It is not unusual in the chemical industry to initiate the first design study within one week of a chemist's, biochemist's, microbiologist's or polymer scientist's discovery of a new reaction system or a new material. The goals of this first design are to see if the process is profitable, to see if environmental problems will be encountered, to determine the best few process flowsheets, and to guide the development of a database that is needed for a final design. Thus, the focus is on understanding the problem, rather than obtaining a rigorous solution. For this reason, short-cut models normally are used for the evaluations in the initial study, and then rigorous design calculations are introduced in subsequent design studies after the best alternatives have been identified. The method can be approximated on the computer by using hierarchical methods, going from a conceptual design to a detailed design to a final design. This work involves automatic implementation of the first step, the conceptual design phase of the procedure. In this stage, the designer identifies the possible process alternatives, selects process units, identifies possible pollution problems, etc. This conceptual design problem is often characterized by a very large number of possible processes that can use a given reaction chemistry to transform the raw materials to the desired product (and some by-products). Both a large number of structural alternatives (that is, equipment selection and streams), as well as values of the design variables (which effect the stream flows and equipment sizes) are encountered. Rules based on artificial intelligence will be developed to choose between these numerous alternatives. Specifically, the PI plans to develop new synthesis procedures for the separation of vapor-liquid-liquid-solid mixtures that will make the hierarchical decision procedure applicable to the conceptual design of most chemical processes. The code he plans to develop will include a systematic procedure for exploring process alternatives, as well as the quick screening of these alternatives; and, be a good starting point for more detailed algorithmic synthesis studies (mixed integer nonlinear programming and simulated annealing). The code will also indicate when pollution problems will result from the design decisions and suggest process alternatives to avoid environmental problems. In addition, the software will be shared with other undergraduate and graduate institutions (it is to be distributed by the CACHE Corporation) to be used for the teaching of process design.
