📚 Water Contact Time and Extraction in Espresso: an Experiment
💡 Newskategorie: AI Nachrichten
🔗 Quelle: towardsdatascience.com
Coffee Data Science
A short experiment to isolate variables
Coffee is a marvelous beverage that, like tea, requires some amount of steep time to extract the goodness. This steep time or water contact time needs to be balanced with the flavors drawn out of the coffee. In espresso, this is particularly challenging because water is flowing through the coffee at pressure in a shorter steep time than other brew methods.
There is a theory that a higher water contact time will cause a higher extraction in espresso. Some have argued against this notion, but like many things, this theory is testable. The key is a good Design of Experiment (DOE), so let’s design an experiment and collect some data to start answering the question.
Total Dissolved Solids (TDS) is measured using a refractometer, and this number combined with the output weight of the shot and the input weight of the coffee is used to determine the percentage of coffee extracted into the cup, called Extraction Yield (EY). This is the primary variable we are trying to measure.
DOE: Modifying Grind Size
The simplest experimental design is to modify grind size which will slow the flow. However, this conflates two other variables:
- The smaller grind size will extract faster at a local level.
- The smaller grind size is more susceptible to channeling causing under-extraction at a global level.
Between these two variables is the extraction yield volcano. So if you used this experiment, you could see a rising extraction yield as grind went smaller until some peak, after which extraction yield would drop. One may conclude from such an experiment that longer water contact time doesn’t lead to higher extraction, and the conclusion would be flawed.
So the experiment must be better and reduce variables.
DOE: Use Flow Control
With a flow control espresso machine (like the Decent Espresso (DE) that I have), one could use a single grind and do 3 different flow profiles. This would control for the extraction rate changing due to the particle distribution.
However, this could still be affected by the grouphead design. Many machines push the water to the edges as the water comes into the basket contributing to side channeling. The DE has a unique water dispersion due to the water dispenser which causes water to come in slightly faster on the left side than the right.
DOE: Flow Control + Mixing Spent Coffee
To reduce the water input variable, we can use some spent coffee. Spent coffee is not impacted by channeling as easily because it has no CO2 and little extractible solubles. So we can make a mix of a single coffee grind and use multiple flow profiles.
An additional improvement is to pull a salami shot so that the extraction rate can be better understood.
Spent Coffee for Testing
For this test, I made some spent coffee as I have for previous tests. However, I thought I could use the opportunity to get a little data before the larger experiment. The grounds came from multiple spent pucks, and I mixed them together thoroughly. Then I aimed to pull a 2:1 shot split into two cups. However, the first shot ran a little longer, so I hit the same ratio for the other two.
This result showed a higher flow rate reduced EY, and generally supports the notion that more water contact time is better (i.e. a slower flow rate).
I then ran a 4 ml/s 105C profile on these pucks until the liquid was clear. I dried out the pucks, and I put them through a 800um Kruve sieve to remove any clumps.
So let’s get to the big experiment.
The Big Experiment
This experiment was done using the Decent Espresso machine with 3 profiles of flat flow rate control. I pull these shots at 90C because I wanted to avoid too much steam as the effects of steam on coffee are still being discovered.
I then mixed freshly ground coffee with spent coffee and a simple WDT plus auto-leveling tamping. 8g of fresh coffee went with 13g of spent coffee.
I pulled a control shot on the spent coffee at 1 ml/s to remove the impact of the spent coffee from the other measurements.
Extractions were then done using 5 cups for the three flow profiles. For TDS, the trend of higher TDS continued for each sample even though the samples for 4 ml/s were a little longer. This is normalized by EY.
The 4 ml/s flow as hard to get the same output weights. I also removed the influence of the spent coffee grounds from these cumulative EY numbers.
These results show that as the flow increases, the extraction yield decreases. This holds true throughout all of the shot ratios, and this supports the idea that longer water contact times extract more coffee.
Coffee extraction in espresso is challenging as there are so many variables influencing each other. I hope this experiment helps shed light on how a single variable can be isolated and examined. This experiment shows that longer water contact time causes higher extraction, and previous thinking about contact time was conflated with the effects of channeling in finer ground espresso.
This experiment could also be repeated using only fresh coffee, and it would be an interesting experiment to conduct. I’m not sure how other variables might conflate the measurement.
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:
Collection of Espresso Articles
A Collection of Work and School Stories
Water Contact Time and Extraction in Espresso: an Experiment was originally published in Towards Data Science on Medium, where people are continuing the conversation by highlighting and responding to this story.
...