PERFORMANCE ASSESSMENT OF CONTROL LOOPS: controller evaluation for frequent set-point changes

The high seasonality and variability of electricity production from renewable sources has enormous impact on both production and distribution networks. Among the many aspects, several traditional power plants, especially, combined cycle plants, but also coal-fired plants, now operate intermittently with variable set points due to fluctuations of energy load which they are requested to deliver [1]. Thus, a specific analysis of control system performance during these transient cyclic phases, as reference changes, operations of start up and shutdown, is highly desirable. Generally speaking, monitoring and assessment of performance of control systems of industrial plants are important topics in process control. The deterioration in performance is, in fact, a fairly common phenomenon and manifests with sluggish or oscillating trends of control variables. Oscillations in control loops can cause many problems which affect normal operation of process plants. Typically, fluctuations increase variability of product quality, accelerate wear of equipment, move operating conditions away from optimality, and, in general, cause excessive or unnecessary consumption of energy and raw materials [2]. This paper introduces a technique for the analysis of performance of basic control loops when process is subject to changes of operating conditions. The method employs the well-established approach of Internal Model Control, IMC. After establishing lower limit for the absolute value of the integral (IAE) of control error and the total variation (TV) of control action, such limits are assumed as reference values for a control considered “optimal”, or anyway “good”. A performance index is thus based on IMC and is properly defined with respect to lower limit of IAE and TV. With this approach, the validity of tuning of PID-type controller in response to any reference change can be assessed. In particular, one can successfully evaluate closed-loop performance for setpoint changes, as steps, ramps, or generic trends, as for the common case of preset programs of variable load of power plants