Dear Sirs,

I studied basic control engineering some 30 odd years ago. I did not come across the view that, a temperature system with a set point range anywhere between 0 to 1000 degrees C, would require the different PID values dependent upon the set point within that range. Further, a different set of PIDs for lets say 20 to 250 degrees C another set for 250 to 500 degrees C another for 500 to 900 etc etc.
I work with those that think this is a normal situation in control engineering? Surely, within most system s the loads and lag times remain constant throughout the control range therefore the PID values need not alter.

I am told that if you set Proportional values for let’s say a temp' increase from 20 deg to 600 degrees, then obviously these values would cause a massive overshoot if used to raise the temp from 20 deg' to 100 degrees, as the former gain would be far to high for the latter 80 degrees step change.

I just know this is wrong but can not argue the case.
Can anyone pass comment? I would be deeply grateful.
Thank you for your time in reading this missive.

Regards to all,
Ted

 

Ted,

There is indeed some merit to what your friends are telling you, but I believe there's actually a subtle difference... they may want to have different gains depending on the set point differences, not merely the temperature range.

Often times I like to control when the integral component turns on, and have different control scenarios depending on the delta between the actual at set point. For example, I might force the integral component to 0 until I'm within say 5% or cross-over of the set point during initial heat up. Otherwise, the integral component will be wound up unnecessarily.

Other times, I use varying integral values depending on how large the difference is.

There are other schools of thought which get around some limitation of more simple system which heat up large systems in steps that only raise the temperature by 20% at a time, for example, to avoid this windup / overshoot.

It really depends on what you have a temperature control unit and its sophistication. I can tell you from experience that B&R Automation's controls have very sophisticated temperature control instructions built-in that offer a lot of flexibility...

Good luck!
Matt

 

Yes indeed, this is a very normal situation with control engineering.

Let's study an analogy of driving a truck.

If you are driving this truck between 0 and 20 mph, you can react to situations (changing lanes or speeds, avoiding children or potholes) fairly aggressively.

Driving between the speeds of 60 and 80 mph, if you were to react in the same aggressive manner, you would likely crash the vehicle.

This is a very simple situation where the PID values that you are mentally using are adjusted for these different ranges. (To complicate the situation further you could also add cargo load to the truck, slick roads or slightly sloppy steering. You can quickly see how the PID must be adjusted for different operating ranges.)

Given the wide range of temperature that you wish to control (0 to 1000 degC) it is easy to imagine that you will require different PID values throughout.

 

You've forgotten the effect of process gain - if you halve the measurement range, you are effectively doubling the process gain and need to halve the controller gain in response to get the same total loop gain.