Good PLC for Temp Control

According to Beckhoff overview description of TF4110 Temperature Controller - "Simple commissioning through self-adjustment of the controller (auto-tuning) is included."

TF4110 Documentation

Just one other I remember was a bank of Carl Schnell mixers, the clients spec was to use Eurotherm controllers for all temperature control, this proved to be a real problem as the product turned thixotropic, shallow vessels so as the product rose in temperature the vapour injectors would blow holes through the product, the controllers by their nature would increase the flow, this just made it worse, talked to Eurotherm, they sent out a specialist, no joy, I then removed the controllers, built in a profile of steam flow into the recipes, this worked like a treat, so PID in normal terms is not always the answer, yes I could have built in a profile of flow into the controllers, tried this still a little iffy, my idea worked & it removed another not required piece of equipment & so reduced the possibility of failure.

parky said:

Just one other I remember was a bank of Carl Schnell mixers, the clients spec was to use Eurotherm controllers for all temperature control, this proved to be a real problem as the product turned thixotropic, shallow vessels so as the product rose in temperature the vapour injectors would blow holes through the product, the controllers by their nature would increase the flow, this just made it worse, talked to Eurotherm, they sent out a specialist, no joy, I then removed the controllers, built in a profile of steam flow into the recipes, this worked like a treat, so PID in normal terms is not always the answer, yes I could have built in a profile of flow into the controllers, tried this still a little iffy, my idea worked & it removed another not required piece of equipment & so reduced the possibility of failure.

Click to expand...

You have pointed out why I hedged my recommendation for using stand alone controllers. Stand alone controllers are great for static applications. A PLC is much more flexible but it requires that one understands the process. Another example is controlling PH. Ph is not linear. However, everyone seems to treat Ph control as if it were linear.




When the application gets "complicated" it requires knowledge of the system. There are no short cuts.

Totally agree, I liked the Eurotherm controller, this had self adaptive tuning and came in useful in an application where over time the injector core efficiency changed due to build up in the injector core, the product varied a little as well, this worked well, originally the techs used to increase the pressure controller a little every day, then do a clean of the core at weekends, SAT negated the need for changing the pressure SP, we even put the temperature controller in the system in SAT, cleaning was still required at weekends but the pressure change was not.
NPID controller has the standard structure of the linear PID controller, however, controller gain and integral time are not fixed but are designed to be nonlinear functions of the control error.

The process is a small furnace, about 25gal volume, with three zones set at 1000C, 1020C, and 1040C. Each zone has a thermocouple that will of course be inputs to the PLC thermocouple module. The load is steady state. The furnaces are just cooking product at those steady temps for days. Temperatures to be held constant. As far as the accuracy required, I want to hold it right at those three set points.

Eurotherrm temp controllers are doing the job now and their holding perfectly steady. I had to open my big mouth and replace controls with a PLC for new furnaces arriving soon without thermocouples.

It's a new job for me and I really want to make this work. I like to avoid looking like a friggin idiot whenever possible. but that goes triple at a new job. I will have a backup plan involving temp controllers but I really want to challenge myself to pull this off.

I thank you all for your previous messages and for any info that can help.

*they're holding perfectly steady

eaglewood said:

...The load is steady state.

...new furnaces arriving soon without thermocouples.

...I thank you all for your previous messages and for any info that can help.

Click to expand...

Minor (second-order) nits, but things to keep in mind if the project does not follow expectations:

• How will the new furnaces be controlled without thermocouples?
• The Eurotherms and PLCs do not control the temperatures of the ovens, they control the temperatures of the bimetallic junctions of the thermocouples, which junctions we assume are adequately representative proxies for the oven temperatures.
• No process load is steady state
  • I assume by "steady state" OP means that, once the furnaces and product are at a uniform temperature throughout, all process disturbances (HVAC cycling, diurnal variations, doors and windows opening and closing) are small enough to be ignored.
  • Certainly at startup, where furnace and product have thermal gradients, the "load" will be slowly decreasing, even if the thermocouples have reached setpoint.
• The query about accuracy has not been answered
  • 10xxdegC are the setpoints
  • +/- 1.0degC?
  • +/- 0.1degC?
  • What is the read noise of the thermocouples? Will there need to be some filtering?
• Is there crosstalk (heat transfer) between the three zones?
  • Does the controller of the 1040degC TC do most of the work, and the controllers of the other TCs are trim?

In the end, since PID behavior is pretty straightforward (less so the tuning, to be sure), anything the Eurotherms are doing any PLC should be able to do, apart from autotuning. The PLC might be able to do even more e.g. for startup, leave the PID in manual, set the output to 100%, and have it switch to automatic when the temperatures are close. Not that that necessarily changes anything for the better compared to a PID on auto the whole time, but it might be possible to leave it at 100% longer: a PID in auto will start to drop the output at a time determined by its tuning parameters and the dynamic behavior of the TC junctions in the furnace.

*they're holding perfectly steady

Click to expand...

This is likely to be the standard by which the final result will be judged.

drbitboy said:

How will the new furnaces be controlled without thermocouples?

Click to expand...

Correction: The new furnaces will arrive without temp controllers not without thermocouples. There will be three type-s TCs.

drbitboy said:

I assume by "steady state" OP means that, once the furnaces and product are at a uniform temperature throughout, all process disturbances (HVAC cycling, diurnal variations, doors and windows opening and closing) are small enough to be ignored.

Click to expand...

Assumption correct. It's a vacuum chamber kept at steady pressure, no doors, no product cycling through, sitting in a temperature controlled indoor environment.

drbitboy said:

What is the read noise of the thermocouples? Will there need to be some filtering?

Click to expand...

I'm not sure how to determine this explicitly but judging by the fact that the temps stay right at setpoint, I'd say we're good. But variations such as a long run of thermocouple extension wire is coming in the future. Right now the thermocouple wires are about six feet. They will be extended to about 25 feet in the future. Noise is something I need to be prepared to deal with.

drbitboy said:

Is there crosstalk (heat transfer) between the three zones?

Click to expand...

Yes

drbitboy said:

Does the controller of the 1040degC TC do most of the work, and the controllers of the other TCs are trim?

Click to expand...

No, they all share the load fairly evenly.

drbitboy said:

+/- 1.0degC?
+/- 0.1degC?

Click to expand...

drbitboy said:

This is likely to be the standard by which the final result will be judged.

Click to expand...

No doubt about it. The accuracy is more to match the performance of the temperature controllers so the temp controllers won't make me look bad. (My inability to control the process will make me look bad, more like, but I'll take it out on those temperature controllers .) Fluctuations of a few degrees C won't hurt the product. But I want those temp gauges pegged to their setpoints.

EL3318, 8-Ch Thermocouple input module.

If you want to go the PLC route, use the above in conjunction with TF4110 temperature control library linked previously.

I understand that it's a new job, and nobody wants to gaffe in a new job, but the best way to learn something is to jump in and start doing it. That's how I learned the Delta Motion and Beckhoff stuff. It was well worth the effort (and some growing pains). You can download TwinCAT for free and start programming and configuring it now, before your new machinery arrives. That way you have a good head start when commissioning time comes.

eaglewood said:

The accuracy...

Click to expand...

Okay, let me ask it this way:

• How many digits to the right of the decimal point do any and all current displays show for temperature, degC?
• What is the resolution, in degC, of the input A/D module (I know it's non-linear, but it will be constant within a factor of 1.00x or so at operating temperature)?

E.g. if the former is 0 and the latter is 0.05degC, then there can be several 0.05degC-sized movements before the display changes, and that makes it look steady.

eaglewood said:

No doubt about it. The accuracy is more to match the performance of the temperature controllers so the temp controllers won't make me look bad. (My inability to control the process will make me look bad, more like, but I'll take it out on those temperature controllers .) Fluctuations of a few degrees C won't hurt the product. But I want those temp gauges pegged to their setpoints.

Click to expand...

Seems not all that complex of a system. As stated, I used a couple of function blocks in the TF4100 Controller Toolbox to control temperature in a fluid tank. I'm no expert in PID, but just through mostly trial and error, I have it holding within +- 3° C, and that is with a fair amount of disturbance in the system (The fluid is pumped out of the tank and onto a large, steel, flat plate (big heat sink) and then flows back into the tank via gravity.

My assumption is that the function blocks in the TF4110 Temperature Control library would work even better. Looking at those, they certainly have many options for temperature control applications, but either library (4100 or 4110) would work for your application. TF4100 has quite a few different PID type function blocks, and function blocks for PWMing, so could work for other applications too. Whereas, the TF4110 Temperature Control library appears to be best suited for temperature control applications specifically and may even be 'overkill' for your application (?).

busarider29 said:

I understand that it's a new job, and nobody wants to gaffe in a new job, but the best way to learn something is to jump in and start doing it. That's how I learned the Delta Motion and Beckhoff stuff. It was well worth the effort (and some growing pains). You can download TwinCAT for free and start programming and configuring it now, before your new machinery arrives. That way you have a good head start when commissioning time comes.

Click to expand...

Absolutely. I'm embracing the challenge. I mean, after a bit of regret for speaking too soon, not ever having tuned a PID loop, I am looking forward to taking it on and working until I nail it.

drbitboy said:

• How many digits to the right of the decimal point do any and all current displays show for temperature, degC?
• What is the resolution, in degC, of the input A/D module (I know it's non-linear, but it will be constant within a factor of 1.00x or so at operating temperature)?

E.g. if the former is 0 and the latter is 0.05degC, then there can be several 0.05degC-sized movements before the display changes, and that makes it look steady.

Click to expand...

The temp controllers are displaying whole numbers only. Not sure of the resolution right now, but I see what you mean. Thx.

It is unlikely that they use decimal places for a temperature in the thousand degrees range, the one thing you have not mentioned is the control is it on/off or analogue (I suspect it will be on/off time prop).
For that application I would plump for dedicated controllers personally.
I once modified a lead smelting system for battery plate manufacturing, this was on I think a PLC2, it was existing but the two burners were upgraded, very little info from the customer & I was commissioning this on Christmas Eve, nobody else on site except the guys fitting the new burners who wanted to get home, I did my own time proportioning algorithm but the burner people said due to pressure problems I could only run one burner not the two, it seemed to work well, however, when the customer returned after the break he informed me that when both burners ran it would go way over temperature even in low flame, I suggested he switch off one burner when close to the required temperature in the software & it worked a treat, I sent him an updated program to download, the response was it worked perfectly & controlled within +- 2 Degrees C, with just a small overshoot on start up.

There is some debate over the interpretation of NADCAP and aerospace heat treat requirements that are covered in AMS2750F regarding the resolution of the setpoint and process variable (section 3.2.3.2 covers the temperature resolution).

Some interpret it to mean only the recorded data needs to be displayed to a tenth of degree, even at temperatures above 1,000, others interpret it as the controller must also have tenth of degree resolution as well.

But, I would think that if the process required compliance to AMS2750F, that the requirement for such would have been made evident to the integrator/implementer.