Meso-scale oriented simulation towards virtual process engineering (VPE) --- the EMMS paradigm
Chemical Engineering Science, 2011, 66(19): 4426-4458
Wei Ge, Wei Wang, Ning Yang, Jinghai Li*, Mooson Kwauk, Feiguo Chen, Jianhua Chen, Xiaojian Fang, Li Guo, Xianfeng He, Xinhua Liu, Yaning Liu, Bona Lu, Jian Wang, Junwu Wang, Limin Wang, Xiaowei Wang, Qingang Xiong, Ming Xu, Lijuan Deng, Yongsheng Han, Chaofeng Hou, Leina Hua, Wenlai Huang, Bo Li, Chengxiang Li, Fei Li, Ying Ren, Ji Xu, Nan Zhang, Yun Zhang, Guofeng Zhou, Guangzheng Zhou
With the dramatic development of computational science and technology, computer simulation is playing an increasingly important role in scientific research and engineering practice, and is believed to bring about a profound revolution to the mode and means of these activities. For chemical engineering, it will promote the transition from an experience-and-experiment-based research and developmentmode to the one based on virtual process engineering (VPE). However, such a revolution still requires tremendous improvements in the Accuracy of physical modeling and numerical methods, the Capability of the computing hardware and software, and the Efficiency of the simulation activities, or in short, ACE. This article will systematically review the 3-decade endeavors at IPE, CAS (Institute of Process Engineering, Chinese Academy of Sciences) on upgrading ACE by establishing a multi-scale computing paradigm focusing on meso-scale structures. We also report recent developments in this direction with projections on future work.
Meso-scales refer to the intermediate scales at which the discrete elements in a system interact to shape the global behavior of the system. Reasonable description of the structures at such scales is a bottleneck for reliable and accurate modeling of global behaviors. To address this problem, we started from gas–solid flow by proposing the energy minimization multi-scale (EMMS) model to quantify the hydrodynamics of meso-scale structures in gas-solid fluidization. This model, that is, its stability condition, was thoroughly analyzed and verified with the pseudo-particle modeling (PPM) method. Other systems, such as turbulence and gas–liquid flow, were then studied following the same strategy, revealing the ubiquity of stability conditions for meso-scale structures as a result of the compromise among different dominant mechanisms, and leading to a generalized model mathematically formulated as a multi-objective variational (MOV) problem. Meanwhile, along with wider industrial applications of the EMMS-based simulation methods, common features were recognized that ACE requires structural consistency among the simulated system, the physical model, the numerical software and the computing hardware under the umbrella of MOV, which we have named as the EMMS Paradigm.
The EMMS Paradigm was first implemented from the software level using traditional computing hardware to achieve Accuracy. Since 2007, with the development of general-purpose GPU (graphic processing unit) computing, CPU (central processing unit)-GPU hybrid computing was deployed to implement the Paradigm from the hardware level, boosting the Capability to Petaflops range. In applying the Paradigm to the development of an industrial petro-chemical process (maximizing iso-paraffins, MIP), enabling technologies such as pre- and post-processing, error-tolerance, network-based computing and seamless integration of different simulation methods are now under development, which will greatly improve the Efficiency. A demonstration virtual laboratory featuring the comparison and interaction of online measurement and real-time computing is under construction. With these endeavors, an industrial process could be simulated globally in almost real-time, and different levels of details could be traced from the global distribution of flow parameters in a reactor to the inner-channels of catalytic particles at a perceivable evolution speed. The realization of VPE is, therefore, in the foreseeable future.
Complex system, GPU computing, Meso-scale, Multi-phase flow, Multi-scale, Virtual process engineering