Software Review: Loop Tuning Simplified

LIBRARY MENU


	Table of Contents
	Performance Supervision
	PID Loop Optimization
	Basic PID Tutorials

MORE RESOURCES
	Webinars
	Training
	Newsletter

"Excellent video! Very nice! I remember how surprised I was the first time I encountered a dramatic disparity between the load response and the setpoint response. It’s not at all obvious that that could happen." Vance Van Doren, Consulting Editor, Control Engineering

Software Review: Loop Tuning Simplified

Reproduced with permission of CONTROL ENGINEERING, May, 1995.
© 1995 by Cahners Publishing Company

PID loop tuning can be a complicated operation. Analytical tuning techniques require a mathematical model of the controlled process and a series of arcane calculations to derive the appropriate values for the P (proportional), I (integral), and D (derivative) parameters. There are as many tuning techniques as there are applications subject to PID control, and each involves a different set of rules for deriving the desired parameter values. Although several very powerful software packages available for performing the necessary calculations, the user typically needs a college education in feedback control theory just to understand the results.

The engineers at ExperTune Software (formerly Gerry Engineering, Hartland, WI) are trying to change all that. Their most recent PC software package, ExperTune for Windows, takes the user through a loop-tuning operation step-by-step without a lot of analytical mumbo jumbo. The results are PID parameters that will give the loop in question optimal set-point tracking or a slow, fast, or faster response for disturbance rejection.

ExperTune does all the work. All the user really needs to do is identify the location of the loop to be tuned and then follow ExperTune's directions. ExperTune first asks the user when and how much to "bump" the process input so as to determine the process response to a control action. Once the process has finished reacting to the bump test, ExperTune terminates data collection, generates a model for the process, and calculates the tuning parameters required to meet the user's performance specifications. ExperTune also estimates the performance improvement that should result once the loop is re-tuned. Once user approves of ExperTune's recommendations, then they can download the new parameters to the loop and resume automatic control operations.

For the more ambitious user, ExperTune provides several additional analysis tools. A "robustness plot" is available to show how a change in the process dead time or gain would affect the stability of the closed-loop system as it is currently tuned. The current process parameters are shown as a point on a two-dimensional plot of normalized dead time vs gain. All points that would correspond to stable closed-loop operations are shown within a "stability region." If the current operating point lies well within the stability region, the closed-loop system is said to be "robust." That is, it is unlikely to become unstable if the process dead time or gain should happen to change slightly. The dimensions of the stability region depend upon the current values of the PID tuning parameters. Hence, the robustness plot can provide the user with a graphical image of the risk of instability that a given set of tuning parameters would entail.

ExperTune can also provide a dynamic simulation of the closed-loop system based upon the observed behavior of the controlled process. This feature allows the user to check the closed-loop performance of ExperTune's recommended tuning parameters even before they are downloaded to the PID loop. I would find this feature particularly useful if none of ExperTune's three recommendations were quite right for my application. I would start with the recommendation that provides the closest approximation to the disturbance response I require, then fine tune the P, I, and D values by trial-and-error. The simulator would allow me to test each new set of parameters without actually running the process.

These and ExperTune's other analysis tools can all be useful for analyzing the behavior of a closed-loop control system, but only if ExperTune's internal process model accurately reflects the behavior of the controlled process. The accuracy of that model depends on the integrity of the process data collected during the initial bump test. To insure data integrity, the bump test must start and stop with the system at rest, it must not be conducted during periods when appreciable load disturbances are possible, and it must include data from a linear range of operation. I would find these conditions difficult to meet in applications where the process lacks a steady state, the process is subject to load disturbances at unpredictable intervals, or the process responds differently to positive and negative step inputs.

In these situations, I would recommend more advanced modeling and tuning techniques or the services of an expert control engineer. ExperTune Software just happens to have a staff of engineers available to help tune such difficult loops. There is even a standard facility built-in to ExperTune for collecting the necessary set-up information and process data on a disk that the user can send to the company for analysis.

ExperTune for Windows runs on an IBM PC-AT or compatible computer under Windows 3.1 or higher. The host PC must have at least 2M bytes of available memory, a VGA display, and a parallel printer port. Several interface protocols for acquiring plant data are supported including DDE; an A/D converter available directly from ExperTune Software; and drivers for Allen-Bradley, Modicon, Siemens, Moore Products, and GE Fanuc controllers.

Vance VanDoren
Consulting Editor