Bi-level Algorithm Of Evolution Method Handle Structure Optimization And Memetic Pso Handle Continuous Parameters In Large-scale Heat Exchanger Networks

Chemical process system engineering plays an important role in the reduction of energy consumption, and heat exchanger networks synthesis(HENS) has a domain role in the chemical process system engineering. So, the heat exchanger networks synthesis problems have been one of the most extensively studied synthesis problems in chemical engineering. In the mathematical programming framework, HENS is in general formulated as mixed-integer non-linear program(MINLP) models, in which continuous variables represent process parameters(e.g. heat exchanger duties) and integer variables represent discrete decision(e.g. heat exchanger matches). However, in the case of large-scale problems, the implementation of MINLP algorithms can give birth to many computational difficulties. As there are a large number of match possibilities of specified number of hot and cold streams. This paper is divided into two parts: the first part is to improve the Powell Method which is belong to the deterministic methods and aim to impend global optimum. In the second part, A bi-level optimization algorithm for optimal design of heat exchanger network is proposed.In the first part, the performance of Powell Method was improved in term of control parameters preference, the influence of convergence, the convergence criterion, and the conjugacy of search direction. In the optimization of small-scale or medium-scale HENS problems, there is a big improvement of calculation efficiency and solution accuracy. With the combination of “principle of process streams arrangement”, The proposed bi-level optimization method has been applied to several cases taken from the literature and the results are very encouraging, among which the large-scale problem.And then, the standard Particle Swarm Optimization(PSO) was applied to the HENS. A lot of benchmark function are implemented to illustrate the excellent search ability of global convergence. A mall case is implemented to illustrate the realization of standard PSO algorithm in HENS problems step-by-step.The adjustment of inertia weight factor, velocity limitation, acceleration coefficients, particle population, and the starting points result in a better optimum. The paper proposed a new topology termed as “ring +random topology” combining with the advantage of calculate efficiency of random topology and capability to converge to better optimum of ring topology. We proposed a Memetic PSO(MPSO) algorithm which combines PSO with local search techniques to overcome the deficiency that they cannot perform a refined local search to compute the optimum with high accuracy once there. Four cases were used to demonstrate the performance of MPSO.