Anti-reset windup (ARW) protection is a standard feature of industrial PID controllers. In some DCS, ARW limits are adjustable besides output limits. The ARW limits may not be at their best values. ARW default values may not match up with output limits as output scale and engineering units change. While we normally think of the ARW limits being set equal to the output limits, this has not been the case before the advent of the digital positioner and is not the case when external reset feedback or an enhanced PID is used. Here we take a closer look at the past and future use of anti-reset windup options and smarter reset action.
This is the first in a continuing series of practical and useful questions on PID tuning raised by Brian Hrankowsky a knowledgeable process control specialist in the pharmaceutical industry after having been present at a panel session on controller tuning at ISA Automation Week 2010 developed by Michel Ruel. Brian is not representing his company in these posts. The questions and answers in these blogs will be addressed in much greater detail in the long overdue 4th edition of my book Tuning and Control Loop Performance being published by Momentum Press.
Since anti-reset windup algorithms are proprietary and complex, the following is a view based on experience with a just a few different algorithms. The adjustment of settings or the selection of options should be carefully tested and monitored because surprises are the norm due to limited documentation and understanding by the user and even the provider of hardware and services.
First and most important, if you have adjustable ARW limits make sure the range of the ARW limits is consistent with the range of the output limits. When output scales are set different from 0-100% due to engineering units in the output or operating limits, the ARW limits may not automatically change, remaining at the default of 0-100%. For cascade control loops, the primary PID output may be in the engineering units of the secondary loop depending upon the vintage and type. Also, the primary output limits may be set to match the setpoint limits of the secondary PID. In an example provided by Danaca Jordan, an ISA Mentor program protégée, a primary temperature PID output scale range was configured to be 0 to 200 degrees but the ARW limits remained at the default of 0-100%. The result was a disruptive fast reset action when the primary PID output was decreasing between the ARW and output limits.
In some DCS if the ARW limits are inside the output limits, integral action is 16 times faster when the output departs from the output limit than when the output approaches the output limit. For the temperature loop example, the ARW high limit being at 100 and the output limit at 200 causes the integral action to be 16 times faster when the primary temperature PID output is decreasing between 100 and 200. In this case, the high reset action promoted undershoot and cycling.
The question is should the ARW be set inside, equal to, or outside the output limits? The answer depends upon the application and the DCS implementation of the integral mode.
Pneumatic positioners were often out of calibration a few months after being in service. Also excessive backlash and incorrect bench settings and actuator sizes prevented the valve from being fully closed at 0% or fully open at 100%. Since there was no position readback in the control room unless position transmitters were installed, it was anybody's guess as to the actual valve position. To make sure a valve was fully closed, the low output limit was set at -10% and sometimes to make sure the valve was fully open the high output limit was set at 110%. Setting the ARW limits inside the output limits (e.g. 0% and 100%) enabled the output quickly to recover to the throttle range via 16x reset action. For some tight shutoff valves, the high breakaway torque prevented the valve from opening until it reached 15%. In this case a low ARW limit of 15% was used to get the valve open sooner. This was important for surge control.
Digital positioners on control valves originally designed for throttling service will have an accurate calibration and readback of actual position. The output limits can be set at 0% and 100%. In fact when digital positioners first appeared, the digital circuit went to sleep if the low output limit was -10% or less due to low current supply. The low output limit should be set at 0% for the modern control valve for best response to and from the closed position. The exception is the on-off valve posing as a control where the positioner is lying due to excessive backlash and position feedback being on the actuator shaft instead of the internal closure member as discussed in the November 2012 Control article "Is your Control Valve an Imposter"
Another case for setting the ARW inside the output limit is when a temperature controller output at its high output limit for an excessive period of time on heat up causes burnt material. Hector Torres, ISA Mentor program protégé, correctly surmised that setting the high ARW limit well inside the high output limit of 100% would allow the temperature to rise fast by hitting the limit but would enable the output to come off the high output limit faster. Hector also put a rate limit on the setpoint change to reduce the initial increase in the output. If the process variable gets ahead of the rising setpoint, the ARW limit quickly decreases the output from 100%.
Setting the ARW limits at the output limits does slightly help get the output off the limit but the effect depends heavily upon the tuning. Basically, Hector could have chosen to schedule the reset setting based on output to prevent hitting the output limit. If the ARW limit is set equal to the output limit, decreasing the reset action as the error decreases will help get the output off of the limit sooner.
If your DCS has the positive feedback implementation of integral mode and consequently the external reset feedback option as shown in slide 15 in the Emerson 2012 Exchange presentation "Effective Use of PID Features for Control and Optimization", you have a more effective method with no adjustment. If you turn on external-reset feedback (e.g. dynamic reset limit), and have the external reset signal correctly configured, you inherently have anti-reset windup. In this case, the ARW limits should be set outside of the output limits to get the effect of the ARW out of the picture. Your output will come off of the output limit when the proportional contribution to the output becomes less than the integral contribution with the total contribution inside the output limits. To make the output come off of the output limit sooner you can either increase the reset time or increase the controller gain. The more conservative choice is to increase the reset time since this increases stability. Most process loops on vessels and columns have too much reset action and would benefit from a larger reset time anyway. Also increasing the reset time in integrating processes, such as level, gas pressure, and batch temperature helps get the controller gain above the low limit gain as shown on slide 21.
The enhanced PID developed for wireless enables external-reset feedback and has a filter recovery parameter that allows the user to adjust when the controller output should come off of the output limit. For pressure relief loops and many override controllers, it is desirable that the output not come off of the limit until the process variable (PV) crosses setpoint. However, for surge control and most other loops in a saturated condition, the output needs to come off of the limit before the PV reaches setpoint to prevent overshoot. This can be achieved by the simple adjustment of the recovery filter.
A general method for all PID could use the computation of the rate of approach to the setpoint and the setting of the reset time to make the integral contribution less than the proportional contribution when the PV is within a specified error per the equations and discussion on pages 14-16 in the ISA 2012 Automation Week paper "Effective Use of Key PID Features" If the near or true integrating process gain can be identified, the method can automatically achieve fastest rise time and minimum overshoot.
The reset setting can also be made smarter based on a future PV calculation and whether the PV is overshooting or faltering in its approach to setpoint. This would deal with nonlinearities, changes in dynamics, a large spectrum of dynamics, and non saturation conditions. This subject has been briefly discussed in the Control Talk Blogs "The ABCs of Controller Tuning" and "Future PV Values are the Future" and will be demonstrated in a future blog.