Optimization Of Mixed Refrigerant Proportion In Natural Gas Liquefaction Process

Natural gas is recognized as clean energy in the 21 st for its high efficiency and low pollution.LNG is a good way of storage and transportation of natural gas and has also made great progress,so that its energy status has become more and more important.Mixed refrigerant liquefaction processes,especially C3/MRC liquefaction processes,were widely used among the many natural gas liquefaction technologies all over the world.The reasonable proportion of mixed refrigerant was a difficult problem in liquefaction process of mixed refrigerant,because the cooling effect of refrigerant and complexity of proportion were mutual restraint.There are no clear standards of the composition and proportion in mixed refrigerant natural gas liquefaction process,so based on the thermodynamic properties of alternative refrigerant mixed composition and the characteristics of natural gas liquefaction process of refrigeration,preliminary optimize of alternative mixed refrigerant components in this paper.Through the establishment of mixed refrigerant liquefaction process HYSYS model,the function of the primary components in natural gas liquefaction in cooling process is briefly analyzed.To find reasonable proportion of mixed refrigerant in mixed refrigerant gas liquefaction process,choose mixture test of quadratic polynomial regression model,optimize the ratio of mixed refrigerants.The optimization results are found that the optimal value and analog value deviation is more than 50%.It proved that the optimized result is not ideal and the method is not suitable for mixed refrigerant proportioning optimization problem.Analyze the deficiencies and problems existing in the quadratic polynomial regression model for mixture test and improve specifically the optimization method.After analysis,this paper chooses of Support Vector Machine(the SVM,short for the Support Vector Machine)regression model with machine learning ability and optimizes the ratio of mixed refrigerants.In the third chapter of this article,linear kernel function and spline kernel function of support vector machine regression method are used to train a small amount of simulated data regression to get the objective function.The objective function is checked through the test samples and the deviation is between 0 and 1.94%.Then the objective function is optimally solved,eventually get the optimal matching and the unit energy consumption.The optimization effect is so significant that the biggest one is 11.9% lower than before.But the optimal value obtained by this method cannot be proved whether it is optimal.In the fourth chapter,this paper optimizes the ratio of mixed refrigerant with the method of genetic algorithm and HYSYS secondary development technology.The results show that optimal effect is so remarkable that the biggest one is 14.4% lower than before.By comparing optimal ratio to the optimal ratio of the two power consumption,the gap is found small and is within the range of 3%.By comparing two kinds of optimization method,it can prove that these two methods can be used to solve the mixed refrigerant proportioning optimization problems in the mixed refrigerant gas liquefaction process.The optimal proportion obtained by optimization is the optimal ratio of mixed refrigerants in C3/MRC natural gas liquefaction process.