Help with temperature control?

[lschiavo]

This is not brewing related although it looks like it could be. I wonder how that happened?

Here is the setup. The heat comes from two cheap 1500W hot plates. The intent is for the kettles to batch out oil and lye for soapmaking. So the initial heating could vary between three and seven batches. They are 5 and 10 gallon kettles. Soon to be both 10. *edit* Goal temperature is around 125.

The problem I am having is getting the controller to cope with the thermal mass of the cast iron hot plate. It is causing crazy overshoot.

I started with PID's. I autotuned several times (full volume and single batch) and could not get the results I was looking for. I did not get into manual tuning yet.

I figured that before I spent too much time on the PID's, I would try the DSPR120. I just swapped one of those in yesterday. I heated the oil kettle in mash mode with only one batch in it and overshot 50 degrees. I set the offset and tried again. It worked the second time but I wasn't crazy how the dspr handled it. It seemed to just limit the power to the element for the entire heating. That seems like it is not fully utilizing the heat source and will add a lot of time to heating.

In short, if any of you folks would care to offer some advice:

- What controller would you use for this situation?
- should the system be tuned to a full kettle, smallest batch volume or somewhere else?
- how best to tune?

Thank you for having a look and any help you can offer.

 

[Yooper]

I have no idea how to help, since you're the electrical engineer and I'm just a soapmaker- but I have to say that you rock.

When you get it figured out, you can easily use the same system for cheesemaking you know!

 

[lschiavo]

I have no idea how to help, since you're the electrical engineer and I'm just a soapmaker- but I have to say that you rock.

When you get it figured out, you can easily use the same system for cheesemaking you know!

Thanks. It helps just to know you're rooting for me.

I have to admit I'm not as patient with figuring out these controls as I should be. I'm hoping some of the experts around here can save me some more gray hairs.

Cheese eh?

 

[BrunDog]

The problem is likely not the controllers - it is the temp probe location/response assuming there is no recirculation and viscous liquids, the convective currents alone probably aren’t enough to encourage mixing and uniform temperatures. Can you share more about the temp probes and their locations?

 

[lschiavo]

BrunDog, I was hoping you might see this. You seem to be on top of this stuff.

The probes are PT100. I had to locate them low in the kettles to allow for batching. I'm actually surprised at the visible convention with the oil. Stirring did not seem to drastically affect the reading at least not enough to be a concern.

You can see the hot plate in this pic. I wonder if I would be better off without the cast iron type?

Here is the dspr120 after sitting for a few hours with only one batch of oil.

I realize that batching is going to cause some difficulties with this setup. Maybe it will have to be a little more hands-on than I hoped.

 

[BrunDog]

I am surprised stirring didn't help. Are you sure you stirred frequently enough?

Is the DSPR a legitimate PID? Does it have an integrator maximum?

 

[lschiavo]

I am surprised stirring didn't help. Are you sure you stirred frequently enough?

Is the DSPR a legitimate PID? Does it have an integrator maximum?

I'm still just in the testing stage with this. We heated three batches of oil on the first test. Once everything was liquified enough, I stirred occasionally to see how evenly the heat was being distributed. When stirring, I only noticed a degree or two of difference. The pot is triclad. It must do a decent job of distributing the heat.

The dspr is an Auber easy boil. I'm not that very well versed in how it actually works.

 

[SleepyCreekBrews]

I'm having a hard time imagining how a couple pounds of cast iron could cause "crazy overshoot" when heating 35 pounds of oil ... Exactly how far is it overshooting?

If the PID is working (and tuned) properly, it should supply full and constant power until it gets to within 95% of the target, and then start duty-cycling lower and lower until it eases into the target temp, with minimal overshoot (1%)

I would expect some hysteresis from the mass of the cast iron, but not huge.

 

[lschiavo]

I'm finding it very frustrating too. The pid with auto tuning was way off. It initially heated to the setpoint, shut down and the temperature kept climbing at least 20 degrees overshoot. This was with 3 batches (~30lbs) of oil. I let it run the rest of the tuning cycle and it overshot again next time. Tried many times with varying amounts of liquid and have yet to get usable results.

I think my problem is going to be the batching out from the kettles. If/when I get something tuned, I feel like the system will have more and more overshoot as material is drawn from the kettles.

I don't know much about manually tuning PID's. Do you think that should be necessary here?

 

[BrunDog]

Try turning the I off (making it a PD controller). Set I to zero if there isn't another way to do it.

 

[lschiavo]

Try turning the I off (making it a PD controller). Set I to zero if there isn't another way to do it.

I'll try that next chance I have to mess with it.

 

[TexasWine]

I don't think the EZBoil is a traditional PID based controller and I don't think you can adjust the values for P, I, or D.

But please check because I'm going off of memory after a very long day of traveling. Austin to Houston to OKC, back to Houston. I'm ready for bed.

 

[doug293cz]

Sounds similar to a problem I faced early in my career with hot plates. The high thermal mass of the cast iron (was thick Al plates in my case) of the hot plates causes slow response to changes in heating input. Once the liquid in the kettle gets to the set point, there is enough heat held in the cast iron to cause continued heating, and thus overshoot.

A DSPR does not use a PID algorithm, and is unlikely to work well in your application. I think you will need to manually tune your PID's to account for the long thermal time constant of your system. If you want to avoid temp overshoot, you will have to achieve what is referred to as a "critically damped control response." Under damping will result in temp overshoot, and over damping will result in excessively long heat up times. In any case, once you get it tuned well, you are likely to be unhappy with the heat up times.

You can improve the response time by going to heating plates with very low thermal mass tops. I don't know if this is an option for you or not.

It's been a long time since I studied how to tune PID's. I'll do a little reading in the next few days.

Brew on

 

[lschiavo]

I don't think the EZBoil is a traditional PID based controller and I don't think you can adjust the values for P, I, or D.

But please check because I'm going off of memory after a very long day of traveling. Austin to Houston to OKC, back to Houston. I'm ready for bed.

You are correct. I started with PID's and swapped in one of the easy boil's to see if it might work. It's not working that well.

 

[lschiavo]

Sounds similar to a problem I faced early in my career with hot plates. The high thermal mass of the cast iron (was thick Al plates in my case) of the hot plates causes slow response to changes in heating input. Once the liquid in the kettle gets to the set point, there is enough heat held in the cast iron to cause continued heating, and thus overshoot.

A DSPR does not use a PID algorithm, and is unlikely to work well in your application. I think you will need to manually tune your PID's to account for the long thermal time constant of your system. If you want to avoid temp overshoot, you will have to achieve what is referred to as a "critically damped control response." Under damping will result in temp overshoot, and over damping will result in excessively long heat up times. In any case, once you get it tuned well, you are likely to be unhappy with the heat up times.

You can improve the response time by going to heating plates with very low thermal mass tops. I don't know if this is an option for you or not.

It's been a long time since I studied how to tune PID's. I'll do a little reading in the next few days.

Brew on

Thank you Doug. Critically damped. The terminology alone should help. The PID's are from Auber if that makes any difference.

I have a standard coil element style hotplate arriving in a few days. I think that will help. I only got the cast iron because they seemed like they would support the weight better. Never thought about the control issue.

 

[doug293cz]

This site gives some practical methods for manually tuning a PID based on the natural resonant frequency of the system.

Brew on

 

[BrunDog]

This site gives some practical methods for manually tuning a PID based on the natural resonant frequency of the system.

Brew on

Just note these directions are somewhat “backwards” when they state “decrease proportional band” - this actually means increase the proportional coefficient. Odd nomenclature!

 

[lschiavo]

There are some magic numbers. I can't find them. I set I and d to zero. P goes from zero to 10000. Where do I start? I need manual tuning a pid for dummies. This sucks.

 

[BrunDog]

I would set P around 20 as a start.

 

[augiedoggy]

I would try it with the cast iron removed if you havent already... for the same reason electric ranges are often frowned upon by chefs because the coil holds too much residual heat for delicate temp control and timing. (according to a cooking special I heard on the radio a few months back.)

Bayou classic kettles are not really made of 304 stainless and in this case when it comes to induction its a plus because they work fine with induction without any additional plates. If you are using the kettles with the triclad bottoms you may not be able to control this unfortunately... The cheaper non triclad ones will actually work better for temp control.

 

[augiedoggy]

I'm having a hard time imagining how a couple pounds of cast iron could cause "crazy overshoot" when heating 35 pounds of oil ... Exactly how far is it overshooting?

If the PID is working (and tuned) properly, it should supply full and constant power until it gets to within 95% of the target, and then start duty-cycling lower and lower until it eases into the target temp, with minimal overshoot (1%)

I would expect some hysteresis from the mass of the cast iron, but not huge.

see above^ When I was listening to a special on news radio about chefs using various types of stove tops they stated that just the coil from the (Non inductive) electric stoves hold so much residual heat a chef has to remove the pot/pan from the coil itself when turning it off or the residual heat will mess up the desired amount of heating.. this is a big factor when making things like glazes or marange (I'm sure I am spelling that wrong)
the extra material being heated is also the reason boil overs are harder to control with a triclad bottomed kettle since it holds the residual heat too much and the liquid continues to boil for a bit even after the heat source is extinguished.

 

[fragged]

From a guy who has tried to cheat 1000s of projects, I'd suggest cutting the losses early and scrapping the hot plates personally. Or get comfortable with the long heat up times. Once you're dialed to as efficient with time as possible, you're still waiting on the latent heat in the cast iron and triclad to finish the last few degrees to your target.

Why not just use an element in the kettle as we do for beer? A heat stick/hot rod should do pretty well here I'd think. I'd lean toward the hot rod personally as most heat sticks I've seen are made with pvc, which I'd guess wouldn't appreciate the frequent exposure to the lye. The hot rod is stainless and silicone from what I recall, should be perfect.

Disclaimer: I've made all of two batches of soap in my life, so there maybe a nuance or two I'm missing