Nuclear Engineering and Design 119 (1990) 67-95 67
North-Holland
QUANTIFYING REACTOR SAFETY MARGINS
PART 4: UNCERTAINTY EVALUATION OF LBLOCA ANALYSIS
BASED ON TRAC-PF1/MOD 1 *
G.S. LELLOUCHE and S. LEVY
s. Levy Incorporated (SLI), 3425 S. Bascom Ave., Campbell, CA 95008, USA
and
B.E. BOYACK (LANL), I. CATTON (UCLA), R.B. DUFFY (INEL), P. GRIFFITH (MIT),
K.R. KATSMA (INEL), R. MAY (SLI), U.S. ROHATGI (BNL), G. WILSON (INEL),
W. WULFF (BNL), N. ZUBER (NRC)
Received 12 January 1989
The Nuclear Regulatory Commission (NRC) has issued a revised EmergencyCore Cooling System (ECCS) rule allowing the
use of best estimate computer codes for safety analysis. The rule also requires an estimation of the uncertainty in the
calculated system response and a comparison of the resulting bound with the acceptance limits of 10CFR Part 50. To support
this revised rule the NRC and its consultants and contractors have developed and demonstrated the Code Scaling,
Applicability and Uncertainty Methodology (CSAU). The last of the three elements of the methodology - Uncertainty
Evaluation - is described in this paper.
1. Introduction
This paper summarizes the results of an uncertainty
evaluation of a Pressurized Water Reactor (PWR) Large
Break Loss of Coolant Accident (LBLOCA) analysis
and completes the presentation on the CSAU methodol-
ogy started with the preceding papers. This paper is in
four parts:
L Calculation of TRA C uncertainty (CSA U steps 3-12)
This section discusses the establishment of the
TRAC calculational matrix and reprises some of
the work presented earlier [1,2]. It considers how to
reduce the number of TRAC calculations to a
feasible number and how we develop the Response
Surface (RS) using Regression Analysis (RA). Fi-
nally, it discusses the establishment of a probabil-
ity distribution function (pdf).
11. Calculation of biases (CSA U steps 8-10)
This section deals with a number of phenomena
* Work supported by the USNRC office of Nuclear Regu-
latory Research, under DOE Contract No. DE-AC07-
761301570.
IlL
IV.
which are considered significant but are not mod-
eled or mismodeled in TRAC. We also consider the
effect of various parameters not considered indi-
vidually significant which, because of large uncer-
tainties in their variation, could become so, or
which in combination may become so.
Combining the uncertainties and biases (CSA U steps
13-14)
In this section we present the conditions under
which the biases and the TRAC pdf may be com-
bined and establish the final LBLOCA uncertainty
table.
Comparison with experimental data (CSA U step 13)
In this final section we consider the direct compari-
son of TRAC with data. This is done at the indi-
vidual transient level and for blowdown and late
reflood at the level of the distribution functions.
2. Conclusions
This uncertainty evaluation for a PWR LBLOCA
analysis shows that:
0029-5493/90/$03.50 © 1990 - Elsevier Science Publishers B.V. (North-Holland)