I am always intrigued by the control system involved in surge control protection of rolling machines. The term which often come across is Anti reset windup controllers, normally used for variable speed compressors.

I'll be obliged if someone demolish the obscurities involved in this action and briefly statinig the exact action.

ranaehtasham @ hotmail. com

 

You can figure out what this action is in all sorts of control systems if you just understand what wind-up is and why it needs to be limited or reset on occasions.

When a control loop is in control with a minimal error, then wind-up is not a problem. Wind-up is only a problem when a control loop is temporarily taken out of control. This can happen in various ways. For example, an external agent might temporarily disable control feedback. Or, a large change in the control setpoint may be introduced. In any case, if a control loop has to be grossly out of control for a period of time, the integral portion of the control feedback (the "I" term in a PID loop), can rise to an arbitrarily high value. Now when the control loop tries to regain control, this huge integral term can take a long time to come out of saturation. This will lengthen the time it takes to get the control loop back in control with minimal error. The term "wind-up" is a way of viewing the integral term as a spring that takes a long time to "unwind".

There are several ways to reduce the negative effects of integral wind-up. One is to enforce saturation limits on the integral term so that it can�t get too far out of whack. Another solution is to recognize the special circumstance in which the control loop is forced out of control and force the integral term to a default value at those times. This is what is meant by "resetting" the wind-up.

Robert Scott
Real-Time Specialties
Embedded Systems Consulting

 

ExperTune has a good definition in their glossary, and go into some more detail (about PID) in the second citation. http://www.expertune.com/glossary.html http://www.expertune.com/artCE87.html

controlguru touches on it a bit http://www.controlguru.com/2008/021008.html

Details of how ARW is implemented varies between manufacturers, and often, even with one manufacturer, may be different from one model and/or generation of controller to the next.

The basic theme is to limit or prevent integration when the output device (whatever that may be) is in saturation, that is, at 100% output.

When output is saturated having the integral term trying to pour more coal onto the fire, so to speak, doesn't do any good, but it does "wind up" it's contribution to the controller's output. Once measured value drops back down towards setpoint the "wound up" integrating term forces output to be higher than it otherwise would be until it "unwinds".

I can't comment on how it applies to your particular application, but can give a different example to illustrate the point, and, with any luck, will do it without misleading you too much.

In a plastics extruder barrel zone temperature control scenario the first two zones typically create a lot of shear heat as polymer pellets are converted into a melted pool, and this can require considerable cooling. Say we have cast-in heater elements using water cooling, a setpoint (SP) of 300°F, and, even with 100% cooling command, and the cooling system otherwise in good condition, we aren't able to get below 340°F measured value (MV) for a particular extruder screw speed, and other conditions (material characteristics, etc) as they obtain.

This deviation of 40°F between SP and MV will pour on more integral correction at the integrating rate (typically, an integrating time on the order of 7 to 9 minutes is suitable for a 6 inch extruder), and after several hour of operating like this the integrating term will get "wound up".

Operationally, things will be fine (except for the 40° deviation) as long as the underlying process remains stable. However, if the rate of shear heat generation drops off to the point where 100% output is no longer required the PID controller's output term, which is now wound up, will remain higher than it needs to be under these new conditions. If the drop in shear heat generation was rapid and/or shifted by a large value (for instance, a 'melt block' formed), then the wound up integrating term will cause overcooling - the more wound up it gets; the more overcooling occurs - and, in the case of an extruder, this can lead to oscillation, and other undesirable consequences.

In this case, if setpoint is simply set too low, there is never enough cooling output under normal conditions to obtain the desired measured value, and good machine operation can be obtained at the higher temperature then considering a higher setpoint is a viable solution.

The other thing is to introduce anti-reset windup.

In some controllers this is expressed as a percentage of proportional band, and, if this is the case, if we have a +/-50 degree prop band in the above example (100° overall - heating start to shave off at 250°; cooling reaches 100% at 350° in pure proportional mode; i.e. - with integrating, derivative, and other esoteric tweaks turned off) then a 20% ARW setting (20°) would limit integration's contribution to loop output once measured value fell outside of the 290° to 310° window (+/- 20°), and heating/cooling was at 100% output.

 

Rana,

I'm not certain that this is what you mean but here goes.

Reset Windup refers to the way the reset action will wind output all the way to 100% or 0% at times when the process is out of service. For example a flow loop but the pump suppling the water is off line. The Reset action will cause the valve to ramp to 100% open. With Anti Reset Windup the Reset action is disabled. The valve may open due to the gain but when the pump re-starts the valve will revert back to the correct position preventing overshoot.

This is common practice.

Regards,
Roy