You asked us: ‘Can you design an experiment to test whether fines migration actually occurs when you distribute your coffee by tapping?’ Here’s what we did.
In a previous post, we looked in detail at the theories around the idea of fines migration. We looked separately at the way particles move in dry coffee during puck preparation, and at the way they move while the shot is running, carried by the flow of water.
While particles of different sizes do often separate themselves out in granular materials, we didn’t find much evidence for the idea that fines migrate in dry coffee. Static forces bind fines strongly to larger particles, limiting their movement (Kuhn et al., 2017), which is why sieving out fines is not always effective. The one published experiment we could find on a related subject, from Socratic Coffee (2016), found that layering particles of different sizes in a filter basket had no effect on shot time.
Having said that, we weren’t able to conclusively rule it out, either. The particle sizes being considered in Socratic’s experiments were much larger than what would normally be considered fines, so it wasn’t looking at these particles directly. And while sieving might not be effective in removing a large proportion of fines, we considered it possible that the movement of even a small proportion of fines would be enough to affect the way an espresso runs.
The simplest possible experiment does show that at least some small particles are mobile: just tapping a filter basket on a sheet of white paper can be enough to release a few small particles. However the holes in VST baskets are approximately 300 μm in diameter (RM Aloe, 2019), rather larger than fines which are typically considered to be under 100 μm, and it’s not clear that this sieving-like effect has any relationship to what happens in the rest of the puck, or if it has any effect on the espresso itself.
To try and shed a bit more light on this, we set up a simple experiment to see if tapping a bed of coffee had an impact on the flow rate of espressos made with different parts of the coffee bed.
We chose to use the indirect measurement of shot time, rather than analysing how tapping affected the particle size distribution directly, as we know from other experiments that any changes in the particle size distribution big enough for us to measure are linked to huge variations in shot time, much bigger than we’d expect to see in this experiment.
We dialed in an espresso grinder, then ground 100g of coffee into a container and stirred thoroughly, in a similar manner to the Weiss Distribution Technique, using a thin metal tool, to evenly distribute any clumps or variations in grinding. We used this coffee to fill two steel containers, with the same approximate diameter as a filter basket (58mm), right to the brim. A metal straight edge was used to shave off any excess coffee to ensure neither container was overfilled. We then placed a card over one of the containers, and inverted it on top of the other, then removed the card from between them.
With the two baskets firmly held together, the coffee was then subjected to one of three treatments: tapped horizontally 100 times (with the taps performed in a similar way to Barista Hustle’s recommendations); tapped firmly 100 times on the bench vertically; or not tapped. The order of treatments was randomised.
Following the treatment, the card divider was re-inserted to separate the coffee at the top from the coffee at the bottom. Each sample of coffee was poured into a separate container and shaken then stirred to break up any packing and compression that might have happened during tapping. We then poured 18g from the top and bottom container (again in a random order) into a brew basket, and pulled shots using a standardised protocol. Three sets of shots were pulled for each treatment.
The shot times for each pair of treatments were as follows:
|Treatment||Shot 1||Shot 2||Shot 3|
|Horizontal taps (top)||28||22||23|
|Horizontal taps (bottom)||26||25||24|
|Vertical taps (top)||28||27||24|
|Vertical taps (bottom)||28||28||23|
The overall shot times varied over the course of the experiment as the grinder heated up, so to allow a meaningful comparison/average, we compared the distance from average shot time for each pair.
|Treatment||Distance from pair mean||Average distance from mean|
|Horizontal taps (top)||+1||-1.5||-0.5||-0.33|
|Horizontal taps (bottom)||-1||+1.5||+0.5||+0.33|
|Vertical taps (top)||0||-0.5||+0.5||0|
|Vertical taps (bottom)||0||+0.5||-0.5||0|
If fines migration was significant, we would have expected to see slower shots from coffee in the bottom container. Looking at the data this way shows that there is very little overall difference between the top and bottom container in any of the treatments.
There did seem to be a bit more variability in the shot time between containers with the horizontal taps, but one wasn’t consistently slower than the other, and the average difference in shot time was only a third of a second, suggesting that even if a slight difference exists, it’s not enough to significantly affect shot time.
Bearing in mind the small differences involved, despite the fact that we were tapping 100 times, we’re happy to conclude that tapping as a distribution method doesn’t appear to have any affect on the flow rate in espresso, and therefore it’s unlikely that significant fines migration takes place in the dry coffee during puck preparation.
RM Aloe, 2019. Espresso Filters: An Analysis (Blog post). Available online at https://towardsdatascience.com/espresso-filters-an-analysis-7672899ce4c0
M Kuhn, S Lang, F Bezold, M Minceva, and H Briesen, 2017. Time-resolved extraction of caffeine and trigonelline from finely-ground espresso coffee with varying particle sizes and tamping pressures. Journal of Food Engineering doi:10.1016/j.jfoodeng.2017.03.002
Socratic Coffee, 2016. Exploring the Impact of Particles on Espresso Extraction (Blog post). Available online at http://socraticcoffee.com/2016/06/exploring-the-impact-of-particles-on-espresso-extraction/