This paper describes a methodology to automatically optimize the settings of the controllers of a plant. This approach is applied to a process dynamic simulation representing a real operated oil field. The controller parameters are tuned for specific process disturbance or setpoint change scenarios, with the aim of providing a faster control action and bring the plant back to normal operating conditions in the shortest possible time.
The proposed method described in this paper uses a genetic algorithm to perform a plant-wide optimization of the PID controller parameters. The optimized parameters values, tuned with the algorithm, are particularly efficient in managing the process disturbances, upsets and setpoint changes scenarios. The controller parameters can be further optimized by including in the method other operating scenarios of interest and those critical for the considered plant simulation.
The genetic algorithm, thanks to its better convergence rate with a high number of parameters and constraints compared to other optimization methods, allows to achieve an accurate tuning of the controllers of larger plant sections with a limited computing power. It should also be noted that the code structure of the method presented in this paper is easily modifiable, allowing the controller parameters optimization to be readily extended to other plant sections or even applied to completely different systems.
Given the high volatility of oil price in recent years, petroleum industry workers are trying to optimize production more than ever, to maximize profits and avoid any kind of financial losses. For this reason, nowadays it is vital to manage the performance of each asset in the best possible way.
After bringing a plant to its optimal configuration, the main problem is to keep it in that state despite the occurrence of disturbances or other changes that affect the asset. It is therefore necessary to monitor the key parameters of the plant so that, in case they move away from the chosen configuration, the control system bring them back to steady state as soon as possible.
