Development And Application Of Atomic Kinetic Monte Carlo Simulation Program For Radiation Damage Of RPV Steel

The role of nuclear energy in saving energy and reducing emissions,optimizing the global energy structure and realizing green development has been widely recognized worldwide.At present,nuclear energy has exerted a significant impact on many fields such as national defense,scientific research,medical treatment,industry,agriculture,aviation and ocean.With the continuous development of modernization,China’s energy demand will further increase in the future.At the same time,once the nuclear power plant accident will cause serious radioactive pollution to the ecological environment.Therefore,the operation safety of nuclear power plants has always been the focus of attention.Nuclear reactor pressure vessel steel is the key equipment which cannot be replaced in nuclear power plant.Its service safety determines the operation safety and service life of the whole nuclear power plant.The long-term exposure of high-energy neutron to RPV steel will induce the formation of nanoscale Cu-rich clusters in the material,and these clusters will lead to hardening and embrittlement of RPV steel by hindering dislocations,which significantly reduces the macroscopic properties of RPV steel.These nanoclusters have a wide range of sizes and have structural differences at different temperatures,Observation by experimental means is relatively complicated and difficult.In recent years,with the rise of the field of scientific computing and the rapid development of high-performance computing in China,the use of computer simulation technology has opened up a way to verify theoretical models and study the mechanism behind experimental phenomena.As a representative simulation method,kinetic Monte Carlo can complete large-scale numerical simulation at atomic level over a long time scale,which is very suitable for studying the radiation damage mechanism of RPV steel.In this paper,Open KMC program based on atomic kinetic Monte Carlo method is developed and optimized,and applied to simulate the radiation damage of RPV steel,the interaction between impurity atoms and radiation-induced point defects was studied in depth theoretically.Finally,the large-scale parallel scalability test was completed on "Sunway Taihu Light".The specific research contents and conclusions are as follows :(1)The embedded atom method energy calculation model is developed in Open KMC program,so that the program can use EAM potential to complete the long time RPV steel evolution simulation process.Two potential models,Pair potential and EAM potential,were adopted respectively,and the range of interaction between atoms under different potential functions was fully considered.A vacancy selection optimization algorithm is proposed.which improves the efficiency of the program by reducing redundant vacancy updating operations.TAU tool is used to locate the communication hot spot function of the program,and the reasons that affect the communication performance are analyzed.Finally,the corresponding communication optimization strategy is given.(2)Fe-Cu binary alloy is used as RPV steel material,and a certain number of vacancy is introduced.The optimized Open KMC program is used to complete the numerical simulation of250 million particles in the long time scale under the Pair and EAM potential energy models.The precipitation phenomena of pure Cu clusters and Cu-Va complex clusters are verified by simulation results.It is found that the structure of Cu-Va complex clusters is composed of vacant clusters wrapped by Cu atoms from the outer layer.Then,the average size and number density of clusters in different size ranges showed an increasing trend.At the same time,the larger the size range,the smaller the number of clusters,but the coarsening of clusters is more obvious,and the fluctuation of the average size and number density is stronger.Different number of vacancies were introduced into the system for comparative analysis.More vacancies accelerated the aggregation process of Cu atoms,and the increase of vacancies had little effect on the overall number density of clusters,but could promote the coarsening of larger clusters.Finally,the precipitation propulsion coefficients at 663K～773K are calculated and compared with the experimental results.It is found that the simulation results have the same variation trend with the experimental results.When the temperature increases,the simulation results corresponding to the EAM potential energy model have a good fit degree with the experimental results.(3)A trillion-particle scale parallel scalability test has been completed on the domestic supercomputer "Sunway Taihu Light".The Pair potential and EAM potential were used for scalability tests.In the strong scalability test,the total number of cores is gradually increased from 10,000 cores to 2.6 million cores with a fixed problem size.When the number of cores was less than 130,000 and 65,000 cores respectively,the test results corresponding to Pair potential and EAM potential both showed the superlinear acceleration ratio.In the weak scalability test,the workload of each process remains unchanged,the total number of cores gradually increased from 65 to 5.85 million,and the maximum size of simulated particles reached 1.44 trillion.The test results of Pair potential and EAM potential always maintain the ideal parallel efficiency of about 80%.