Alireza Behroozsarand - Faculty of Chemical Engineering Urmia University of Technology Urmia, Iran
Mohammad Divband - Faculty of Chemical Engineering Urmia University of Technology Urmia, Iran

In this paper, an ethylene production plant by using the steam cracking technology with naphtha feedstock is modeled by a multilayer, feed forward, fully connected neural network. Mass flow rate of different feeds include of C4 raffinate, naphtha, LPG, and ethane were the input variables to the network. These input variables will help to generalize the model. The output variables of the network were mass flow rate of different products consist of propylene, ethylene, cracked fuel oil, petroleum gas, C4 cut, hydrogen, and fuel gas. The optimum topology of network was an eight layers network with eight neurons in hidden layer. The network outputs were in agreement with the experimental values

Industrial Olefin plant; ANN; Feed forward back propagation network; Modeling