Monte Carlo-linked depletion methods have gained recent interest due to the ability to model complex three-dimensional geometries using continuous-energy cross sections. The integration of CINDER90 into the MCNPX Monte Carlo radiation transport code provides a completely self-contained Monte Carlo-linked depletion capability in a single Monte Carlo code that is compatible with most nuclear criticality (KCODE) particle tracking features in MCNPX. MCNPX depletion tracks all necessary reaction rates and follows as many isotopes as cross-section data permit. The objective of this work is (a) describe the MCNPX depletion methodology dating from the original linking of MONTEBURNS and MCNP to the first public release of the integrated capability (MCNPX 2.6.B, June 2006) that has been reported previously, (b) further detail the many new depletion capability enhancements since then leading to the present Radiation Safety Information Computational Center (RSICC) release, MCNPX 2.6.0, (c) report calculation results for the H. B. Robinson benchmark, and (d) detail new features available in MCNPX 2.7.A.

Each version of MCNPX depletion starting from MCNPX 2.6.A leading to the official RSICC release of MCNPX 2.6.0 and the new beta release MCNPX 2.7.A included significant upgrades that addressed key issues from earlier versions. This paper details these key issues and the approach utilized to address the issues as enhancements for MCNPX 2.6.0. The MCNPX 2.6.0 depletion capability enhancements include (a) allowing the modeling of as large a system as computer memory capacity permits; (b) tracking every fission product available in ENDF/B VII.0; (c) enabling depletion in repeated structures geometries such as repeated arrays of fuel pins; (d) including metastable isotopes in burnup; and (e) manually changing the concentrations of any isotope during any time step by specified atom fraction, weight fraction, atom density, or gram density. These enhancements allow better detail to model the true system physics as well as to improve the robustness of the capability.

H. B. Robinson benchmark calculations were completed to assess the validity of nuclide predictability of MCNPX 2.6.0. The results show comparisons of key actinide and fission products as compared to experiment and the SCALE-4 SAS2H sequence 27-group cross-section library (27BURNUPLIB) results. MCNPX 2.6.0 depletion results are within 4% of the experimental results for most major actinides.

Two major depletion enhancements are available in the MCNPX 2.7.A beta release: improved 63-group flux querying and parallelization of the burnup interface routines in multiprocessor mode. Fixing the energy group querying routine does correctly tally the energy flux for use with isotopes not containing transport cross sections; however, results show <1% change in nuclide prediction for the benchmark test case. MCNPX 2.7.A parallelizes the depletion interface routines and running of CINDER90 so that different burnable regions of a given depletion system can be preprocessed, burned, and postprocessed on separate slave processors. The parallelization involves minimal communication between processors and therefore leads to significant computational performance enhancement.

The combination of new enhancements and testing of the MCNPX 2.6.0 depletion computational system make this capability a valuable Monte Carlo-linked depletion tool. Additional testing and feature enhancements are under development to further improve the usefulness of the computational tool.