Random sampling method is adopted in many fields of engineering and is currently been used to propagate uncertainties from nuclear data to behavior parameters of nuclear systems. It is also of great importance to know the effect of uncertainties inherent to physical components in order to calculate the total uncertainty associated with a model simulation. The propagated uncertainty is part of criticality safety calculations and guide vendors about accepted tolerance limits of the parts. In this work random sampling method for quantifying manufacturing uncertainties is accomplished by determining variance on neutron multiplication factor due to physical parameter uncertainties. Sample size and computational uncertainty were varied in order to investigate sample efficiency and convergence of the method. Random sampling efficiency was improved through the use of an algorithm for selecting distributions. Mean range and standard deviation range were compared with reference true values in order to verify the sampling. Transport code MCNPX was used to simulate a benchmark experiment and allow the mapping, from uncertain inputs to uncertain outputs. Replication-based approach with internal measure of accuracy was defined as convergence criterion to the method. For a fixed computational time, in order to reduce the variance of the uncertainty propagated, it was found, the sample of size 186 is the best alternative among the tested cases. It serves as a reference sample size for future use in sampling based method. If this method is used to propagate uncertainties to keff in determining the upper limit of criticality for example, the value of result is lower than the value obtained by mean of the conservative method, allowing design optimization.

Full text: IJEIR_1699_Final.pdf