Multi-function and generalized intelligent code-bench based on Monte Carlo method (MagicMC) for nuclear applications

摘要: The Monte Carlo (MC) method offers significant advantages in handling complex geometries and physical
processes in particle transport problems and has become a widely used approach in reactor physics analysis,
radiation shielding design, and medical physics. However, with the rapid advancement of new nuclear en-
ergy systems, the Monte Carlo method faces challenges in efficiency, accuracy, and adaptability, limiting its
effectiveness in meeting modern design requirements. Overcoming technical obstacles related to high-fidelity
coupling, high-resolution computation, and intelligent design is essential for using the Monte Carlo method
as a reliable tool in numerical analysis for these new nuclear energy systems. To address these challenges,
the Nuclear Energy and Application Laboratory (NEAL) team at the University of South China developed a
multifunctional and generalized intelligent code platform called MagicMC, based on the Monte Carlo particle
transport method. MagicMC is a developing tool dedicated to nuclear applications, incorporating intelligent
methodologies. It consists of two primary components: a basic unit and a functional unit. The basic unit,
which functions similarly to a standard Monte Carlo particle transport code, includes seven modules: geometry,
source, transport, database, tally, output, and auxiliary. The functional unit builds on the basic unit by adding
functional modules to address complex and diverse applications in nuclear analysis. MagicMC introduces a
dynamic Monte Carlo particle transport algorithm to address time-space particle transport problems within
emerging nuclear energy systems and incorporates a CPU-GPU heterogeneous parallel framework to enable
high-efficiency, high-resolution simulations for large-scale computational problems. Anticipating future trends
in intelligent design, MagicMC integrates several advanced features, including CAD-based geometry modeling,
global variance reduction methods, multi-objective shielding optimization, high-resolution activation analysis,
multi-physics coupling, and radiation therapy. In this paper, various numerical benchmarks—spanning reactor
transient simulations, material activation analysis, radiation shielding optimization, and medical dosimetry anal-
ysis—are presented to validate MagicMC. The numerical results demonstrate MagicMC’s efficiency, accuracy,
and reliability in these preliminary applications, underscoring its potential to support technological advance-
ments in developing high-fidelity, high-resolution, and high-intelligence MC-based tools for advanced nuclear
applications.