Coffee Data Science
The Coffee Bean is Not Homogenous: Sifted Salami Espresso
Measuring extraction rate across grind size diving into the bean
The biggest hinderance in understanding espresso is an understanding of the coffee bean. Typically, we assume the coffee bean is homogenous in taste and extraction potential. When I started developing the staccato shot, I had remarked that coffee using only large or smaller particles felt like it lost some taste. I didn’t know why aside from the potential that homogeneity was incorrect.
In this article, I study extraction yield across sifted particle size, and I did extra. I split out the inside fines from the outside so I could measure extraction yield for both. This should help explain why an inside-out staccato espresso performs better than a regular staccato shot.
Background
A year ago, I studied extraction rate across grind size, and that study greatly informed me about the limits of extraction. Along with some other studies, I was inspired to study the same thing across sifted particle size. Particles don’t extract like you would think they do because the coffee bean is not homogenous.
Around the same time last year, I had found that the inside of the bean was different from the outside. One could sift a coarse grind resulting with the inside fines. Taste tests confirmed these fines were different from those produced by grinding the boulders finer.
In fall of 2021, I worked on the inside-out staccato shot as well as the lazy staccato shot as a way to utilize inside fines. Both shots were interesting, but of course, more processing steps were added. It was unclear exactly why separating fines would help aside from a nebulous statement like “fines from the same part of the bean (at the same particle size) should extract more evenly.”
Equipment/Technique
Espresso Machine: Decent Espresso Machine
Coffee Grinder: Niche Zero
Coffee: Home Roasted Coffee, medium (First Crack + 1 Minute)
Shot Preparation: Staccato Tamped
Pre-infusion: None
Infusion: Constant flow
Filter Basket: 7g VST
Other Equipment: Atago TDS Meter, Acaia Pyxis Scale, Kruve Sifter, Fellow Shimmy
Design of Experiment
For this test to be successful, other variables needed to be well controlled, namely solubles and CO2. Solubles and CO2 gas released during extraction influences extraction. To isolate those variables, I made a lot of spent coffee grounds. They were clumpy at the end, so I used a coffee sifter to remove all the clumps, and I remixed both the fine and coarse grounds after sifting.
Each shot would be mostly spent coffee grounds with around 11% sifted coffee. I also used a 7g VST basket so that I mixed 1.5g of fresh coffee with 4g of spent coffee grounds. On top of this layer was 8.5g of spent coffee, so that 14g of coffee was in the basket. I topped this off with a metal mesh screen.
Then I used a constant flow (4 ml/s) profile on the Decent Espresso machine as a benchmark for all the shots. This ignored other optimizations like Pre-infusion, pressure profiling, pressure pulsing, and blooming, but those would be interesting to study at a later time.
Metric of Performance
I use extraction yield based on a refractometer to evaluate performance. 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).
Data
Here is the route I took to produce the different particles using two grind settings:
Just to note, I used the Fellow Shimmy to filter out < 300um because it is faster, and then I used the 500um Kruve screen. The Shimmy is rated for 200um sifting, but because it uses square holes, it behaves very close to the Kruve 300um screen.
I ended up with this distribution.
I then pulled some salami shots using a sample from each bin.
All the shots including the control shot had a 20 second shot time, and all the split times were within 1 second of each other. This indicates that the fresh grounds did not impact flow, which was the intent of this investigation.
First, I pulled a control sample to measure how much is being extracted from the spent grounds. While the value is not high, it does end up affecting the measured TDS for the other tests. I was aiming for spacing from 0.5 to 3 in 0.5 intervals on the output ratio, but the shots ran so fast that it was difficult to manage.
For each salami shot, I adjusted the EY by the control which is 2 to 3%.
Let’s take a look at the final result. Here we can see the inside fines extract much faster than the outside. This shouldn’t be surprising if we assume the inside of the bean is more brittle than the outside of the bean.
In looking at how quickly the different sizes completely extract, the finer layers extract almost immediately. By the 1.5 ratio, they are mostly extracted. However, the coarser particles take much longer. This could explain why an Allongé needs a 3:1 or longer ratio to hit the highest extraction yields.
We can use this information to make a theoretical puck if the previous sifting distribution was in a puck with the same profile. This theoretical closely matches the mid-range particles [300um to 500um] from the outside grinds.
I compared it to two shots (regular and staccato), which used my regular profile. Their EY trends were higher which merely suggests using pre-infusion, bloom, or pressure pulsing helps improve over the theoretical.
Another View on the Data
I wanted to change how I was viewing the data, so I did a line graph across the sifted particle bins for each shot ratio. They’re roughly the same shot ratios, and I like the way this data comes out. I’m still working on the next steps of what I will do with this new information.
This data shows how coffee is not homogenous, and it helps give a baseline for how extraction develops over time. This experiment could easily be repeated for different temperature settings, pressure settings, and other considerations, and it could be used to help better characterize coffee. Hopefully, the methodology could prove useful in developing more advanced espresso shot profiles.
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.
