화학공학소재연구정보센터
Chemical Engineering & Technology, Vol.26, No.8, 823-827, 2003
Interactivity and automated process design
Automated process design tools are often based on the use of optimization to identify the best process flowsheet, typically in the early stages of design. Due to the non-linear models that are frequently required, the resulting optimization problem is difficult to solve. Difficulties arise both in identifying good initial solutions and in finding the global optimum. In some cases, even feasible initial points can be difficult to find. Furthermore, the objective function and the feasible search space may be non-convex. As a result, automation alone can be insufficient for the solution of difficult process design problems. The engineer can, and arguably should, be involved in the process of design, and tools that encourage this direct involvement are desirable. In early design, exploration of alternatives can be useful.Jacaranda is a system for automated design [1]. It is based on the use of discrete programming, in conjunction with implicit enumeration and branch and bound techniques, to solve process design problems with complex models. Jacaranda has been designed to support interactivity to ensure the maximum effectiveness from the combination of the computer and the engineer [2]. The aim is to allow the engineer to perform the tasks that a human can do best while ensuring that the computationally complex and repetitive steps are undertaken by the automated design tool. In this context, Jacaranda provides visualization and data mining support for the engineer [3,4].Examples demonstrating the potential of an interactive automated design approach are presented. These include heat integrated separation sequence synthesis, reaction/separation with environmental issues, and a process optimization problem with small feasible regions. Interactivity is used to support an iterative refinement approach to process synthesis or to target optimization procedures to improve the quality of the solutions obtained. Visualization is shown to be a key underpinning technology, which also aids the engineer in understanding the results obtained by automated design tools.