Coffee Data Science
Debugging Vaporexpress Profile Espresso Shots
Renaming steam pre-infusion profile
A few months ago, I was able to more pinpoint why a lever machine (Kim Express) outperforms a Decent Espresso (DE) machine, namely steam pre-infusion. I started exploring this variable more, and steam pre-infusion breaks a few things we normally associate with espresso like pressure. Most of my shots have had pressure below 2 bars but with good extraction. So I wanted to push the boundary and further develop the profile.
After 40 iterations, I renamed the profile. Vaporexpress pulls in a few words that describe the profile. Vapore is the word for steam in Italian. The profile is based on steam pre-infusion, which I discovered using the Kim Express. The first variant would run as fast as a turbo shot with the grind of a regular shot, hence the extra meaning on express. Current incarnations have slowed down slightly.
I looked at a few variables to better understand where to go with the profile because the profile is new. There is no other information on where or how to improve it nor is there information on how to troubleshoot it. I used the following metrics:
- 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). Intensity Radius (IR) is the radius when converting the TDS vs EY control chart to polar coordinates.
- Grounds TDS (gTDS) is measured by placing coffee grounds on a refractometer with some water to measure the amount of solubles left in the coffee. Higher extraction during espresso leads to a lower gTDS reading.
Data
This is a good sample shot. The flow goes to one side, and in another discussion, I root-caused it to the shower screen.
In looking at this puck, it is hard to see any dark spots to give an understanding of water flow inside the puck.
So I looked to grounds TDS (gTDS) to give more answers.
In looking at this shot, I pulled a sample from each side and the center of the puck. I cut each sample in top, middle, and bottom.
The right side had a clear amount amount of unextracted solubles, but the top of the puck was mostly extracted. I plotted this a few ways here:
I needed to better understand how to fix that issue, but I also wanted to better understand how steam was affecting the coffee because this had a big impact on temperature.
Handling Temperature
I dropped the temperature after the first two steps from 105C to 98C, but the temperature keeps rising in the mix. In this graph, all the temperature values are 8C higher than shown (due to a calibration offset). I suspect as new, colder water is add, that water flows down but the steam bubbles up. I’m not sure how else to explain the increase in temperature.
Inside this puck is even stranger. Typically, the top is more extracted than the middle, and the middle is more extracted than the bottom, but in one of my shots, measurements of gTDS from the center and the side show an increase. This is different than the shot above.
Only Steam
So I looked at just steam-preinfusion. I did this by using a flow of 0.2ml/s at the hottest water temperature. This turns into steam, but for how long this stays in a steam state is not known. I started with 10, 20, and 30 seconds. Water didn’t make it all the way through, and major channeling started once the water was about half way through the puck.
In looking at the center and side cores, it is interesting to see that the side always extracts on the top and mostly in the middle. The center seems to have a lot of trouble. This is partially due to the shower design and basket shape that tend towards side channeling. I even tried rotating the puck at 30 seconds to try to counter-act the slight unevenness between the right and left sides in terms of water flow, but it was challenging to accomplish and the channeling seemed to have already been done.
In terms of actual shot TDS/EY/IR, I didn’t get a clear indication of where to go with the profile.
I figured out what was going on by mistake.
Video Evidence
I decided to hack my Kompresso underneath my Decent Espresso machine. I wanted to find a way to get answers.
The first video had too deep of a puck so water didn’t flow quite well:
However, I used a paper towel in the second video, and that worked much better. I was able to see that steam pre-infusion turns into low pressure pre-infusion very quickly.
As a result of this video and these investigations, I changed my profile to do steam pre-infusion for 2 seconds and then a slow ramp.
I enjoy this profile, but it has been a trudging process. I know I will get to something great, but each time I find failure and dig deeper, I learn a little bit more.
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.
