حسین رهیده

Pressure swing adsorption is one of the most common ways for producing Oxygen from air and has been studied by many researchers through recent decades. Because of the huge cost of doing physical experiments and limitations of changing effective parameters due to them, modeling of this process has always been noticed. Different analytical and numerical techniques have been used by the researchers and compared with experimental data. With developments of estimation algorithms such as Artificial Neural Network, some researchers have noticed to their applications in adsorption process and few works have been done in this field. In this work, it is tried to program the most completed model for pressure swing adsorption process fundamentally. For this purpose, three different numerical techniques such as Differential Quadrature, Finite Difference and Upwind Schema have been used for removing partial differentials in length dimension; next the equations have been integrated through time with Finite Difference method. The resulted system of equations has been solved with Newton and Simplified Newton methods and the results have been compared with each other. The validation of the results has been also checked with available experimental data in both one bed and two beds cases. The profiles of different variables such as gaseous and solid concentrations, bed and wall temperature, velocity and pressure through the bed in both adsorption and desorption steps and due to a cycle have been studied. In the next step, effects of different parameters such as bed length and its inner radius, feed temperature, high and low pressure and tortuosity on the purity and recovery of the product have been studied. And finally, the usage of Artificial Network in estimation of modeling results with independent input data is studied and the results have been presented. By the results presented in this work, the Differential Quadrature method with 12 nodes has been considered as the best method f