Monte Carlo simulations made better with new CMGC model

DOI: 10.1007/s41365-020-00793-8

One sentence summary:

A newly developed computer-aided design model to Monte Carlo geometry converter (CMGC) promises a highly efficient constructive solid geometry (CSG) representation.

CAD STEP标准文件到蒙特卡罗构造实体几何的自动转换方法，设计了一种全分解转换策略，开发了CAD到蒙特卡罗几何转换程序CMGC。

http://www.nst.sinap.ac.cn/newsDetails/112/3101/en/

The Novelty (What)

This study elaborates the algorithms, implementations, and applications of a new computer-aided design (CAD) model to Monte Carlo geometry converter (CMGC). This newly developed conversion code promises highly efficient Monte Carlo simulations by generating appropriate constructive solid geometry (CSG) representation. As a result, this automatic conversion can produce heuristic strategies for solid splitting, reduce the time taken for particle transporting, provide low relative volume error, and has a friendly graphical user interface. At the end of the CMGC conversion, the converted models show high accuracy and improved efficiency in particle transport calculations.

The Background (Why)

The computer-aided design (CAD) model to Monte Carlo geometry conversion is recognized as the most effective automatic conversion method compared to the conventional boundary representation model. Recent researches have been heavily focused on improving the efficiency of large-scale and detailed Monte Carlo modeling. Among the areas of focus are to reduce manual work, increase conversion speed, and improve conversion accuracy. This study aims to provide an in-depth analysis of a  new conversion method  and demonstrate its improved Monte Carlo simulation efficiency.

The SDG impact (Big Why)

The Monte Carlo methods are very  common and preferred when designing sophisticated nuclear systems due to their capability of dealing with complex problems with high accuracy. Radiation technology and computational medicine physics, among other fields, are known to benefit from an efficient and safe interface such as the CMGC model mentioned in this study. Therefore, the detailed description in this study would serve as a good reference for future infrastructure upgrading with increased resource-use efficiency and accuracy, as proposed in the 9^th UN Sustainable Development Goal: Industry, Innovation and Infrastructure.

蒙特卡罗辐射输运方法是核科学与技术领域重要的数值计算方法之一。为了达到高效的粒子径迹计算，蒙卡方法普遍采用基于基本体元及其布尔运算的构造实体几何CSG来描述三维计算域，建模难度大且容易出错，复杂实体建模成本超过计算成本。为此，清华大学工程物理系和中物院高性能数值模拟软件中心联合团队研究了CAD STEP标准文件到蒙特卡罗构造实体几何的自动转换方法，设计了一种全分解转换策略，开发了CAD到蒙特卡罗几何转换程序CMGC。全分解转换策略使用直接分解面、间接分解面和辅助分解面将实体分解成一系列子实体，并尽量降低子实体的数量，从而生成有利于提升蒙特卡罗输运计算效率的CSG表达式。利用实体表面之间的邻接关系、曲率、凸凹性等信息，以及实体表面顶点与曲面的相对位置关系实现分解面的选取，减少布尔运算，提高转换成功率。此外，我们只为包含凹曲面的子实体补充辅助面，从而降低了辅助面的冗余度。目前CMGC支持面向MCNP、Geant4、FLUKA、JMCT和MCShield的CSG模型转换，CMGC提供友好的图形用户界面，用户利用CMGC可以实现CAD模型和装配信息显示、重复结构提取、材料信息设置、CSG模型并行转换等功能。

我们使用一系列包含不同特点的CAD模型对CMGC进行了验证，并使用蒙特卡罗粒子输运程序JMCT对转换得到的CSG模型进行辐射输运计算，计算结果显示CMGC转换正确、转换得到的CSG模型具有较高的输运效率。图中所示是一个加速器模型，可见其三维CAD模型与转换得到的三维GDML模型一致，二维切面与转换得到的MCNP模型二维切面一致。此外，某光学组件转换模型的粒子追踪效率比基于特征的手建模型的追踪效率提升7倍。

后续，我们将持续优化和改进CMGC的算法和功能。