ML18092A514

Evaluation of Slow Response Reactor Temp Detectors on Salem Unit 1 Safety Analysis.

ML18092A514
Person / Time
Site:	Salem
Issue date:	02/25/1985
From:	WESTINGHOUSE ELECTRIC COMPANY, DIV OF CBS CORP.
To:
Shared Package
ML18092A512	List: ML18092A510 ML18092A514
References
NUDOCS 8503050216
Download: ML18092A514 (7)

Text

EVALUATION OF SI.CW RESRJNSE RTDs. ON SALEM UNIT 1 SAFETY ANALYSIS Introciuction During the refueling outage fer Salem Unit 1 Cycle 6, PSE&G replaced *.the Rosemont RTDs with RdF RTDs. The RTDs are use to* measure- average coolant -

teinperature (Tavg)in the core. The measured value of Ta~g is used as input-to

. both the central and protection systems. The effect on the protectiol'! system due to the new RTDs must be evaluated in order to show that the safety analysis presented in the FSAR remains valid. Although the control system response may also be affected, -operation Of the control system is not asst.med .in the safety analysis to mitigate* a transient. Thus, the impsct on the control systems* need not be considered.

The origirsl Rosemont RTDs are fast acting _RTDs with a response time of approximately* .5 seconds, whereas the RdF RTDs _have a respon5e time of about 3 seconds. _ Since the new RTDs have a mu:h slcwer response* time, those protection system f1.11ctions which rely on measurement of Tavg may also have a slower response time. The*resil:>nse *times assuned_in the safety analysis are reflected in the plant Technical S~cifications. These must be measured on a pariodic basis. With the increased response time of the RdF RTDs, the ~ssibility exists that the maximun allowable* response times listed in the Tech Spacs cannot be met, thus ~tentially invalidating the safety analysis. The Tech Spec response times are defined as the time interval from when the monitored paraneter exceeds its setpoint 1.11til the rods are free to fall or _the Engineered Safety Feature is capable of parforming its f1.11ction.

Since the Rosemont RTDs have a fast response time, the signal output is processed through an electronic filter to reduce noise. The combined response of the fast RTD with the filter is approximately equal to_ that of the slew response RdF RTDs without a filter. Thus, by removing_ the filter, the overall protection system response times which are based on T input will be similar *

.~~~~----=-=-=-~==-==-:::=~~~~* avg r 8503050216 850225 l PDR ADOC~ 05000272 p PDR

Saiety Eyalyation The T signal is used as input to the overtem~rature/overi::ower delta-T trips avg . .

and to the steanline break protection system signal of high steam fl~

coincident with lcw-Tavg* This signal actuates reactor trip, safety .injection, steamline isolation, and feedline isolation, and initiates auxiliary feedwater.

Thus, those accidents which rely on this signal must be evaluated with respect to a slGJer protection system response time. The evaluation presented here assumes that the overall response time of tile protection signals is increased by 3 seconds over the times assuned in the FSAR safety *analysis. This is a conservative assun~ion, particularly if tile filters on Tavg measurement are removed.

Steamline Rupture - The steamline rupture analyses presented in the Salen FSAR

.do not take credit fer the high steam flGJ/lGJ Tavg signal. In the core analy si~, the high steam flGJ coincident with lGJ steat1 pressure signal is assuned in the analysis of tile double ended rupture (Section 15 .4.2). For *the accidental depressurization of the main steam system (Section 15.2.13), the lGJ

. pressurizer pressure signal is used. These signals are also considered in the contairme.nt analysis presented in Section 15 .4 .8.2 of tile FSAR. In addition, contairment pressure signals are considered. Thus, the_ increased. response time of tile coincident high steam flew/ lGJ Tavg signal will not, imI=Sct the safety analysis.

Note that this additional delay, if considered, would not have a significant imI=Sct on the safety analysis. With respect to the core analysis, the.

steamline break event is a cooldown transient and is primarily dependent on the core reactivity paraneters and the blGJdown rate of the faulted loop. For both steanline break transients presented in the FSAR, it is shown that the DNBR remains above the limit value of 1.30. For the containment analysis, the containnent pressure and tem~rature resp:rnses are a fl.l1ction of.the mass and* energy released during the blcwdown.

In evaluating the core response, the effect on the minimun DNBR must be considered. Prior to steamline isolation, all four loops blGJ down at the same rate, providing uniform cooling to the core. Thus, there is no assymmetric

e e cooling which results in a non-unifcrm power distribution in the core. If steamline isolation is delayed, the time. at which non-uniform cooling begins will also be delayed,- resulting in a more L11ifonn power distribution. This reduces the Il)Wer i:eaking in the core and results in a higher DNBR.

On the other hand, an increased delay in feedwater isolation will cause more feedwater to be delivered to the steam generators and increase the blcwdown. _

. This increased blcwci.own will result in more cooling of the primary siQ.e and consequently, a higher return to power.* This may result in a lcwer DNBR. An evaluation of the increased feedwater isolation time shows that an additional three* seconds will not have an ll'lacceptable effect on the minimun DNBR for the double ended rupture. This evalilation is based upon sensitivity studies presented in WCA.P-9226, "Reactor Core Response to Excessive Secondary Stean Releases" (proprietary), rriere commonly known as the steambreak topical.

Al.though the steanbreak topical was not used in support of. the Salem FSAR (it post-dates the FSAR), the conclusions presented are generic in nature and as such can be applied to Salem. Thus, an increase of three seconds in the delay in feeawater isolation is acceptable fer the core analysis.

The containment analysis performed for Salen assunes both the failure of an MSIV and a feedwater regulating valve. These single failures have a ~imilar effect as an additional three second delay in steamline or feedline isolation since both the single failures and the delay result in more mass being released to the. containment. Hcwever, the limiting transients fer both tani:erature and pressure are not those for which the single failure is either the MSIV or feedwater regulating valve* failure. Rather, the limiting single failure is the failure of a containment safeguards train. Thus, the additional delay is* not is not expected to impact the containment response.

The delay of safety injection will also have a negligible impact on the resul~ts

.of both the core and containment analyses *. The increase in core power ~s tenninated by the dryout (cessation *of bla.Jdbwn) Of the faul ~ed. loop, not by the addition of boron to the core by the safety injection system. The total mass released to the containment is unaffected by the time of safety injection. Thus, the delay of boron by three seconds will have no impact.

Note also that an analysis assuning boron concentration reduction in the BIT

.* has been perfc:nned fer Salen wh~ch shows acceptable results (reference 1). The

. delay of boron injection into the. core fer tilis analysis is more severe than that i:ostulated due to the additional three second delay of the protection system.

Finally, the delay of auxiliary feedwater addition will not irilµact the results because the auxiliary feedwater system is conservatively ass1.111ed to be on at*

the initiation of the transient. This maximizes both the cooldown fer the core analysis and the mass/energy releases to containment. No time response fer auxiliary feedwater initiation is asst.med. Thus, a delay in auxiliary

• feedwater wOuld be a benefit fer both the core and containment analysis.

. In sunmary, the increased protection system response time will not impact the steamline break analyses presented in the FSAR because the Ta vg signal is. not used in the analysis. If the delay is assuned independent of this consideration, there is sufficient margin and/or conservatism in the analysis such that the safety limits will continue to be met.

• Bank Withdrawal at Power .;. This transient relies upon the high neutron flux and overtemi:erature delta-T trips fer protection to ensure that the minimun DNBR remains above 1.30. This event is analyzed in Section 15.2.2 of the FSAR. A range of reactivity- insertion rates I caused by the bank with draw al is considered from several initial p'.)Wer levels to demonstrate that these two trips provide

. adequate protection.. These analyses are also i:erformed assuming minimun and maximun -reactivity feedback. Although the increa5ed resp'.)nse time or' the RTDs . *

• will not impact those cases which trip .

due* to a high neutron flux signal, . those cases resulting in an overtemi:erature del ta-T trip will be delayed. Reactor ti'ip may instead occur due to a high neutron flux signal or by the delayed.

ov~rtemi:erature delta-T signal. For these cases, the minimun DNBR reached during the transient wili be lcwer. Thus, this event must. be reanalyzed to -

show that the DNB design basis continues to be met.

The transient was reanalyzed employing the same digital computer codes and*

assumptions regarding instrunentation and setpoint errors used for the FSAR.

The FSAR assuni;tions on reactivity feedback were *also ass1.111ed. Hcwever, instead of assuming a 6 second delay time for reactor trip on overtani:erature

.del ta-T, a 9 second time delay was assuned. The 9 seconds includes the time to heat the RTD bypass lines, the sensor resp:mse, logic, voltage decay and gripper release of the rods.

Figure 1 shows the minimllll DNBR as a ft11ction of reactivity insertion rate fer the bank withdrawal initiating *from 100 percent power for both maximllll and minimun feedback. S:imilar results were obtained fer the transient initiating from 60 percent and 10 percent power. These* are 1 ess limiting than the results from full power, thus the detailed results are not presented. As expected, mere cases results in a trip due to a high neutron flux signal because the overtanperature del ta-T signal is delayed.. However, fer all cases, the minimllll_

DNBR remains above 1.30~ Thi.ls, the _additional delay of the res!X)nse time is acceptable.

Recently; Westinghouse performed a safety analysis for Salem with a positive moderator tanperature coefficient (reference 2). This* affects the minimun .

reactivity feedback cases _of the bank with draw al at power event. Al though t.his has not -yet been approved by the NRG, an analysis incorporating a +5 pcm1°F moderator tanperature coefficient was al.sO performed asstining a 9 second delay. The results of this analysis also. show that the DNBR design basis is*

met.

In addition, Westinghouse_ also performed safety analyses incor!X)rating an increased Doppler-only power coefficient (reference 3). This* new -coefficient was based on calculations of the coefficient: by PSE&G fer the Salem Unit 1 Cycle 6 design. In this analysis, t.he bank withdrawal at power was reanalyzed fer maximun feedback to account for the increased Doppler coefficient,.

'Iheref cre, in order to complete the evaluation of the increased protection system res!X)nse time, t.he bank withdrawal at power event with maximun feedback and the PSE&G calculated Doppler-only power coefficient was reanalyzed assun+/-ng a 9. :Second response time *.

'Ihe results of the analyses fer the positive moderator tem~rature coefficient and the increased Doppler feedback are presented. in Figure 2 for the event initiated from full power. Again, the minirnun DNBRs reached when the transient initiates f_rom intermedia_te !X)wer levels are less limiting. The minimun

f°eedback cases are those assuning a !X'Si tive moderator tanperature coefficient. The maximun feedback cases are those incor1X>rating the Doppler-only !X>Wer coefficient as calculated by PSE&.G. For all cases, the minimt.m DNBR is above 1.30.

sunmary An evaluation of those accidents relying on measuranent of Tavg for actuation of the protection system has been performed. These accidents are stecrnJ.-ine rupture and uncontrolled bank withdrawal at 1X>Wer. The results of the evaluations and analyses show that all safety limits will be met if an additional 3 second delay is considered in the response time of the protection system.

References

1. Letter PSE*00-528, Noon (J:l) to Librizzi (PSE&G), 11Feasibility Report fer BIT Concentration Reduction/BIT Elimira ti on fer Salem Units 1 and 211 ,

September 1980

• Letter PSE-81-509, Noon (J:l) to Uderitz (PSE&G), March 1981, contains additional infermatian.

2. Letter 84PS*-G-024, Croasdaile (J:l) to Brenner (PSE&G), "Positive Moderator

• Temperature Coefficient Study", March 1984.

Letters 84PS*-G-057, 84PS*..c;;..072, and 84PS*-G-074 contain additional infermation on the study.

3. Letter 84PS*-G-058, Croasdaile (J:l) to-Brenner (PSE&G), "Increased Doppler Coefficient .study*i, June 1984.