Abstract: "In this paper, the second in a three part series considering the multiperiod design of azeotropic separation systems, we focus on the development of simplified models for azeotropic distillation design. Several shortcut design techniques from the literature are reviewed and key aspects of a successful model for use in multiperiod azeotropic distillation design are identified. Simplified models for azeotropic design which rely on parameters derived from distillation simulations involving rigorous thermodynamic models are developed. These models, which separately approximate the separation tasks and the design and operating conditions necessary to carry out the tasks are combined with shortcut costing correlations to arrive at an economic measure for flowsheet designs. Several examples are presented."
Abstract: "In this paper, the final in a three part series, we explore an agent based solution approach to the multiperiod design of azeotropic separation systems. Asynchronous Teams, a software organization that has been successfully applied to combinatorial optimization problems in other disciplines, is applied to this chemical process system design problem. Asynchronous Teams is a framework in which multiple design techniques can work harmoniously in the solution of a problem to [sic] demanding for any one method. We introduce a simple architecture for constructing Asynchronous Teams which incorporates the use of persistent data. Several approaches for generating improved multiperiod azeotropic design solutions are developed and are used simultaneously in a large scale demonstration. The software organization is found to be capable of developing and improving a population of preliminary designs for the multiperiod azeotropic design problem."
Abstract: "In this paper, the final in a three part series, we explore an agent based solution approach to the multiperiod design of azeotropic separation systems. Asynchronous Teams, a software organization that has been successfully applied to combinatorial optimization problems in other disciplines, is applied to this chemical process system design problem. Asynchronous Teams is a framework in which multiple design techniques can work harmoniously in the solution of a problem to [sic] demanding for any one method. We introduce a simple architecture for constructing Asynchronous Teams which incorporates the use of persistent data. Several approaches for generating improved multiperiod azeotropic design solutions are developed and are used simultaneously in a large scale demonstration. The software organization is found to be capable of developing and improving a population of preliminary designs for the multiperiod azeotropic design problem."
The Adaptive Computing in Design and Manufacture conference series has become a well-established, largely application-oriented meeting recognised by several UK Engineering Institutions and the International Society of Genetic and Evolutionary Computing. The main theme of the series relates to the integration of evolutionary and adaptive computing technologies with design and manufacturing processes whilst also taking into account complementary advanced computing technologies. Evolutionary and adaptive computing techniques continue to increase their penetration of industrial and commercial practice as awareness of their powerful search, exploration and optimisation capabilities becomes ever more prevalent, and increasing desk-top computational capability renders stochastic population-based search a far more viable proposition. There has been a significant increase in the development and integration of commercial software tools utilising adaptive computing technologies and the emergence of related commercial research and consultancy organisations supporting the introduction of best practice in terms of industrial utilisation. The book is comprised of selected papers that cover a diverse set of industrial application areas including engineering design and design environments and manufacturing process design, scheduling and control. Various aspects of search, exploration and optimisation are investigated in the context of integration with industrial processes including multi-objective and constraint satisfaction, development and utilization of meta-models, algorithm and strategy development and human-centric evolutionary approaches. The role of agent-based and neural net technologies in terms of supporting search processes and providing an alternative simulation environment is also explored. This collection of papers will be of particular interest to both industrial researchers and practitioners in addition to the academic research communities across engineering, operational research and computer science.
