Model Predictive Control of a Gas Refinery Boiler
Client: Bidboland Gas Refinery
Director project: Dr. Babak Tavassoli
Project type: Simulation study
Project collaborator: K.N. Tossi University of Technology (KNTU)
Summary of project:
In the gas refinement process, the required steam refinement unit is supplied by a boiler. A boiler system is a nonlinear and multivariable system involving several interacting control loops. In this process, the PID controller is normally used to control the pressure and liquid level of the vessel. However, it is not easy to achieve an acceptable performance with PID loops and avoid fluctuations. An improved control algorithm for replacing the PID controllers is the model predictive control algorithm. This algorithm has a high computational load which can be reduced by applying the multi-parametric MPC control. Based on this, the boiler controller in the gas refinement unit is studied and a nonlinear model of the boiler has been obtained and validated using the data obtained from experiments at steady state. Then, the multi-parameter explicit predictive controller has been used to control the level and the pressure of the boiler. It is shown that the controller performance improves considerably.