Coffee DOE

I wasn't t really sure where to put this. I call "designs of experiments" a tool so I'm throwing it here. I'm about a quarter of the way through my 16 treatments. I'm trying to determine the best French press procedure for my morning cup. I make coffee every day so I figured why not learn something about it. I plan to run this one and then another follow up based on these results. For those that could even understand this here is the factor relationship diagram. It's kind of like a map showing the factors and the level settings for each treatment.

One pot a day, so maybe another 2 weeks. I'll let you guys know the results once I'm done, however I think coffee is a pretty personal thing. Hence why I'm running my own experiment! 😁

Find more posts tagged with

What About Tools?

Comments

wolfiegirl

I had me some book learnin' as a child. Now I feel like one of those apes who can ask for a banana in sign language......

torridice_498271

Ha, I do a lot of statistical experimenting for work, so this is just one for fun at home.

JasonWirth_68226

Can you explain the experiment?

The picture is a little hard to see. What are the factors?

Roast

Grind

Weight

Temp

Time

torridice_498271

The black part is called the design structure, those are the factors and yes you got them right. Roast, weight in grams, grind size, temp, and steep time. The red is called unit structure and helps define the error term. I am rating two cups per pot and called them repeats. My measurement is the typical attribute type stuff, aroma, body, bitterness/acidity, flavor, aftertaste. All on a 10 point scale.

Sometimes I laugh at myself when I hear people like kenji or Alton or the mythbusters talk about isolating one variable. I'm sure they have to be somewhat aware, but we can do a whole lot better in industry then learning about one variable at a time.

matthewmicahhall

How did you construct this chart? Is it a DOE-related add-in/extension for Excel or hand drawn?

torridice_498271

Since I have to make these all the time for work we use a custom macro in Excel.

lfmichaud

Your idea is interesting but from my experience it won't be conclusive because many of these factors are independents. The way that I found work best is that most of these variables are continuous and can be thought of as parabolic. (Grind size, temperature, infusion time) Going under or over the optimum value will decrease the quality of the end result. But changing one of the other variable will change that optimum.

Since we know that good coffee must aim to get about 20% extraction and that grind size, temperature and steep time are all correlated with extraction that will help you limit your solution space.

I'm no coffee professional but I like to tinker. My recommendation to you would be:

Try to put your test range in the 90-96C. I've rarely made good coffee outside of that range.
Consider different beans/roast independent.
The easiest way to get multiple data point is to take a sample every minute from you pot. Usually the best results will be somewhere between four and eight minutes.

Once you discovered the best time for that combinaison of grind/temperature/quantity you need to think what would improve it.

Too bitter-> Try coarser grind or lower temperature
Too acid -> Finer grind or higher temperature

This way after a few tuns you'll probably have a pretty good idea of your preferred recipe.

Good luck and keep us informed

torridice_498271

Really great information! And thanks for the link, I think this is going to help me in the follow up experiment. This initial fractional factorial would never find an optimum, it's just meant to find statistically significant factors and provide general direction. To find an optimum I'd run a central composite design which could account for any curvature in the effects or any factor interactions like you mentioned.

I didn't know these charts existed, this will help me get closer to the optimum faster and hone right in. Thanks for the tip on extraction.

Now I can't wait to get through this one to update my factors and levels in the next one.

lfmichaud

Happy to have helped a bit. There's some good information on the Wikipedia coffee extraction page. There's also a lot of good information in the ChefSteps coffee class (Free and Premium)The only thing I can add but you're probably conscious of is that you need to empty your French press when you've finished your steep because even if it's pressed down you're still infusing and that will give bitterness to your following cups.

torridice_498271

Thanks, I'll check those out. Yea I've noticed that happen if I leave it in there. I also notice a reduction in body on the second cup too.

eathyssen

I find things like coffee and eggs sooo very subjective -- not like you could order them rare/med-rare/ etc. I prefer my coffee cold-brewed at home, and the concentrate then diluted with hot water. Mellower, less acid but very full-bodied. And my eggs? Once over with all the whites firm BUT a runny yoke. Invite me for breakfast?