ML20003E748
| ML20003E748 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 04/06/1981 |
| From: | Hukill H METROPOLITAN EDISON CO. |
| To: | Reid R Office of Nuclear Reactor Regulation |
| References | |
| RTR-NUREG-0737, RTR-NUREG-737, TASK-2.K.3.02, TASK-2.K.3.07, TASK-TM L1L-102, NUDOCS 8104100402 | |
| Download: ML20003E748 (18) | |
Text
r i
-b Metropolitan Edison C,pany Met-Ed GPU
~ ' = * *
- Middletown, Pennsytvania 17057 vWriter": Direct Deal Number April 6,1981 IlL 102 g4j b[
Office of Nuclear Reactor Rwulation g
g Attn:
R. W. Reid, Chief 7
Operating Reactors Branch b
~.
U. S. Nuclear Regulatory Cou sion g,
9 \\
_9 Washington, D.C.
20555 cd
-i o
w/
h.4*
Dear Sir:
Three Mile Island Nuclear Station, Unit 1 (TMI-1)
Operating License No. DPR-50 Docket No. 50-289 Response to NUREG-0737 Items II.K.3.1. II.K.3.2, and II.K.3.7 Enclosed please find B & W Generic Report which addresses Items II.K.3.2 and II.K.3.7 of NUREG-0737. We have reviewed this report and have determined that it is applicable to TMI-l with the exception of the following two items:
the unreliability assumed for the Emergency Feedwater (EFW) System is higher than the TMI-l EFW System; and number of openinks of the PORV assumed to occur for steam generator tube rupture is excessive.
In spite of these two exceptions, the report is conservatively acceptable for TMI-l in that the assumptions overestimate the probability of PORV opening and a loss of coolant accident via the FORV.
Based on the results of this B & W Report there is no need for a PORV automatic Isolation Sy c -m as identified by Item II.K.3.1 of NUREG-0737.
Sincerely,
)
N. D. Mukill Director, TMI-l HDH:LWH:Ina l
Enclosure cc:
H. Silver L. Barrett R. Jacobs l
D. DiIanni O
B. H. Grier I
//
8104100 Mecochtan Ed: son Company s a Memoer of the @eral Put'c Ut.L!:es S,s'em
A l
l REPORT ON POWER-OPERATED RELIEF VALVE I
OPENING PROBABILITY-AND JUSTIFICATION FOR PRESENT SY5 TEM AND SETPOINTS l
- Submitted to Satisfy Requirements of l
NUREG-0737, Items II.X.3.2 and II.X.3.7 Document No. 12-1122779
(
i i
+
(
k m
y ma
-m w
w
I e
REPORT ON POWER-OPERATED RELIEF VALVE i
OPENING PROBABILITY AND JUSTIFICATION FOR PRESENT SYSTEM AND SETPOINTS f~
[f Submitted to Satisfy Requirements'of i
l NUREG-0737, Items II.K.3.2 and II.K.3.7 i
Occument No. 12-1122779 7
s.
~ (
f i
(.
be a
+
e~-
,y.
-,,v_.,
.-r,3
,-.4,.
,.,,..%y.-,4-w,.,w
-.... =
p,,.4,,,
,-,,,,w-e,y.,,,,,,,,9,-=,,e
,,,,.*'r---,eeevr--e-+
,w+e-a---'es,e4es v
q- - -
Table of Contents 1.0 Introduction & Summary 2.0 Discussion 2.1 Evaluation of PORY Opening Probability During Overpressure Transient 2.2 Evaluation of PORY and Safety Valve Reliability 7-i 2.2.1 Safety Valve Failure Rate History i
2.2.2 Evaluation of Small Break LOCA Probability /Need for PORY Isolation System 3.0 Conclusions i
P I <
i i
e 9
l l
,,~,._.m-_
m..
1.0 INTRODUCTION
AND
SUMMARY
NUREG-0737, " Clarification of TMI Action Plan Requirements," November 1980, required that a report be submitted which providas the information identified in Items II.K.3.2 and II.K.3.7.
Specifically, NUREG-0737 req;ested the following information/ justifications:
1.
II.K.3.2 Compile operational data regarding pressurizer safety valves to o
determine safety valve failure rates Perform a probability analysis to determine whether the modifica-o tions already implemented have reduced the probability of a small
'i break LOCA due to a stuck-open PORY or safety valve a sufficient amount to satisfy the criterion (<10-3 per reactor year}, or
\\
whether the automatic PORY isolation system specified in Task Item II.K.3.1 is necessary.
I i
2.
II.K.3.7 Perform an analysis to assure that the frequency of PORV openings o
is less than 5% of the total number of overpressure transients.
This report is submitted in compliance with NUREG-0737 and demonstrates that the requirements of NUREG-0737 are met with the existing Power-0perated Relief Valve (PORV), Safety Valve and 'digh Pressure Trip Setpoints and that no automatic isolation system is required.
w----
e--
<w r--
g y
.e
2.1 Evaluation of PORY Opening Probability During an Overpressure Transient An evaluation of the probability of PORY opening has been performed. Two separate analyses have been performed. The first is an analycical estimate, the second is an analysis based upon operating experience.
2.1.1 PORV Opening Probabilitv Based Upon Analyses A series of calculations have been completed using best estimate numbers to estimate the probability of PORV opening. Wherever possible, these 6
F "
calculations were based on operating plant data in an attempt to provide i
realistic estimates for the analyzed events. The following paragraphs sunmarize the results-and -calculational basis for the analysis.
The probability of the PORY lifting during a loss of feedwater (LOFW) or turbine trip is approximately 3.9x10-6/Rx-Yr for plants with a PORV setpoint of 2450.psig and 3.9x10-3/Rx Yr for plants with a PORY t
b setpoint of 2400 psig. The latter setpoint is presently applicable only
(~
to Davis-Besse 1.
These probabilities are based on the assumptions that the high pressure trip setpoint is 2300 psig with a standard deviation of 1.4 psi and that the actual setpoint at which reactor trip occurs is a i
random variable which is normally distributed. The small standard deviation is based on the fact that the PORV and RPS actuation points are ir not completely independent; i.e., they share a common source; i.e., sensor
, j and instrument string.
Thus, these parts of the string errors are l
l perfectly correlated and cancel one another in the analysis. Other parts of the relevant string error are not correlated and it is upon these that the 1.4 psi standard deviations are based.
In a similar fashion, the i
I
,-.n-p.-,.
,-n,-,
.,m
.,-r,.,,-,,n.,,-a,
.,._w._
actual opening setpoint of the PORV is also assumed to be a random variable with a normal distribution. The assumption of normality for the actuation of (Ither the high pressure trip or the PORY is just an assumption; no data is available to justify or deny the validity.
The RCS pressure rise above the RPS high pressure trip setpoint (hence referred to as " pressure rollover") during a LOFW or turbine trip was detennined by a combination of plant data and engineering analysis. Pressure rollover data from the operating plants (Table 2.1-1) was compiled from available data. However, these data points represent situations in which the PORY could open, thus decreasing the amount of pressure overshoot.
Therefore, it was necessary to correct for the PORY opening, since we are interested in the situatien in which it remains closed.
This was an accomplished by benchmarking the CADO code to a transient in which the i
PORV was isolated. After satisfactory duplication of this transient, the code was rerun modeling proper functioning of the PORV. The resulting pressure correction to the rollover data was 17.4 psi. The rollover data itself was tested and is statistically acceptable as normally distributed.
It has a mean of 9.2 and a standhrd deviation of 27.52 psi. The presence i
of negative values in this data set indicates that the RPS trip setpoints have frequently been set low. Since the data reflects actual operating experience, the use of the negative values can be justified in the I
analysis.
I r
Using the above data and assunptions, a Monte Carlo simulation of the relation PORY - RPS - EXCESS - BIAS = SAMPLE i..
a
was conducted. The terms in the above relation are defined as follows:
RORY - PORY setpoint, a normally distributed random variable RPS - High pressure trip setpoint, also a normally distributed random variable
[
EXCESS - Pressure follover, a randomly distributed normal variable 9
BIAS - A constant (17.4 psi) defined by analysis which compensates the rollover data for the fact that l
the PORY will remain closed.
P Six thousand sample values of the above alogriths expression were calculated using the SAMPLE code. A negative value of the above exprescion implies the PORY opens.
In the computer trials, no negative values in 6000 instances were observed.
t It was then assumed that the random variables described above are independent in the probabilistic sense, so an analytic approach was applied. The sum or difference of several independent normal distributions is also a normal distribution with mean equal to the algebraic sum of the means and standard deviation equal to the square root i
of the sum of variances.
In this case, the mean is 2450 - 2300 - 9.23 - 17.4 = 123.37 (except 08-1, = 73.37) and standard deviation is
'E (1.4)2 + (1,4)2 + (27.52)2 27.59 (for 08-1,= 27.59)
=
e
The probability that the PORY will open during an overpressure transient is 3.9X10-6/Rx-Yr (for 08-1 this value is 3.9X10-3/Rx-Yr). The statistics show that we can be 99% confident that at least 99.99% of all LOFW and turbine trip high pressure transients will not open the PORV for the PORY set at 2450 psig. Fo-a setpoint o? 2400 psig, the statistics indicate a 99% confidence that more than 99.4< of the overpressure transients will not result in opening the PORV.
2.1.2 PORV Opening Probability Based Upon Operational Data NUREG-0667, " Final Report of the B&W Reactor Transient Response Task Force," contained a listing of reactor trips (148) with PORV actuations prior to the TMI-2 accident. Since the accident at TMI-2 approximately 59 trips have occurred on B&W designed plants. Approximately 42 of these trips would have lifted the PORY with the old setpoints. Of the 190 trips that would have lifted the PORV with old setpoints, three of these events would have lifted the PORY with the new setpoints.
In addition the modifications that have been made to the plants since those transients would have precluded PORY actuation given the same initiating events on those plants and the new setpoints. Based on these data, it is estimated that the present PORY opening probability is less than 1.6% for an overpressure transient, which is less than the 5% requirement stated in II.K.3.7 of NUREG-0737.
t L
1 we c
---.---,,n,
,,-w-,
g w
m e
g.
e
-.,-n--
e-e
TABLE 2.1-1 PRESSURE ROLLOVER DATA Trip #
Power, %
Peak Pressure, psig Rollover, osig 1
95 2355 0
2 90 2385
+30 3
25 1,400
+45 4
20 2385
+30 5
90 2390
+40 6
32 2345
-10 t'
7 40 2360
+5 8
40 2352
-5 9
92 2375
+20 10 15 2365
+10 l
11 35 2400
+45 12 13 2370
+15 13 14 2355 0
v.
14 38 2380
+25
(
[
15 98 2410
+55 16 72 2400
+45 f
l 17 100 2340
-15 18 100 2340
-15 19 100 2390
+35 20 100 2330
-25 L
21 98 2325
-30 22 15 2355 0
i 23 9
2370
+15 24 30 2345
-10
[l~
25 99 2350
-5 f
26 16 2295
-60 L
u e
e e--+vrw-
-.-,,,r-,-----w,e-r-
-r y-y
=r-------
-++-g y,---
w
- - ---94 o
I 2.2 Evaluation of PORY and Safety Valve Reliability 2.2.1 Safety Valve Failure Rate History There have been three cases where pressurizer safety valves were lifted on B&W plants. None of these cases resulted in failure of the safety valve to reseat. Ber.ause of the few data points, no estimate was made of the safety valve failure rates.
l 2.2.2 Evaluation of Small Break LOCA Probabilities /Need for PORY Isolation System The contribution to the probability of a SB LOCA from an open PORY was
(
estimated by two methods. The first was an analysis effort, the second r
was based strictly upon operational data. The results are discussed l
below:
1 2.2.2.1 Small Break LOCA Probability Calculations f
The probability of a stuck open PORY is the product of the probability of being demanded open times the probability of failing open on demand.
The raising of the PORY setpoint has reduced the number of demands and l
thus the probability of being in the stuck open state. The point estimate for PORY SB LOCA probably (variation not estimated) is calculated to be 5.04 x 10~4per reactor year which complies with II.K.3.2 requirement that the probability of stuck open PORY SB LOCA i
does not significantly impact the probanility of 58 LOCA from all causes (1 x 10-3 per reactor year). The initiators of PORY actuations have been grouped into five categories along the associated frequency of each b-category. Details on how tne values are calculated are contained in i
Table 2.2.2-1.
L
%e
- 1. PORY opening on overpressure transient 3.9 x 10-6/Rx-Yr 2.
PORY opening on transient with delayed 1.4 x 10-3/Rx-Yr aux. feed 3.
PORY opening on operator action under 1.58 x 10-2 Rx-Yr
/
i ATOG guidelines 4.
PORY opening due to instrumentation 5 x 10-3/Rx-Yr control faults 5.
PORV opening from additional 1.8 x 10-3/Rx-Yr consideration from II.K.3.7 TOTALS 2.40 x 10'2/Rx-Yr 2.61 x 10-2/Rx-Yr(08)
(
This total is then multiplied by the probability of the POR. sticking open t'
I on & mand.
t' Note that all plants except Davis Besse (Crcsby PORV) have Dresser valves; however, the entire B&W operating plant experience was used to arrive at a generic PORY sticking open probability as follows: There have been ten stuck open PORV events, five of which could be classified as mechanical failure of the PORY (the other five were basically installation errors).
I Using all these five failures in determination of future frequency is considered conservative since two of the failures (OC-3,6/13/75 and CR-3, 11/75) were rectified by design changes, another (TMI-2, 3/28/79) cause is c
unknown. OC-2,11/6/73 could be considered as a burn-in failure and the l
l.
r-l I
1 AW
08-1,10/13/77 event is a Crosby valve. Using five failures in 250 demands retults in a value c" 2 x 10-2 to fail to reclose on demand. This value is considered conservative not only due to the inclusion of all five failures but also the number of demands is-l probably much higher than 250. There have been 148 documented PORY openings on reactor trips; however, there is not a listing of PORV demands when the reactor did not trip (e.g.. ICS runback) nor is
[
consideration given to transients that could have actuated the PORY numerous times during an event. The value of 250 demands is
(
onservatively used here. An analysis was also performed to include values for other than mechanical failure that keep the PORV open. The
- al contributor (2 x results of this analysis is stened with the mechat I
10-2/d) to arrive at the value for failure to reclosa on demand (2.1 x 10-2/d).
t Probability of PORV small break LOCA equais:
(2.4 x 10-2) (2.1 x 10-2/d) = 5.04 x 10-4/Rx-Yr (2.61 x 10-2)(2.1 x 10-2/d) = 5.48 x 10-4/Rx-Yr (DB) 5'
'i 2.2.2.2 Small Break LOCA Probability Based Voon Operational Data As discussed ir. Section 2.1.2, there have been three events which with the revised setpoints would have actuated the PORY. However, the plants have been reconfigured (e.g., upgrades on aux. feedwater, l
control circuitry of PORY, NNI power sources, AC power sources) so as to reduce the probability of these PORV actuations.
Conservatively s-estimating that one event could occur in the 45 years of B&W plant L~
operation, yields a probability of occurrence of 2.22 x 10-2/Rx-Yr.
i 1.
w
-v ey
The previous section gave a PORY failure probability of 2.1x10-2/d.
Therefore the probability of a PORY small break LOCA equals:
(2.22x10-2 /Rx-Yr)(2.1x10-2/d) = 4.7x10-4/Rx-Yr d
which is less than the 1.0x10-3/Rx-Yr criterion.
3.0 CONCLUSION
Both the analytical prediction and the estimate based on historical data result in values for a stuck open PORY for all causes which meet the l
requirements given in II.K.3.2.
Note that r
no credit has been assigned for the operator closing the block valve given an open PORV. Analytical predictions (given proper auxiliary feedwater response) result in a value less than.01% of PORY openings for overpressure transients (taking into account the most limiting non-anticipatory trips) and historical data shows the frequency to be less f
than 1.6% which satisfies the criterion (less than 5%) specified in ll.K.3.7.
Since the requirements of II.K.3.2 and II.K.3.7 are met with the current PORY configuration and set snint it is not necessary to address the f
requirement for an automatic block valve closure system per II.K.3.1.
m
(
4
Table 2.2.,23 1.
The probability of a PORY opening on an overpressure transient fron Sectton 2.1.1 for plants with PORY setpoint of 2450 3.9 x 10-6 px-Yr
/
for plants with PORV setpoint of ?400 (08) 3.9 x 10-3/Rx-Yr f
2.
The PORY opening probability in a trusient with delayed aux. feed f
[
A value of 1.C was assigned for PORV p
opening probability if aux. feedwater was
(~
not supplied. A value of I
1.4 x 10-3/Rx-Yr for loss of all feedwater was referenced from a B&W calculation
[.
which used average unavailability as calculated in the generic aux. feedwater reliability studies (BAW-1584) in conjunction with generic EPRI data on loss of main feedwater frequency and loss of offsite power frequency.
f On completion of the ongoing aux. feedwater reliability analysis (AP&L, SMUD, FPC) more specific values can be applied to those plants.1.4 x 10-3fgx.yp i
3.
The PORV coening probability on operator action under ATOG guidelines
~
There are 3 events that call for operator opening of the PORV:
a) Loss of All Feedwater. This contribution is already counted in 2 above; b) Small LOCA.
Not applicable to u
Table 2.2.2-1 (Cont'd) this calculation since the plant is already in a small LOCA; c) Steam Generator Tube Rupture (considered sinaller than small LOCA as defined in II.K.3.2 su argument of b) does not hold): The demand on the PORY i
given a tube rupture varies depending on f
whether offsite power is available or lost.
i If offsite power (Reactor Coolant Pumps) is available, only one PORY opening is required, I
whereas in the loss of offsite power scenario as many as 23 PORY openings are required.
[
The value calculated assumes that the probability of Steam Generator Tube Rupture considered with a LOOP event is small (no f
causal effect of LOOP or Stean Generator Tube Rupture) and therefore, the WASH-1400 of 1 x 10-3 for a LOOP given a reactor trip is used in the calculations. There have not been any tube ruptures in the cumulative B&W experience, due to the limited number of years experience.
A Chi-square 50% confidence value with 0 failures is rather high (1.54 x 10-2 Rx-Yr).
1 i.
G W e'8
Table 2.2.2-1 (Cont'd) 1.54 x 1r;-2 gx.Yr x 1 demand (offsite power available) 1.54 x 10-2/Rx-Yr.
f 1.54 y 10-2/RX-Yr x10-# offsite power 1.oss/ event x23 demands (offsite peer lost) 3.54 x 10-4/Rx - Yr.
1.58 x 10-2/Rx - Yr.
P In the final calculation of probability to reclose, it should be noted that no adverse effects of the 23 demands in the loss of offsite power case on PORY operability is assuned.
4.
PORY opening due to instrunentation control l
faults This has been estimated at 5 x 10-3f
~
reactor year. This value assumes that power supply faults and other control deficiencies have been corrected by each
(
(
utility.
5 x 10-3/Rx - Yr.
5.
PORY opening probability from additional t
considerations from II.K.3.7 There are overcooling transients L
that initiate HPI and opirator failure to
(
throttle or terminate flow before the PORV setpoint is reached.
There have been 3 overcooling transients that initiated C
b9
-.e
Table 2.2.2-1 (Cont'd)
HPI in 392 reactor trips.
The current frequency of reactor trips is 6 trips /
Rx-Yr per plant.
In this event sequence, the operator has approximately 4 minutes from time of HPI initiation until PORY setpoint is reached.
The operator failure rate to terminate or throttle HPI flow is based on having ATOG in place (1.5x10-2/d - based on NUREG-CR-1278 with moderately high stress). The overall probability of t
this sequence is therefore estimated to be 6 trips /Rx-Yr x 8/392 overcooling events / trip x 1.5x10-2 1,3 x 10-3 Rx-Yr N.A. for 08 TOTALS 2.40 x 10-2/Rx-Yr.
2.61 x 10-2/Rx-Yr.
t
[
Note that these values are dominated by the conservative analysis of steam
(
i generator tube rupture. Analytical studies could be performed to obtain a more realistic value. Also note that the calculation for category 4 did not
~
include operator or maintenance induced faults, such as the 08 event of 10/27/80.
O e
O
- m.
.