OpenMC Tutorial | Build, install & run nuclear simulations

[@maxtrap1947]

Thanks Luke! Wearing my CA t-shirt here in Seattle😎

[@joelnyumuah6310]

I am working on rector burnup simulations and this software is new to me. I will like to get in touch with you so you could share more knowledge with me. I am a master's student in Nuclear Energy Engineering Studying in China Thank you

[@thedancingscientist8180]

Hi there Luke! I'm so glad I found this. I'm an applied physics degree student doing a paper on neutronic calculations using OpenMC. I know what Monte-Carlo is but I'm still a newbie at it. That brings me to ask, with regards to the code itself, does it define the dimensions of the reactor? Follow up question: Would I be able to manipulate the dimensions of the reactor through the code itself?

[@CopenhagenAtomics]

Hi! Thanks for the question. So OpenMC in itself, is always a 3D-code - and so you do define your reactor in 3D in all cases. That said though, you may restrict the geometry by means of reflective surfaces. This could be used to emulate a 2D or even a 1D-problem quite easily and efficiently. Hope that helps, otherwise let us know :))

[@thedancingscientist8180]

@@CopenhagenAtomics thanks a lot! Also, if I wanted to extract numerical information (i.e multiplication factor, cross-sections, buckling, reactor core dimensions) once OpenMC has completed the simulation, would I be able to?

[@PhilipRhoadesP]

I came here to learn about MSRs and found JuNest! - I am mostly a Fedora Linux person and am familiar with Docker / Podman but JuNest is interesting and I will have a closer look . .

[@muhammadimron5849]

That unsymmetrical flux maybe related to the statistical nature of Monte Carlo method. Have you tried to use larger number of particle histories? ideally the standard deviation should be less than 1% if you want to get near symmetric flux. Also the was flux tally done in mesh-wise? If the meshes are very small, then you would need larger number of histories to get lower flux uncertainties. That because the smaller the mesh, the fewer collisions were tallied in each mesh (I am assuming you used collision based estimator to calculate the flux).

[@CopenhagenAtomics]

Thanks for the suggestions. This was a mesh tally, but since the asymmetry is fairly diffuse, I wouldn't expect increasing the number of histories to help. I also think it is too pronounced to be purely stochastic. To confirm this, you can change the seed for the random number generator (via openmc.Settings.seed), and see that the asymmetry persists. I've since looked further into it, and I think it is more likely a consequence of the asymmetric geometry in the core. There are seven detector tubes in the fuel region on the side of the reactor with reduced flux relative to the other.

[@catnadas]

Thank you! Way to go!

[@AJ5]

Thank you for making this tutorial and putting it out there! Is there any benefit to running an Arch-based distro over a Debian-based one like Ubuntu?

[@CopenhagenAtomics]

The particular benefit in this case is a simpler installation procedure than most others currently. Installation of OpenMC, its dependencies, as well as nuclear data download and the CAD_to_OpenMC mesher are configured and installed with essentially one command via the Arch package manager `yay`. If you prefer to use other distros, we have also put together build scripts for Debian & Ubuntu here: github.com/openmsr/openmc_install_scripts which perform manual source installs.

[@bamideleebiwonjumi2547]

I see that you have MOAB working with your arch based openmc installation, is it also easy to get dagmc working with my archbased installation?

[@CopenhagenAtomics]

Yes! This installation includes the DAGMC toolkit as a dependency, along with MOAB. So both will be built and installed automatically along with OpenMC. Both examples in the video are in fact using CAD geometry, and thus rely on DAGMC to run.

[@mr_happygolucky7095]

-> Failed to install the following packages. Manual intervention is required: openmc-git - exit status 4 How do I intervene manually?

[@mortennygaard5335]

Hi Luke! Do you have good recommendations for a cheap PC for a student, that will still run OpenMC well?

[@CopenhagenAtomics]

Hi @mortennygaard5335. Thanks for reaching out! For student purposes, I think any PC is fine (Mac is trickier but still doable). Windows now supports Linux environments (Windows Subsystem for Linux). Depending on the number number of particles used, it can have a sizable memory footprint. I think 8-16GB is fine for most basic purposes. If you plan on meshing CAD geometry or running larger simulations, you might want more. Parallelization can really reduce runtime, but I think most laptop PCs are going to have a similar number of cores anyway. If it’s a desktop, you might wanna go for a higher number of cpus.

[@raygarrh4510]

Hello ! Thank you for the tutorial, I followed all the steps but unfortunately when I try to run simulations I have "ModuleNotFoundError" for modules 'CAD_to_OpenMC' and 'openmc', do you have any idea why and how can I solve it ? Best regards

[@farukapece]

Manually Install missing dependencies and retry installation using yay -Syy openmc-git. I faced the same problem and resolved it this way.

[@CalegarioLeal]

Hey, any tips on how to manually install? @@farukapece

[@masalotaibi7448]

@farukapece can you explain in details please 😊

[@masalotaibi7448]

@@farukapeceplease give me the steps if you can 😊

[@masalotaibi7448]

Did you solve it? If you did it please give me the steps if you can 😊