Skip to main content
Open Access Publications from the University of California

UC Berkeley

UC Berkeley Electronic Theses and Dissertations bannerUC Berkeley

Validations of Computational Codes of Molten Salt Reactors


As one of the six advanced reactor concepts selected by the 2002 Generation IV roadmap as a technology meriting future research, Molten Salt Reactors (MSRs) attracted broad attention and multiple private and public entities are working towards its commercialization. Facing stringent regulatory requirements, validations of computational codes used to calculate and prove the safety characteristics of MSRs play a key role.

This project developed the world-first, MSR-related reactor physics benchmark basing on the series of zero-power experiments of the Molten Salt Reactor Experiment (MSRE) for the International Reactor Physics Experiment Evaluation Project (IRPhEP) handbook, in order to fill the knowledge gap of MSR benchmarking.

The role of the MSRE was to demonstrate the practicality of this high-temperature fluid-fuel reactor concept. Design of the MSRE was initiated at the Oak Ridge National Laboratory (ORNL) in 1960. It was a successful experiment, demonstrating key features of the liquid- fuel MSR, enhancing confidence in the practicality and performance of MSRs and leaving numerous experimental data from nuclear operation.

To reconstruct the MSRE, a three-dimensional high-fidelity benchmark model was developed by Monte Carlo neutron transport code Serpent2 with new methods developed to account for the unique feature of fuel salt motion in the core. The calculated effective multiplication factor, keff, for the criticality experiment, when 235U was progressively added to the fuel salt in order to achieve criticality with stationary salt and isothermal conditions, was 1.02132 (±3 pcm). The total uncertainty for experimental keff was estimated to be 420 pcm. The calculated keff is 2.154% larger than the experimental and benchmark model value, which is approximately 5 times the benchmark model uncertainty. It is to be noted that, for systems containing large volume of graphite (or other carbonaceous materials), Monte Carlo codes tend to overestimate the keff of the benchmark model by 1% to 2%. The bias is possibly due to uncertainties in the impurity content of the graphite blocks, the accuracy of the neutron capture cross section of carbon and the accuracy to model the nuclear-grade graphite.

The calculated reactivity coefficient of 235U concentration on the MSRE benchmark model (0.2228±0.0014, represented as the change of reactivity over the relative change of 235U mass in loop) matches well with the experiment value (0.223±0.006), strengthening the confidence of accurate representation of the fuel salt composition in the MSRE benchmark. Most of other reactivity effect calculations, including the control rod bank worth, reactivity effects of fuel circulation and isothermal and fuel temperature coefficients show good agreement with experiment values (within 1σ) as well.

This dissertation also illustrates the scenario of building a series of neutronics benchmarks for the Fluoride-salt-cooled High-temperature Reactor (FHR), a conceptual MSR without fueled experiments to compare. This is a code-to-code verification benchmark and can be used to verify the credibility of neutronics codes in modeling reactors with TRISO particle type fuel and a pebble bed geometry.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View