Molten Salt Delayed Neutron (MoSDeN) is a tool used for reconstruction of delayed neutron precursor groups in molten salt reactors.

This tool had a previous version in this repository accessible with git hash b56528a4. That version has a publication associated with it, given here:

MoSDeN operates in three distinct stages: preprocessing, processing, and post-processing. Preprocessing loads in data and builds necessary files for processing. Processing runs the main bulk of the calculations for generating parameters. Post-processing handles figure generation and data analysis.

Preprocessing should be run to generate any data needed (if it does not already exist as processed data). This data is dependent on the energy of the irradiating neutrons as well as the fissile nuclide target.

The exact organization of raw, unprocessed data is flexible, with some notable exceptions:

• OpenMC chain files to be should all be in a subdirectory labeled with "omcchain" (see preprocessing.py for all keywords)
• ENDF NFY data should all be in a subdirectory labeled nfy
• ENDF NFY files should be named "nfy-.csv", so 235U would be nfy-092_U_235.csv.
• IAEA beta-delayed neutron emission data should be in a directory iaea and be called eval.csv (default when downloading data).

Data can be collected from different sources:

• OpenMC depletion chains: these give half-lives and independent fission yields (linearly interpolated energy dependence)
• ENDF data: these give (currently) cumulative fission yields (with energy dependence based on nearest energy)
• IAEA data: these give emission probabilities and half-lives

Processing consists of three steps:

1. Generate concentrations (or collect fission yield data).
2. Generate the delayed neutron count rate.
3. Fit a set of delayed neutron precursor group parameters that best fit the count rate.

The generation of concentrations varies based on the model used. The simplest model is the 0D scaled model, and uses cumulative fission yields. The concentration of each DNP is calculated as the cumulative fission yield over the decay constant of that DNP. The 0D flow model (not implemented as of 2025-09-02) uses OpenMC to incorporate decay chains and parasitic absorption effects, offering a better model of the DNP concentrations at each point during the irradiation and subsequent decay.

The generation of the delayed neutron count rate and non-linear least squares fitting methods do not change between different models.

Postprocessing handles plotting and data analysis from the processed results, including analysis of each step.

Download the repository from GitHub. The environment will also to be created by running conda env create -f environment.yml. This should be followed with conda activate mosdenv to activate the environment. Run pip install -e . to make the package available to use on the command line as mosden in the mosdenv environment. Download the data used in tests by running bash download_data.sh. Check that tests pass by running pytest or pytest -m "not slow" for the faster version. Use mosden -a <input.json> to do a full run, mosden -pre <input.json> for preprocessing, or mosden -post <input.json> for post-processing.

The input file contains the majority of parameters of interest. The command line arguments describe what stage of MoSDeN to run (preprocessing, processing, or post-processing), while the input file describes what should happen during each of those stages. The default.py and input files in examples can be used as a guide for formatting and what parameters can be included.

One of the parameters is the log level, which can be useful for collecting additional information about the simulation. This can be configured in the input file, but the default level of 20 is also the suggested level for collecting useful information while not overcollecting various debug outputs.

• [<10] is the debug level
• [<20] is the info level (This is the suggested level)
• [<30] is the warning level
• [<40] is the error level
• [<50] is the critical level