Effect of Correlation Uncertainty Distribution on Design Limit CHFR

The design limit CHFR is evaluated deterministically with the correlation uncertainty alone, and the actual CHFR of the fuel is estimated conservatively determined with a hot channel factor (HCF) to consider the uncertainty of the heat flux from the fuel. This approach gives clear insight about the probabilistic information, and also decreases unnecessary conservatism of estimation. The statistical determination of design limit CHFR depends primarily on the distribution of correlation uncertainty, which is assumed on the actual experimental data. Since the experimental data does not coincide with the ideal distribution of normal or uniform, the proper assumption on the distribution is critical for the statistical determination of design limit CHFR. In this paper, the effect of the correlation distribution assumption on a design limit CHFR is examined for two different CHF correlations, Mirshak and Kaminaga. The design limit CHFR is estimated for different CHF correlations and different assumptions on the uncertainty distributions. Mirshak and Kaminaga correlation are selected as the CHF correlation and the design limit CHFR are estimated for them. The uncertainty distribution on Mirshak correlation is more similar to uniform than normal, therefore the uniform assumption on the distributions gives proper estimation on design limit CHFR with the correlation. The uncertainty of the correlation need to be defined for predicted/measured since the formulation of random normalized CHFR has the correlation uncertainty factor at the numerator. Defining the uncertainty for predicted/measured, the uncertainty distribution of Kaminaga correlation can be assumed as a normal distribution, and the design limit CHFR is less conservatively estimated comparing to that defined for measured/predicted.