Benoit Forget is the KEPCO Professor and Department Head of Nuclear Science and Engineering at MIT. Since joining the faculty in 2008, he has led the Computational Reactor Physics Group (CRPG), known for developing the open-source codes OpenMC and OpenMOC—tools advancing high-fidelity reactor simulations on cutting-edge computing platforms. He teaches undergraduate and graduate courses in reactor physics, engineering, and radiation transport, and regularly co-organizes industry-focused short courses on reactor safety. Before MIT, he worked at Idaho National Laboratory on reactor methods development and fuel cycle analysis. Prof. Forget earned his Ph.D. in Nuclear Engineering from the Georgia Institute of Technology in 2006. He holds a Master of Engineering (M.Eng.) in Energy Engineering (2003) and a Bachelor of Engineering (B.Eng.) in Chemical Engineering (2001), both from École Polytechnique de Montréal. A Fellow of the American Nuclear Society, he chaired its Reactor Physics Division in 2011–2012 and received the ANS John Landis Young Member Engineering Achievement Award in 2013.
Monte Carlo neutron transport methods have long been considered a reference solution since they model explicitly the random walk of neutrons in a nuclear system using the fundamental nuclear data. Recent advances in computing power bring these methods closer to performing realistic core analysis and design. CRPG has developed an open-source Monte Carlo code, OpenMC, to perform large scale reactor analysis on modern computing architecture. CRPG focuses on the development of novel algorithms for improving the parallel efficiency of Monte Carlo methods, new communication schemes to reduce network load for massive tallies, new algorithms to capture time dependence and improved nuclear data models to reduce memory requirements and increase fidelity.
In addition to the Monte Carlo work, CRPG has also developed new concepts to attain high-fidelity 3D deterministic simulations. This work has led to the development of OpenMOC, a 3D method of characteristics code designed for efficient multi-threading and low memory requirement, as well as novel approaches such as Random Ray MOC and on-the-fly ray tracing for unstructured mesh geometry discretizations. This project seeks to reduce current bottlenecks limiting the applicability of such methods to full core analysis on both CPU and GPU architectures, and develop new acceleration schemes for more efficient analysis.
Multiphysics coupling
CRPG also focuses on the development of methods enabling coupling of reactor physics codes with fuel performance or thermal-hydraulics codes for both steady-state and transient analysis. Work has focused on the coupling of OpenMC with external frameworks using novel techniques to transfer information from varying geometrical representation and developing capabilities to use continuously varying material properties during simulation to avoid the need for fine discretization. This work also focuses on the development of new approaches for dealing with the temperature dependence of nuclear data as to provide accurate feedback.
Uncertainty Quantification
CRPG is also exploring the area of nuclear data uncertainties by looking at the sensitivity of resonance parameters and the propagation of uncertainties on any nuclear system. This information will not only provide a measure of nuclear data uncertainty for quantities of interest, but can also better inform nuclear data evaluators for future data evaluations.
22.212 Nuclear Reactor Physics II
22.05 Neutron Physics and Radiation Transport
22.211 Nuclear Reactor Physics I
22.213 Nuclear Reactor Physics III