PID controller performance assessment based on closed-loop response data

Process control engineers and instrumentation technicians have the daunting task of maintaining and troubleshooting control systems. The performance of control systems typically degrades as the process ages for a variety of reasons such as scaling and fouling of process equipment, nonlinear process behavior, and deactivation of catalyst. Ensuring that control systems are near design specifications requires the regular collection and analysis of large quantities of raw process data. Often it is not feasible for operating personnel to manually monitor all the control systems in a plant. Few tools exist to aid plant operating personnel in maintaining control systems. Given the importance of ensuring that control systems perform as designed and the difficulties that plant operating personnel face in this task, there is a need for automated and reliable methods to assess and monitor the performance of control systems.; Two simple methodologies have been developed for assessing the performance of PI and PID feedback control loops with regard to setpoint tracking and load disturbance attenuation. The methods are based on dimensionless performance indices that are determined by dividing closed-loop response characteristics by the apparent time delay of the process. Extensive simulation results show that the dimensionless indices of several different closed-loop systems that represent a wide range of dynamic process behavior are independent of the process model. Thus with limited knowledge of the underlying process, the performance of PI and PID controllers can be assessed.; The two new methodologies were applied to six published multi-loop PI control tuning methods for the Wood-Berry Distillation Column Model. A closed-loop identification technique to estimate the apparent time delay from a single closed-loop response was proposed. The closed-loop response data used to assess performance can also be used to determine the apparent time delay, which is required by the developed performance assessment techniques. The proposed closed-loop identification technique and the proposed methodology for assessing setpoint tracking performance were successfully applied to an industrial distillation column. The performance assessment resulted in the identification of two poorly tuned control loops. New controller settings were implemented and their performance assessed, which resulted in significant improvements in performance.