Coffee Data Science
Thermal Pre-Infusion: A Paradigm Shift in Understanding Espresso
I have been looking at espresso from a different lens the past few months. This all started because my Decent Espresso (DE) machine greatly underperformed my Kim Express machine. Just let me to eventually creating the pump and dump profile for the DE, which evened out the competition. I was satisfied with this until I started testing on the Meticulous Home (Met). The particular prototype I had came with a broken heat coil unknown to me or to them. This prompted another look at heating.
So I ended up doing an experiment where I wanted to separate water flow and heat movement through the puck. I separated flow and heat by locking in the puck into the machine and letting it soak in the heat for 30 minutes. Then I pulled the shot using a flat profile. I was shocked to see how even the shot looked and pulled.
This is Thermal Pre-Infusion (TPI).
I have previously found that the thermal effects of hot water going into cold coffee, had influence side channeling. This was particular interesting with the Met because since the heater was broken, I had to do everything I could to heat the coffee without losing too much temperature.
Theory: thermal side channeling precedes water flow channeling. By starting with even thermal temperature closer to the brew water temperature, water flow will be more even and extract more efficiently.
Coffee grounds are usually around 22C in my house, and if I use water that is 100C, the water will quickly drop in temperature. The heat profile for the coffee on the top and bottom will probably be different as a result.
Another group did a temperature investigation, and their temperature data was useful in understanding how quickly the coffee at the top and bottom of the puck heat during the shot. From this perspective, the bottom of the puck won’t get up to the temperature at the top of the puck by the end of the shot (35 seconds), and that means the top will have a much higher extraction potential as a result. This conclusion is matched by other extraction data I have which showed the top of the puck was almost completely extracted vs the bottom.
When I started pulling espresso on my Kim Express, I didn’t track any data. Even when I did start recording data, I didn’t record when I locked in the portafilter. It’s quite possible one of the early benefits in my shots was locking it in a minute or more before the shot.
The biggest advantage of the Kim over the DE was steam pre-infusion and higher brew temperatures. I didn’t know exactly how it worked, but I knew coffee was gaining some moisture or heat very quickly throughout the puck.
I couldn’t quite replicate this on the DE, but I got close with the Pump & Dump. The profile started with some steam and it took close to 4 minutes for a shot. The latest iteration was 90 seconds of nothing. So what I thought was trying to emulate steam pre-infusion was also doing thermal pre-infusion.
Additionally, I look back to staccato espresso because it has more even water flow by design, and I suspect it also improved heat flow through the puck. I’m curious how heat moves through different sized particles or layers with more or less density.
One other area to reflect on is microwaving coffee beans. This was an experiment many tried a few years ago. It seemed to improve extraction yield and had some effects on taste. I theorized it was due to higher ground temperature as some of my data showed my afternoon shots were slightly better than my morning shots, which I suspected was from a slightly warmer climate as the day wore on.
The beauty of this profile is how easily is can be applied across espresso machines. It is easier to do with a boiler, but one could find other ways to pre-heat the coffee grounds. I don’t recommend a microwave.
- Prepare your puck as usual.
- Lock the portafilter into the machine.
- Wait 5 minutes
- Pull the shot.
- Aim for a lower output closer to 1:1 because you risk over-extraction if you go higher.
I use a bloom ramp profile where I move on to the bloom phase when the bottom of the filter is covered.
I would also suggest exploring temperatures and adjusting output yield. I have had consistently higher extraction yields so I have had to do a 1:1 otherwise risk over-extraction.
Coffee: 4 month old coffee from Chromatic
Shot Preparation: Staccato Tamped
Profile: Flat 1 ml/s without pressure restriction
Filter Basket: 20 Wafo Spirit
Metrics of Performance
I used two sets of metrics for evaluating the differences between techniques: Final Score and Coffee Extraction. I focus here strictly on extraction metrics while acknowledging the importance of taste. The most exciting part of this discovery is that it allows a drop in water temperature to dial in taste that wasn’t possible without greatly affecting flow.
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).
I used a flat rate 1 ml/s profile for these tests with no pressure limit. This is not the ideal profile, but it is constant and easy to understand. I then pulled the shot as a three part salami shot to understand extraction trends (1:1, 1.5:1, and 2:1). All shots used the same coffee that was four months off roast at the same grind setting, dosed at 22g. For the soak tests, the portafilter was locked in for 30 minutes to assure the temperature was as homogeneous as possible. I later found that temperature equalizes pretty quick, within a few minutes.
This first graph is too much data, so let’s simplify.
I split this graph into regular temperature and after 30 minutes of thermal soaking. Thermal soaking is the winner.
Let’s better understand the trend using a cumulative bar chart. Pay attention to the x-axis starting at 18% and not 0%. From here, we can observe an anomaly at 100c where it doesn’t follow the trend. However, we can also see that soaking the puck at 60C outperforms 90C in the 1:1 ratio (First 22g).
Another way is to look across the trends, and temperature soaked (dotted lines) always outperforms in EY.
In peak pressure, temperature soaking has resulted in lower pressure. I’m not sure why this happens, but it was a curious trend that I have also seen with the Pump & Dump profile. As a result, little crema is formed.
The first criticism will be “how does it taste?” This method has improved the taste of my shots, which can be explained for three reasons:
- Higher extraction efficiency
- Lower brew temperature (I dropped from 105C to 80C)
- Less side channeling
I didn’t want to do a taste comparison until I learned more about the temperature drop.
I wanted to share this sooner rather than later because it was such an easy method to implement. More testing needs to be done to more deeply understand why this works, does the tasty point have a specific temperature, and how that temperature could potentially differ for different roasts or across roast age.
I’m currently exploring 80C with similar or higher extraction yield, which leads me to believe more will be revealed.
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.