NOC-AE-14003082, Alternate Source Term Dose Analysis: an Estimate of Risk Attributed to GSI-191, Attachment 3

Alternate Source Term Dose Analysis: an Estimate of Risk Attributed to GSI-191, Attachment 3

ML14086A387
Person / Time
Site:	South Texas
Issue date:	03/11/2014
From:	Morton D, Tejada J, Zolan A South Texas, University of Texas at Austin
To:	Office of Nuclear Reactor Regulation
Shared Package
ML14083A383	List: NOC-AE-14003082, Alternate Source Term Dose Analysis: an Estimate of Risk Attributed to GSI-191, Attachment 3
References
GSI-191, NOC-AE-14003082, STI133816627, TAC MF2400, TAC MF2401 STP-RIGSI191-RAI.1, Rev 1.3
Download: ML14086A387 (9)

Text

NOC-AE-1 4003082 Attachment 3 Alternate Source Term Dose Analysis:

An Estimate of Risk Attributed to GSI-191 STP-RIGSI191-RAI.1

South Texas Project Risk-informed GSI-191 Evaluation Alternate Source Term Dose Analysis:

An Estimate of Risk Attributed to GSI-191 Document: STP-RIGSI191-RAI.1 Revision: 1.3 Date: March 11, 2014 Prepared by:

Alexander Zolan, The University of Texas at Austin Jeremy Tejada, The University of Texas at Austin David Morton, The University of Texas at Austin Reviewed by:

Ernie J. Kee, South Texas Project Zahra Mohaghegh, University of Illinois at Urbana-Champaign Seyed A. Reihani, University of Illinois at Urbana-Champaign

Alternate Source Term Dose Analysis:

An Estimate of Risk Attributed to GSI-191 Alexander Zolan, Jeremy Tejada, and David Morton The University of Texas at Austin Abstract We estimate the rate, in terms of events per calendar year, of an alternate source term acceptance criteria exceedance (AST ACE) event, which is attributable to GSI-191. To do so, we use output from the CASA Grande simulation model to estimate the rate at which a sump failure occurs as a function of time since a loss of coolant accident. At the time the first sump fails, we assume all containment sprays are lost. We couple this estimation with a dose calculation as a function of the time at which containment sprays are lost, at the Technical Support Center (TSC) and the Control Room (CR). If the dose exceeds 5.0 remi when containment sprays are lost at the TSC or the CR then an AST ACE event is said to occur.

1 Sump Failure Frequency for a Given Time Threshold The dose at the Technical Support Center (TSC) and the Control Room (CR) depends on how long the containment spray system (CSS) operates after a loss of coolant accident (LOCA). We refer to [4]

for such doses including a detailed description of how they are calculated and their applicability to an AST evaluation. If required to operate following a LOCA, each CSS pump train would pull water through its sump that may fail due to filter plugging. At the time the first sump fails, we assume all other trains of CSS fail. And, we define an alternate source term acceptance criteria exceedance (AST ACE) event as a sump failure that occurs before the dose drops below 5.0 rem. For each size of a LOCA--small, medium, and large-and each pump state, we use output from the CASA Grande simulation model to estimate the conditional probability of a sump failure up to a given point in time, conditioned on that LOCA size and pump state. Then, the frequency of a sump failure event, as a function of time, can be calculated using the frequency of a LOCA of that size, multiplied 1b3y the joint probability of being in a pump state and having a sump failure occur by that time, summed over all pump states and LOCA sizes. Note that we only attribute the AST ACE event to GSI-191 if the pump state has at least two containment spray pumps initially available.

In this section we restrict attention to estimating the frequency with which we observe GSI-191 sump failures as a function of time after a LOCA. In Section 2 we estimate dose as a function of the time at which the CSS is lost. Section 3 combines these two calculations to estimate: (i) the frequency with which AST ACE events occur that are attributable to GSI-191, and (ii) the fraction of GSI-191 sump failures that lead to an AST ACE event. We begin formalizing our calculations as follows.

Indices and Sets:

k = 1 ... , N index for set of pump states K C {1, ..., N} subset of pump states for which at least two containment sprays are available; i.e., K consists of pump state cases from Table 2.2.11 of Volume 3 [2] in which the "Working CS Pumps" column entry is at least 2 1

Events:

SL small LOCA ML medium LOCA LL large LOCA PSI pumps in state k S sumIp failure Paramieters:

fLOCA frequency of a LOCA event, LOCA = LL, ML, SL (events/CY) t index of time elapsed from LOCA event (minutes) fs(t) frequency of sump failure, by time t, for pump states indexed by K (events/CY)

Random Variable:

T time elapsed between LOCA and sump failure WVe calculate the frequency of a sump failure by time t for pump states in which at least two containment sprays are initially available, fs(t), by equation (1):

fs(t) [fLL " IP(PS,) ' P(T < tIPS, LL)+ (la)

I kEK fA1L Z IP(PS,). IP(T *< tIPS,, MlL) + (.b) kEA7 fSL " IP(PSk)

• P(T < tIPSk, SL)] . (1c) k rK K

We obtain fLL, fAIL, and fSL from the "GSI-191 PRA" column of Table 4-1 of Volume 2 [1].

We derive P(PSk) from Table 2.2.11 of Volume 3 [2]. Simulation output from CASA Grande yields estimates of P(T < tlPSk, LU, IP(T _ tIPSk, ML), and P(T < tlPS.,SL) for each LOCA category.

Figure 1 displays the conditional cumulative distribution function (cdf), P(T < tIPSk, LL), for several pump states, including the bounding pump states used in the analysis of Volumes 2 and 3 (see [1, 2]).

However, for the analysis we report here, we do not use bounding pump states. Rather, we use output from CASA Grande for all pumlp states in which at least two containment sprays are available. For the simulation results we have from CASA Grande, we observe no sump failures after a small or medium LOCA event; i.e., P(T < tIPSk, SL) = P(T < tiPSk, ML) = 0 Vt, k G K. Hence, the only contribution to fs(t) in equation (1) is from the large LOCA term in equation (1a).

By estimating the conditional cdf for each pump state in which a sump failure would be attributable to GSI-191 (i.e., those pump states indexed b1y, K, in which at least two containment sprays are available), we can apply equation (1) to estimate the frequency of a sump failure, for each time threshold t. The results are showrn in Figure 2.

2

CDF of Sump Failure Time, by Pump State 0.020

-Case 1 0.016

-Case 2

-Case 9 0.012

--- Case 22 II 0.008 -Case 26

-Case 42 0.004

........ Case 43 0.000 0 500 1000 1500 2000 Time (Minutes)

Figure 1: Conditional cumulative distribution function of sump failure for time elapsed from a large LOCA event, for seven example pump states. No sump failures occur after a small or medium LOCA event in CASA Grande output.

Sump Failure Frequency as a Function of Time Since LOCA Event Occurs 2.5E-08

• -2.OE-08 1.5E-08

-, 1 .OE-08 5.OE-09 O.OE+00 0 500 1000 1500 2000 t(Minutes)

Figure 2: Estimated frequency (events/CY) of a sump failure for time threshold t.

3

2 Determining the Time Threshold for an AST ACE Event We find an appropriate time threshold t for operation of the CSS after which the dose drops below 5.0 rem. By using conservative engineering calculations for (lose data as a function of time at the TSC and the CR from [4], we fit an exponential function to close with respect to CSS operation time, where our exponential function has the following form:

D(t) = Ae-Bt + C. (2)

Table 1: The first three columns contain CR and TSC dose data from engineering calculations [4]. The first column indicates the time (hours) since the LOCA event occurred. The second and third columnns indicate the CR dose and the TSC dose (rem). Thus, if the spray pumps fail at time 4.000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br />, the CR close is 4.760 rem and the TSC dose is 5.040 rem, according to these data. Columns four and five give the same values according to the fitted exponential functions. The final two columns indicate the squared error, with the bold values at the bottom indicating the sunm of squared errors (SSE).

Time CR Dose TSC Dose CR Fit TSC Fit CR SSE TSC SSE 2.185 6.700 6.330 6.699 6.330 1.82E-07 L.00E-09 4.000 4.760 5.040 4.761 5.040 8.82E-07 1.80E-07 4.250 4.620 4.950 4.622 4.950 3.81E-06 2.49E-08 5.000 4.310 4.750 4.309 4.749 1.14E-06 1.46E-06 6.500 3.980 4.540 3.978 4.540 4.43E-06 6.18E-08 7.600 3.870 4.470 3.866 4.472 1.23E-05 3.80E-06 10.000 3.770 4.420 3.774 4.417 1.48E-05 9.83E-06 15.000 3.750 4.400 3.746 4.401 2.OOE-05 9.58E-07 16.000 3.740 4.400 3.745 4.401 2.33E-05 3.78E-07 8.08E-05 1.67E-05 We determine the unknown parameters A, B, and C in equation (2) by minimizing the sum of squared deviations between the given data points and the estimates of those data points provided by the fit exponential curves. The first three columns of Table 1 present the close data used for the fitting procedure. In order to determine the best parameter values, we fit the exponential function given in equation (3) to all nine data points for both time series. Equation (:3) gives the fit equation for the Control Room, and equation (4) gives the fit equation for the Technical Support Center:

0 0 DCR(t) = 10.676655

• e .5878 29t + 3.744 (3)

DTSC(t) = 7.283738. e-0.6078874t +-4.400. (4)

The fourth and fifth columns in Table 1 give the values of equations (3) and (4) at the times in the first column, and the final two columns give the squared deviations between each data point and the fit equation. Figure 3 displays the dose as a function of the time the CSS operates (in hours).

An AST ACE event is said to occur if a sunip failure occurs, and hence we lose containment sprays, at a time when dose is 5.0 rem or more in either the CR or the TSC. We can use equation (2) to solve 4

CR and TSC Dose - Fitted Exponentials 8.0 CRDose 7.0 CR Fit 6.0 a TSCDose 5.0 .. TSCFit 5.0~

4.0 C3.0 2.0 1.0 0 2 4 6 8 10 12 14 16 CSS Operation Time (hours)

Figure 3: Fit exponential curves for the dose as a function of the time the CSS operates in the Control Room and the Technical Support Center.

for the time at which the CR and TSC curves reach a dose of 5.0 rem as shown in equation (5):

t B -In (5.0A C) (5a) 1n(5.0- 3744N tCR = -0.5878029-1 In 10.6766 ) = 3.641 hours0.00742 days <br />0.178 hours <br />0.00106 weeks <br />2.439005e-4 months <br /> = 218.4 minutes (5b)

S 1 6(50-4400 tTSC -0.6078874-1 In 5.0-4.400) = 4.107 hours0.00124 days <br />0.0297 hours <br />1.76918e-4 weeks <br />4.07135e-5 months <br /> = 246.4 nminutes. (5c)

Thus, an AST ACE event is said to occur in the Control Room if the sump fails at 218.4 minutes or earlier after a LOCA, and an AST ACE event is said to occur in the Technical Support Center if the sump fails at 246.4 minutes or earlier after a LOCA. Because we require both the CR and the TSC to have dose levels below 5.0 rein, we record an AST ACE event in the simulation model if we observe a sump failure by 246.4 minutes after a LOCA event.

3 Estimate of AST ACE Event Frequency We see from Figure 2 that if the threshold t is large (i.e., we assume that the dose exceeds 5.0 rein for large values of t), that the asymptote of fs(t) approaches 2.004E-8 sump failure events per calendar year. However, because the CSS is operational for the requisite 246.4 minutes in most cases for the dose to drop below the 5.0 rem limit in both the Control Room and Technical Support Center, we can see from Figure 2 that the frequency (value of the y-axis at an x-axis value of 246.4 minutes) is instead fs(246.4) =3.724E-9 events per calendar year. Thus, the Emergency Core Cooling System works well 51

enough that 3724E-9 2.004E-5 18.6% of the GSI-191 sump failures, when at least two containment sprays are initially available, lead to an AST ACE event.

4 Estimate of Initial Dose In order to gain perspective for the values reported in Table 1, which are based on the engineering calculations in [4], we review historical leakage data from STPNOC. The total containment leakage is continually updated per the STPNOC Local Leakage Program [3], and results froom containment penetration test measurements are recorded. A query of the results for both STPNOC Units from September 2002 to December 2013 (approximately 700 values) was analyzed to understand the vari-ability and magnitude of the values, and the results are summarized in the box plots of Figure 4.

The maximum measured value between both units in that time framne was 106,800 sccm (standard cubic centimeters per minute), approximately the same as two standard deviations above the mean rate of the data set. The containment leak rate used in the calculations discussed in Section 2 is 199,6:33 sccm, a factor of 1.87 larger than this maximumn historical value. We note that this value of 199,633 sccm is STP's Technical Specification limit (see page A2 of [4]).

Boxplot summary of STPNOC contairinment test measurements (sccm)

Unit I v zi i~

I I I I I 80000 85000 90000 95000 100000 105000 Observed Leakage Rate (sccml Figure 4: Boxplots of containment penetration test measurements from both STPNOC units. The threshold for an AST ACE event is a factor of 1.87 larger than the maximum observation across both units.

6

References

[1] Wakefield, D. and D. Johnson (2013, October). South Texas Project Risk-Informed GSI-191 Evalu-ation, Volume 2, Probabilistic Risk Analysis: Determination of Change in Core Damage Frequency and Large Early Release Frequency Due to GSI-191 Issues. Technical report, STPRICSI191-VO2, Revision 2, ABSC Consulting Inc.

[2] Letellier, B., T. Sande, and G. Zigler (2013, October). South Texas Project Risk-Informed GSI-191 Evaluation, Volume 3, CASA Grande Analysis. Technical report, STP-RICSI191-V03, Revision 2.

[3] C. Bradshaw and R. Stark, "Local Leakage Rate Test Calculations, Guidelines, and Program,"

STPNOC Procedure OPSPl1-ZA-0005, Revision 20, October 11, 2011.

[4] "Control Room, TSC, and Offsite LOCA Radiation Doses", STPNOC Engineering Calculation NC6013, Revision 17, 2014.

7