Operational Optimization of Binary Distillation Column to Achieve Product Quality using Imperialist Competitive Algorithm (ICA)

Quality of product in the chemical process must be controlled and optimized to achieve the best quality and minimize energy consumption. Quality in the process of a distillation column that performed by using optimization techniques is not easy because distillation column is non-linear. The Distillation column is a class NLP (Non-Linear Programming) which can be solved by using a stochastic method to achieve the global solution. ICA has several advantages that can solve the complex problems, increasing the speed of converging value than other algorithms. In this research, binary distillation column model was built using neural network MultiP-Layer Perceptron (MLP) with Non-Linear autoregressive with external input (NARX) structure and learning algorithm using Levenberg-Marquardt. This paper focus on optimization of product quality and quantity by changing the operational conditions of the column, i.e. molar flow feed, a composition of feed (methanol & water), molar flow reflux, re-boiler, and condenser heat duty. The output of optimization provides the better quality and quantity of product.