Advances in Environmental Research 7 (2003) 549–562 1093-0191/03/$ - see front matter  2002 Elsevier Science Ltd. All rights reserved. PII:S1093-0191 Ž 02 . 00026-6 IDEAS approach to the synthesis of globally optimal separation networks: application to chromium recovery from wastewater Ahmad M. Justanieah, Vasilios Manousiouthakis* Department of Chemical Engineering, UCLA, Los Angeles, CA 90095, USA Accepted 15 March 2002 Abstract In this work, we present an optimal separation network synthesis methodology within the novel Infinite DimEnsionAl State-space (IDEAS) framework. The resulting network synthesis problem is formulated as an infinite dimensional convex (linear) optimization, where the objective function is a weighted combination of the cost of utilities associated with the network operation and the capital cost of the separation units employed. The IDEAS methodology is illustrated on a chromium recycling case study. The globally optimal cost is identified and compared with that of a conventional (tree) design. The IDEAS design is less than half as expensive as the conventional tree design, exhibits no need for waste disposal, and employs a smaller number of units.  2002 Elsevier Science Ltd. All rights reserved. Keywords: IDEAS; Global optimization; Separation networks synthesis 1. Introduction Process network synthesis aims at identifying the optimal interconnection and sizing of process units, so that the overall processing system yields minimal waste generation and maximum profitability. The problem of process network synthesis can be divided into sub- categories such as heat exchange network (HEN), mass exchange network (MEN), reactor network, and sepa- ration network synthesis (SNS). Two classical problems arise in network synthesis, minimum utility cost (MUC) and total annualized cost (TAC). MUC designs are, by construction, the most efficient users of energy and material resources. On the other hand, TAC designs take into consideration a weighted combination of both capital and operating costs. Other optimality criteria have also been used such as attainability, which deter- mines whether a certain task can be achieved using a set of technologies (for example, the attainable region problem in reactor networks). *Corresponding author. Tel.: q1-310-825-9385; fax: q1- 310-825-2394. E-mail address: vasilios@ucla.edu (V. Manousiouthakis). Within this context, development of systematic pro- cess network design methods has been the focus of many researchers. Traditionally, three major approaches have been employed to address the SNS problem. The first, so-called heuristic approach, is based on examining a large number of network configurations (alternatives) generated based on a set of {rules of thumb’. The second approach addresses the SNS problem by devel- oping first principle-based design techniques, even though they do not address the optimality of the under- lying networks. These methods often employ thermo- dynamic principles to establish properties andyor limitations of the design problem at hand. The third approach employs a mathematical description of the design problem and mathematical programming tech- niques to identify the optimal structure of the network. There have been numerous studies that employ one of these three approaches. Next we provide a, by no means exhaustive, sample of such works. 1.1. Heuristic based methods One of the early works was that by Thompson and King (Thompson and King, 1972) in which they con-