Optimization of a Prefractionation Column Using Improved Firefly Algorithm

Crude oil refining is an important part of the chemical industry. The separation of different components in crude oil is realized by distillation columns. To optimize distillation columns, operating variables need to be set properly. Heuristic optimization algorithms are suitable for a variety of optimization problems, and more easily extended to large-scale optimization problems. The firefly algorithm is inspired by the luminescence and attraction behavior of fireflies in nature. In order to solve the problem that standard firefly algorithm is easy to fall into a local optimum, a dynamic switching strategy is introduced to the new improved firefly algorithm. The location update strategy could dynamically switch between focus on global optimization and local optimization. The switch is triggered by the consecutive generation times of unchanged optimal value of objective function. The improved firefly algorithm is implemented to optimize a prefractionation column which is modelled from a real refinery using Aspen Plus and MATLAB, and its performance is compared with standard firefly algorithm.