Tarawipa Saurod. Process modeling, dynamic data reconciliation and control of acetylene hydrogenation reactors . Master's Degree(Chemical Engineering). Chulalongkorn University. Center of Academic Resources. : Chulalongkorn University, 1998.
Process modeling, dynamic data reconciliation and control of acetylene hydrogenation reactors
Abstract:
The mathematical model of the industrial acetylene hydrogenation process is developed and formulated on SPEEDUP program which is a dynamic simulation program. The fixed bed acetylene hydrogenation reactor is modeled as the CSTRs connected in series. The data using in modeling, the actual data from the ethylene plant are reconciled first using material and energy balance redundancy. The best number of CSTRs is found to be twenty which is given the best agreement in temperature profile and the output acetylene concentration of each bed. The six kinetic models with the different reaction mechanism are derived and selected for the best one by comparing their predicted outputs with the actual data. The kinetic model which its reaction mechanism are the reaction rate depends on the first order of H2, H2 break to free atoms before react, the product are not adsorbed on the catalytic surface, and the catalytic activity depend on the accumulation of the inlet acetylene, gives the best agreement of the predicted result with the actual data. The dynamic data reconciliation using the material and energy balance constrains of process is performed. The best time history horizon is found to be ten steps. It can reduce the standard deviation of the actual data and the simulated data to about 40-70% and 90% in series. The noised data and the reconciled data are then used to obtain the new parameters of the model, i.e. the reconstruction of the model. The noised model gives 2.13% of temperature error and the reconciled model gives 0.16% of temperature error. The obtained model of acetylene hydrogenation reactor is used to demonstrate the design, implementation, and performance of Dynamic Matrix controller by simulation. The Dynamic Matrix controller is tuned for best performance with the control horizon, U=2, the prediction horizon, V=3. The integral error of Dynamic Matrix controller is only 3.33% of the PID controller's error. The ethylene loss is reduced by 80-98% by using Dynamic Matrix controller over PID controller. The degree of the benefit of using the Dynamic Matrix Control is illustrated