Monday, November 9, 2015
Exhibit Hall 1 (Salt Palace Convention Center)
Petroleum refinery is an industrial process plant for processing of crude oil to more useful products. Optimum functioning of the Crude Distillation unit (CDU), believed to be a vital process of petroleum refinery, is crucial for good product quality and economic benefits in the petroleum industry. The present work is aimed to investigate the general design of the CDU and to achieve optimum conditions using simulation. A heuristics based Crude Distillation Unit (CDU) is modeled in ASPEN PLUS® and used to obtain the data points which act as a substitute of working plant data to provide the process insights. A steady state model is developed in MATLAB® using the MESH equations, which predicts the temperature distribution and concentration of Naphtha over different trays. The developed model predicts the Naphtha distribution and withdrawal tray correctly, ascertaining the model accuracy. This model is then simulated for different sets of input conditions in order to obtain the specimen data points. An Artificial Neural Network (ANN) model is generated using the MATLAB neural network tool, which is further trained using sets from the specimen data points. This trained network is capable of predicting the output of a sample data point. Using this prediction capability, feed flow rate and steam flow rates are compared with the Naphtha product ratio. This provides the optimum value of feed and steam flow rates at which the Naphtha production is maximum.
Keywords: Crude Distillation Unit (CDU), MATLAB, Artificial Neural Network (ANN), Aspen Plus, MESH equations.