Report 140-1, "PUSCo Thermal Hydraulic Vedel Oualification , Volume 1 (PFTPANI," was submitted to demonstrate NUSCO's technical corrpetence to use the PFTPAF computer code to perform systems transient analysis for the Paddam Neck Plant. This submittal was made in response to NDC Generic Letter 02-11.
The subiect topical report was submitted by NUSCO in a letter dated Auaust 1, 1984 The licensee's ob,4ectives were to demonstrate the adequacy o' the plant modeline technioues and the technical competency of the NUSCO staff to apply the RETRAN code and thereby obtain approval for the PETRAN models to be used for non-LOCA licensino analysis for the Haddarr Neck Plant.
In response to staff reovests for additional information, additional supporting material was provided in NUSCO's letters of Pay 7, 1987, November 19, 1987 and February 11, 1988. The staff has completed its review of the NUSCO submittels with technical assistance from its consultants, International Technical Services, Inc. (ITS). The staff evaluation is addressed below. A technical evaluation report prepared by ITS to support this m aiuation is attached.
8808030329 080726 3 DR ADOCK 0500
{
.. 'u
?
2.0 EVILUATIOP NUSCO developed a fcur-loop nodalization for the Haddam Neck Plant. The pressurizer was modeled as a non-eculibrium volure connected to a loop with a separate surge line volume. The POPV's and safety valves were modeled using RETRAN valve locic and the pressurizer spray as a positive fill with its conditiens tied to the cold leg conditions through the control system. The core whs represerted by one volure using point kinetics with one beat conductor, and a separate volurte was used to represent a by-pass region. The upper head volume was modeled as a single volume connected to the annulus ard the upper plenum, and two junctions were used to represent flow paths throupb the upper guide structure and upper annulus. The stear generator was modeled using two volumes to represent the primary side of the U-tubes, two plenum volumes (inlet and cutlet), one secondary side volume and six heat sisbs.
Nt!SCO has used the RETRANCP/P0003 code in the topical report. This version of the code, under staff revicv, has been updated to include error corrections previously identified in the P0002 version. Neverless, for the purpose of demonstrating the licensee's technical corpetence for using the PETRAN code, it is ecceptable to the staff to use MOD 03 of the code.
NUSCO selected the limitino transient and established a base case from each of the following major classes of transients:
- 1. Peactivity and power distribution anomalies
- 2. Decrease in reactor coolent system flov rate
- 3. Decrease in heat removal by the secondary system
- 4. Increase in heat renoval by the secondary system The topical included 5 sensitivity studies, with emphasis on stean line break and 9 corrparative studies. Our review examined:
a) the nodalization
3 b) the use of phenomenological models such as the bubble rise model in the steam generators and the non-equilibrium in the pressurizer model c) the user generated models such as the stebm generator, the reactor vessel upper head and the boren transport. In each case, NUSCO demonstrated its competence to prepare code input, its under-standing of the code and its models, and its understanding of the transients being examined.
3.0 CONCLUSION
The staff concludes that t'l'SCO has demonstrated its knowledge for the use cf the PFTPAN computer code. The tcpical report NUSCO-140-1 and supplemental information may be generally referenced with regard to nodali etion and phenomenological models to support future licensing subnittals 'or the Haddar Neck Plant. We fine that NUSCO bas demonstrated its technical competency to perform ncn-LOCA transient analysis usino tFe FETRAN computer code in compliance with Generic letter 83-11. The liter.see, however, is required for future licensing, to justify, on a transient-by-transient basis the folicwing:
a) nodalization, phenomenological model selection ard control
' systems for transients not analyzed in the terical report tll!SCO-140-1 or the supplemental information and, b)licensingconservatism.
Principal Contributor: D. Katze Date: July 26, 1988
. ITS/NRC/88-2 Technical Evaluation of NUSCo Tooical Reoort 140-1 "NUSCo Thermal Hydraulic Model Oualification Volume 1 (RETRAN)"
1.0 INTRODUCTION
The NUSCo Topical Report 140-1 entitled "NUSCo Thermal Hydraulic Model Qualification Volume 1 (RETRAN)" (1) was submitted to demonstrate tha technical competence which NUSCo has developed for performing transient systems analysis with the RETRAN computer code. NUSCo presented 5 sensitivity studies in the report, 9 comparative studies; 4 with FDSA/FSAR (2) and 5 with operational data. In addition, NUSCo performed some limited RETRAN model refinement studier for steam generator (SG) modeling for steam line break (SLB) and upper head modeling for natural circulation. A matrix of these transients is provided in Table 1.1.
The licensee's stated objectives for this topical were: (i) to demonstrate the adequacy of the plant modeling techniques and the competency of the NUSCo staff per Generic Letter 83-11 (3); and (ii) to obtain s safety evaluation report (SER) for the RETRAN (and VIPRE [4]) base models to be used for non-LOCA licensing analysis; so that NUSCo would have models in place for the Haddam Neck plant (HNP) which comply with current requirements, are consistent with the state-of-the-art, independent of fuel characteristics.
NUSCo stated that they intend to continuously maintain and upgrade their models. In addition, the topical report presented:
- 1. Sensitivity studies to assess the effects of modeling options and to verify reasonable functioning of these models.
- 2. Model refinements to incorporate results of special sensitivity l
1 i
. 1 studies and benchmark studies.
- 3. Benchmark studies against Facility Description and Safety Analysis (FDSA) computations to refine base models to be used for licensing analyses.
- 4. Additional benchmark studies against selected operational transients or actual plant data.
NUSCo selected the limiting transient from each of the four major classes of transients:
- 1. Reactivity and Power Distribution Anomalies (dropped rod),
- 2. Decrease in Reactor Coolant System Flowrate (loss of flow),
- 4. Increase in Heat Removal by the Secondary System (steam line break).
2.0 TECHNICAL EVALVATION This review examined: (i) the nodalization; (ii) the user selection of phenomenological models from among those programmed in the code; and (iii) the user generated input which models the plants control systems. Each of these is transient dependent. NUSCo has not addressed its control system modeling separately; however, this was evaluated by examination of the transient analysis (1,5,6,7]. In each case, NUSCo demonstrated its competence to prepare these code input, its understanding of the code and its models, and its understanding of the transients being examined. In that regard, this topical report can be used as a reference for future licensing analyses.
2.1 Nodalization NUSCo has developed a four-loop nodalization for the Haddam Neck Plant to provide maximum flexibility as a base case. The pressurizer was modeled as a non-equilibrium volume connected to a loop with a separate surge line volume.
2
The PORV's and safety valves were modeled using RETRAN valve logic and the pressurizer spray as a positive fill with its conditions tied to the cold leg conditions through the control system.
In the base case model used by NUSCo in this report, the core was represented by one volume using point kinetics with one heat conductor, and a separate volume was used to represent a by-pass region. The upper head volume was modeled as a single volume connected to the annulus and the upper plenum, and two junctions were used to represent flow paths through the upper guide structure and upper annulus. The steam generator was modeled using two volumes to represent the primary side of the U tubes, two plenum volumes (inlet and outlet), one secondary side volume and six heat slabs.
Reasont.ble justification, except to meet conservatisms required for licensing computations, of the base case nodalization through sensitivity studies was provided for its use in the SLB, loss of load, loss of FW, loss of flow, and dropped rod transients.
2.2 Phenomenoloaical Models The licensee justified its selection of phenomenological models through sensitivity studies and demonstrated the adequacy of the selected models, options, and values for each of five transients for which sensitivity studies were performed.
2.3 Control System Modelina The licensee has made extensive use of the control block in RETRAN to modil control systems; for example, modeling of boron transport in the SLB transient, computation of control rod cluster insertion rate for the 30% load rejection event, and cooldown rate of upper head temperature in the natural circulation test analysis. The licensee has therefore demonstrated its understanding of the control system and simulation of such systems with use of the RETRAN control block.
3 l
l
- - -_- , =_ - _ _ . . _ ,
2.4
- Sensitivity Studies The licensee has performed a series of sensitivity studies to justify model selection; particularly for the steam line break accident. NUSCo's particular selection of which parameters were varied in the sensitivity studies was reasonable. By this parametric approach, NUSCo was able to assemble useful information on identification of some dominant parameters for different transients and gain confidence in the base case model developed fer use in a wide range of transients.
2.4.1 Steam Line Break Sensitivity studies were performed with respect to (1) secondary noding, (2) primary U-tube noding, (3) phase separation, (4) vessel head modeling, and (5) plant parameters and options which may impact the computation of the reactor coolant system (RCS) minimum temperature.
Steam Generator Modelina NUSCo performed a series of sensitivity analyses examining: the number of U tube heat ccnductors; the number of nodes in the primary side of the U tube; the blowdown back pressure; use of the temperature transport option; use of various options for break flow modeling (such as los; coefficient variation); use of the vertical junction index; selection of the various optional flow equation types built into RETRAN; use of the enthalpy transport option and the two-phase friction option, including a separate node for and including friction losses of the surge line; and various secondary side heat transfer models. NUSCo found that; (1) Increasing the number of primary side U-tube volumes in the broken SG from 2 to 6 raised the minimum RCS temperature by 0.4*F, thus decreasing the return to power; (2) Use of the temperature transport option with 3 or 10 meshes raised the minimum RCS average temperature by 1.9'F and 2.1*F respectively; (3) Using the HEM break flow model raised the oinimum RCS temperature tj 0.2*F compared to 50ody and stretched the transient out 2 more seconds; 4
-(4 ) Deletion of the surge line node lowered the minimum RCS average temperature by 0.3*F and raised the minimum pressurizer pressure by 23 psi (these are competing effects since the lower temperature increases the reactivity insertion while the higher pressure reduces DNB); and (5) Using the low flow heat-transfer option instead of the forced convection correlation in the steam generator reduced the minimum RCS average temperature by 0.4'F.
In addition, NilSCo conducted a sensitivity study of the SG nodalization and model selection and fcund the following:
(6) Increasing the secondary side detail by using 3 nodes and using 8 nodes instead of one, in each case increased the minimum computed RCS temperature by 2.7'F. This was attributed, in both cases by NUSCo to the fact that the higher detail models had larger void fractions in the tube bundle region of the SG secondary and therefore reduced the primary to secondary heat transfer; and (7) Comparison of RETRAN prediction to the Containment System Experiment (CSE) experiments using RETRAN's dynamics slip model and various numerical values for bubble rise velocity led NUSCo to conclude that using bubble rise velocities between 1 and 2 ft/sec best matched the CSE tests. In addition, NUSCo compared computations for a ster.d- -
alone SG model between the RETRAN bubble rise model comput>tions and computations made with their 5-equation model NULAP computer code and found that: (a) pressure responses of the RETRAN model with bubble rise velocities varying from 0 to 3 ft/sec were miniaal, but an essentially infinite bubble rise velocity minimized the inventory depletion rate by assuring pure steam exited the break; and (b) reasonably good agreement was obtainer,between NULAP and RETRAN when RETRAN bubble rise velocities between 1 and 3 ft/see were used, and when the RETRAN dynamic slip model was used. However, in cases where NUSCo used multiple stacked nodes in RETRAN in the tube bundle region, pancaking (layers of water separated by layers of steam, an artifice ;f the code's computation) was observed.
5
_, FinaTly, NUSCo conducted a study, using the single node secondary model, of the impact of bubble rise velocity on the predicted SLB transient results and found:
(8) finite bubble rise velocities predicted more rapid depressurizations but did not depressurize as far as the infinite bubble rise model computations; and (9) similarly, finite bubble rise velocities predicted more rapid initial cooldown than an infinite bubble rise velocity, but the infinite bubble rise velocity permitted the removal of more total heat and therefore reached a lower RCS minimum temperature.
These results are caused by the fact that the infinite bubble rise velocity causes only pure steam to be released from the break, while use of a finite bubble rise velocity permits two phase flow out of the break thus permitting some secondary fluid to escape without absorbing the heat of vaporization.
Although NUSCo concluded "unrealistically large bubble rise velocities will significantly alter the plant response and should not be used", in their licensing submittal NUSCo elected to assume "perfect moisture separation in the steam generator" which they accomplished by using an infinite bubble rise velocity. This choice, in combination with a U-tube primary to secondary heat slab height which was chosen to be artificially low to assure that the tubes were treated as if always covered, maximized both the heat transfer rate and the total cool down and resulted in a conservative analysis.
These parametric studies of the SG modeling give us reasonable assurances that NUSCo has adequately investigated the impact of SG nodalization and model selection upon the computed transient results to be able to select nodalization and parameters leading to a conservative calculation and provide a solid example of the NUSCo staff's understanding of the code and the transient.
Primary Looo Modelina The primary loop model consists, in addition to the SG and vessel models, of one volume for each hot leg, SG inlet, SG outlet, loop seal, a pump and cold leg. The primary side of the SG is modeled as two 6
- vol unies . The pressurizer volume is connected to a separate surge line volume.
Sensitivity studies of the primary loop modeling to the RCS temperature during a SLB found that (i) removal of the surge line volume resulted in a 0.3'F lower minimum RCS average temperature and a 23 psi higher minimum pressurizer pressure than the base case which included the surge line volume and (ii) increasing the number of U-tube volumes in the broken SG from two to six raised the minimum RCS temperature by 0.4'F thus decreasing return to power.
Since neither of the parameters showed significant impact on the RCS temperature, in the model used for the licensing SLB, a surge line volume was modeled and the two volume SG U-tube model employed. Although small, the net effect of these choices is in the conservative direction.
Reactor Vessel Modelina An important factor in SLB computations is the method used for determining core fluid temperatures. Since the SLB is a relatively long transient which extends over several loop cycle times, cold fluid entering the reactor vessel from the affected SG near the beginning of the transient will pass through the core, be mixed to some degree with warmer fluid from the unaffected loops, pass around the SG loops again and return to the core before the transient is over. Thus the modeling of mixing in the reactor vessel and the fluid temperatures used by NUSCo in computing moderator temperature for use in the reactivity feedback are very important. NUSCo presented a SLB computation using their base case model, which has a single channel core and therefore results in perfect mixing of the fluid from the broken SG with that of the intact SG loops as they pass through the core.
NUSCo submitted additional analysis using a split core model [6,7] containing horizontal junctions joining otherwise separate core inlet plenum nodes and a similar junction joining the core outlet plenum nodes together with a bypass flow (4.5%) model which goes through a common bypass channel. In this submittal NU!Co demonstrated a thorough understanding of the RETRAN code and of the SLB transient.
In addition, since the SLB transient can empty the pressurizer causing the upper head to void and act like a pressurizer, modeling of the upper head is l also important for this transient since departure from nucleate boiling (DNB) 7
is a function of the pressure. NUSCo compared their RETRAN upper head modeling to Westinghouse predictions and to limited data obtained from a 1981 Natural Circulation test at HNP. Those comparisons were used by NUSCo te select flowrates between the upper plenum, annulus and upper head. NUSCo concluded that for natural circulation conditions, an upper head flow rate of between 2 and 10 lb/sec best agreed with the experimental data (which consisted of temperature measurements of the vessel head). They further concluded that with the reactor coolant pumps (RCPs) running, flow should go from the annulus into the upper head at a flowrate on the order of 50 lb/sec, and when RCP's are tripped the flow should reverse and steady out at natural circulation rate of approximately 2 lb/sec. NUSCo then computed with the RETRAN model that they obtained 40 lb/sec with the RCPs running and after trip the flow reversed and settled out at 2 lb/sec.
However, NUSCo pointed out that the standard RETRAN initialization routine will initialize the RC head as a Tcold plant which is inconsistent with their oelief that Haddam Neck is a Th ot plant. NUSCo cautions that for analyses where upper head voiding is important, the RC head must be initialized to the proper temperature and that this can be done by isolating the RC head prior to the start of the transient and opening valves after transient initiation.
NUSCo has informed us that within 2 seconds after numerical opening these valves a steady flow was computed and that initiation of the St.B computation begins 6 seconds after opening of the valves, and that therefore this does not caused any artificial transient. By this method, NUSCo's computation tends to initialize to the correct initial upper head temperature while preserving the recirculation rates, and thus would be expected to yield the proper pressure conditions for the DNBR computation done separately with VIPRE-01 based upon RETRAN output. This further demonstrates NUSCo's understanding of the behavior of the reactor during a complicated transient and of their use of the RETRAN code options to model.
Pressurizer Modelina The pressurizer was modeled as a non-equilibrium volume connected to a separate surge line volume. Adequacy of this modeling during an outsurge was demonstrated by comparison to plant data during an inadvertent actuation of the emergene; core cooling system (ECCS) which occurred on March 8
. 27, '1980 at Haddam Neck. During this event, the pressurizer pressure and level decreased initially in response to reactor trip and then increased because of the liquid delivery by the charging system. RETRAN underpredicted the pressurizer pressure by roughly 20 psi. Thus NUSCo has demonstrated that for the simulation of SLB, RETRAN is conservative in predicting the primary pressure.
Boron Transoort Model As required by the RETRAN SER [8), the boron transport model used by NUSCo during the SLB analysis was reviewed. Injection of borated water was assumed to prevent the reactor from returning to power after the reactor trip. NUSCo did not attempt to benchmark their modeling of this effect with plant data. The core boron concentration for the 4-loop case was calculated in the control system block in RETRAN using:
- 2. The flow rate in each of the RCS loops;
- 3. The boron concentration of the safety injection system;
- 4. The safety injection flow ra:e into each of the RCS loops; and
- 5. Transport times for the RCS water to pass from one locetion to another in the RCS. The transport tines are estimated based on the volumetric flow rates and the volumes of the nodes.
For the 3-loop case, a simplified model was used which assumed a uniform boron concentration in the RCS. The HPSI injection was delayed to account for the boron transport time.
Although NUSCo modeled the boron transport in a best-estimate approach, they did use it in a conservative fashion. During the SLB, for conservatism NUSCo assumed only one safety injection pump and only one charging pump available.
2.4.2 Loss of load NUSCo performed a sensitivity study to investigate the impact of the change in several parameters described in Table 2.1 on the peak RCS pressure. The base 9
l
m case'nodalization was used for a 100% loss of load transient in which AC power was assumed to be available throughout the transient and a scram signal was assumed to occur on the high pressurizer (PZR) pressure signal. A summary of the results from NUSCo's four cases is as follows:
Table 2.1 Loss of Load Sensitivity Study Results Parameter Amt of Change Peak RCS Pressure
- of Steam Line + Header Volume x5 -0.33 psi Temperature Transport Option +
- of meshes affected from 3 to 10 +3 pst PZR spray option + Deactivated Spray Flow vs on -24 psi PZR Rainout Velocity from 1 to 10 ft/s no change NUSCo reached the following conclusions from this study: (1) use of temperature transport option is not supported; (2) use of one volume to
- represent the steam header and the steam line is adequate;, (3) a pressurizer rainout velocity ranging from 1 ft/sec to 10 ft/sec is appropriate; and finally (4) modeling of the pressurizer sprays is important for predicting the RCS pressure.
2.4.3 Loss of Feedwater This sensitivity study was performed by NUSCo to examine the impact of certain parameters (described in Table 7,2) have on the pressurizer PORV actuation and SG dry out time. For this transiert analysis, it was assumed that there was a loss of offsite power (LOOP) and a failure of the AFW system. The reactor 10
scrammed as a results of LOOP, which also resulted in a loss of RCPs, pressurizer heaters, circulating water pumps and main FW pumps. The resalts of four parametric cases run are as follows:
Table 2.2 Loss of Feedwater Sensitivity Study Results Parameter Amt of Change PZR PORY SG Dry Out Time Actuation Secondary Side Flow Area +20% -3 see no change
- of U-Tube Heat Slabs + 100Y. +3 sec +5 sec
- of Primary U-Tube Nodes! x 3.0 -19 sec +5 sec U Tube Heat Slab Height +1 ft -58 sec +15 sec
-1 ft +130 see -80 sec NUSCO found that when the number of primary U tube node volumes was increased from two to six, a substantial change in the natural circulation flow rate was observed at 1000 seconds; flowrate decreased by BY. and loop delta T increased by a corresponding amount.
NUSCo concluded that the nominal values for secondary side flow area vi i secondary side heat slab height should be used. Use of three heat slabs per stack was found to be acceptable when using the local condition heat transfer option. Two primary U-tube nodes per SG was also found to be acceptable.
However, NUSCo concluded that when natural circulation flows are important, l six primary U-tube nodes will be used to ensure conservatism.
2.4.4 Loss of Flow NUSCo investigated the impact of changes in certain parameters (described in Table 2.3) upon the minimum DNBR. The transient was initiated by a less of 11 l
l l
E.
, power to all RCPs. The scram signal was generated due to a drop in the RCS flowrate to 85% of its nominal value. Offside AC power was assumed available.
A control block was created to compute the HDNBR during the transient using the W-3 correlation. Results from five parametric cases are as follows:
Table 2.3 Loss of Flow Sensitivity Study Parameter Amt of Change MDNBR Timing Pump Inertia +20% +0.1%
-20% -0.3% 0.4 see Junction Inertia +20% no change 0.1 see
-20% no change Rod Insertion Tima +20% -0.1% 0.1 sec
-20% +0.1%
Scram Signal +1 sec 2.4%
Form Loss Coefficient +20% 0.2%
-20% +0.2%
NUSCo reached the following conclusions from the sensitivity studies: (1) each of pump inertias, junction inertias and RCS loss coefficients has minimal effect on the results; (2) a mild sensitivity to rod insertion time and a large sensitivity to scram delay time was found.
2.4.5 Droceed Rod NUSCo conducted a series of parametric studies te ?etermine the impact of certain parameters (described in Table 2.4) upon the minimum core power and fuel temperature in a dropped rod transient. The dropped rod was assumed to 12 l
have'a worth of 40 cents, the reactivity was ~ assumed to be inserted linearly over a 2.4 second period, and it was assumed that no scram signal was generated. The turbine runback feature was assumed to be inoperable, but credit was taken for the rod block signal to prevent automatic rod withdrawal.
Because of the nature of this transient parameters related to core conductor properties were selected.
Table 2.4 Droppeo Rod Sensitivity Study Results Parameter Amt of Change Fuel Temperature Core Power Clad Thermal Conductivity +20% -4'F
-20% +7'F Fuel Meshes -50% -5'F
+50% +1'F Clad Meshes 50% -2*F
+150% +1'F Fuel Gap size +20% +46'F -3MW
-f'. -39'F +9MW Fuel Gap Conductivity +20% -39'F +9MW
-20% +58'F -4MW
- Fuel Thermal Conductivity +20% -40*F +12MW
-20% +94*F +3MW i
NUSCo concluded that a moderate effect was obtained when fuel rod gap, gap conductivity, fuel thermal conductivity and fuel heat capacity were varied, and therefore in performance of licensing transient analysis values of these parameters would be selected to provide conservative results.
13
2.5 Transient Analysis Nine comparative analyses were performed by the licensee in support of their effort to validate their model, Four of these were compared with analyses contained in the FDSA to demonstrate NUSCo's understanding of licensing type analyses and NUSCo's capability to reproduta design basis ant. lyses using RETRAN02.
NUSCo also presented five separate comparisons of RETRAN computations to limited plant data, all of which obtained reasonably good results. Although this set is rather limited because of the lack of transient plant data, NUSCo demonstrated understanding of the plant and the events, and indicated their ability to model these transients well.
2.5.1 Rod E_iection NUSCo compared their RETRAN computation with that contained in the FDSA, using, to the extent ascertainable, the same assumptions. The transient initiator was assumed to be a failure of a control rod drive mechanism housing so that a control rod could be ejected rapidly from the core, (case analyzed assumed 0.18% ejected) when the reactor was at the full power at the beginning of life (BOL). The calculation was carried out for 2 seconds.
The rod ejection caused a 40% prompt jump in nuclear power which was predicted by both codes. For the next I second, RETRAN's computed reactivity due to density changes was about one half of that computed in the FDSA while the difference in computed Doppler reactivities was much smaller. Thus, the FDSA computed nuclear power was about 5% higher than predicted by RETRAN. NUSCo attributed the difference to differences in the physics methodology.
Since this is a physics dominated transient, the treatment of reactivity feedback is most important. NUSCo's conclusion is reasonable that the most likely reason that these two calculations do not agree well (core average heat flux differed by a factor of 2) is because of the different physics methodologies used in these calculations; RETRAN is based on a single 14 l
v
integrated kinetics and thermal hydraulics package and the FDSA used two computer codes performing the physic and thermal hydraulics separately.
2.5.2 Four Pumo Coastdown NUSCO compared a RETRAN computation of the plant response to a four pump coastdown to that contained in the FDSA. In this transient, the power-to-flow mismatch created by the pump coastdown can result in DNB. The simultaneous loss of power to all RCPs is assumed to initiate this transient.
The pump coast down curves and the timing of the nuclear power peak compared well; however, the computed leval of nuclear power did not. Although the FDSA computed peak power reached roughly 110% of the nominal power at roughly 3,4 seconds and dropped to about 30% power by 5 seconds, RETRAN computed the peak power to be about 117.5% and dropped to about 20%, both at about the same time as FDSA. This discrepancy was reasonably attributed by the licensee to the differences in assumptions with respect to decay heat rate and control rod worth.
2.5.3 Loss of toad NUSCO compared an analysis of a complete loss of load transient to analysis contained in the FDSA. For conservatism, the following parameters were assumed: initial power level and coolant temperature at their maximum values, RC pressure ninimum, most positive moderator coefficient for BOL, least negative beppier reactivity. These parameter were selected in such a way to ,
result in the maximum power difference and the minimum margin to core protection limits. 1:e 4ddition, NUSCo made assumptions that the initial core flow was the das;3a value, and took no credit for charging or letdown.
The RETRAN results computed the same trends and timing as the FDSA, with the peak pressurtzer pressure and water level and loop average temperature, of the FDSA being slightly higher than RETRAN. NUSCo attributed these differences rather generally to differences in the primary to-secondary heat transfer or to the use of different rod worths. These are both plausible sources of the 15
differences.
2.5.4 Steam Line Break NUSCo compared the RETRAN computation of a break in a 24" steam line between a non-return valve and a steam generator to that contained in the FDSA. The 1 core reactivity was computed for the end of life (EOL) with initially unborated reactor coolant. A boron transport calculation is performed in a control block and fed into the RETRAN feedback system. Only one safety injection (SI) pump and one charging pump were assumed to inject borated water during this transient. Minimal information was provided regarding the FDSA assumptions and computation, and it is therefore difficult to assess the value of this comparison.
The results indicated global agreement. Since the first run with NUSCo's initial estimate of the SG inventory did not agree as well with the FDSA result, NUSCo ran a second case with increased SG inventory which resulted in improved agreement, particularly for total steam flow and pressurizer pressure.
. 2.5.5 30Y. Load Reiection A spurious control rod bottoming signal occurred at Haddam Neck on September 9,1980, which resulted in a turbine runback to 707. of normal steam flow over a 36 second period. Due to variations in loop average temperature, the reactor control system caused insertion of rod Bank B from step 300 to 158.
There was maximum pressurizer spray flow and no steam bypass during the event.
In the analysis, maximum spray flow of 200 gpm was assumed and the spray temperature was assumed to be 500'F to account for heat losses in the spray line. NUSCo modeled the output controlling actuation for automatic rod cluster control, which is the difference between the Tave signal and the Tref-The insertion rate was determined by the magnitude of this signal.
RETRAN computed parameters agreed well with the plant data.
16 l
2.5.6 Partial loss of Feedwater A low-level alarm on the heater drains tank initiated a chain of events which resulted in a trip at Haddam Neck on December 22, 1981. First, the heater drains pumps tripped and SG feed pump "A" started cycling due to low suction pressure. The operator reduced the load to compensate for the reduction in FW flow. The mismatch between heat generation and heat removal in the primary and secondary sides caused the reactor to trip on high pressurizer pressure.
The RETRAN prediction of the trends for the system response agreed well with the plant data.
2.5.7 H3tural Circulation Tests Two natural circulation tests were used for comparison in this topical report:
one condue:ed in 1981 and the other in 1967.
- 1. 1981 Test TF.e purpose of this particular natural circulation analysis was to demonstrate that the predicted upper head flow rate response was reasonable. Toward this end, the base case model with valves to isolate the upper head until cooldown was used for this analysis. The purpose of use of valves to initialize the input deck was discussed above. Cooldown was simulated by a control system which computed the steam flow from the SG based upon the upper plenum temperature. For this transient, the resistances in the head flow rate junctions were adjusted to produce a reverse flow of 10 lb/sec.
Although the RETRAN computed upper head temperature and head flow rate exhibited oscillatory behavior, they captured the trend quite well and the head temperature comparison was good. It is reasonable to assume, as has NUSCo, that the oscillation is due to the lag time built in the control system.
17
- 2. 1967 Test A natural circulation test was performed to determine if natural circulation I was evident following a plant trip. Two methods of determining natural circulation were employed in the test; the transport time method where actual data was used; and an analytical method where a combination of theoretical information and actual data were used. These were then compared with the RETRAN calculational results, indicating reasonable comparison considering the crudeness of the method of computing from plant data.
2.5.8 Inadvertent Actuation of ECCS A spurious actuation of the high containment pressure isolation system on March 27, 1980 caused a safety injection actuation signal which resulted in a turbine and a reactor trip. The difference between this event and a normal plant trip is that for a spurious SI signal, both charging pumps come on with suction switched to the refueling water storage tank (RWST), and letdown is isolated. In addition, the operators manually tripped the RCPs.
Because of the non equilibrium nature of pressurizer behavior during the insurge portion of the transient, NUSCo tried running RETRAN with two different interfacial heat transfer coefficient to examine its impact on the predicted system pressure. RETRAN tends to overpredict the pressure during the insurge because heat transfer between the steam and the pressurizer walls is not modeled (and cannot be accurately modeled in non-equilibrium volumes in RETRAN). Results show that the RETRAN predictions using the non-equilibrium pressurizer model agreed well with the plant data, and NUSCo therefore suggested that this may be one way in which this effect could be simulated.
However, in general, NUSCo stated that they believe that the RETRAN non-equilibrium pressurizer model is inadequate in simulation of this type of phenomenon; however, they also stated that use of this model is in the conservative direction since it is an over-prediction of pressure during an insurge which docs not usually present a DNB related problem. This conclusion would not be accurate for any reactivity insertion transient in which the primary loop inventory would heat up causing an insurge and creating a 18
simultaneous DNBR challenge, since over-prediction of the pressure would operate in a nonconservative fashion with respect to the DNBR computation.
3.0 CONCLUSION
AND RECOMMENDATIQM NUSCo has demonstrated its knowledge of the RETRAN computer code and models in its analysis of parametric sensitivity studies for SLB. NUSCo has adequately demonstrated its technical competency and NUSCo 1401 and the suppl)ments d contain sufficient material to serve as a document which may be generally referenced to support future NUSCo licensing submittals with one exception:
the report does not contain sufficient material to provide a basis for assessment of adequacy of their plant nodalization for the full spectrum of transients. Use of this model in future licensing sulmittals would therefore require the licensee to justify, on a transient-by-transient basis:
- 1. nodalization and control system modeling for transients not analyzed in this report and the supplements thereto referenced herein; and
- 2. modeling of licensing conditions and controls and how its model selection may be impacted in the licensing calculations.
L Although RETRAN02/ MOD 003 has not been formally approved by the NRC, there exists a reasonable assurance that analysis using this version of RETRAN is acceptable as used by NUSCo for these transients unless future review of that code determines to tne contrary.
4.0 REFERENCES
- 1. "NUSCo Thermal Hydraulic Model Qualification Volume I (RETRAN),"
NUSCo 140-1, August 1, 1984.
- 2. Connecticut Yankee Atomic Power Company, Haddam Neck Plant, "Facility Description and Safety Analysis," Docket No. 50-213.
- 3. Letter from D.G. Eisenhut (NRC) to Licensees, "Licensee Qualifi-i cation for Performira Safety Analyses in Support of Licensing
. Actions (Generic letter No. 83-11)," February 8. 1983.
19
- 4. "Safety Evaluation by the Office of Nuclear Reactor Regulation Regarding NUSCo Topical Report 149 2 VIPRE-01 Connecticut Yankee Atomic Power Company Docket No.50 213 Haddam Neck Plant " October 1986.
- 5. E.J. Mroczka letter to F.M. Akstulewicz, "Haddam Neck Plant Response to Request for Additional Information on RETRAN," May 7, 1987.
- 6. "Haddam Neck Plant Additional Information Reanalysis of Non LOCA Design Basis Accidents," E.J. Mroczka (NUSCo) letter to U.S. NRC, November 19, 1987.
- 7. "Haddam Neck Plant Response to Requests for Additional Information on RETRAN," letter from E.J. Mroczka to U.S. Nuclear Regulatory Commission, February 11, 1988.
- 8. "Safety Evaluation Report on the RETRAN Computer Code," U.S. Nuclear Regulatory Commission, July 1984.
P 20 l
_-m. , . . _ , _.. ___ _ - _ . - , _ . _ _ _ - _-
O' o
. TABLE 1.1 Musco Transients Transient Description objective Consnents Steam Line Break sensitivity secondary noding phase sopration Loss of Load sensitivity Loss of Feedwater sensitivity Loss of Flow sensitivity Dropped Rod serwitivity Natural Circulation model refinement upper head modellrg Rod Ejection conparison FDSA/FSAR Four Ptsp Coastdown conparlson Test Loss of Load comparison FDSA/FSAR Steam Line Break conparison FDSA/FSAR 30 % Load Rejectior. comparison operational date Partial Loss of Fea %eter copperison operational data Natural Circulation Tests comparison operational data 1961, 1967 Inadvertent Actuation v' ECCS conparism operational date 21
_ _ _