Hail The Mighty Thumper

[haggy]

I am a drinker of whiskey and gin, but too old to be a distiller. However, I have developed something that may be of interest and help to some forum members; an Excel computer model of the pot/thumper operation.

I am a retired Chemical Engineer whose work for 40 years in the industry involved the design, startup and optimization of various commercial operations. A major part of this work was to develop a computer simulation model of the chemical process and use it to guide the work.

These last 10 Covid months of I have been going back and redoing some of this work, and then in July, I stumbled on batch distillation and became interested in ethyl alcohol (moonshine) distilling calculations. I first modeled the pot still operation and got good agreement with the calculations of this forum and with Hugelwilli of hobbybrennan.ch.

Then I became hooked on the thumper operation and for the last two months have been trying to develop a believable Excel computer model for it. I joined this forum in September and have read many, many posts and found about 20 posts with some good data on pot/thumper runs to verify the model. Key contributors of good run data were Kareltje, Usge, Beerswimmer, Corene1, Midnight Brewer, Truckinbutch and Moonshiner67.

I will discuss the model details and run data examined in future posts for those interested. It is based on basic chemical engineering theory for mass transfer, heat transfer and VLE. It predicts the pot and thumper gallons output, instantaneous %ABV and cumulative %ABV, pot and thumper liquid contents,volume and temperature, pot KW and more during the course of the run. Input data is only starting pot and thumper gallons and % ABV, and the run length (time).

The data and model covers this range:
Pot: 4 to 13 gallons and 0% to 40% ABV
Thumper: 0.5 to 12 gallons and 0% to 40% ABV

Yes, the pot can be used as a steam stripper. Need some run data on this though. Many combinations of the above have worked. No prediction of the cuts is made, but from the current %ABV made one could foresee when the tails might start. The model agreed with %ABV data +/- 2% of the value and agreed with gallons out within 5% (ie, 3 gal +/- 0.15gal) on average.

The model can be used to plan runs, estimate gallons out and %ABV out vs run time, see if thumper pukes, see the effects of changes in starting gallons and %ABV, see what KW is required, etc.

I will provide nice graphs of the results in the next post. It is neat to see how fast or slow the %ABV decreases during the run depending on the starting conditions. Only the pot, thumper and worm/condenser are included, no columns.

Of course, experience is the best teacher and most of you distillers have that and use it. But there are many novices whom this model might help, and the experienced users can test new ideas.

So, I hope to hear from the forum if there is interest in pursing this model forward. More run data is very helpful, please supply starting conditions as above and some data on gal and %ABV condensed along the way. Other information on pot and thumper volume (gal) at end is good also.

The model will be updated/improved. People need to test it and approve it, etc. Ultimately, if all goes well we could include it in the Calculations section of the forum.

For now, I would be glad to run and test the model for anyone that requests it and get results back to them.

Thanks,
haggy

I better give some results.
Input data: Pot 6 gal at 11%ABV mash Thumper 3gal at 11%ABV mash
See the figures. Do not have labels on the graphs - See the top heading and figure out the symbols. Hope the graphs show.

[Probus]

Good on you Haggy. I'm interested in your excel calcs for a thumper. I have a similar background with some experience in process calculations using excel. May I suggest you rethink your decision not to still. I started this year, age 77. Thanks to the great information from the good people on this site, I am enjoying making and drinking G&Ts.

[haggy]

Thanks for the interest. I have three strikes, 82 yrs old, bad legs, and a conservative wife.
So I will keep on thumping on the computer.

[Twisted Brick]

Haggy,

How generous of you to share your knowledge to create such a valuable tool for us!

I am closing in on the completion date of a two-keg thumper rig and would be keenly interested in the modeling you can provide. As I get experience with it, I will be more critical of the data I collect (volumes, temps, ABV) towards this end.

[The Baker]

Thanks for the interest. I have three strikes, 82 yrs old, bad legs, and a conservative wife.
So I will keep on thumping on the computer.

My legs are okay and I am only seventy-nine....

I may not well understand your work but it is really good to have scientific people
study distillation.
Thank you, for advancing our hobby.

Geoff

[Kareltje]

I am certainly interested in your model.
I made and developed one myself, but based only on the vaporizing and condensing of the mixture.
Maybe we can exchange models?

Oh, the key is the formula from https://faculty.kfupm.edu.sa/CHE/binous ... harthi.pdf

[haggy]

I was hoping you would reply. Thanks.
Yes, we can compare models. I will write up some of the concepts in my model today so you can get an idea of what my model does.

I have found some of your run data and my results are in line with them.

You did a Pot 6gal Thumper 4gal both at ABV12% run all the way down to 4% and I come fairly close to those results.

You did a Pot 9.5gal Thumper 5gal run all the way down to 5% and I come close to those results. I used Pot 8% ABV
and 7% ABV Thumper. Not sure that is exact ABV you ran with.

I saw your model graphs for a 5-5 gal both at ABV 12%. My model comes pretty close to those curves also. I get a little higher starting Thumper vapor ABV but we both end up with 1.6 gal distillate at 35% end point. I have a boost in the thumper vapor ABV at the start probably due to some reflux with colder exit pipes. I saw this in many other run posts.

So our models are not far off. Let's keep in touch.

Thanks again

[haggy]

Kareltje,

You posted your model result on July 2017 in the "What does a Thumper Actually Do" Forum. Maybe you have changed it since then but this is something we can compare results on.

Starting conditions were Pot: 5 ltr at 12% ABV Thumper : 5 ltr at 12% ABV

I ran my model on those same conditions ( sorry, I used gallons instead of liters). I tried to make my time steps so that 6 minutes equal one of your time steps. There is a 30 minute thumper heatup time.

Attached are my results. You can compare yours with them. I get a higher %ABV vapor out early and it stays higher for a while. Also the thumper fills up some with mostly water and the alcohol in the liquid is almost gone at the 120 minute time. Vapor rates from the pot and thumper are in close agreement.

[Kareltje]

I made another run with more or less the same conditions.
Did not compare our results, for that is difficult switching from screen to screen, but tomorrow I can print your graphs and compare them with mine.

The left one is about volumes, the right one about %ABV.
Glossary:
ketel = boiler
bijvat = thumper
product = product
ketel uit = leaving boiler
ketel rest = staying in boiler
Destillaat = product (cumulative)

The head is a kind of switchboard, where several variables can be chosen.

[LWTCS]

What is thus far the take away from this data?

[haggy]

LWTCS,

Thanks for inquiring. A fair question.

Sometimes, it is good to know what might be expected before making a run. A model calculation could be a way to do that.

Of course, many distillers have experience and know what to expect. But if you want to try something different, this could help lead the way. The model can be used to plan runs, to decide what feed volume and %ABV will give close to what is desired.

Novice distillers might find this very useful if they are not sure what will happen and when.

An estimate of the expected gallons and %ABV made vs run time is useful. It could help you watch for when the tails start, if you know the %ABV of the tails.

You can try different thumper charges to see if the thumper is likely to puke. You can see the effects of changes in starting gallons and %ABV. The model could tell you how long the initial higher ABV will last before it starts dropping rapidly. After a run, you could check out if things went as expected.

There are probably more uses, we will keep looking.

Knowledge is power, and a good model calculation can give you that power. We hope to develop a good, reliable model.

[Tummydoc]

Haggy, probably over my head on this, but i think your input variables could be changed. Volumes and ABV are critical input variables, but your other input is run time. Most of us don't decide run time and use that to determine required energy input. Much more commonly we set the power as a constant, and the run time is a result of power input. I think if this were an app to estimate time and product output (both volume and ABV), you'd want to input the power.

[haggy]

OK, thanks a lot, a good suggestion, I will work on this.

The model does calculate the kW needed to achieve the output rate calculated. So kW are in the model but not initially specified. In my November 23 post, see the third graph from the top, it gives the pot kW, pot volume(gal) and pot vapor (gal) during the run. You can see the time on the bottom goes on for over 300 minutes. I did specify a run time to get to 10% of the charge %ABV for that run.

The power starts off at about 1.8 kW ( high to achieve a shorter time to heatup the thumper ) , then is turned down to about 0.8 kW and each incremental time is set to give a constant kW power setting.

I have seen in a post ( maybe Pinto ) that about 25-30 ml/min product rate out the condenser is a good run target. I do have in the model now an option to use a calculated run time to give this desired rate. That is an option in the model if you specify zero run time.

So I think the required power to set is already there, it is just backed out from the other calculations. A run time to give 25-30 ml/min can be set if the run time input is zero. I could make as input data a desired output rate, that would then trigger a different run time and required power setting for that rate.

Let me caution now that I have not seen run data on power settings used. My kW calculations need to be verified with some data. Maybe you can get back to me with your run experiences.

I will look to see how power can be set as an input data.
Thanks again for the interest. I will continue to improve the model and I hope the above method can work for you.

[Kareltje]

I made the model because I could not see what happened inside and what went on between boiler and thumper.
Now, like haggy said, I use the model to warn against the emptying of the boiler or the puking of the thumper.

@haggy: Ah, now I understand why you had the wattage as a result of your calculations. I set the wattage as a parameter.
So your model on some parts at least works reverse compared with mine. Interesting!

[haggy]

Yea, I set a decrease in the liquid % ABV (in moles) in the pot still for each increment and set the time for each increment. Then use the Rayleigh equation to calculate the decrease in volume and then amount of vapor made from the pot. Input to that equation also is the relative volatility and the non-ideal activity coefficient which is also used to get the %ABV of the vapor. The kW for the pot then results from the latent heat needed to vaporize the vapor made for that time increment.

Maybe I could turn this around, but we will see.

Kareltje will understand this. Don't worry.

[Kareltje]

I understand half of it, to be honest. The Rayleigh equation is quite easy, but the relative volatility and the activity coefficient did not really work for me. So when I found equation 9 in the article of Binous & Al Harthi I jumped to that, as it has a sufficient fit with the real graph.
I did not want to build a automated still, so a rather crude fit would do.

I am trained as an agricultural engineer and our equations did not have to be 100 % perfect. We build a formula, say for distance between drainage pipes and then we use variable constants to make the results fit with reality.

I started this model with defining the amount and strength in both boiler and thumper.
The strength of the vapour coming from the fluid in the boiler can be found via equation 9 and from the heats of evaporation we can then find how much of this vapour is formed by the energy in one step. This vapour is substracted from the contents of the boiler (water and ethanol) so we can start line 2.
It also is added, ethanol, water and heat, to the contents of the thumper. When these contents are heated sufficiently (I took a fixed setpoint of 80 dgr C for convenience) the strength of the vapour can be found by equation 9. The amount of the produced vapour can be calculated from the available heat minus the losses.

To this basic model I added several tuningpoints. Like the change in losses from the thumper and the possibility to calculate with additional trays. When a thumper is almost empty, there might be considerable passive reflux and one can also add a column after the thumper. The starting temperature of the thumper is a variable too. Later I made it possible to stop the boiler halfway and go on firing the thumper.

Of course this is a rough model. One of the premisses is, that all vapour coming from the boiler is first condensing in the thumper, rather than bubbling through the fluid and disappearing to the condensor. And a lot of the assumed constants are not constant, of course.
But to me it works. It shows me the two greatest dangers: boiling dry of the boiler and overflowing of the thumper.

[Kareltje]

I took some data from a run in april 2019.
8 liter of 9,95 % in the boiler and 2 1/4 liter of the same in the thumper. Ran with 607.6 Watt on my stove. Boiler reached 90 dgr C at 14:30, first drops fell at 15:00, I stopped running at 19:30. Result was 2.186 liter of 45.5 %ABV, rest in the boiler was 4 liter, rest in the thumper was 3 3/8 liter.
This is the graph of the results:

I detemined the strength of the transferred vapour based on the temperature of the bridge between boiler and thumper.

Yesterday I tried to make the model fit as good as possible. Could not exactly make it right, but this was satisfying enough:

[haggy]

Thanks for some more data. I will look at it tomorrow.

I certainly understand where you are coming from and what your needs are. I am impressed with your work and commend you for it.
My objective is to produce a reliable pot/thumper computer model that other people could use to help guide their work. I need to keep active during this shut-in period and doing this helps.

Here is some good news. Our different calculations of the pot/kettle are in very good agreement. You use equation 9 and I use a chemical engineering approach and we get the same results. I arranged my model to give a constant 1800 watts power for the first 5-5 case and the vapor produced agreed with your calcs for the volume and %ABV produced. Also the pot volume remains agreed.

The thumper calcs are a little different for the 5-5 charge case. Maybe they will be closer for the November 26 case of 8-2.25 charge. I get the vapor ABV higher initially and lower toward the end of the run and the ABV runs out sooner than your calcs. We got similar vapor out volume so that is good. My model was fit to 15 or so runs from different posts and may reflex some refluxing initially due to cold piping.

I see the need to set the power as input data. I had to set my increments in %ABV to get a constant 1800 kW power. This is only for the pot/kettle. Since our models agree there, it may be good for me to use a hybrid model; your pot calcs where power is an input and my thumper model. Or I may find a way to change my pot model for a power input. We will see.

So, keep on with providing some data and I hope to get more run data from others. Your five/eight liters charge is smaller than most but seems to be ok. I will check out the data of Nov 26 and get back to you. Got to go for some turkey now.

[Kareltje]

I should correct my last graphs.
In the graph of the real run the curve of the strength of the transfer should start a bit earlier and a bit higher. Forgot to correct the range of data.

And when I set the number of trays for the thumper to 1, the red curve of product is more like the real curve, it only starts about 10 %ABV lower.

[haggy]

Your model serves you well and works for you. I have been helped by your work and your data. I am trying to build a universal model for others so both models are OK.

Good news
I can now input the kW power to my model and it will predict the run time along the way as the %ABV of the product decreases. An end time for a run can be determined. This was a good suggestion by Tummydoc and is better than my other way of backing out the kW power.

More good news
I ran your 8 liter -2.25 liter data on my model with kettle/pot power as input data. I scaleup to 8 gal - 2.25 gal charges so I get output in gal comparable to your output in liters. The 600 kW power input scales up to 1.66 kW when 20-25% radiation etc losses are accounted for. My %ABV results were very close to your data. See the attached curves. The x on the curves is your % ABV data.

The run volumes and times were also very close to the data from the first data curves. So I am well on the way to a good universal model thanks to your help. I will certainly want more data to check things out and finalize this.

Maybe we are done for a while. Send more run data when you can. We will keep in touch. I hope others can also supply some run data.

[Kareltje]

I can try to dig up some run data. But the point is: I have several possible constellations for boiler and thumper.
3 10-L kettles, of which 2 can be used as boiler and 3 as thumper. And this year I got a 15-L kettle, that can be used either as a boiler or as a thumper. Each combination has its own variables, I think.

[Kareltje]

Found some runs with thumper and different loads.
Can I upload a excel-file here? Let us try.

Model bijvat-vergelijk.xls

(330 KiB) Downloaded 145 times

It seems to have worked.

Note: we use a comma as separator between units and tenths, I do not know if that is automatically corrected.

[haggy]

Yes, it worked. Thanks a lot.
I will look it over, run my stuff, and get back to you.
Some runs with thumper (dreunvat) and some with pot still (ketel) only.
Nice to have the kW and time data.
What is potje?
Is the % transfer ABV based on a temperature calculation? This may not be accurate at less than 30% ABV?

[Kareltje]

Yes, it worked. Thanks a lot.
I will look it over, run my stuff, and get back to you.
Some runs with thumper (dreunvat) and some with pot still (ketel) only.
Nice to have the kW and time data.
What is potje?
Is the % transfer ABV based on a temperature calculation? This may not be accurate at less than 30% ABV?

These three runs are with both boiler (ketel) and thumper (bijvat). (Like bijkeuken (second kitchen), bijvrouw (second wife) etc. )
Potje is little jar. I use smaller ones (1, 2, 3...) of 100 to 200 ml and larger ones (I, II, III, IV ....) up to 500 ml.
The Greek theta is for thermometer and these are a bit off: some show 2 dgr C too low, some 2 dgr C too high. Just for getting an idea of how far the process has gone. I measure the %ABV in the pot with an alcoholmeter, but not always at the right temperature! And it is of course the percentage of the total amount collected in that jar. The %ABV of the transfer from boiler to thumper I can not measure, alas. I tried to calculate this from the temperature, but then I first have to correct the reading and then calculate the %. Two very uncertain steps.
Oh, I named my kettles, so I know which ones I used. Kees, Koop and Kock are 10 L, Kolos is 15 L.
I start the run full throttle, that is about 2,3 kW. At the mentioned time I switch to lower wattage.

When you can use these data, I can send some more.

[haggy]

I have the first four cases done. Took a while to figure out your spreadsheet and transfer the data to mine. For each case, I had to find the power (watts) setting to give a run time where about 5% was left in the product distillate. Power is input data.

First Run 5/14/2017
This run had 4 liters of 12% ABV in the pot but an empty thumper to start, something different. Of course there will be bypassing (blow by) of the pot vapor through the thumper, but how much. I spent a while adding bypassing and an empty thumper to the model. But then saw that the final distillate was about the same amount and ABV of the vapor from the pot. So, maybe100% bypass, not enough condensing on the walls to get full vapor/liquid contact. The first 76% ABV result may be a hot sample error or may be some contact on the walls. Power setting was 620 watts after an initial 2 kW for heatup. At 42 minutes, the pot produced 1 liter vapor, just above the amount collected. Here are results: Pot conditions vs product data. Good agreement, confirms we have a good model for the pot.

Second run. ( same day)
This run had 4 liters in the pot and 4 liters in the thumper, both with 11.2% ABV. Run took about 125 minutes and the power to give this was 620 watts, just like above first run. Model results are in very good agreement with the data. Next are the thumper results for this run. The amount of distillate made is also in good agreement.

Third run. 7/30/2017
This run had 4 liters in the pot and 3.125 liter in the thumper, both with 8.9% ABV charge. Run time was about 95 minutes and the power set to the model for this run time was 750 watts. It would be best to know the actual power used - but this was not available. Model results agreed with the data.

Fourth run ( same day )
This run starting conditions were very similar to the third run. This time 800 watts were set. Similar results.

I will check the other runs next. Tell me if I have misinterpreted the run data . Next run has 6 liter water in the thumper. Wow.

[Kareltje]

Oh shit: I wanted to upload another sheet! I intended to sent a simpler file, without all the calculations and just the data.
I am sorry!

But this has more different runs, so no harm done, I hope.
But I can imagine it took you a lot of time to grasp the sheet I sent.
I have to study your results.

[haggy]

I figured the sheet out and was glad to see all the data and calculations. I used the Google language translator.
Only problem was I had to move all the graphs. It is great data with a variety of runs conditions.
The next set of runs has the pot remains data, I can use that to help get the kW input for the run.
The % transfer ABV data ( from the temperature ) seems to be too high at less than 35% ABV. I will not include those points.

[haggy]

The rest of the runs in Kareltje's data set includes data on the remaining Pot volume at the end time of the run. This data and the time to get to 5% ABV of the distillate will be used to determine the kW for each run. The kW is input data.

Runs 5 and 6 8/25/2017
These runs had similar initial conditions, pot at 8l liter and 21.5% ABV and the thumper with 6 liters of water, 0% ABV. Much different than the other runs. Run 6 had a higher initial burn rate it seems since the thumper start time was lower. The power input that fit the runs was 710 watts and 680 watts.

The initial thumper distillate ABV of the data was 67% and 55%, I suspect the 55% of Run 6 may be too low. The model had 67.5% initially for a while but tracked the ABV down very well below 60%. Also, the model closely predicted the distillate rate and thumper ending volume and pot ending volume. Cumulative ABV was higher in the model. But in Run 6, the low start ABV data could be wrong giving low cumulative.

So the model can work ok starting with only water in the thumper.

[haggy]

I continue to post the results of comparing my pot/thumper model with the good run data from Kareltje.

The next two runs are in this post. Run 7 was conventional with alcohol in both the pot and thumper to start. Run 8 was a good test of the model, it was a steam stripping run with only water in the pot creating steam to heatup and boil over the thumper contents. This is useful if solids are in the mash. The thumper is not externally heated and should not scorch the solids.

Run 7 9/29/2017
This run had the pot start at 5 liters and 5.6 %ABV with 740 watt power input. The thumper started at 4 liters and 10.7% ABV. About 2.1 liters of product were made and the thumper ended up with 4.75 liters. The model predicted these values closely. The %ABV made along the run and cumulative %ABV tracked the data.

Run 8 9/27/2017
This was a steam stripping run. The pot started with 5 liters of WATER, the thumper had 3.25 liter of 10.7% ABV. The run only went for 86 minutes and 740 watts were estimated to fire the pot. The model had comparable results with the data so it can be used for this type of application.

CAUTION: In all these runs, the estimated power (watts ) may be different than the actual watts used.

If the actual power is different, the %ABV and the volume of product from the model would be somewhat greater than or less than the data. The end time of the run was used to find the input power. That also was the time where the model gave a low (5%) ABV of the product. Runs with actual power data need to be examined, but we are close with this good data.

[haggy]

I have examined all 12 of Kareltje runs and will report the next four run results soon. Some interesting options were run in the last four cases. My pot/thumper model did well.

Some observations I see from running the model using the Kareltje data are reported here.

A steady pot power applied over the time of a run should equal the amount of pot material boiled off times the average latent heat of vaporization (it increases during a run since water latent heat is much greater than alcohol latent heat) of the alcohol/water mixture. An equation that says this is:

Watts * Time (min) = Pot gal boiled off * Density lb/gal * BTU/lb * 0.293 watts/(btu/hr) * 60 min/hr

So, if we know the starting pot gal, %ABV and if steady power is applied, we can specify a % of the pot boiled off and find the run time. Likewise knowing run time, we can estimate the power used.

The pot/thumper model was used to find the average latent heat for several runs and then that was correlated to the starting ABV. It is to be used for about 5% to 30% alcohol in the pot. It is not for water only, use 945 Btu/lb there. So to find the steady power in watts from the pot run time data, divide the above equation by the run time:

Watts = Pot gal * % Boiloff / Run time min * 8.1 lb/gal * ( 860-680 * %ABV ) * 0.293 *60 min/hr

For example, see run 7 form above posts. Pot 1.32 gal at 5.6% ABV and 60% of pot was boiled over. Run time was 125 minutes.

Watts = 1.32 * 0.6 / 125 * 8.1 * ( 860-680 * .056) * .293 *60 = 742 watts

So 740 watts was input data to the model for run 7. The accuracy here is about +-7%, depends on the accuracy of the average latent heat estimate.



However,
for most home distiller runs, you start with setting a power input, so run time can be estimated from:

Run time (min) = Pot gal * % Boiloff / watts * 8.1 lb/gal * ( 860-680 * %ABV ) * 0.293 *60 min/hr

A typical run for a home distiller might be this: Pot 6 gal,12% ABV , Thumper 4 gal at 12% ABV, and 65% pot gal boiled over with 2000 watts power input. The end run time estimate to get to <15% ABV final distillate is:

Run time (min) = 6 gal * .65 / 2000 watts * 8.1 * (860-680* 0.12 ) *.293 * 60 = 216 min

The pot boiloff amount can range from 50% to 80% depending on the pot and thumper charges. The 65% figure is an estimate for runs with pot charge less than 10 gal, with thumper charge less than the pot, and both have ABV charge less than 13% . The final distillate at 216 minutes run time should be less than 15% ABV.

If more alcohol (gal and %ABV) is charged to the pot or thumper, the % pot boiloff and run time will need to be longer to get to <15% final ABV. Putting 80% boiloff into the equation should give you a safe maximum run time provided the watts used are constant and correct.

The pot/thumper model has pot power as input data and we see the expected distillate results at all run times. The accuracy in run time and %ABV results should be good. The run time to get to 40% ABV distillate, where the tails might start, can be determined from the model.

Of course; smelling, tasting and measuring during a run is always done and is best. The concepts above and the pot/thumper model can be used before a run is made to know what you might expect and when.

Feedback and data on these observations are welcome.