I was having a conversation with some IPA brewers who said they like to get their sulfate levels into the 250-350ppm levels. I asked if they were adding gypsum to the kettle and some were but others said, "No, just the mash and sparge water". I asked how on Earth their pH could be lined up in the ideal range with all that gypsum in the water and one said, "I don't worry about low mash pH". This is an experienced and long-time brewer so I am wondering about this. High mash pH can cause harshness, the pulling of tannins from the grains, haziness, a darker wort, etc. What does LOW mash pH do for you?
What are the effects of a low mash pH?
- Thread starter kenlenard
- Start date
Help Support Homebrew Talk - Beer, Wine, Mead, & Cider Brewing Discussion Forum:
Calcium doesn't move pH that much. At 350 ppm with 0 alkalinity knockout pH is only shifted, per Kolbach's rule of thumb, by -0.17 pH so if your brewers don't use acid and don't use a lot of dark crystal then the pH shouldn't fall below the nominal desired range.
As for what would happen if it did: some speak of all manner of dire consequences. Some say they brew with those pH's all the time without negative consequences. I can't say as I've never been there.
As for what would happen if it did: some speak of all manner of dire consequences. Some say they brew with those pH's all the time without negative consequences. I can't say as I've never been there.
kenlenard
Well-Known Member
Interesting. I used my local water in EZ_Water and added enough CaCl and CaSO4 to the mash to get to 350ppm of sulfate and my RA was something like -300 and the pH was suitable for beers between zero SRM and zero SRM. This leads me to the idea that the pH would be ultra low. And... what if the beer being made was a darker IPA... say SRM 15 to 20 or something?
I don't know how much CaCl2 you might have added in addition to the gypsum but the amount of calcium from gypsum that goes with 350 mg/L sulfate will lower the RA to -100 in the absence of any alkalinity.
Color of the beer doesn't really have much if anything to do with it (though many apparently still think it does).
Color of the beer doesn't really have much if anything to do with it (though many apparently still think it does).
kenlenard
Well-Known Member
Sorry AJ, can you expand on that? The color doesn't have much to do with the RA or mineral content... not sure what you were referring to. Cheers & thanks.
We all know that Pilsner beer originated in a place where the water is low in mineral content (low RA), that Dunkles is brewed in a place where the water is high in mineral content, particularly bicarbonate, so the RA is high, that pale ales are brewed with high mineral water but with (relatively) low bicarbonate implying low RA etc. This caused John Palmer to conclude some years back that RA and beer color were correlated which conclusion he verified by plotting the colors of beer vs the characteristics of the water they were brewed with and fitting a curve to the data. He then published first nomographs (bak cover of his book) and then spreadsheets which could be used to 'design' brewing water based on color using this correlation. Home brewers loved this. It seemed there was a simple way to deal with this water chemistry business after all and led people to make statements like the one you did in #3 where you say "...my RA was something like -300 and the pH was suitable for beers between zero SRM and zero SRM". Calculators and spreadsheets based on this model were released by other authors and this model became quite prevalent.
There were several problems with his approach
1. As he didn't know what the actual colors of most of the beers were he generated color data from what he could find out about the grain bills.
2. Much of the published water data is bad
3. Even if one knows what the water chemistry in a region is one does not know what the chemistry going into the mash tun is.
4. The 'fit' wasn't very good.
With respect to the last point, if you look at a curve fit with r = 0.66 you might conclude that it is a pretty fair fit but in fact only 43% of the variation comes from the model. Fifty seven percent is 'noise'.
The result of all this was that a generation of brewers were blithly adding tablespoonsfull of chalk to their mash without ever stopping to think that this might be excessive.
Recognition of the problems has caused several spreadsheet authors to 'tone down' their residual alkalinity recommendations and so the situation is a little better but most who understand how things really work have dropped beer color as a 'design parameter' for beer and the idea of an 'RA requirement' to match a particular color seems to be fading.
There is, of course, a correlation between malt color and the amount of acid it produces and this can be used to help calculate predicted mash pH to the extent that this correlation is tighter (r closer to 1).
There were several problems with his approach
1. As he didn't know what the actual colors of most of the beers were he generated color data from what he could find out about the grain bills.
2. Much of the published water data is bad
3. Even if one knows what the water chemistry in a region is one does not know what the chemistry going into the mash tun is.
4. The 'fit' wasn't very good.
With respect to the last point, if you look at a curve fit with r = 0.66 you might conclude that it is a pretty fair fit but in fact only 43% of the variation comes from the model. Fifty seven percent is 'noise'.
The result of all this was that a generation of brewers were blithly adding tablespoonsfull of chalk to their mash without ever stopping to think that this might be excessive.
Recognition of the problems has caused several spreadsheet authors to 'tone down' their residual alkalinity recommendations and so the situation is a little better but most who understand how things really work have dropped beer color as a 'design parameter' for beer and the idea of an 'RA requirement' to match a particular color seems to be fading.
There is, of course, a correlation between malt color and the amount of acid it produces and this can be used to help calculate predicted mash pH to the extent that this correlation is tighter (r closer to 1).
kenlenard
Well-Known Member
Well, I know that homebrewers are looking for a tool (or tools) that will help them with this and I admit to liking the idea of getting the mash and water profiles for various styles correct and anything that will help the masses is good. Of course, it's not good if the product doesn't work, the information is misleading, incomplete or inaccurate. If you say that many of the spreadsheets are causing problems or just not filling in all the blanks, I would feel like I needed to look elsewhere for this. Another brewer recently said to me, "I'm trying to make good beer, not balance numbers on a spreadsheet" and I said, "But balancing numbers on a spreadsheet could result in making good beer!". Maybe I was wrong.
Be careful or you'll get me started on a long discourse on modeling. A spreadsheet represents a model. Some models are better than others but none are any good if you don't put the right data into them. For example, I believe I know how to predict/control mash pH quite closely because I have a pretty good model. But I can't get the data I need to go into that model without doing a lot of hard, dull, repetitive work in the lab. This doesn't mean the model doesn't have value. I can use it to tell me about where mash pH will wind up and about how much acid I need to get it there but I'll use a fraction of the acid and check pH in the mashtun before I add it all.
I got a good perspective on modelling from a internet weather report prior to the big recent snow storm. The meteorologist said snowfall would be between 1 and 23". The eye candy asked why the range was so broad. The meteorologist said that whereas only a few years ago we had but two predictive models to choose from there were now 9. The least pessimistic model predicted 1", the most pessimistic, 23". The actual snow fall reached 30" in some places.
People are such suckers for anything that comes out of a computer.
Consider the following scenario. You are a young researcher with a family to feed working at an eastern liberal (tautology) university on long term climate models using government funding. Your model shows that increasing solar constant is a larger factor in long term warming that anthropogenic CO2. Maybe you should review your model.
I got a good perspective on modelling from a internet weather report prior to the big recent snow storm. The meteorologist said snowfall would be between 1 and 23". The eye candy asked why the range was so broad. The meteorologist said that whereas only a few years ago we had but two predictive models to choose from there were now 9. The least pessimistic model predicted 1", the most pessimistic, 23". The actual snow fall reached 30" in some places.
People are such suckers for anything that comes out of a computer.
Consider the following scenario. You are a young researcher with a family to feed working at an eastern liberal (tautology) university on long term climate models using government funding. Your model shows that increasing solar constant is a larger factor in long term warming that anthropogenic CO2. Maybe you should review your model.
kenlenard
Well-Known Member
I get that. The products are only as good as the data and even then they should be used as a tool, not a magic bullet. I feel like the spreadsheets do an okay job determining (at least) the zip code of where the mash pH will be. The small adjustments can be done later with acid or whatever. But there is always a swirling vortex of information as it relates to mineral additions (mash, sparge, kettle), level of minerals in any given beer, water and the role it plays, etc. I keep thinking that I may have made better beer before I got into any of this. But I also keep thinking that I will make EVEN BETTER beer if I can get a handle on all of these pieces. I kind of feel like I have all the pieces of the puzzle but I haven't put the puzzle together yet. The results of many of the latest things I have picked up are now sitting in primary and secondary so fingers are crossed that some recent changes will create some better results.
Hermit
fuddle
Establish that everyone is using the same protocol for pH testing and then do a beer swap?
Kaiser
Well-Known Member
Let me add my point of view here since this is something I have worked on for a few years now.
based on my research I have reworked the SRM to mash pH formula and it does seem to work fairly well once one makes the distinction between color from dark roasted malts and color from other malts. The new formula takes into account that roasted malts don’t contribute nearly as much acid for the same color compared to base or crystal malts. That formula (http://braukaiser.com/wiki/index.php?title=Beer_color_to_mash_pH_(v2.0)) ended up being a mix of marrying the color and mash pH formulas as well as corrections for practical observations. The latter takes care of some of the inconsistencies.
I agree. Brewers, especially home brewers, like the idea to plan their brewing as much as they can. That’s why they developed IBU prediction formulas and that’s why there are mash chemistry calculators that take a stab at predicting mash pH. I would say that a given spreadsheet or program is only “broken” when it causes you to make bad beer.
As an example. When I noticed that John Palmer’s RA spreadsheet didn’t get the expected alkalinity from chalk I first made an experiment to check if that “bug” actually made a noticeable difference in mash pH. Interestingly enough it didn’t and many more experiments followed to better understand what was going in. While it was not John’s intend to implement chalk with only 50% of its alkalinity contribution it worked well in the grand scheme of things and the spread sheet was not really “broken”. But it may have caused brewers to add excessive amounts of chalk to their beer. Based on a recent discussion with A.J there is the possibility that this can harm your beer even if only a small portion dissolves in the mash.
Reading your paper on RA and mash pH got me started on characterizing various grains and looking into the effects of water and grain more systematically. While I think that there are still mechanism at work that we don’t fully understand or are difficult to characterize I believe that I have developed a model and have broad enough data for a reasonably accurate mash pH prediction. Granted, there are outliers, but based on the feedback I have gotten and data from my own batches it’s doing pretty well. When I come across outliers I may spend the time to investigate them and find the assumptions or data input that was not correct. That may point me to something that needs to be accounted for in the future if it can be characterized.
But I do like the idea of mash pH prediction al lot. when I brew I oftentimes have lots of other things to do around the house. I don’t want to add some acid, test, add more acid and test again. If I can predict that I need 2% acid malt to get to 5.35 and I end up between 5.3 and 5.4 I’m happy. Even if I end up between 5.2 and 5.5 I may not do anything about it. I’m going to be less happy in that case, mostly b/c my mash pH prediction failed me.
Because of the complex interactions between water and malt mash pH prediction models need to be based on a large amount of experiments. It is nice if these experiments align with under understanding of the chemistry that is involved. But even w/o knowing the chemistry one could develop mash prediction models as long as enough data exists that represents the effects of the parameters the model is considering.
The suitability of any given model also depends on the accuracy of prediction that we expect. If we expect to predict mash pH within 0.01 pH units, we don’t have a model or even data to make this happen. If we are fine with +/- 0.1, I think we are there for most practical brewing cases.
That’s the intend. no more and no less.
I’d like to draw an analogy to IBU prediction. Here we have the same issue. Many factors affect the IBU level of the final beer yet the predictors we are using only consider initial AA content, boil time, wort gravity and hop product type. Some can also adjust for hop age. And many brewers know that the predicted IBU level is not necessarily the actual IBU level and that they may have to aim for a higher or lower predicted IBU level to get the desired bitterness. But we are not having much discussion about the accuracy of these predictions. Mostly because ordinary homebrewers can’t easily measure IBUs. But we can do that for mash pH which is why I believe there is more scrutiny on the pH prediction model.
There is more to brewing water modification than mash pH. Furthermore, the spreadsheets only make an attempt at predicting mash pH. They do not attempt to predict the mash pH that would be best for your beer. Just like the IBU estimators won’t tell you what IBU level to shoot for when making an IPA. That’s where brewers need to build up experience and knowledge: what is a pH that works best for this particular beer and how accurate do I have to control this pH.
Kai
based on my research I have reworked the SRM to mash pH formula and it does seem to work fairly well once one makes the distinction between color from dark roasted malts and color from other malts. The new formula takes into account that roasted malts don’t contribute nearly as much acid for the same color compared to base or crystal malts. That formula (http://braukaiser.com/wiki/index.php?title=Beer_color_to_mash_pH_(v2.0)) ended up being a mix of marrying the color and mash pH formulas as well as corrections for practical observations. The latter takes care of some of the inconsistencies.
I agree. Brewers, especially home brewers, like the idea to plan their brewing as much as they can. That’s why they developed IBU prediction formulas and that’s why there are mash chemistry calculators that take a stab at predicting mash pH. I would say that a given spreadsheet or program is only “broken” when it causes you to make bad beer.
As an example. When I noticed that John Palmer’s RA spreadsheet didn’t get the expected alkalinity from chalk I first made an experiment to check if that “bug” actually made a noticeable difference in mash pH. Interestingly enough it didn’t and many more experiments followed to better understand what was going in. While it was not John’s intend to implement chalk with only 50% of its alkalinity contribution it worked well in the grand scheme of things and the spread sheet was not really “broken”. But it may have caused brewers to add excessive amounts of chalk to their beer. Based on a recent discussion with A.J there is the possibility that this can harm your beer even if only a small portion dissolves in the mash.
Reading your paper on RA and mash pH got me started on characterizing various grains and looking into the effects of water and grain more systematically. While I think that there are still mechanism at work that we don’t fully understand or are difficult to characterize I believe that I have developed a model and have broad enough data for a reasonably accurate mash pH prediction. Granted, there are outliers, but based on the feedback I have gotten and data from my own batches it’s doing pretty well. When I come across outliers I may spend the time to investigate them and find the assumptions or data input that was not correct. That may point me to something that needs to be accounted for in the future if it can be characterized.
But I do like the idea of mash pH prediction al lot. when I brew I oftentimes have lots of other things to do around the house. I don’t want to add some acid, test, add more acid and test again. If I can predict that I need 2% acid malt to get to 5.35 and I end up between 5.3 and 5.4 I’m happy. Even if I end up between 5.2 and 5.5 I may not do anything about it. I’m going to be less happy in that case, mostly b/c my mash pH prediction failed me.
Because of the complex interactions between water and malt mash pH prediction models need to be based on a large amount of experiments. It is nice if these experiments align with under understanding of the chemistry that is involved. But even w/o knowing the chemistry one could develop mash prediction models as long as enough data exists that represents the effects of the parameters the model is considering.
The suitability of any given model also depends on the accuracy of prediction that we expect. If we expect to predict mash pH within 0.01 pH units, we don’t have a model or even data to make this happen. If we are fine with +/- 0.1, I think we are there for most practical brewing cases.
That’s the intend. no more and no less.
I’d like to draw an analogy to IBU prediction. Here we have the same issue. Many factors affect the IBU level of the final beer yet the predictors we are using only consider initial AA content, boil time, wort gravity and hop product type. Some can also adjust for hop age. And many brewers know that the predicted IBU level is not necessarily the actual IBU level and that they may have to aim for a higher or lower predicted IBU level to get the desired bitterness. But we are not having much discussion about the accuracy of these predictions. Mostly because ordinary homebrewers can’t easily measure IBUs. But we can do that for mash pH which is why I believe there is more scrutiny on the pH prediction model.
There is more to brewing water modification than mash pH. Furthermore, the spreadsheets only make an attempt at predicting mash pH. They do not attempt to predict the mash pH that would be best for your beer. Just like the IBU estimators won’t tell you what IBU level to shoot for when making an IPA. That’s where brewers need to build up experience and knowledge: what is a pH that works best for this particular beer and how accurate do I have to control this pH.
Kai
Please be aware that I am not against modeling. I made my living doing it for many years. Perhaps because of this I am hypersensitive to the limitations of models and all too aware that sometimes the customer doesn't like what the model shows. I really want people to be aware of the limitations. It looks, from the previous couple of posts, as if they are.
Kaiser
Well-Known Member
A.J, I didn't see it like that. My response was rather long b/c I saw it as an opportunity to outline my position on this. I agree that users should understand the limitations and don't trust the models too much.
There seem to be two sides to this. A water calculator needs to be simple enough to gain acceptance. But if it is used by brewers with a wide range of experience it will also be used by some who expect it to be correct all the time and trust it over their own measurements.
It think that this education will come over time. It did for the IBU calculators.
Kai
There seem to be two sides to this. A water calculator needs to be simple enough to gain acceptance. But if it is used by brewers with a wide range of experience it will also be used by some who expect it to be correct all the time and trust it over their own measurements.
It think that this education will come over time. It did for the IBU calculators.
Kai
I like Kai's observation that the RA vs beer color is workable when you factor in the roast content. He pointed that out to me a few weeks ago and I agree that with that roast content caveat, a more simple approach could be cobbled together. The only reason I don't go in that direction is that I like having all the pieces (malt components) in the calculation so that I can follow what is going on. That is the way Bru'n Water is formulated and I think that is the way that Kai's on-line calculator is working also. But Kai's simple approach is potentially more elegant.
I disagree with AJ's position on models. I use them all the time in my engineering and rely on them with my reputation and license. I actually do believe that a brewing chemistry model can be quite accurate in pH prediction. The data sets that I've gathered, indicate that the response is very linear and repeatable. The biggest problem is that we can't rely on the consistency of the malts we use and in some places, we can't rely on the water quality. Those are the biggest shortcomings in using these models. GIGO cannot be overcome with even the most accurate model. Using a model that can't even help a brewer tell if there water inputs are faulty is just asking for problems. However in any case, putting malt and water is going to provide a wort and it will produce beer. Being off with these calculations only means that the opportunity for great beer was dashed and maybe only good beer was produced. Its not a disaster.
There is no way around the GIGO limitations. That applies whether you are relying solely on a model or monitoring pH and having to chase a pH goal. However, the utility of a model is that more users can create more success with a smaller equipment investment and less hassle than with using a pH meter. And even with a meter, guessing at acid or caustic additions is a fool-hardy endeavor. Having a model of a mash that allows a brewer to anticipate and calculate those additions is FAR better than having a pH meter and having to learn by trial and error (and probably screwing up a few batches). And all that trial and error only gets you tuned in to the requirements for that brew you just brewed and rebrewed. When a brewer moves on to another style, they are back at square one. It is quite apparent to me and hopefully nearly every other reader of this post that models are the best way to improve your brewing when you aren't going to brew the same thing time after time.
There are people who are forward thinking and embrace new tools and methods. And then there are people whose names are Ned Ludd.
I disagree with AJ's position on models. I use them all the time in my engineering and rely on them with my reputation and license. I actually do believe that a brewing chemistry model can be quite accurate in pH prediction. The data sets that I've gathered, indicate that the response is very linear and repeatable. The biggest problem is that we can't rely on the consistency of the malts we use and in some places, we can't rely on the water quality. Those are the biggest shortcomings in using these models. GIGO cannot be overcome with even the most accurate model. Using a model that can't even help a brewer tell if there water inputs are faulty is just asking for problems. However in any case, putting malt and water is going to provide a wort and it will produce beer. Being off with these calculations only means that the opportunity for great beer was dashed and maybe only good beer was produced. Its not a disaster.
There is no way around the GIGO limitations. That applies whether you are relying solely on a model or monitoring pH and having to chase a pH goal. However, the utility of a model is that more users can create more success with a smaller equipment investment and less hassle than with using a pH meter. And even with a meter, guessing at acid or caustic additions is a fool-hardy endeavor. Having a model of a mash that allows a brewer to anticipate and calculate those additions is FAR better than having a pH meter and having to learn by trial and error (and probably screwing up a few batches). And all that trial and error only gets you tuned in to the requirements for that brew you just brewed and rebrewed. When a brewer moves on to another style, they are back at square one. It is quite apparent to me and hopefully nearly every other reader of this post that models are the best way to improve your brewing when you aren't going to brew the same thing time after time.
There are people who are forward thinking and embrace new tools and methods. And then there are people whose names are Ned Ludd.
Kai,
I didn't think you did but I just wanted to be clear on that point.
Martin,
I went out of my way to be clear that I do appreciate the value of models and in fact made my living developing and using them for a good part of my career and yet you disagree with my position so I am not sure at this point what it is you are disagreeing with. I will say that the models in question here are definitely not linear, even though we may solve them by piecewise linearization, so perhaps that's it.
I didn't think you did but I just wanted to be clear on that point.
Martin,
I went out of my way to be clear that I do appreciate the value of models and in fact made my living developing and using them for a good part of my career and yet you disagree with my position so I am not sure at this point what it is you are disagreeing with. I will say that the models in question here are definitely not linear, even though we may solve them by piecewise linearization, so perhaps that's it.
