using Chat GPT for brewing calcs - for example, Calculating priming sugar addition in keg

[Unicorn_Platypus]

There isn't really a great priming sugar calculator for kegging beer. I fed chat GPT my detailed process so it could calculate the final volumes of CO2.

It took a little bit of tweaking, to get the prompts to work, but I think it finally got it right. This beer is a Belgian IPA, so I'm going for high volumes (to be served on long lines)

It's also been using it to help calculate lactic acid additions in finished beers (where there aren't standard calculator). Pretty wild stuff!

 

[doug293cz]

Seems like a waste of energy intensive computing power, and also a really over-complicated carbonation process. Why do partial forced carbonation if you want to do natural (from fermentation) carbonation?

Brew on

 

[Tony B]

I’ve been using it lately for all kinds of brewing and non brewing info. Pretty freaking cool. Of course I don’t trust any single source 100% and weigh the input of several sources before making any decisions. That said, my experience with chat so far has been enjoyable.

 

[Unicorn_Platypus]

Seems like a waste of energy intensive computing power, and also a really over-complicated carbonation process. Why do partial forced carbonation if you want to do natural (from fermentation) carbonation?

Brew on

The goal of the intermediary steps #1-#5 isn't to carbonate. It's to keep O2 out & uss yeast as an O2 scavenger at those various stages, when dry hopping beers. I am picking up CO2 along the way as a byproduct and was curious to know how much. For my non dry hopped beers these steps are bypassed.

Steps #2-3 I cold crash and pull out the yeast prior to dry hopping. The CO2 in the headspace is used to keep positive pressure when cold crashing.

For step#4 I warm back up after pulling the yeast so I can dry hop on the brightest beer possible. Yeast in suspension is bad for retaining hop oils in finished beer. When yeast flocs it puls the oils with it which I am trying to avoid. At step #4-5 (really a single step to articulate the dry hop stage to chat gpt), my dry hop time is super quick, only 24 hours to reach full extraction from the hops without pickup up unwanted astringency.

The small sugar addition in #5 is used to scrub O2 introduced with dry hops and wake the yeast back up for keg conditioning. My kegs are generally carbonated 3-5 days after adding promises sugar (including passing VDK test for diacetyl. I don't normally VDK test, but I have done so a few times to check the process. Never had diacetyl once)

Steps #6-8 are byproducts of me filling the keg with headspace and shaking the keg to mix the beer. The goal of these steps isn't to achieve carbonation, but carbonation is again a byproduct.

At step #6, I shake the keg so I can take a uniform PH reading of the beer. I often opt to acidify dry hopped beers depending on how much PH spikes up during the dry hopping.

Steps #7-9 are really a single step, for after I inject the priming sugar via my gas post. I broke it down into substeps so ChatGPT could understand it better. You always want headspace pressure in a sealed keg. So that headspace also is not added with the intent of adding carbonation, but is a necessary byproduct to avoid O2 ingress.

I hope I answered your question and please don't be a hater

You won't meet another homebrewer who has better shelf life and less hop burn than I do on hoppy beers


P.S. Doug I love all the math you have done over the years in various threads over the years regarding gas laws and purging kegs. We should revisit some of those calcs with ChatGPT and see how close it gets to your calcs.

 

[Unicorn_Platypus]

I’ve been using it lately for all kinds of brewing and non brewing info. Pretty freaking cool. Of course I don’t trust any single source 100% and weigh the input of several sources before making any decisions. That said, my experience with chat so far has been enjoyable.

Exactly, you can't trust the output it gives you initially completely. It requires additional clarifying queries to refine the responses with better details. That's why I have this particular request broken down into so many step / queries.

 

[dmtaylor]

ChatGPT is really really bad at mathematics. Experts on AI will tell you the same thing. It should only be used for summaries of language, not numbers. I wouldn’t trust ChatGPT with any calculations.

 

[Unicorn_Platypus]

ChatGPT is really really bad at mathematics. Experts on AI will tell you the same thing. It should only be used for summaries of language, not numbers. I wouldn’t trust ChatGPT with any calculations.

I'm not full relying on it for the calcs, but am testing how it does against my normal process and measurements. For the post fermentation acid adjustments it has actually gotten pretty close as long as I am specific enough. It does better with some calcs than others

 

[Unicorn_Platypus]

ChatGPT is really really bad at mathematics. Experts on AI will tell you the same thing. It should only be used for summaries of language, not numbers. I wouldn’t trust ChatGPT with any calculations.

I think that depends on the complexity of the calculation. And it certainly shouldn't be used a primary source for a critical calc. As a secondary check it's pretty cool though

 

[doug293cz]

ChatGPT is really really bad at mathematics. Experts on AI will tell you the same thing. It should only be used for summaries of language, not numbers. I wouldn’t trust ChatGPT with any calculations.

Yeah, math should never be a "black box," as there is no way to verify that it's using the correct math, making reasonable assumptions, using appropriate boundary conditions, etc. I show my math so people can understand what's going on, understand my assumptions, recreate with different parameters, and call me out for mistakes (if I make any.) If you have to do the calculations yourself to see if the AI got it right, why bother with the AI at all, other than for amusement? You also get no understanding of the underlying principles if you just ask the AI to do something.

Brew on

 

[Unicorn_Platypus]

Yeah, math should never be a "black box," as there is no way to verify that it's using the correct math, making reasonable assumptions, using appropriate boundary conditions, etc. I show my math so people can understand what's going on, understand my assumptions, recreate with different parameters, and call me out for mistakes (if I make any.) If you have to do the calculations yourself to see if the AI got it right, why bother with the AI at all, other than for amusement? You also get no understanding of the underlying principles if you just ask the AI to do something.

Brew on

AI does show all of math for each step of the calcs. I left that out of my screenshots.

So it's not really a black box. It details all assumptions and lists out the formulas just like you do!

You could easily audit the math and review the formulas if you are familiar with the subject matter.

I'm not familiar with the chemistry math, but I did refine the assumptions I gave it in the prompts.

The more specifically you are with assumptions and what sources you direct it to use, the better it does.

Since I'm not a chemistry expert, its definitely better than me trying to do all the math on my own. Of course it wouldn't do as good a job as you though

Here's some examples of the math Chat GPT shows as part of the responses.

 

[hungupdown]

AI might give good figures, that can be manually verified. But it can't be relied upon, as it could just as easily produce complete garbage the the next time, depending on what new training source material has been injested.

 

[Unicorn_Platypus]

AI might give good figures, that can be manually verified. But it can't be relied upon, as it could just as easily produce complete garbage the the next time, depending on what new training source material has been injested.

If you carefully direct it to use certain specific sources it can be relatively reliable though.

The specificity of the prompt you give it can help minimize potential garbage

You need to talk to it like a child with explicit instructions every time

 

[doug293cz]

There isn't really a great priming sugar calculator for kegging beer. I fed chat GPT my detailed process so it could calculate the final volumes of CO2.

It took a little bit of tweaking, to get the prompts to work, but I think it finally got it right. This beer is a Belgian IPA, so I'm going for high volumes (to be served on long lines)

It's also been using it to help calculate lactic acid additions in finished beers (where there aren't standard calculator). Pretty wild stuff!
View attachment 871869View attachment 871871

The more I look, the more problems I see with this analysis:

Step 1: The CO2 content resulting from spunding at 4 psig and 73°F should be 0.98 - 0.99 volumes. I use equation 2.1 from the attached .pdf for carbonation calculations.

Step 2: 12 hours is not enough time to reach equilibrium when absorbing CO2 from the headspace into the beer, so this calculation isn't even possible, unless you have absorption rate as a function of temperature data for your specific system.

Step 3: Was the pressure 6 psig at the start of the step or the end? If it was at the end, then you could calculate the amount of CO2 absorbed during steps 2 & 3, if you know the headspace volume and beer volume accurately.

Step 4: Raising the temperature will cause the headspace pressure to increase, not decrease. If it did in fact decrease from step 3 to 4, that means the pressure could not have been 6 psig at the end of step 3.

Step 5: If you spund at 3 psig and 62°F, the carbonation level will be ~1.1 volumes.

Steps 6 & 7: If you add pressure and disconnect the CO2, you have to measure the pressure at the end of the shaking to know how much CO2 got absorbed. There is no way to calculate it reliably. If the CO2 stayed connected at the specified pressure, then you need to know how much the weight of the vessel increased during the step. You would be measuring weight increases of 10's of grams (single digit oz) on a total weight of ~50 lb (for a 5 gal batch.)

Step 8: If you start at 26 psig, and keep the keg at constant temp, the pressure will not drop to 1 atm (0 psig.) It's possible to calculate where it actually comes to equilibrium (if you know beer starting carb level, beer volume, and headspace volume.)

Brew on

 

[mac_1103]

If you carefully direct it to use certain specific sources it can be relatively reliable though.

Then why not just use those certain specific sources yourself?

Steps #7-9 are really a single step, for after I inject the priming sugar via my gas post. I broke it down into substeps so ChatGPT could understand it better. You always want headspace pressure in a sealed keg. So that headspace also is not added with the intent of adding carbonation, but is a necessary byproduct to avoid O2 ingress.

I put priming sugar into an empty sanitzed keg and purge it with fermentation gas, then closed transfer the beer into it from the fermenter. The keg stays sealed with no added pressure and no oxygen ingress.

 

[Unicorn_Platypus]

Then why not just use those certain specific sources yourself?

I put priming sugar into an empty sanitzed keg and purge it with fermentation gas, then closed transfer the beer into it from the fermenter. The keg stays sealed with no added pressure and no oxygen ingress.

Question #1 I'll ignore because it's rhetorical

This is now going off the intent of the topic which was to give an example on AI. I wasn't asking for a critique on my process, but am intrigued by yours.

Regarding your process if I understand correctly...

you are relying on primary fermentation to purge an empty keg (i.e. Not one filled with sanitizer)? The priming sugar is staying in there for whole duration on fermentation?

I've got 3 gallons of headspace in my fermzilla then another 5 gallons empty space in a keg. Fermentation gas could probably purge 8 gallons of air? Has anyone run calcs on that?

I'm more comfortable filling to the brim with sanitizer and purging to know I got everything out.

Are you keeping a spunding valve on the empty keg receiving keg to keep positive pressure after fermentation calms down?

I would also worry about keeping priming sugar in an empty keg for that many days without it being inoculated for risk of infection. Although chance of infection would likely be pretty small. Your process seems sound, but I'm still skeptical.

If I were confident the empty keg would be fully purged I'd be open to trying it. I'd still probably dose the receiving keg with priming sugar at packaging though.

Here's what ChatGPT says on the matter. Does @doug293cz agree with this math


Question?

Can fermenting 5 gallons of 1.050 wort purge 8 gallons of headspace completely of air?

To determine whether fermenting 5 gallons of 1.050 wort can purge 8 gallons of headspace of air (i.e., replace it with CO₂), we need to estimate how much CO₂ the fermentation produces and compare it to the volume of the headspace.

---

Step 1: Estimate CO₂ Production​

Fermentation converts sugars into alcohol and CO₂. A rough rule of thumb in brewing is:

• 1 mole of glucose → 2 moles of ethanol + 2 moles of CO₂

We can estimate the amount of CO₂ based on the original gravity (OG). A 1.050 OG wort contains about 0.050 × 8.34 lbs/gal = 0.417 lbs of extract per gallon, or roughly 2.1 lbs in 5 gallons (ignoring precision of extract yield).


But here's a better brewing-based estimate:

Approximate CO₂ produced:​

• A 5-gallon batch at OG 1.050 fermented to FG 1.010 produces roughly 0.9 to 1.0 lbs of CO₂.

Using molar mass:

• 1 mole CO₂ = 44 g
• 1 lb = 454 g → ~10.3 moles of CO₂
• At standard temperature and pressure (STP), 1 mole of gas occupies 22.4 liters, or 0.79 ft³

So:

• 10.3 moles × 22.4 L = ~231 L CO₂ = 8.15 ft³ CO₂

Since 1 gallon = 0.1337 ft³:

• 8 gallons = 1.07 ft³

---

Conclusion:​

Yes, fermenting 5 gallons of 1.050 wort can easily purge 8 gallons of headspace. It produces over 8 ft³ of CO₂, which is more than 7× the volume of the headspace. So all the air (O₂ + N₂) in that space would be replaced with CO₂ quite early in fermentation, assuming proper sealing and venting through an airlock or blow-off tube.


Let me know if you’re trying to use fermentation as a CO₂ source (e.g., for purging a keg or tank) — the setup matters too.

 

[mac_1103]

I've got 3 gallons of headspace in my fermzilla then another 5 gallons empty space in a keg. Fermentation gas could probably purge 8 gallons of air? I'm more comfortable filling to the brim with sanitizer and purging to know I got everything out.

Haven't you ever read Doug's Epic Post on CO2 Generation?

I would also worry about keeping priming sugar in an empty keg for that many days without it being inoculated for risk of infection.

What is it going to get infected with? It's cleaned, sanitized and sealed; the only thing going into the keg is fermentation gas. If there's anything to worry about, I would think it'd be yeast from the fermenter getting into the keg and eating the priming sugar before its time. But I'm not the only person who does this and it works.

 

[doug293cz]

I've got 3 gallons of headspace in my fermzilla then another 5 gallons empty space in a keg. Fermentation gas could probably purge 8 gallons of air? Has anyone run calcs on that?

Yes. See the link that @mac_1103 posted above.

I'm more comfortable filling to the brim with sanitizer and purging to know I got everything out.

Nothing wrong with that.

Here's what ChatGPT says on the matter. Does @doug293cz agree with this math

No. Again see the link.

Brew on

 

[mac_1103]

Here's what ChatGPT says on the matter. Does @doug293cz agree with this math

Probably trained on Doug's post.

 

[Unicorn_Platypus]

Wow! I wasn't expecting you to check all the math, but I'm impressed

There are some details I gave the prompts to clarify the steps. ChatGPT didn't spit those out in the final table, so you wouldn't have been aware of the assumptions. I'm not expecting you to analyze the math on any of these steps, but see some additional comments below just for color.

The more I look, the more problems I see with this analysis:

Step 1: The CO2 content resulting from spunding at 4 psig and 73°F should be 0.98 - 0.99 volumes. I use equation 2.1 from the attached .pdf for carbonation calculations.

this looks off, I agree

Step 2: 12 hours is not enough time to reach equilibrium when absorbing CO2 from the headspace into the beer, so this calculation isn't even possible, unless you have absorption rate as a function of temperature data for your specific system.

I am adding 12 PSI of headspace just before I cold crash. Once the temp gets to 38F, I burp the headspace to 6 PSI (so I don't pickup the extra carbonation. when I dry hop, I don't want the beer to be especially carbonated and foam).

Step 3: Was the pressure 6 psig at the start of the step or the end? If it was at the end, then you could calculate the amount of CO2 absorbed during steps 2 & 3, if you know the headspace volume and beer volume accurately.

burp to 6 PSI after it hits 38F and then just leave it there for 1.5 days undisturbed

Step 4: Raising the temperature will cause the headspace pressure to increase, not decrease. If it did in fact decrease from step 3 to 4, that means the pressure could not have been 6 psig at the end of step 3.

I'm burped the headspace to 5 PSI after it warms up

Step 5: If you spund at 3 psig and 62°F, the carbonation level will be ~1.1 volumes.

will the 24 hours be enough to release all the previously absorbed CO2 from solution though? You are assuming the duration is long enough to fully reach equilibrium which it probably isn't

Steps 6 & 7: If you add pressure and disconnect the CO2, you have to measure the pressure at the end of the shaking to know how much CO2 got absorbed. There is no way to calculate it reliably. If the CO2 stayed connected at the specified pressure, then you need to know how much the weight of the vessel increased during the step. You would be measuring weight increases of 10's of grams (single digit oz) on a total weight of ~50 lb (for a 5 gal batch.)

that makes sense, I could use a pressure gauge to measure the headspace pressure after shaking. What if we assumed I shook it enough to reach equilibrium as a conservative estimate?

Step 8: If you start at 26 psig, and keep the keg at constant temp, the pressure will not drop to 1 atm (0 psig.) It's possible to calculate where it actually comes to equilibrium (if you know beer starting carb level, beer volume, and headspace volume.)

that sounds complicated, but makes sense

Brew on

In practice, I take notes from previous batches for dosing amounts in future batches. I try to keep the amount of headspace pressure the same each time to limit variables.

In my process, I find that 60g dextrose gets me a reasonably carbonated beer (good for stout or malty lagers) & 120g gets me closer to Belgian carbonation territory. I measure via a gauge the CO2 pressure after carbonation and 10 days sitting in the keezer at 32F.

I figure by then its reached equilibrium. At that point the reading I get on the gauge should give me the final CO2 volume when compared to a carbonation chart. Is that a fair assumption?

I don't use chat gpt in practice, but was having fun messing around with it. The purpose of this thread was to explore its use, not to endorse its use.

 

[Unicorn_Platypus]

ChatGPT approves of Doug's work: "In conclusion, Doug293cz's analysis provides a scientifically sound framework for keg purging using CO₂ from active fermentation. His approach is both theoretically robust and supported by practical applications within the brewing community. Therefore, I find his analysis to be well-founded and agree with his conclusions on the efficacy of this method."

--------------------------------------------
do you agree with doug293cz's analysis on keg purging? (see post #3 in the below link)https://www.homebrewtalk.com/threads/keg-purging-with-active-fermentation.628658/#post-8004741

ChatGPT said:​

Doug293cz's analysis on keg purging during active fermentation offers a detailed and methodical approach to utilizing the CO₂ produced during fermentation to effectively purge kegs of oxygen. His insights are grounded in scientific principles, particularly the behavior of gases and the dynamics of CO₂ production relative to sugar consumption during fermentation.Brülosophy+5Brülosophy+5YouTube+5Home Brew Talk
The core of his analysis addresses two main points:

1. Determining the Volume of CO₂ Needed to Purge a Keg: Doug293cz delves into the calculations required to understand how much CO₂ is necessary to reduce oxygen levels within a keg to a minimal concentration. He explains that each pressurization and venting cycle decreases the oxygen concentration in the headspace, and multiple cycles exponentially reduce the remaining oxygen. This method, while effective, is noted to be gas-intensive. Home Brew Talk+1The ****************+1The ****************+1Home Brew Talk+1
2. Calculating CO₂ Production During Fermentation: He provides a scientific explanation of the relationship between sugar consumption and CO₂ production, allowing brewers to estimate the amount of CO₂ generated during fermentation. This understanding enables brewers to harness naturally produced CO₂ for purging kegs, offering a resourceful alternative to using commercial CO₂ tanks. Home Brew TalkBrülosophy+6Home Brew Talk+6Home Brew Talk+6

The method of purging kegs with CO₂ from active fermentation has been practically applied by various brewers. For instance, an experiment compared beers packaged in kegs purged with fermentation-derived CO₂ versus those purged with commercial CO₂. The results indicated no significant sensory differences between the beers, suggesting that using fermentation-produced CO₂ is an effective purging method. Home Brew Talk+15Home Brew Talk+15Brülosophy+15Brülosophy
Additionally, discussions among homebrewers have highlighted practical considerations, such as connecting the fermenter's gas output to the keg's liquid post to ensure CO₂ enters at the bottom, effectively displacing oxygen upwards and out of the keg. Implementing a spunding valve or blow-off tube on the keg's gas post can further facilitate this process. Home Brew Talk+7Brülosophy+7Home Brew Talk+7Brülosophy+10Home Brew Talk+10Brülosophy+10
In conclusion, Doug293cz's analysis provides a scientifically sound framework for keg purging using CO₂ from active fermentation. His approach is both theoretically robust and supported by practical applications within the brewing community. Therefore, I find his analysis to be well-founded and agree with his conclusions on the efficacy of this method.

 

[Unicorn_Platypus]

Haven't you ever read Doug's Epic Post on CO2 Generation?

What is it going to get infected with? It's cleaned, sanitized and sealed; the only thing going into the keg is fermentation gas. If there's anything to worry about, I would think it'd be yeast from the fermenter getting into the keg and eating the priming sugar before its time. But I'm not the only person who does this and it works.

I'll give it a shot on my next batch. It wouldn't be fair if you were the only person doing it.

one flaw I see with this though is that a corny keg generally requires head pressure to stay properly sealed. If you are allowing all the pressure to go out via a blow off tube, the lid may not be totally sealed up. I guess, as long as you disconnect it an pressurize it after fermentation starts to slow it should be OK

 

[Unicorn_Platypus]

Now, can anyone produce a calculator for adjusting PH on finished beer based on OG, FG? That sounds trickier. But I would vouch that 3ml 88% lactic acid typically drops ~0.1 PH for 4.5 gallons based on numerous batches I've done. More variability to it depending on starting PH, ABV, etc. 3ml is a good reference/starting point.

 

[day_trippr]

Wouldn't that calculation actually have to go back to the grain bill to have a legit idea of the pre-pitched wort pH?

Cheers!

 

[mac_1103]

one flaw I see with this though is that a corny keg generally requires head pressure to stay properly sealed

This hasn't been my experience. I can seal most of my kegs airtight without adding any pressure. And when I can't get one to seal properly adding pressure doesn't help. But I do recall some rather heated exchanges on this topic in a few threads.

 

[Unicorn_Platypus]

This hasn't been my experience. I can seal most of my kegs airtight without adding any pressure. And when I can't get one to seal properly adding pressure doesn't help. But I do recall some rather heated exchanges on this topic in a few threads.

I think if they are newer kegs with good seals its probably OK. Older jankier kegs might be a problem.

I'll give it a shot on my next batch, I think my seals are pretty good as I mostly have newer kegs.

I'll plan to pressurize to ~5 psi and disconnect from the fermenter once 60%+ attenuation is reached. That way I can be confident it's kept at positive pressure until I'm ready to transfer. I can connect the spunding valve to the fermetor at that point also at a very at a low pressure.

 

[Unicorn_Platypus]

Wouldn't that calculation actually have to go back to the grain bill to have a legit idea of the pre-pitched wort pH?

Cheers!

yeah there's probably a ton of other variables that are needed including minerals, grains, etc.

Martin has said its a rather difficult calculation once alcohol is in the equation and recommends to just do it by taste post fermentation. There isn't really an ideal final PH for all beers due to variability. 3ml lactic after a big dry hop charge tends to work great for me for most beers.

I'll take a sample and PH read, and a taste before I dry hop. Then I take another taste and another read after I dry hop. If the PH post dry hopping is above >4.55 (or the beer tastes dull) its generally a good idea to bring it down below 4.5. Hoppy beers get quite dull if you approach 4.6. 4.35-4.45 is generally the sweet spot I find most hoppy beers taste right. Some beers I won't touch if the PH post dry hop lands in that range and they taste good as is. Some yeasts are bigger acid producers than others

If I like the way the beer tasted before I dry hopped, bringing it back down to that pre dry hop PH range usually works well.