Evaluation of Memory Optimization in a Large-Scale Structural Finite Element Pre-processor

As one of important tools in the analysis of the design of nuclear power plants, finite element analysis methods have been widely used in the reactor structural analysis. ATLAS is a program for rapid analysis of finite elements of large-scale structures developed to analyze complex structures in nuclear engineering. ATLAS pre-processor has the basic realization of various functions of finite element pre-processor, the functions of importing geometry, meshing, etc. To avoid frequent system calls which can cause performance degradation, the memory pool algorithms are implemented to reduce memory fragmentations, especially for the large units of data and node data. The simulation results indicate that the designed memory pool can improve the speed of memory allocation and release by two digits in some cases. Although, for the detailed operations and support functions, there is still a large gap between ATLAS pre-processor and commercial finite element software, with the more understanding of the complex structures in nuclear engineering, ATLAS pre-processor will be upgraded in the future.