By modeling a group of neighboring real particles as a single coarse-grained particle(CGP),discrete particle method(DPM)is now capable of simulating industrial-scale particle-fluid systems.However,a systematic approac...By modeling a group of neighboring real particles as a single coarse-grained particle(CGP),discrete particle method(DPM)is now capable of simulating industrial-scale particle-fluid systems.However,a systematic approach to determine the CGP properties and develop their interaction models is still lacking,which casts uncertainty on the predictivity of the method.In this study,collisions between predefined particle groups are analyzed to construct kernel functions for modeling the CGPs and then the model parameters are determined by equating the statistical properties of the CGPs and the real particles in the physical process studied.This approach is implemented for homogeneous cooling of granular gas,then demonstrated effective in simulating experimental fluidized beds.展开更多
The reactor-regenerator loop is the core facility of the maximizing iso-paraffin(MIP)process.Although the discrete particle method(DPM)simulation can provide detailed information at the particle scale,it has been unab...The reactor-regenerator loop is the core facility of the maximizing iso-paraffin(MIP)process.Although the discrete particle method(DPM)simulation can provide detailed information at the particle scale,it has been unable to simulate such a complex loop system due to limitations of coarse-grained(CG)models,computing software,and hardware.In this study,a newly proposed soft-shell CG-DPM model with a CG ratio of up to 800 is used to simulate a 3.5 Mt/a industrial-scale MIP reactor-regenerator loop.The solid fraction distribution obtained is found to agree well with in-situ measurements.Hydrodynamic properties including the distribution of solid fraction,gas and solid velocity,standard derivation of solid fraction with time,temporal distribution of the flow field,and particle residence time distribution are measured in the simulation,which are meaningful to better design and operate such systems in the future.展开更多
Large-scale atomistic simulation of low-dimensional silicon nanostructures has been implemented on a heterogeneous supercomputer equipped with a large number of GPU-like accelerators(GLA).In the simulation,an innovati...Large-scale atomistic simulation of low-dimensional silicon nanostructures has been implemented on a heterogeneous supercomputer equipped with a large number of GPU-like accelerators(GLA).In the simulation,an innovative parallel algorithm was developed for the combined utilization of the dynamic neighbor and static neighbor list algorithms aiming at the different regions of the nanostructures.Furthermore,some optimization techniques were performed for the computationally intensive many-body force evaluation between atoms,such as SIMD vectorization,manual loop unrolling,pre-calculation of memory addresses and reordering of data structure etc.Finally,the simulation achieved the excellent weak and strong scalabilities in the parallel implementation,where up to 805.3 billion silicon atoms were simulated.This development suggests an exciting future of predicting the thermodynamic properties of low-dimensional nanostructures.展开更多
基金supported by the National Key Research and Development Program of China(grant No.2020YFC1908805)the National Natural Science Foundation of China(grant Nos.22293024 and 22078330)the Youth Innovation Promotion Association,Chinese Academy of Sciences(grant No.2019050).
文摘By modeling a group of neighboring real particles as a single coarse-grained particle(CGP),discrete particle method(DPM)is now capable of simulating industrial-scale particle-fluid systems.However,a systematic approach to determine the CGP properties and develop their interaction models is still lacking,which casts uncertainty on the predictivity of the method.In this study,collisions between predefined particle groups are analyzed to construct kernel functions for modeling the CGPs and then the model parameters are determined by equating the statistical properties of the CGPs and the real particles in the physical process studied.This approach is implemented for homogeneous cooling of granular gas,then demonstrated effective in simulating experimental fluidized beds.
基金supported by the National Key Research and Development Program of China(grant No.2020YFC1908805)the National Natural Science Foundation of China(grant Nos.22293024 and 22078330)the Youth Innovation Promotion Association,Chinese Academy of Sciences(grant No.2019050).
文摘The reactor-regenerator loop is the core facility of the maximizing iso-paraffin(MIP)process.Although the discrete particle method(DPM)simulation can provide detailed information at the particle scale,it has been unable to simulate such a complex loop system due to limitations of coarse-grained(CG)models,computing software,and hardware.In this study,a newly proposed soft-shell CG-DPM model with a CG ratio of up to 800 is used to simulate a 3.5 Mt/a industrial-scale MIP reactor-regenerator loop.The solid fraction distribution obtained is found to agree well with in-situ measurements.Hydrodynamic properties including the distribution of solid fraction,gas and solid velocity,standard derivation of solid fraction with time,temporal distribution of the flow field,and particle residence time distribution are measured in the simulation,which are meaningful to better design and operate such systems in the future.
基金financially supported by the Beijing Natural Science Foundation under grant No.JQ21034the Major Research Program of Henan Province under grant No.201400211300+1 种基金the National Natural Science Foundation of China(NSFC)under grant Nos.21776280,22073103 and 91934302the Strategic Priority Research Program of Chinese Academy of Sciences under grant No.XDC01040100。
文摘Large-scale atomistic simulation of low-dimensional silicon nanostructures has been implemented on a heterogeneous supercomputer equipped with a large number of GPU-like accelerators(GLA).In the simulation,an innovative parallel algorithm was developed for the combined utilization of the dynamic neighbor and static neighbor list algorithms aiming at the different regions of the nanostructures.Furthermore,some optimization techniques were performed for the computationally intensive many-body force evaluation between atoms,such as SIMD vectorization,manual loop unrolling,pre-calculation of memory addresses and reordering of data structure etc.Finally,the simulation achieved the excellent weak and strong scalabilities in the parallel implementation,where up to 805.3 billion silicon atoms were simulated.This development suggests an exciting future of predicting the thermodynamic properties of low-dimensional nanostructures.
