Synthesis And Optimization Of Heat Exchanger Networks Using RWCE Based On Non-Structural Initial Frameworks

The heat exchanger network(HEN)reduces external energy expenditure by reasonably matching cold and hot streams in the process system,which plays an important role in improving the energy efficiency and economy of the system.The HEN optimization problem includes integer variables(existence of heat exchange units)and continuous variables(heat exchange loads,split ratios),which together,belong to mixed integer nonlinear programming problems,and their global optimization has become a difficult and hot spot in the field of process system integration.As the scale of the problem continues to grow,the number of feasible network structures increases and the number of local extrema in the objective function increases.Finally,changes in the parameters of the optimization model,the stream matching modes,and the evolution method of the optimization algorithm are necessary,which are closely intertwined in the course of acquiring optimal global solutions for complex networks.However,the currently widely used optimization models(structural models and node-based nonstructural models)limit the solution domain to some extent due to fixed matching methods.In dealing with the problem of achieving global optimization of large-scale heat exchanger networks,the optimization methods of integer and continuous variables in the current heuristic algorithms,and the suboptimal solution avoidance mechanisms need to be further refined.Therefore,it is imperative to develop efficient optimization models and algorithms as they play an important role in obtaining the global optimal solution,improving energy utilization,and increasing the benefits of the production process.The main work and innovations of this thesis are as follows:1.The influence of the fixed number of split groups in the node-based nonstructural model on the optimization performance for the heat exchanger networks is analyzed,and it is found that the existing heat exchangers weaken the connection freedom of unmatched nodes.On this basis,the dynamic split group node-based nonstructural model(DSG-NNM)is proposed.By way of this model,new split groups are dynamically inserted in process streams during the optimization process,which can expand the solution domain in real time and improve the degree of freedom of node connections.The results show that DSG-NNM can effectively compensate for the limitation of fixed nodes in the basic node-based non-structural model on the optimization quality,thereby enhancing the flexibility of the model and improving the optimization ability of the algorithm for the synthesis of heat exchanger networks.2.DSG-NNM achieves heat transfer matching through the random connections of discrete nodes,which cannot guarantee a completely sufficient solution domain.Therefore,this thesis proposes a non-structural model with stream splitting(NSM-SS),which quantifies the exchanger positions based on continuous real number intervals,randomly determines the split regions,and randomly connects the hot and cold streams to form a heat transfer matches.NSM-SS greatly broadens the solution domain of the problem,laying a better foundation for efficient optimization algorithms to give full play to their effectiveness.3.Based on the characteristics of the NSM-SS,a compulsive evolutionary random walk algorithm is chosen for heat exchanger network optimization.In order to improve the match between the algorithm and the model,a structural update strategy and a utility merging strategy have been added to the algorithm.Under the multi-parameter coupling of NSM-SS,the structural update strategy effectively improves the degree of freedom in the evolution of the optimized parameters,allowing the algorithm to explore more matching relationships and obtain better results in a defined solution space.The utility merging strategy can combine some external and internal utility exchangers in the network,and reduce the number of heat exchange units in the network.While effectively reducing the annual overall cost,NSM-SS provides more exchanger connection options and more reasonable heat transfer matching,and further improves the quality of optimization results.In order to further broaden the solution space using NSM-SS,this thesis proposes three modifications.In the NSM-SS1 version,multiple heat exchangers per split branch and submixes are allowed to explore a better HEN configuration in terms of lower annualized costs.The NSM-SS2 variant uses the stream sub-splits as another topological possibility to enhance the network heat recovery.In order to make better use of the temperature differences,the NSM-SS3 considers the influence of the bypass connections in grassroots design of the heat exchanger networks from the perspective of network flexibility and controllability,and provides a theoretical basis for the subsequent actual HEN application and transformation.In summary,this thesis mainly focuses on the construction and improvement of the non-structural models of the heat exchanger network synthesis problem,and improves the node-based non-structural model so that it can broaden the solution domain in real time during the optimization process.In addition,due to limited flexibility of the nodebased unstructured model,the non-structural model with stream splits(NSM-SS)is constructed for the first time and a forced evolutionary random walk algorithm that matches the characteristics of the model is proposed.The NSM-SS has been improved from the aspects of model characteristics,widening of the solution domain,actual engineering requirements and optimization modes.Through several examples,it is verified that using the NSM-SS proposed in this thesis to optimize the heat exchanger networks can obtain high-quality optimization results,and provide theoretical support for the development of efficient models and algorithms in the future.