χ−MeRA: A Physics Based Adaptive Mesh Refinement Method and Program for Mesh Based Monte Carlo Tallies

Abstract

The solution to the coupled multi-physics problem is of keen interest to the reactor physics community. Many methods have been developed to solve this system; however, they all have one common aspect: they require a large amount of computational time and memory. Additionally, these simulations rely on large-third party codes that have a steep learning curve. χ−MeRA was built to solve the problem of large computational time and memory required by traditional methods for mesh-based tallies.

χ − MeRA couples a Monte Carlo (MC) code with an adaptive mesh refinement (AMR) algorithm. By coupling these codes, χ − MeRA can take advantage of the time and memory savings inherent to an AMR algorithm, and the accuracy of the MC code. Several test cases were used to assess the performance and accuracy of χ − MeRA compared to a fully refined mesh as is the typical method. For the first test case, χ −MeRA used 1,370.6% less memory and 625.5% less computational time. For more detailed test cases, χ − MeRA made it possible to obtain mesh-based tally results at the desired refinement level. Additionally, the tally results pulled from χ − MeRA are also compared to reactor performance data, where great agreement between the shape of the tally results and the experimental data was observed. Thus, χ − MeRA helps solve the coupled multi-physics problem in a faster, more computationally efficient manner, and the final mesh created contains accurate results that can be passed onto the next physics code to complete the multi-physics coupling.