Coffee Data Science
DF83v Coffee Grinder RPM Analysis
By: Robert Aloe and Aaron Criss
Aaron the Burr Collector (otherwise known as Aaron Criss) on the Home Espresso Aficionado Discord had been curious to test RPM in terms of particle distribution. I offered to produce the analysis for any samples he collected, and he sent me images based on my protocol. RPM does have an impact on both particle distribution and shape, so let’s see what we can make from the data.
The testing was done using JBC Roasters Nemba Burundi light roast beans. The grinder and the catch cup were cleaned before and between each test. The exact same RDT process with the exact same amount of water each time. Each grind test was with 20g of beans. All tests were done with a hot start and the same feed rate.
The imaging SOP helped with insuring clear and accurate images with minimal particulate clumping. Since the DF83V has an RPM range of 300–1,600 RPM, 5 tests were done at 300, 600, 900, 1200, and 1600 RPM.
The particle distribution shifted to the right as the RPM went up, which means that the same burr gap resulted in a coarser distribution with a higher RPM. 600 and 900 RPM seem to not follow the pattern exactly, but we didn’t dig further into those results.
Looking at the average particle size, you can see an upward trend as RPM goes up. There is probably a connection to the time spent inside the grinder since the higher RPM would put a larger force on particles to exit the burrs.
Looking at the percentage of fines, this steadily decreases, which makes sense since the average particle size is going up.
Then we looked at particle shape using LBP and K-means clustering. The result is being able to compare the same binning of particles created by each RPM. A potential theory is that particle size changes due to RPM but particle shape doesn’t which would mean that there would not be shape differences in same sized particles.
This shows a matrix showing the similarity for particles centered around 100 um, 200 um, etc. Some particle shapes show not much difference (yellow), but then there are particles like 500 um where each RPM is very different than each other. For the smaller particles, they are mostly the same in shape except for 1,600 RPM.
What is going on inside the burrs is very interesting but there is some remaining mystery. There is not much public data on how burr design, feed rate, and RPM affect the grind as well as the brew. These numbers also weren’t tie in taste metrics, which is ultimately the most important aspect of understanding grind settings.
It would be interesting to test RPM and replicate this shape experiment, but dial in each RPM setting to have as close of a match as possible in terms of particle distribution. While a higher RPM makes larger particles, the fix is to shorten the burr gap, and by doing that, do the particles better match in terms of shape and the brew in terms of taste?
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