FG calculations

I'm playing with an open source brewing calculator tool and am thinking about improving the system that estimates the final gravity for a particular recipe. Currently the system uses an attenuation percentage entered by the user to make the calculation but I have seen other systems that make this estimate themselves.
My understanding is that the important factors in this calculation are the specific grains used in the recipe, in particular the speciality grains that contain unfermentable sugars. In addition there are some mash related factors (durations and temperatures) but these are much harder to calculate.
I'm having difficulty finding both examples of how this calculation could be done and the specific data on unfermentable components of malt types.
Can anyone give me some pointers so I have some idea what to do?

Thanks

-Base malts => Fermantability limited by the yeast attenuation

-Lactose & Dextrins => 0% fermentable

-Crystal malt => ?, reason why I'm conducting the following experiment
https://www.homebrewtalk.com/f128/testing-fermentability-crystal-malt-208361/

-Attenuation variation with mashing temp = check my post at
http://beertech.blogspot.com/2009/12/considerations-when-predicting-final.html

Hope this help

I guess you could start by going to the maltsters' websites where they typically give dry basis FG HWE numbers for their products. Home brewers never get near this so you would have to allow the user to dial in his anticipated efficiency and then apply a second multiplier to that running linearly from 1.0 at say, 142 °F to say 0.8 at 160 °F (I'm guessing wildly here - maybe it's not even linear) to account for saccharification temperature. There's a large uncertainty from the yeast most of which are specified at something like 70 - 78% attenuation but you would have to give the user the opportunity to put that in.

This is one of those things that can easily get over modeled - precise calculations made based on inaccurate inputs are inaccurate (but precisely so). It would seem simpler to just let the guy measure his OG, specify an attenuation or perhaps an attenuation range and calculate FG and estimated alcohol (why else would he care about FG) or a range by which I mean if you let him put in yeast suppliers estimated attenuation range (e.g. 70 - 80%) you might want to display the corresponding FG (e.g. 4 - 3 P) and ABV ( e.g. 4.8 - 5.2%). All numbers just pulled out of the air for illustration.

I don't want to poop in your cheerios, but FG calculations are just guesses at best. There is way too much that factors into it for brewing software to give you a number to shoot for.

I mean, if you want to go for it, then go for it, but this could end up as a rat hole.

Walker said:

I don't want to poop in your cheerios, but FG calculations are just guesses at best. There is way too much that factors into it for brewing software to give you a number to shoot for.

I mean, if you want to go for it, then go for it, but this could end up as a rat hole.

Click to expand...

Oh absolutely, I'm aiming for a sort of approximate number that gives some vague idea of what effects changing your recipe will have.

nilo said:

-Attenuation variation with mashing temp = check my post at
http://beertech.blogspot.com/2009/12/considerations-when-predicting-final.html

Click to expand...

Love it when people take the trouble to collect some data. So I had a look at it. A bit difficult to read the numbers off the graphs esp. temperatures which range on the graph from 52 to 58.5 which I doubt were the actual temperatures in either Celsius or Fahrenheit but as long as they are linearly related to temperature it doesn't matter. I did a linear fit and came up with ADF = 163.09 - 1.4702*T where T is the temperature in the units of the graph whatever they may be. Pearson's r was -0.83 (not a bad fit but not a good one either) and chi-squared 87.6. I wondered why you used the a +b/(T-c) form for your fit. It works but no better than a straight linear and c brings nothing to the party. You can set it to 0 and get just as good a fit (though a and b will be different).

This data is illustrative of what I was on about in my last post. At 155 (the middle of the range) one can be 95% confident that the attenuation will be between 75% and 90% based on these data with a linear fit (slightly - 1% reduction in span relative to the reciprocal function). The one-sigma (68%) band is 79-86% ADF - again at the mid temperature.