(242m) Practical Strategies for Using Signal Filters with Available Industrial Controllers

Signal filters improve controller performance by reducing the impact of noise or random error in the measured process variable. These filters may be placed upon either the measured process variable or the controller output. Directly or indirectly, signal filters limit the large controller output moves caused by derivative action upon noisy measurements. Even if the noise appears not to cause performance problems, the filter reduces fluctuations in the controller output that wear the final control element. This paper uses mathematical analysis of closed-loop transfer functions combined with real-time experimental data and simulation studies to demonstrate practical strategies for using available industrial controllers with controller output signal filters either internal or external to the controller.

Many industrial controllers include a signal filter as the fourth mode of a proportional-integral-derivative (PID) with Filter controller. These exist in a variety of forms depending upon the placement of the derivative and signal filter terms. In the ideal and interacting forms, the controller consists of a single ordinary differential equation (ODE) including all four terms. In the parallel form, the controller consists of a system of two ODEs with the proportional and integral terms comprising one ODE and the derivative and signal filter terms comprising a second ODE. The output of the parallel controller form equals the sum of the outputs of the two ODEs. The parallel form filters controller output changes resulting only from derivative action while the ideal and interacting forms filter changes resulting from proportional, integral, and derivative action combined.

Novel contributions of this work are derived Internal Model Control (IMC) tuning correlations for the parallel form of the PID with Filter controller for both self-regulating and non self-regulating systems. These derived correlations are necessary to accurately and dependably tune commonly-used parallel form industrial controllers manufactured by ABB, Bailey, Emerson, and Honeywell, among others. These new correlations allow parallel form controller users to employ IMC tuning methods similar to those already available for the ideal and interacting forms.

This paper demonstrates the use of signal filters both internal and external to the controller through mathematical analysis of closed-loop transfer functions supported by experimental and simulation results. If a four mode PID with Filter controller is not available in an industrial process, the paper demonstrates how a three mode PID controller combines with a signal filter external to the controller to essentially create a four mode PID with Filter controller. The advantages and disadvantages of each of the PID with Filter controller forms are detailed using real-time results from a non self-regulating pumped tank experiment along with simulation studies.