. . . . . DRAFT Evolutionary Computation for Industrial Design

A background briefing on the use of evolutionary computation to optimise industrial designs Executive Summary Nature has a process that optimises the design of living organisms. This process is called evolution. Evolution is based on a set of key principles including the use of a genetic code and Darwinian natural selection. Todayy s scientists and engineers can use the same principles to solve difficult design optimisation problems using computers to simulate evolution. So-called evolutionary algorithms are now routinely used to solve real-world problems that are impractical to solve by traditional analytical methods. A particular application area where evolutionary algorithms can shine is the creation of industrial designs. As an example relevant to WA industry, take the design of a processing plant for a mine. The quality of this design can have a major economic impact on a mining operation. A small percentage improvement represents a great deal of added value, and any mistakes mean lost value. Yet the process of creating and improving flowsheets for plant is largely intuitive, relying critically on the experience of the individual engineer. An automated tool that could improve the quality and consistency of flowsheet design would bring continuing benefits and competitive advantage. We believe that such tools can be developed using evolutionary algorithms. In this position paper, we outline the principles of these algorithms and describe case studies showing the benefits achieved in industrial design settings. We believe there is an opportunity to develop a flexible automated design tools in partnership with WA companies. This would provide industry with a cost-effective way to benefit from the expertise of researchers in the field, and would provide financial support and a real-world focus for the research and development work. If there is in-principle support for this concept, we would like to meet with company representatives to formulate a proposal for an initial prototype development project. A background briefing on the use of evolutionary computation to optimise industrial designs