Premium
Multiscale Discrete Supercomputing – A Game Changer for Process Simulation?
Author(s) -
Ge Wei,
Lu Liqiang,
Liu Shiwen,
Xu Ji,
Chen Feiguo,
Li Jinghai
Publication year - 2015
Publication title -
chemical engineering and technology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.403
H-Index - 81
eISSN - 1521-4125
pISSN - 0930-7516
DOI - 10.1002/ceat.201400746
Subject(s) - supercomputer , computer science , software , process (computing) , computational science , speedup , simulation software , high resolution , parallel computing , operating system , remote sensing , geology
To address the challenge of virtual process engineering (VPE), a systematic co‐design of the physical model, numerical software, and computer hardware was conducted, eventually leading to the so‐called EMMS Paradigm, which can greatly accelerate simulation speed while reducing its cost. Following this paradigm, a multiscale software framework was developed, coupling continuum and coarse‐grained discrete simulations and a multiscale hardware architecture with different parallelization schemes, which have drawn the dream of VPE a substantial step closer: High‐resolution interactive simulation of pilot‐scale reactors or direct numerical simulation of microreactors, both at quasi‐real time speed for the multiphase hydrodynamics, are now possible on the upgraded Mole‐8.5 supercomputing system. The spectrum of multiscale methods available or to be developed for different purposes in VPE, and the corresponding software and hardware co‐designing from more fundamental levels, are discussed also.