Spring 2018 – Water network synthesis for industrial parks

“Nowadays, the establishment of industrial parks has become the mainstream in process industry for considering symposium among multiple enterprises. As an important resource in process industry, fresh water consumption accounts for large amount of operating cost for a plant. The need for effectively managing the water resource is required since many water-using unit (sink) allow certain amount of impurities in the feed without consuming fresh water. This feed can be obtained from other water-using unit product streams (source). An effective management strategy for water resource is the water integration, aiming at designing the most cost-effective water networks. This water network enables the design of source-sink matches inside each plant and among plants in an industrial park. In this project, we consider a problem of designing industrial parks with water-using unit in each plant. To achieve the optimal design of industrial parks, we apply two integration strategy: direct integration via direct connection between sources and sinks as well as indirect integration via reallocation of water from regenerators which accept streams from other sources. This problem can be further extended to include heat integration among streams in industrial park, multi-period design for long-term water management under uncertainty. These problems can be formulated as nonconvex mixed integer nonlinear optimization (MINLP) problems, which may be computationally expensive for many commercial solvers. The global optimization strategy for this project includes multiparametric disaggregation approach for providing effective linearization of nonlinear terms, nonconvex benders decomposition, lagrangian decomposition and cross decomposition. For decomposition-based method, parallelization among different subproblems can also be considered. “