PID replacement

[JoeM3]

Hello,
After a lifetime of hating PID, I have decided to try to replace it with something else. My first stop is my ever handy Logix 5000 development program. Within it are the IMC, CC, and MMC instructions.

My first shop test will be some heating plates. These are 6x6 plates, with 450 watts of heating, and a thermocouple installed in the plate.

There are two of them and sandwiched between them are components that are relatively non thermally conductive. (IP prevents more detail.) The distance between these plates is approximately 1/4 to 3/8 inch, so there is heat transfer between the two halves.

Has anybody any experience with heating, or better yet, heating like above, with the IMC CC or MMC instructions? Or something else? Bang Bang is not quite accurate enough. Once again, I am not interested in PID.

Thanks

 

[geniusintraining]

I am not interested in PID.

Understood but why do you think the others are going to be better?

This may help with the instructions https://literature.rockwellautomation.com/idc/groups/literature/documents/rm/1756-rm006_-en-p.pdf

It may not be an issue of using a PID or any instruction, I have worked on systems that were such **** poor designs that no matter what was done they would still suck at best

I have done a lot of heating and maybe a simple controller will be better than a PLC, most presses I worked on were large and used thermal oil so PID's worked good on them.

 

[drbitboy]

Is the 450W of heating applied to a plate on one side, to encourage the measured temperature of a thermocouple, in some sort of thermal contact with the plate on the other side, to be near the SP, to some specification? Stated more plainly, where is the heat applied, and where is the thermocouple? Because the process will not be directly controlling the temperature of either plate, it can only control the temperature of the thermocouple junction.

What removes heat from the thermocouple junction (forced convection; natural convection; conduction; radiation; some combination of all four)?

What is the relative thermal conductivity of the plates compared to the sandwiched material? What is their relative thermal inertia?

 

[JoeM3]

Natural convection-conduction for the cooling method.
So there are 3 cartridge heaters about 3/8" in diameter, and probably 4 inches long.
We drill holes in the side of the plate which is about a half inch thick and embed the heaters in these holes. The TC is embedded in a similar hole, which if I had to guess is .125" diameter, and it is parallel to the holes for the heaters.

 

[JoeM3]

geniusintrainging,
Something has to be better. Trying to move the bar and achieve something next level.
30+ years of "tuning" PID has led me to a reality that it can achieve mediocre-to-OK results with most things, yet, I can not say that I have ever gotten to great. Because of its characteristics, I don't think it can be great. Hence a new search begins, with new shop experiments. And there do seem to be alternatives these days.

 

[geniusintraining]

And there do seem to be alternatives these days.

I personally have not used them but I am sure others on here have, I have set up a dual PID and worked good on a couple applications

 

[drbitboy]

Natural convection-conduction for the cooling method.
So there are 3 cartridge heaters about 3/8" in diameter, and probably 4 inches long.
We drill holes in the side of the plate which is about a half inch thick and embed the heaters in these holes. The TC is embedded in a similar hole, which if I had to guess is .125" diameter, and it is parallel to the holes for the heaters.

So the heaters have around 1/16" of plate material between them and the non-conducting sandwiched material?


Can we assume the plates are very conductive, at least relative to the sandwiched material?


Are the heaters bang-bang controlled (on or off)? can they approach an analog-ish behavior via PWM?

 

[JoeM3]

Yes the plates are VERY heat conductive.
No, we have junked the bang bang control and we are using SCRs and a analog output on the PLC

 

[Ken Moore]

Can a human control the temperature within acceptable limits, by manually manipulating the analog output?

 

[JoeM3]

No. Walk-away functionality is needed.

 

[drbitboy]

No. Walk-away functionality is needed.

Respectfully, that is not what Ken asked. Not "will a human control" but "can a human control" this process?

 

[JoeM3]

No

 

[Ken Moore]

If a human cannot control it, then controlling it automatically will be difficult.

 

[drbitboy]

I would expect natural convection will have a fairly stable heat transfer coefficient; is it the 1/2” dimension or one of the 6"dimensions that are vertical?

Is there some other form of cooling or other process transients that will vary so much that a human could not keep up with this?

 

[cardosocea]

Hello,
After a lifetime of hating PID, I have decided to try to replace it with something else. My first stop is my ever handy Logix 5000 development program. Within it are the IMC, CC, and MMC instructions.

CC is coordinated control with different control variables... your application, as far as we know has one control variable. Are you going to engineer another one or two because you hate PID?

IMC requires a process model, you hate PID but somehow will enjoy creating a process model more?

MMC controls two process variables with three control variables, see my comment for CC and multiply by 1.5 to assess complexity.

Perhaps changing your mindset around PID control and potentially find something where you can run simulations and understand the impact of each parameter is best than devoting your time to even more complex methods (some of which seem to revolve around PID as well) without a clear gain from doing so?