Distribution Test Results

April 05, 2015

Testing distribution methods for espresso is turning out be quite a journey. This post is another step toward understanding what’s actually going on between the grinder and espresso machine.

If you’re new here: three weeks ago I sent out a survey about distribution methods. The results and discussion are here. This week I've set out the post as a pseudo-scientific paper. It's not strictly formatted or written as such, because I wanted it to be easily read and understood by everyone. That said, I think it's important to at least loosely structure my experimental posts this way.

 

Abstract

The test I ran this week might seem a bit weird at first, but I’m certain that it is an elegant and meaningful metric. Essentially, the test measures a method's ability to distribute a consistent mass of coffee into a fixed volume of the basket.

 

Introduction

As I mentioned in that previous post, distribution methods should be evaluated by their quality of extraction, consistency, speed and cleanliness. It would be pretty difficult and time consuming to test all of these things in one fell swoop. So I’ve chosen to break it down, and run a series of tests that will (hopefully!) help us understand what’s going on from different perspectives. At its most fundamental level, a distribution method is trying to evenly arrange the coffee grinds in the basket. It should result in a consistent density of grinds from side to side and top to bottom. Working towards a "maximum limit" like this is great for testing because there’s no such thing as “too even”. Once we understand this, then it becomes much easier to design a test that evaluates evenness.

Materials And Methods

The test I designed is quite simple, and doesn’t even involve brewing the coffee. It is purely a test of a method's consistency and the evenness of grinds within the basket. Here’s how it went down: I chose three methods that represent the overarching schools of thought in espresso distribution. I did this because many methods are actually quite similar, and I needed to save time! 1) Just tamp it. - Grind directly into the basket, aiming to distribute the grinds as evenly as possible. - Tap vertically on the grinder forks twice. - Tamp. 2) Manipulate the top layer of grinds with your hand/palm/tool. (For this I chose the Stockfleth method as I’m most comfortable with it.) - Grind directly into the basket, aiming to distribute the grinds as evenly as possible. - Tap vertically on the grinder forks twice. - Move the grinds around with ones finger until the top surface is perfectly smooth and even. - Tamp 3) Tap vertically and horizontally. - Grind directly into the basket, aiming to distribute the grinds as evenly as possible. - Tap vertically on the grinder forks twice. - Tap the side of the portafilter with one’s palm until the grinds are perfectly levelled. - Tamp For each distribution method, I used a little bit too much coffee (0.5-1g above my target of 20.8g) and distributed it in the basket as required. I then tamped very firmly until the grinds stopped compressing. I used much more pressure than I normally would on the bar, because I wanted to remove tamp-related density of grinds as a variable. Here’s a quick chart explaining why.

Tamping Pressure

As you can see, small changes in pressure make a massive difference to the density of the grounds when you’re tamping. This effect is lessened with a medium or ‘standard’ tamp pressure and is almost negligible when tamping very hard. At a certain point, no matter how hard you tamp, the density won't change at all. This is why I chose to tamp so hard - so it could be eliminated as a variable affecting the density of the grounds. Here’s where things get a bit weird, so stay with me. I wanted to measure the how well a method can distribute grinds within the basket. To do this, I needed to isolate a very consistent volume of the basket and weigh it. The consistency of coffee mass within that volume of space should be an indicator of how evenly the coffee is distributed. As I mentioned above, I distributed and tamped 0.5-1g extra coffee in each handle. This left me with some excess grinds in the basket. To achieve that consistent volume of grinds, I used a precise, flat and sharp metal scraper. The depth that the tool scrapes down to is based on the lip of the basket. This means It’s perfectly flat and impossible to screw up. Spinning that tool around inside the basket resulted in 0.5-1mm of grinds being neatly shaved off the top of the bed. Then, finally, I could weigh that volume of grinds and know how much there was.

 

Scraper

That excess coffee doesn't even represent 5% of the total mass, so I'm happy with removing it to satisfy the rest of the experiment. This simple experiment means I could quickly and easily check the weight of a reasonable number of handles to determine consistency.

Results

Here’s the results from 15 samples of each of the three distribution styles: Tap, Stockfleth and Tamp.

Distribution Graph

From these results, it’s pretty obvious that tapping side to side and vertically (purple, if it wasn't obvious) is superior in terms of consistency. It’s also obvious that manually manipulating the top layer of grinds (stockflething: red) is an improvement on just tamping (cyan) whatever you get out of the grinder. Let’s play with some numbers to gain some more understanding. I’ll calculate the median (central tendency, or middle of the road) of each method, and then figure out its consistency from that spot. My favourite tool for measuring consistency is called the “Standard Deviation”. Simply put, Standard Deviation tells you how spread out the results are, or how far away they are from that “middle of the road". In this case, more spread out = more inconsistent. [Dear Math Nerds, I chose not to use Variance or the slightly more accurate Average Absolute Deviation because that wouldn’t display the results in the same units - grams - as the test was conducted in, and would likely confuse matters.] Distribution Spreadsheet

Discussion

I think it’s fair to say that if the weight is very similar across a large number of handles, then that method is consistently distributing grinds around the basket. If the weight is inconsistent then there are areas of lower/higher density that would promote/retard water flow. So inconsistency isn’t just an indicator of inconsistency itself. It’s also telling you that there are areas of unevenness. From that, I think it’s more than fair to assume that a consistent density would promote the most even flow of water through the entire bed of grounds, creating the most even extraction. This test was only conducted with 45 samples. More would of course make it more accurate. I am satisfied that there is a reasonable difference between the methods - if there wasn't I would have collected more samples to make sure. The median column also adds an interesting element to the experiment. Tapping’s median is 0.2g higher than the others. I can’t profess to fully understand what this means, but I’ll give it a crack: If we think of the maximum amount of coffee that can be compressed into our consistent volume of basket, then it would also by default be perfectly distributed. Hitting maximum would mean there would be no areas with less coffee, because pockets of lower density implies uneven distribution. It might then be fair to say that we should be chasing a distribution method that (within the realm of this experiment) can fit the most coffee into that space the most often. The trouble with this is that the first 3 samples of just tamping yielded a higher density of coffee within that space. This leads me to believe that there could be a more superior distribution method lurking around the corner. Just tamping achieved that higher density a couple times, but can’t achieve it consistently. There’s always something better out there.

 

Conclusion

Acknowledging the small size of the sample, tapping appears to be the most consistent method. Stockflething is a reasonable improvement from just tamping. This by no means declares tapping the ultimate distribution winner. It just adds a tally next to Tapping on the distribution scoreboard in a game that will be played for quite some time. I'll be continuing to devise experiments and ideas about distribution and will be inserting them in amongst my regular posts to keep things fresh and interesting! Any ideas on another good experiment? Let me know below!

 

If you have found this useful and want to enjoy delicious coffee with the rest of the community – register for our monthly Superlatives coffee subscription. Or if you just want to keep up with every thing Barista Hustle - sign up to the Newsletter.