Distributed Parallel Computing

The solution of realistic, industrial-scale simulation and optimization problems is computationally very intense, and requires the use of adequate computational resources to be done in a timely manner. High performance computing (HPC) technology, in particular parallel computing, provides the computational power to realistically model, simulate, design and optimize complex chemical manufacturing processes. To better use these leading edge technologies in process simulation requires the use of techniques that efficiently exploit parallel computational resources. One of major trends in this regard is the use of distributed computing systems. Typically, in this sort of system, memory is physically distributed, and communication may be done by message passing through some interconnection network.

The use of parallel processing in chemical engineering has attracted significant attention over the past decade or so. There are a variety of applications for which a distributed approach to parallel computing has proven to be effective. In chemical process systems engineering, some examples, that involve either actual implementation on distributed systems, or algorithms appropriate for distributed computing, can be seen in the field of deterministic global optimization and reliable nonlinear equation solving (e.g., [1,2,3,4,5,6,7,8,9]), nondeterministic global optimization (e.g., [10,11,12]), BB in process scheduling (e.g., [13,14,15,16]), BB in process synthesis (e.g., [10,17,18,19]), and process simulation, analysis and optimization (e.g., [20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39]). There are also a number of important application areas outside of process systems engineering (e.g., [40,41,42,43,44,45]).

The type of distributed parallel system of particular interest here is a cluster of workstations (COW), in which multiple workstations on a network are used as a single parallel computing resource. This sort of parallel computing system has advantages since it is relatively cheap economically, and is based on widely available hardware. Thus, such an approach to parallel computing has become an important trend in providing high performance computing resources in science and engineering.