Model-Based Absorption Refrigeration System Optimization And Control

Absorption refrigeration system is a complex, nonlinear system with strong interactions among each working device. With the expansion of the distributed energy market, it has become the focus of attention again because of its strong energy adaptability and so on. However, it still has the problem of low heat efficiency, which limits its development. This paper mainly studies the modeling and simulation of the absorption refrigeration system and the optimized operation scheme.First of all, the necessity of absorption refrigeration system optimization is analyzed. On basis of analyzing the internal mechanism of components, the parameters, such as component geometry size, thermal exchange coefficients and so on, are lumped to unknown parameters, which can be identified by experimental data using nonlinear least squares methods. The variables which can be measured and controlled easily are selected as the input/output variables of the model. The hybrid models for each components of the absorption refrigeration system are developed based on energy and mass conservation.Secondly, on the absorption refrigeration system experimental platform which is built in the laboratory, steady-state experimental data are collected to identify unknown parameters of each component models by nonlinear least squares method (L-M method). The results of the simulations and the tests validate the effectiveness of the models.Then, in order to optimize the refrigeration system, the simulation platform of the refrigeration system is built by using TRNSYS software. A nonlinear constraint optimization problem of the overall system is formulated by analyzing the variables which affect the system performance. According to the interaction and physical constraints between components, the constraint condition is analyzed.Finally, the maximum performance coefficient of the refrigeration system is considered as the optimization objective function. The hybrid models of components are considered as the equality constraints of whole optimization problems. The boundary conditions of each variable are considered as the boundary constraints. The penalty function method is adopted to transform constraint condition to unrestraint condition. Particle swarm optimization algorithm is used to solve the optimization problems. A detailed calculation procedure and the comparison results of system optimizing operation are given.The results of simulations and experiments show that the hybrid modeling approach take advantages of both physical and empirical modeling approaches, and each of the component models is not only simple, but also can accurately predict their performance. These models are very suitable for the real time control and optimization of industrial processes. The proposed optimization method indeed improves the system performance significantly.