Auth: Zach Hartwig
Mail: [email protected]
Date: 04 Nov 16
There appears to be an issue using a user-defined particle source (via compilation of the user-implemented source.F90 MCNP source code file) with DAGMC geometry. I am not certain if this is a bug (either in MCNP or DAGMC) or something obvious that I am missing/not correctly implementing in MCPN input deck or DAGMC preprocessing steps.
The main issue is the following: When a user-defined source.F90 is used with DAGMC geometry, MCNP tracking fails very quickly within the first few particles with the following message:
bad trouble in subroutine hstory of mcrun
source particle no. 12
starting random number = 148359214513709
the weight of the current particle is zero or less.
Note that if the identical particle distribution is specified via the standard SDEF source cards (replacing and not using the source.F90) then everything works perfectly. Also note that this issue only appears with cells that have material defined for them. The use of the "void" card causes the run to continue to completion without errors.
The included source/source.F90 should be included in the DAGMC MCNP5 source code directory and compiled into the DAGMC executable. Then, preprocessing the model full (See "prepareModel.sh") with DAGMC best practices and running it will trigger the issue (if the MCNP SDEF card has been commented out on the "DAGMCSourceTesting.mcnp" deck
The following scripts are useful
- The "prepareModel.sh" script performs all DAGMC preprocessing
- The "runModel.sh" script runs the DAGMC calculation
- The "cleanPrep.sh" and "cleanModel.sh" delete transient files