Dephlegmator Control via Vapor Temperature?

[Johnnywhiskey]

So I have had a chance to experiment with the PID controller on two runs--one whiskey and one rum distillation. For compressing the heads and taking off most of the hearts it kept the temperature between the set point and two degrees F below the set point, which seems pretty good. On the rum run I set the hearts to come off at 190F (or just below 140p) and most of the hearts came off at 138p (measured by hydrometer and temperature corrected). So that worked pretty good.

The controller did not do a good job of controlling the latter part of the hearts and the tails. There were some fairly substantial swings in temperature above and below the set point. It seems like once the temperature got more than a few degrees over the set point the controller would try to overcorrect and the temperature would too far below the set point. Maybe I can tune that out by adjusting the controller parameters. I tried to use the autotune feature on the controller with little success. I guess you cannot expect too much from a $20 PID controller from Amazon. But I think it is worth trying to work out the kinks. The temperature was rock steady through most of the hearts (along with the ABV), as the temp got close to the set point the water solenoid would momentarily open and then the temp would drop down a degree or two.

For me, building and testing the controller has also underscored the usefulness of monitoring the vapor temperature above the depheg. In my experience, reading the vapor temperature is as reliable, if not better than taking a hydrometer reading. Looking through the data on each run, the abv based on vapor temperature was always within a few % of the actual abv I measured. And any difference may be due to the vapor temperature being an instantaneous measurement, while the hydrometer is measuring the whole parrot of distillate. Previously I had a dial temperature gauge that I had to stand up (god forbid) and look around the side of the still to read. From now on I'll use the PID controller to take temperature reading even if I'm not using it to run the coolant.

[Johnnywhiskey]

I came across this on the Stilldragon forum--a PID build for a Dephleg. I think they are using coolant temperature instead of vapor temperature, but similar in principle. It looks like the build also uses a variable flow valve controlled by the voltage output on the PID.

https://www.stilldragon.org/discussion/ ... water-flow

I have to admit building this controller has got be thinking about going all-electric with my distilling. There are so many interesting ways to setup heating and cooling with the controllers and parts available. I just need to get a 220V outlet to handle a decent heating element.

[discoreaper]

I've been working on vapour temp control on an LM. Using the temp to dictate how much the needle valve opens. I have all the code written, but still working on a way to connect the stem of the valve to the stepper motor. Hasn't said much about it on here as there is quite a negative view of automation. But I was planing on doing a full wrote up once I had finished.

I've been prototyping a few ideas for the build, and am using an ABB Comander 350 to control a fan coil sonenoid, in closed control loop, the temp of the vapours being monitored by a pt100 in a 1/2" gland pack. I'm using the ABB350 for the momnet, but intend on doing the final set-up with a Siemens S7-1200, K600 HMI, thus giving me total control, I know there not much support for automation on the forum, but its what I do and know the commercial distilleries use this to achieve consistency

High quality is still achieveable using automation, plus .....it can be switched to manual!

[discoreaper]

It seems like once the temperature got more than a few degrees over the set point the controller would try to overcorrect and the temperature would too far below the set point. Maybe I can tune that out by adjusting the controller parameters. I tried to use the autotune feature on the controller with little success. I guess you cannot expect too much from a $20 PID controller from Amazon. But I think it is worth trying to work out the kinks. The temperature was rock steady through most of the hearts (along with the ABV), as the temp got close to the set point the water solenoid would momentarily open and then the temp would drop down a degree or two.

.

PID tuning & setup is a dark art in itself, dont be discouraged by the swings and variations, I do this for a living and its taken me weeks sometime to setup commercial processes, often resorting to trial n error which is seriously frowned upon in a validation scenario, in this business we're supposed to be able to document and make everything repeatable ....everytime

[pocomabuono]

This is the vapor temperature controlled by the dephlegmator in my CM still.
Water to the deflator comes through a solenoid valve followed by a needle valve.
The needle valve must be adjusted at the beginning of the distillation according to the water temperature.
[utube][/utube]

[pocomabuono]

Using the dephlegmator control i noticed too that, when the temperature difference of the vapors before and after the dephlegmator becomes 4/5 Celsius degrees, the regulation is no longer so stable.
In fact, the temperature range around the set point fluctuates with an excursion of one degree.
The problem is due to the delay between the solenoid valve control and the temperature variation read by the thermometer.
It often happens, in feedback systems where the corrected signal is delayed, that the system may be brought to a certain instability
I'm now looking for a way to improve dephlegmator control.

[pocomabuono]

I am increasingly convinced that this type of control cannot work fine when the temperature difference before and after the dephlegmator increases considerably.
Simply because the steam that passes after the dephlegmator is reduced too much.
How can I measure the temperature of the steam if it is not in sufficient quantity to be measured?
When the dephlegmation is pushed to the limit, another strategy must be found for its control ...
But I want to try again, placing the thermometer just before the condenser, where the section of the column is reduced from 51 mm to 10 mm and the stream is 25 times faster.
Even with very little steam you could have a flow with a speed high enough to be measured without excessive delays.
It helps to have a thermometric probe that responds quickly.
It also helps increase boiler heating to increase steam.

[Johnnywhiskey]

After messing around with the PID on the defleg, I found that a simple on/off valve is just too touchy. Even with a needle valve upstream from the on/off PID controlled valve, too much cooling water would come through on each on/off cycle. I'm not sure if its a problem with steam flow rate, but there is a significant delay between adding cooling water to the defleg, and a reduction in head temperature--much more than the PID can easily tune for. As I mentioned above, I've seen commercial distillers using a PID on the defleg, but that was on a much larger scale, even at a craft distillery.

Luckily the PID controller is super flexible, and I use it fermentation control now.

I'm thinking a better method for maintaining a consistent head temperature would by using PID on the head temperature, and let the PID set the wattage to maintain that temp. This would obviously be on a reflux column. If the defleg cooling water flow rate maintained constant, it should be able to maintain a constant head temperature. I haven't tried it yet, but should work--not sure how much tuning the PID would require.

Stay Thirsty, JW

[cayars]

Johnnywhiskey, did you still have the needle valve inline?

[pocomabuono]

I'm trying to make a new head, with a coil dephlegmator with a few turns.
Immediately above the coil, the section is reduced from 52mm to 10mm and the thermometric probe should also feel a weak breath of steam.
Within a month I will do another distillation and see how it will work.
Of course, a linear control is definitely better ...
However, when the temperature gap is not high and the vapor that passes the dephlegmator is enough, the control works very well.
Heating control also seems to me an excellent strategy, but in any case I must always be able to measure the temperature without excessive delay.
And finally, when the amount of steam drops to almost zero, the problem remains

[pocomabuono]

But help me to understand.
To maintain, after the dephlegmator, a vapor temperature lower than that of the rising vapors, the mash should be heated below the boiling temperature.
So the vapors would not rise ...

[Johnnywhiskey]

The boiler is at whatever temperature the combination of water and etOH boils at. Can't do much to control that. But you can control how much vapor is being produced.

The dephleg (and the refux taking place in the packing below) cools the vaper to something lower than the temperature of the boiler. If you keep the water flow constant and limited to the dephleg, it can only condense so much of the vaper making it up the column. Increasing or decreasing the wattage into the boiler will increase or decrease the volume of vapor making it to the dephleg. More vapor but the same amount of cooling will equal a higher head temperature--more "watts" of vaper will make it past the dephleg and packing.

What I like about controlling the head temperature is that there is a direct relationship (at a given atmospheric pressure) between vaper temperature and ABV. It's a lot more responsive and accurate during the run than a hydrometer, and no need to adjust for temperature and fluctuations in the temperature of the liquid product coming off. There are a lot of ways to run a still--but this one works best for me.

Stay Thirsty, JW

[LWTCS]

Curious if anyone has seen or used a still that controls the coolant this way. Any downside to running a still this way? I did a few searches here and did not come up with anything.

Best, JW

Sure. This is the same concept that we discussed on Yummy's continuous thread where by low wines vapor temp will govern the beer pump speed. Therefore the beer injection temp ends up being a byproduct of the low wines vapor temp.

My analogy for this is that it is easier to pull a rope than push a rope.
Adjustments are predicated upon the desired end result so to speak. So adjustments are made working backwards, or right to left rather than left to right if that makes sense?

[pocomabuono]

Clear
Even in my case, having a CM without packing, it could work too.
But I fear that my resistance from 1800 is insufficient, considering that at least 600/700 W are needed to keep the boil.
The adjustment range may be poor ...

[cayars]

The boiler is at whatever temperature the combination of water and etOH boils at. Can't do much to control that. But you can control how much vapor is being produced.

You can't change the boil point but you can control the temperature (ie hold it) and vapor rate based on monitoring the temp. Keep in mind the boiler still works the same as when ran in pot still mode. Lower heat and less smearing, more heat than needed to boil and you get faster vapor rates and more smearing. The bulk of heads come from a particular boil range, so do hearts and tails when not being smeared.

Just like on a pot still, I try to avoid overly smearing the vapor even though I can "fix" it with reflux. What's my point?

Personally, I think it's a balancing act of running the proper heat/vapor flow for that part of the run and adjusting the cooling (reflux) to match the vapor temperature range. So by controlling the temperature (like running a pot) I'll allow it to reach at any given point I can have my vapor stop producing toward the end of heads or hearts. This will give you really clean separations but will take a bit longer to run.

That's an oversimplification, but my point is that together (reflux control and boiler/vapor control) is better than locking one in and only driving the still using the other control. Personal opinion of course.

[bluefish_dist]

Interesting thread. Using the pid makes the cm run how I run my vm by hand. Pick a temp and hold it by adjusting reflux. What I find is temperature varies by 1-3 deg even in steady state. Say I am aiming for 185 deg, I just want to hold 180-185.

To program that for a cm I would think you would want a base rate of valve pulses, say 1 second on, 1 second off, then when the temp increases above the set point increase the rate of pulses, if below the dead band then reduce the rate of pulses. You would also need to limit how often the rate can be adjusted as it takes time for the column to react. Ie make a change then ignore temperature for 1-2 min. I often thought of automating my still bu adding a stepper motor to my valve (vm) and allowing a pid to slowly close it as temperature increased.

[cayars]

I often thought of automating my still bu adding a stepper motor to my valve (vm) and allowing a pid to slowly close it as temperature increased.

Funny, I was just looking at a servo from a remote control heli and was thinking, dang I bet I could use this to control a needle valve very accurately.
Then of course it could be used to accurately control a rheostat type control for heat as well. Bypass the simple PID type controllers and move the functionality over to Pi, tablet or computer. Could gain logging and full control pretty easily of heat and cooling.

[Johnnywhiskey]

The boiler is at whatever temperature the combination of water and etOH boils at. Can't do much to control that. But you can control how much vapor is being produced.

You can't change the boil point but you can control the temperature (ie hold it) and vapor rate based on monitoring the temp. Keep in mind the boiler still works the same as when ran in pot still mode. Lower heat and less smearing, more heat than needed to boil and you get faster vapor rates and more smearing. The bulk of heads come from a particular boil range, so do hearts and tails when not being smeared.

Not sure I follow you on this. Are you talking about the vapor temperature or boiler temperature?

[pocomabuono]

To program that for a cm I would think you would want a base rate of valve pulses, say 1 second on, 1 second off, then ...

PWM regulation is very effective because allows linear regulation even with ON / OFF control (if the frequency is high enough).

make a change then ignore temperature for 1-2 min.

Mhhh ... too much.
In my pot still with 20 cm double coil dephlegmator (therefore with high thermal inertia) when the solenoid valve closes the temperature rises by 1 degree celsius in 10 seconds, even with a heating of only 700/800 W.

However, whatever type of regulation is chosen, the control takes place with the temperature of the steam which must be read accurately and fastly.
What I noticed (ie with dephlegmator set at 92 degrees) is that the valve closes, the temperature of the vapors rises a lot (ie 93 degrees) but the thermometer does not notice it immediately (and still reads 92.2 degrees).
When the valve opens the temperature drops (ie 91 degrees) but the thermometer is still measuring a higher temperature ...
It is the typical problem of systems with delayed feedback which become unstable

[bluefish_dist]

To program that for a cm I would think you would want a base rate of valve pulses, say 1 second on, 1 second off, then ...

PWM regulation is very effective because allows linear regulation even with ON / OFF control (if the frequency is high enough).

make a change then ignore temperature for 1-2 min.

Mhhh ... too much.
In my pot still with 20 cm double coil dephlegmator (therefore with high thermal inertia) when the solenoid valve closes the temperature rises by 1 degree celsius in 10 seconds, even with a heating of only 700/800 W.

However, whatever type of regulation is chosen, the control takes place with the temperature of the steam which must be read accurately and fastly.
What I noticed (ie with dephlegmator set at 92 degrees) is that the valve closes, the temperature of the vapors rises a lot (ie 93 degrees) but the thermometer does not notice it immediately (and still reads 92.2 degrees).
When the valve opens the temperature drops (ie 91 degrees) but the thermometer is still measuring a higher temperature ...
It is the typical problem of systems with delayed feedback which become unstable

I think you missed my point. Yes, pwm driving the water flow in the deflag, wasn’t sure everyone knew pwm. You can’t just turn on or off the flow by temp, but I think you could adjust the switch rate. Say your base rate is 30%, as the head temp increases, adjust the rate up to say 32%, then wait a min or two for the column to stabilize again. Might even be longer. If you don’t the control will simply keep upping the deflag flow rate until the temp swings very low. Then high as it’s adjusting too fast for the system. You could also add a maximum change per time to help dampen the swings. A human watching can easily do this.

[pocomabuono]

Now I have understood, sorry but I don't know the language so well ...
The strategy seems to be excellent, updating the PWM duty cycle every minute

[charcoal]

I ran my CM still with a PID 3 times for now. I control the heating element and let the vapour after dephlegmator stay at 78.1C.

I have uploaded two videos from different runs.

I got 92 ABV. Will run it with a longer column in few weeks.

[bluefish_dist]

I ran my CM still with a PID 3 times for now. I control the heating element and let the vapour after dephlegmator stay at 78.1C.

I have uploaded two videos from different runs.

I got 92 ABV. Will run it with a longer column in few weeks.

How are you controlling the temp? Pwm, needle valve? How often do you make adjustments?

[charcoal]

I ran my CM still with a PID 3 times for now. I control the heating element and let the vapour after dephlegmator stay at 78.1C.

I have uploaded two videos from different runs.

I got 92 ABV. Will run it with a longer column in few weeks.

How are you controlling the temp? Pwm, needle valve? How often do you make adjustments?

Via PID (internally PWM maybe). I did not need to make adjustments on my three runs.

[charcoal]

This is the video

viewtopic.php?f=15&t=77452

[Pierrot Lunaire 55]

There is another option for controlling defleg water control that is less fidgy than a needle valve. Try a metering pump.

These are designed to deliver flow at a controllable rate and most designs are unaffected by pressure, etc. I just tried one I bought on amazon for about $42 that is a small peristaltic unit (Search on: 1Ml250Ml/Min Durable Peristaltic Pump, Dosing Pump, 50/60Hz Us Plug 110-240V For Biochemical Analysis) with enough range to cover a slow drip to a good strong squirt. It's self priming and can draw water out of a bucket. Turn the knob and adjust its throughput until you have your desired vapor temperature.

Defleg Metering Pump.JPG

Here's the setup. The bucket holds plain water, and is refillable by opening a valve on the cooling water header. The pump pulls it up and sends it to the defleg inlet. There is a needle valve there, but it's left fully open. The defleg outlet sends its water through the hose up and to the right to get it out of the way. There is also an RTD at the outlet.

I haven't checked, but the suspicion is that the pump's motor speed is controlled via PWM, so it should be adaptable to run from an Arduino. That means it's possible to control it via a PID loop based on vapor temperature, measured using the RTD in a thermowell extending through the elbow in the upper left corner of the photo. (The bicycle wheel is not involved in the process.)