« on: September 06, 2010, 07:35:13 AM »
I emailed A.J. Delange and asked and asked him if he had any additional thoughts on this thread. I felt that almost everything he sent back was relevant to this discussion so here is his reply in its entirety:
Let me preface by saying thank you! I'm really gratified that someone is able to use this stuff.
Now, on to the thread.
I really think you and Martin have pretty much got it covered. There are a couple of points where you guys disagree a bit but they are almost philosophical. One is on the correlation between beer color and RA. There is, of course, a correlation. Everyone knows that styles that originated in places where the water was hard tended to be darker. The question is as to how strong that correlation is. To determine that, of course, you have to have data and that's hard to get. The reason its hard to get is because you don't have good knowledge of either the SRM or the RA for a particular beer. If I were to try to determine what the correlation actually is I would have little choice except to use my own beers and that is hardly a subset which would result in an informative model as I do, for example, Bocks, which are darker than my ESB with water that has lower RA.
The other approach is to look at beers for which I have measured the SRM that are brewed in cities for which I have a water profile. For example Guiness and Dublin and Bass ale iand Burton. Trouble with that is that Guiness is not brewed in Dublin any more and Bass isn't brewed in Burton either. But PU is brewed in Pilsen and Kölsch in Köln so we are perhaps not all wet if we assume that Guiness is still brewed with a Dublin-like profile and Bass with a Burton-like profile (which it clearly isn't - modern Bass isn't anything like as minerally as an ale brewed with "traditional" Burton water). The other problem is, of course, that I have about 6 mineral profiles for Burton. You all have been commenting on the fact that most published profiles are hooey and indeed so are most of these (I think 2 balance fairly well) and the range of RA's is -13 - 85 for the two profiles that balance. Caveats aside, I took 17 beers and did the correlation. It shows that the relationship between SRM and RA is RA = 5.6 + 1.05*SRM. This would predict that, for example, my 70 SRM Irish stout should be brewed with water of RA 80. This is certainly more reasonable than the 300 some number that the popular spreadsheets come up with but we still haven't looked at the tightness of fit. That's measured by "Pearsons r", a measure which indicates how much of the variation in observed data comes from the model (the model here is RA = 5.6 + 1.05*SRM). For the data I played with r = 46% (100% means the model is perfect, 0% means there is no correlation whatsoever between SRM and RA). So the correlation is there but it is weak. If this were a game where you tell me the SRM and I bet on the RA predicted by the model I wouldn't take the bet!
I think what John did when he first came up with these was get data from brewers about water treatment and grain bill. I believe he calculated RA and color from this data and then did the regression. I'm pretty sure about the color part because I remember seeing a comment in a post of his concerning which color model he used. I'm not so sure about the RA. Whatever he did the resulting slope defies common sense (and he acknowledges this).
So given all the pitfalls of the curve fit approach (and who said it had to be linear?) I recommend that the spreadsheet developers take this "feature" out of their spreadsheets.
Imbalance seems to be another question that didn't seem quite resolved. You all said it without saying it explicitly: imbalance represents a measure of the quality of a water report. If the report exhibits large imbalance that means errors were made in measurement, the sample changed while the measurements were going on, bicarbonate was calculated incorrectly from alkalinity, ionic strength was ignored, or some relatively prevalent ion or ions were not measured. The way I often put it is to say that mother nature cannot make imbalanced water and neither can you so if you are trying to match an imbalanced profile you will not succeed.
Finally, the philosophy of owning a pH meter: Given all the variables I do not believe it is possible to accurately predict mash pH and that, therefore, it is essential that you measure it to see if your treatment and grist formulation combine to land you in the right pH range. But I agree that once you have determined that they do you should come back to pretty close to the same pH every time you brew a particular beer and don't really need to check pH. Being the sort I am I do and I also check at the return of each decoction, out of the kettle and throughout the course of the ferment. I don't advocate this for everyone but those pH readings are like familiar landmarks on the road home to me. Each time you see one in the right place you are further assured that your journey will come to a successful end.
Now one thing that has not been mentioned (or emphasized) is that most beers will require the addition of acid in some form to reach proper mash pH. For German/continental brewers this is sauermalz or sauergut (i.e. lactic acid). For British brewers it is "Carbonate Reducing Solution" (a mix of sulfuric and hydrochloric acids).
Hope this is of some help. Feel free to quote it if you like.