Coffee Data Science
Constructing Coffee Brew Water One Mineral at a Time
Another try at producing a good water recipe
Developing a brew water recipe is very challenging because there are so many variables at play. I previously experimented with a brew water recipe and diluted it until I had a recipe that was good. I was particularly interested in how I could use a recipe of drops to add to coffee after the shot rather than modifying the brew tank recipe (on an espresso machine). I think this allowed me to iterate faster. Now, I wanted to try again, but this time, optimize more to each mineral.
For each mineral, I made a solution, and I added one drop (0.03g) to each shot. Then I would dilute this solution and hopefully find a peak taste. I did this for all the minerals, and then I started combining minerals. I used my filtered brew water as a basis rather than distilled because I wanted an improvement from my base brew water.
I evaluated these shots based on taste. Brew water is specific to each person which makes it more challenging because whatever I prefer may not be optimal for you. I used Lotus drops to construct the brew water, which were straightforward to use, but I weighed the drops rather than counting drops.
Here is how I evaluate taste using a Final Score:
Final score is the average of a scorecard of 7 metrics (Sharp, Rich, Syrup, Sweet, Sour, Bitter, and Aftertaste). These scores were subjective, of course, but they were calibrated to my tastes and helped me improve my shots. There is some variation in the scores. My aim was to be consistent for each metric, but some times the granularity was difficult.
I evaluated these shots but had to use a few roasts over a few weeks which led to the introduction of other variables in the process. I’m not opposed to repeating this experiment with more control; this is what I produced from this set of experiments.
Individual Minerals
I started with potassium (K) because it was my favorite in previous tests. I was surprised I didn’t quite find a peak performance, but maybe I didn’t test enough.
Then came Magnesium (Mg), but I generally didn’t like Mg or Calcium (Ca). I did want to try again just in case there was some interaction with Mg or Ca that was important for taste.
The ratio numbers are a bit all over the place because I weighted water as I added it, and I planned on calculating the ratio after the fact.
Sodium did show an improvement at 29:1, and that was sort of expected because I love salt.
I was also surprised to find a level of calcium (Ca) that was enjoyable, but it was just over the baseline.
Mixing Results
I started with two mixes to separate hardness (Mg and Ca) and alkalinity (K and Na). I used the best individual dilution levels as a starting basis.
Alkalinity found an improvement, but this could have been only from Na. I’m not sure, and I would have to try making the final mixture with just Na.
For hardness, this mix showed improvement even with dilution.
Then I combined the best of these two mixes, and I repeated the recipe a second time for some of the tests.
The initial results looked pretty good, but the taste improvement was not a homerun. For only a one point improvement, I’m not sure I would add the hassle of a specific brew water.
Let’s take another look at the data using some line graphs instead of stacked bar charts:
The fun part about data is that there are always different ways to look at it.
While I succeeded at another brew recipe, I feel like I failed to find some holy grail of brew water where the taste became super magical. I will have to search again, and maybe part of the answer is that my previous work on brew recipe showed a much more diluted recipe.
This data was not full enough to show if different coffees needed different brew water to be optimal. It is quite possible.
Other things I would like to try:
- Mix hardness and alkalinity differently: KMg, KCa, NaMg, NaCa
- Repeat some of these tests over more ratios and more coffees.
If you like, follow me on Twitter, YouTube, and Instagram where I post videos of espresso shots on different machines and espresso related stuff. You can also find me on LinkedIn. You can also follow me on Medium and Subscribe.
Further readings of mine:
My Second Book: Advanced Espresso
My First Book: Engineering Better Espresso
