Study On Steady-State Simulation And Multi-Objective Optimization Of Crude Oil Distillation Units

As the key unit in refinery fields, crude oil distillation unit (CDU) is thebasement of refinery enterprises. Its improvance in efficiency has huge impacton the efficiency of units downstream and the total profit of the entireenterprise. As we know, the profit of the entire enterprise is affected by therunning status of the units; in another word, it is affected by the stability andreliability of the control strategy. Aiming at enery saving, profit improving,and process optimization, the research of steady-state simulation andmulti-objective optimization of CDU has great meaning. The main work canbe summaried as follows:First of all, the author made a brief analysis on steady-state simulationand multi-objective optimization. And then, with rigid mechanism, the authorbuilded a steady-state model of CDU. For comparison with flow simulationsoftwares, the author also builded a steady-state model of the CDU with AspenPlus.Secondly, after the detailed research of simulated annealing algorithm,genetic algorithm, and simulated annealing genetic algorithm, and kept the advances and weaknesses of those algorithms in mind, proposed jumpinggenes adaption of adaptive simulated annealing genetic algorithm (ASAGA).And with many proven metrics and test problems, the author had tested theperformance of jumping genes adaption of ASAGA, the results showed thatjumping genes adaption of ASAGA is much better than ASAGA, and on sometest problems, it even better than NSGA-II and RJGGA.Thirdly, based on the rigid mechanism simulation of CDU, the authorselected the right objectives from the view of enterprise, suitable decisionvariables, and constrains. After finish constructing of multi-objective model,the new jumping genes adaption of ASAGA was applied into themulti-ogbjective optimization of CDU. It is observed that current plantoperation is sub-optimal and more profit can be realized for the same energycost using the obtained optimal operating conditions, which are under theconstraints of product quality and total distillate. The simulation resultsdemonstrate that the ASAGA-JG is able to generate non-dominated solutionswith a wide spread along the Pareto-optimal front and good address the issuesregarding convergence and diversity in multi-objective optimization.