Coffee Data Science
Espresso Filter Basket Hole Changes After 2,000 Shots
Comparing old and new data
I have been examining filter baskets for a few years, and as a result, I have some data on my basket when I originally bought it. I thought it would be interesting to see how the basket has changed and if I can measure that change. In two years, I pulled roughly 2,000 shots with the VST basket.
I would expect the holes would wear over time. I am interested to look for general changes as well as specific ones due to flow. If donuting happens consistently, does that cause uneven wear? Is wear measurable? Does the wear affect the shot? I don’t know the answers, and this work is the start of trying to answer these questions.
Data Issues
The biggest challenge is that my camera has changed, and the way I have taken images has also improved. So I have to do some alignment for the distributions. I have images from my VST 20g basket from 2020 and 2022.
I had some data from the top and bottom of the filter basket, but the issue was that I didn’t know how that aligned to the filter holes currently as I didn’t record which the filter orientation when I took the image. This is something I should do now to make sure I can look at this data in the future.
So I manually aligned images from 2020 on the six different orientations that it could be. It was amazing how well the filter rotated to where you could not visually see which angle had the best alignment. I calculated the average error between them to come up with this:
A 120 rotation seemed like the best, so I used that as my baseline.
Hole Distributions
There is a shift in distribution between the top and bottom of the basket. I don’t have a good reasoning for the dip in the middle. The images of the filters with false color to signify hole size didn’t show a bias that indicates an angle with the camera. I am continuing some investigation into this difference.
Now we can compare the top and bottom distributions over two years. The bottom view still has this weird dip for both years.
For the top, there is a slight tail towards larger holes. I’m still not sure if that’s noise, capture error, or real. For the bottom, there is a slight change for the smaller holes and the spike on the right side showing some increase in hole size.
From the false color images, there are big shifts, but there doesn’t seem to be something systematic. Visually, I don’t see a pattern in the shift.
This analysis isn’t quite complete. Better data is needed especially from a coffee shop to truly understand how these baskets change because a commercial machine goes through far more shots in a shorter time than I do. Additionally, many variables of my process has changed like the machine, the preparation, and the shot profile.
This sort of investigation along with shot performance over the time of a filter basket can help understanding the life expectancy of a filter basket.
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.
