Auto tuning of PID control loops

06 September 2011

Many PID control loops are still operated manually. CEE looks at an auto tune solution which offers the benefit of having a clear view of the information on a large-screen HMI.

The three-term local loop proportional-integral-derivative (PID) controller is considered to be the process control workhorse. However, the derivation of the PID constants required for optimal setting of the three-term controller is often still a problem because, in order for control loops to work properly, the PID loop must be properly tuned.

A large number of different imperial/graphical derivation methods are available for the manual tuning of a process loop, including the Ziegler-Nichols system developed in 1942, the Cohen-Coon system developed in 1953, and the Minimum Error-step Load Change developed in 1998. However, for all of these an understanding of the plant transfer function is required to select the appropriate model and settings.

The automated setting of the P, I and D parameters by the controller can offer benefits in most control systems, and is relatively easy to implement. However, as recently as January 2009, an article by Greg Baker, stated that more than 30% of control loops were operated manually, and this was based on the 1993 article by David B. Ender, indicating little change over the last 15 years. In the feedback to the article Mr Baker said: “Until a current and comprehensive study on the root causes of underperforming PID loops exists, we may only agree that the relative magnitude of underperformance, due to incorrect PID configuration, is not insignificant. We’d hope it has improved since the time of the source information, 15 years ago.”

All algorithms/tuning methods will have benefits and drawbacks, and it is often these drawbacks, combined with a lack of knowledge about its potential benefits, that have resulted in many process loops remaining in a manual state.

Adaptive control systems
Fundamental process understanding, coupled with appropriate modelling and development of robust on-line monitoring is necessary to identify which are the critical process parameters that require monitoring and control, for an individual plant system.

To simplify this task Unitronics, a manufacturer of process control PLC/HMI systems, developed a proprietary auto tune PID algorithm to remove the burden of plant tuning. This has been used in the field for over four years and has been incorporated across its product range, starting with four loops, moving up to today’s offering – the V1040 and V1210 HMIs - which can handle 24 independent loops.

The system is based on the Ziegler–Nichols heuristic method of tuning a PID controller, with propriety modifications by Unitronics based on real world results, Ku (the Ultimate gain) and the oscillation period Tu (the plant oscillation period) are used to set the P, I, and D gains. These gains apply to the ideal, parallel form of the PID controller. When applied to the standard PID form, the integral and derivative time parameters Ti and Td are only dependent on the Plant oscillation period Tu, which includes any plant non-linearity and time lags.

The Ziegler–Nichols tuning creates a ‘quarter wave decay’ to give PID loops with best disturbance rejection performance. It does have some disadvantages, as this setting typically does not give very good command tracking performance. The Ziegler-Nichols method also creates an aggressive gain and overshoot, and most applications require the PID system to minimise or eliminate overshoot. Using the step function, bounded system, the Unitronics proprietary algorithm produces a practical and industry accepted PID system for a wide range of plant types.

The auto tune facility requires five inputs from the control engineer for the initial auto tune run, Set point, Process variable low limit, high limit, controlled variable low limit and high limit. The user can set the number of stages the auto tune applies in the plant recognition algorithm. The default setting is three, but can be increased to a maximum of eight.
With minimum information, the PLC controller is connected to the control loop and during the auto tune process the plant reaction to the proprietary stimulus algorithm, is measured and compared in controlled iterations.

The PID Server utility enables the control engineer to auto-tune PID loops for both the Vision and M90/91 controller series. Although it is installed as part of the VisiLogic/U90 Setup, PID Server runs independently of other Unitronics software. It allows users to collect PID data and to use it to troubleshoot and solve control loop problems by displaying the information as a graph on the HMI. It also makes it easy to set up, test and tune a complicated multi-loop system.

Conclusion
There is little double that the loop tuning debate will continue, but as auto tuning algorithms reach maturity it is hoped that industry will accept the reliability and ease of tuning and maybe in the next 15 years the reported percentage of process control loop set to manual will be significantly reduced.


Contact Details and Archive...

Print this page | E-mail this page