Hello everyone, this is the small plate-type core that I modeled. Currently, I am using OpenMC to generate two-group cross-sections for each component and then input these cross-sections into the multi-group mode of OpenMC. However, there is an error of more than 3000 pcm in the results. What could be the problem?
For such a calculation, it may be that 3000 pcm is entirely reasonable depending on the type of materials, dimensions, etc.
Have you tried running a much smaller problem? Maybe begin with an infinite-homogeneous composition where you should be able to get exactly the same answer, then expand to two materials, and go from there. Just to make sure that the mechanics are working.
There could be many reasons for a 3000 pcm delta between continuous-energy OpenMC and multi-group OpenMC. Most of these differences come down to how you generate multi-group cross sections (MGXS). For compact (high leakage) cores you often need higher fidelity spatial homogenization, such as per-region MGXS, instead of per-material MGXS. Many reactors are quite sensitive to the scattering Legendre order (especially thermal systems where scattering off of light nuclei is forward peaked), so you might need to consider increasing the scattering order. Finally, you might just need more groups or an improved group structure to get decent results.
It’s hard to provide specific recommendations without knowing more about the core you’re modelling. My advise would be to check the literature to see if anyone else has modelled a similar system using Monte Carlo generated MGXS and try to adapt their homogenization procedure, group structure, scattering order, etc. to work for this core.
1 Like