U.S.-Brazil Cooperative Science Program: Dynamical Process Simulation on Computers with Parallel Architectures

这项美国-巴西合作科学奖将支持加州理工学院的曼弗雷德·莫拉里、加州大学劳伦斯·利弗莫尔国家实验室的安东尼·斯凯勒姆；里约热内卢联邦大学的埃瓦里斯托·比斯卡亚和巴西南里奥格兰德州联邦大学的阿吉米罗·塞奇。该项目旨在对在具有并行体系结构的计算机和超级计算机上进行化学过程动态模拟的备选方案进行比较分析。化工过程模拟负责小型和大型计算机上使用的许多CPU周期。描述现实复杂工业过程的仿真模型转化为非常大规模的问题，需要高速仿真器才能使这些仿真可行。因此，必须使用高速计算机，并且必须开发适合这些计算机的模拟器。在这个项目中，研究人员将考虑一系列工业问题，以提高所开发算法的性能，并提供与商业动态模拟器更可靠的比较。该项目的互惠互利将来自巴西方面在动态模拟方面的专业知识，以及美方在加州理工大学的计算设施。

This US-Brazil Cooperative Science award will support collaboration of Manfred Morari, California Institute of Technology; Anthony Skjellum, University of California Lawrence Livermore National Laboratory; Evaristo Biscaia, Federal University of Rio de Janeiro; and Argimiro Secchi, Federal University of Rio Grande do Sul, Brazil. The project aims to make a comparative analysis of alternatives for dynamic simulation of chemical processes on computers with parallel architectures and supercomputers. Chemical process simulation is responsible for many CPU cycles used on mini and mainframe computers. Simulation models describing industrial processes of realistic complexity translate into problems of very large dimension, and high-speed simulators are required to make these simulations feasible. Thus, high-speed computers must be used, and simulators suitable for these computers must be developed. In this project, the researchers will consider a range of industrial problems in order to improve the performance of the developed algorithms and provide more reliable comparisons with commercial dynamic simulators. The mutual benefits of this project will accrue from the Brazilian side's expertise in dynamic simulation, and the US side's computing facilities at Caltech.