Simultaneous Synthesis And Retrofit Of Heat Exchanger Network Based On Stage-wise Superstructure

The energy supply situation is growing increasingly intense, which has been the great challenge to countries all over the world. Throughout the world, people have indentified the importance of higher energy efficiency and the effect of low carbon development. Under this background, the synthesis and retrofit of heat exchanger networks (HENs) have attracted significant research in the field of process systems engineering study. Numerous investigations on theory and application have been carried out to improve the performance of heat exchanger network synthesis (HENS). However, in most cases the industrial HENS problems are large scale optimization problems, the computation complexity of which promotes further perfection to the current optimization methods. In recent years the retrofit of existing HENs has become important due to the large savings of the utility consumption. But only a few methods for HENs retrofit design have been proposed. This paper has completed the following research work to solve the problems of the synthesis and retrofit method.1. Based on the concept of hybrid stage-wise superstructure, a modified simultaneous synthesis model of heat exchanger network is proposed. In the proposed superstructure, the candidate networks featuring splits and series structures are considered simultaneously, which can control the solution performance and search space efficiently. The novel model removes the assumption of isothermal mixing in literature and is more suitable for engineering practice. However, the model is more complicated to solve due to additional optimization variables and nonlinear constraints.2. The simultaneous synthesis of HEN model is usually formulated as a complex mixed integer nonlinear programming (MINLP) problem with a great number of optimization variables and constraints, which belongs to a particular difficult class of NP-hard optimization problems. Compared with deterministic methods, stochastic algorithms do not depend on gradients and work on function evaluation alone, and is therefore more suitable for dealing with complex MINLP model. In this work, a novel hybrid algorithm (PSO/GA) is developed to sovle nonlinear programming problems. Some benchmarks are presented to show the efficiency of the proposed algorithm.3. In order to avoid trapping into local optimum prematurely, a two-level method is proposed for optimizing HENS problems. The proposed two-level method is supposed to consider levels and associations between discrete and continuous variables, and can evolve adequately in continuous search space since discrete variables evolve in the same pace. In the upper level of the presented method, GA is used for structural optimization, since it is more suitable for the non-continuous optimization problem than other stochastic algorithms. The information of structures is sent to the lower level, in which, the process parameters such as stream-split flows and heat loads of heat exchangers are optimized by the proposed hybrid algorithm (PSO/GA). This algorithm is demonstrated to be a powerful tool for solving continuous and unconstrained optimization problems. Five case studies are presented to show the applicability of the proposed methodology.4. The hybrid stage-wise model for new designs is extended for Simultaneous retrofits of HEN. The presented model does not rely on pinch analysis, and does not require predetermining minimum heat transfer temperature difference. Through solving this model, investment costs in retrofit and operating costs can be optimized simultaneously. The investment costs in retrofit include reassignment cost of existing exchangers, cost of area for existing exchangers and cost of area for new exchangers. This model is a MINLP problem with more optimization variables and constraints, which can be solved efficiently through the proposed two-level method. Four case studies are presented to show the applicability of the proposed methodology.