Simulation Of DBT And 4, 6-DMDBT Kinetics In The Hydrodesulfurization Process Using Artificial Neural Networks

DBT and 4,6-DMDBT are two inactive compound in HDS process. In order to get the rate constant of DBT and 4,6-DMDBT, artificial neural network which has been applied to chemical engineering for a long time was used In this thesis. but this is the first time to use this method to calculate the kinetics in HDS process. In order to get a good result, the topologies of the neural networks should be determined, thus a great deal of data were needed, so in this work, we extended the data which is gathered from public papers to 6300 for each compound. To determine the topologies of the neural network, the weights and bias in each layer should be calculated. Independent data were used to test the accuracy of the neural networks. All the calculation process was base on Matlab platform. With the topologies obtained in this thesis, similar model could be built to get the constant rate from different original data.In this work, Co-Mo/Al2O3 was selected as catalyst since the Co-Mo/Al2O3 was widely used in HDS process. Additional, the other two crucial parameters temperature and pressure were also chosen. The rate constants of DBT and 4,6-DMDBT was used as the output parameters in each calculation. According to the calculation, we get the topology for DBT:1 input layer with 4 neurons,1 hidden layer with 14 neurons and 1 output layer with 1 neuron, for 4,6-DMDBT: 1 input layer with 4 neurons,1 hidden layer with 11 neurons and 1 output layer with 1 neuron.These topologies can be used to calculate both constant rate of DBT and 4, 6-DBT in a wide condition range to avoid complicated experiments. In the end of this thesis, both topologies and programs were listed.