Study Of Key Techniques Of Beam-target Coupling Simulations In The High-power Dense Granular Flow Target

As an important part of future energy systems,nuclear fission energy has the advantages of small size,high energy density,continuous and stable output,and low pollution compared with new energy and traditional fossil energy.However,the existing nuclear fission energy technologies still face the problems of low utilization rate of uranium resources and difficult disposal of nuclear waste.Accelerator Driven Sub-critical System(ADS)is one of the most promising nuclear energy technologies in the future,which uses the high-energy proton beam generated by the accelerator to bombard the heavy metal spallation target in order to release high-throughput,hardspectrum neutrons to drive nuclear waste transmutation and nuclear fuel proliferation.As the neutron source of ADS,the beam power that the spallation target can bear determines whether the range and the flux of neutron energy spectrum meet the needs of transmutation and proliferation.Because of some technology problems in existing solid or liquid targets,the development of ADS has been restricted.In order to overcome the shortcomings of solid or liquid targets,the Chinese Academy of Sciences innovatively proposed a new type of spallation target based on existing target technologies,called Dense Granular Flow Target(DGT).Since there is no precedent for the construction of DGT,no experiences can be referenced in DGT researches,but the use of computer numerical simulation is a practical way.The dynamical and thermal problems of DGT is one of the key contents of ADS researches,which is related to the long-term stable and safe operation of ADS.In computer numerical simulations,the calculation of energy deposition of grains after being bombarded by a high-energy beam during flow processing is the core problem of dynamical and thermal simulations of granular flow targets.Although the equivalent homogenization method can realize the synchronized calculation of granular dynamics and energy deposition,the calculation results are quite different from those of Monte Carlo particle transport method due to ignoring the uneven distribution of granular mass in space.The discrete energy deposition calculation method can obtain more accurate calculation results,but it is difficult to be synchronously calculated with the dynamics due to the huge amount of calculations.This work is to analyze and optimize the process of marking cells of the energy computing space and the process of calculating energy deposition in the previous proposed discrete energy deposition calculation algorithm based on GPGPU parallel computing,so as to realize the high-performance and synchronous calculation of granular dynamics and energy deposition.In the process of marking cells of energy computing space,through the analyses of the Euler Searching Method used in the previous algorithm,it was pointed out that the performance bottleneck of the Euler Searching Method comes from the large memory access latency caused by frequent granular data accessing,resulting in poor performance of the algorithm.In the optimized algorithm,the Lagrange-Euler Mapping Method was proposed to reduce the frequency and latency of memory access in the marking process.In the process of calculating energy deposition,it was found that the performance bottleneck also comes from the large memory access latency caused by massive data writing,so an improved scheme using thread aggregation writing method was proposed to reduce the amount of data writing.From testing results,the performance of the optimized parallel discrete energy deposition calculation method has been greatly improved,making the granular target simulation performance on A100 GPU and K80 GPU improved by 85.38% and206.92%.In ADS target simulations with scan beams,the improved algorithm can quickly and accurately calculate the energy deposition on grains in the dynamical and thermal simulations of DGT,achieving the study of energy deposition distribution within the DGT effected by scan beam frequency,beam energy,and heat transfer,and verifying the significance of the algorithm for DGT researches of ADS.