I am working on a project that involves designing a pebble bed reactor. I was wondering if anyone here has tried running OpenMC in multigroup mode using homogenized cross sections generated from pebble lattices and how it went. If it’s a rough process, then we can just use AGREE for the multigroup and only generate the cross sections on OpenMC.
Hi there, I know it’s been a long time since you posted this but I am starting a new project that includes modelling a Pebble Bed Reactor using OpenMC. Did you manage to run your code efficiently? How was the experience?
This was for a school project in which we had a lot of freedom, so we ended up switching to a TRISO compact fuel. However, even that faced significant computational hurdles so I used the “Reactivity Equivalent Physical Transformation” method. This allowed us to only have to simulate a single pincell of explicitly modeled TRISO compact and use the eigenvalue as a target for a tuned homogenized pincell, which was used for full core simulations (citation we used: Golesorkhi, S., et al. “Impact of the TRISO Fuel Homogenisation Approximation in the MHTGR.”
Annals of Nuclear Energy, vol. 183, Apr. 2023, p. 109634, https://doi.org/10.1016/j.anucene.2022.
109634.). OpenMC simulations with billions of surfaces is extremely slow (I’m talking 5 particles per second for a single TRISO compact pincell). SERPENT does not use the same method that OpenMC does and can handle high surface counts for more easily, if that is an option for you. If you have to use OpenMC for this, you will need to use some kind of homogenization technique like we did or perform multigroup cross section generation with individual pebbles and build a nodal core, as I suggested in my original post (parametrize cross sections by pebble packing density, assign each node a random packing density within the expected range, ect depending on how complex you want to get). I still have not gotten around to using multigroup mode in OpenMC, so I can’t be of much help if you pursue that. Best of luck.