(239g) MPC Perfomance and Monitoring for a Demethanizer Column in a Natural Gas Processing Unit

The oil, gas, and chemical manufacturing industries and also the academics have been investing heavily in the implementation of advanced process control (APC) applications, of which the most popular form is model predictive control (MPC). MPC bridges process modelling, control and optimisation to enhance the profitability and stability of process operations in a upper layer than PID controlers. An Industry issue to develop the MPC appliance is the monitoring and performance evaluation of the implemented systems. A great part of MPC controllers installed frequently is left behind because it does not follows the plant process profile during their operation time. The performance of the installed MPC system is usually dependent on various factors that affect their performance. The factors found to be most contributing to the poor performance of MPC applications from a practical point of view are: ? Lack of properly trained operators and support personnel. ? Lack of MPC condition monitoring applications. ? Significant process modifications and enhancements. ? Poor controller tuning and inaccurate models. ? Unresolved basic (regulatory and PID) control problems. To address the above issues, academic researchers, practitioners, and control technology providers have developed a keen interest in monitoring the performance of control applications in general, and the condition monitoring of MPC applications in particular. The gas processing plants usually involves multivariable problems which affects the stability of process variable and also the control designed. The demethanizer column in a gas plant unit summarizes this interaction where you can find a distillation column with many feed streams. The column has to take granted the top and bottom composition by temperature inferences. There are side-reboilers and gas-gas exchangers, bringing a high energy interaction in the system what increases the challenge of controlling. The method proposed in this paper consists in a reverse engineering compared to the performance and monitoring system applications, where the plant is simulated, the plant is identified in order to tune the PIDs and MPC controllers, following the selected variable to control and manipulate the system. Thus, finally; mismatches are created to configure each scenario that can be monitored and also tested the performance evaluation and its mismatch identification. Applying this sequence, different methods can be evaluated and validated.