Several months ago, I posted a few brief comments here on the topic of maintaining espresso flow and quality throughout the day in a commercial environment as espresso grinder burr temperatures heat up, and shot flow changes.
I linked to related posts that had come out shortly before by Maxwell on Colonna and Smalls blog (particularly this post), which offered a new explanation for precisely why flow changes, which resonated with me, and which correlated well with my own observations.
This is something that I’ve always wondered about since first becoming aware of the effect, and something I’ve been looking at more closely than ever (and approaching from a different way) for over a year now, since late 2014.
Most of my experimentation and diagnosis on this has simply involved my standard close observation of general practices (shot making, dosing, flow, visual characteristics of crema, tasting, flavour, grinder calibration and settings, etc), along with regular measurement of all the weights wherever necessary as another reference point, and detailed logging of these findings in daily brewing journals.
Basically just employing the standard methods I always utilise, to record and digest the effects of new practices.
But recently, I decided I’d like to add some laser particle sizing analysis to the mix, to both support and enrich my own findings, and to substantiate from my own environment and equipment the theories and explanations conveyed by Maxwell on the blog, which stemmed from the data they were given by grinder manufacturers.
I’ve had the privilege to access laser analysis quite regularly in the past (thanks to the wonderful Mat Smith!). Previously, my focus with this was to use it to look at various coarse, filter grind profiles from several grinders I use, and I have not used it for espresso grounds, until now.
For those unfamiliar with the phenomenon in question, what happens in commercial environments, with conventional espresso grinders and burrs (certainly flat burrs), is that as the day progresses, and burrs heat up from use, espresso flow rate increases (generally, overall). This can require one to make lots of adjustments (of one kind or another) in order to atempt to keep the coffee consistently good.
In this environment, the condition of the espresso (and the grounds) is in continuous flux, and one needs to be constantly vigilant, monitering parameters, and frequently calibrating where necessary, to try to maintain a consistency of quality from a moving target, otherwise shot quality can deteriorate rapidly.
(This of course makes for a completely different, more dynamic and fluid scenario for making and maintaining espresso than is normal in a home or even a usual roastery environment, where burr temperatures remain more consistent for low volume or occasional shot making, and where it’s simply a case of initial dialling.)
But surprisingly (or perhaps not, as genuine, concrete substance is often rare), there was, and is, very little detail available anywhere online about why this actually happens.
In fact, no one (publicly) seemed to know why for sure, even the top coffee professionals seemed to only be able to give vague hypotheses.
Virtually no conclusive evidence or even detailed speculation, that I’m aware of, about what, specifically, causes shots to speed up in this way.
Just some thoughts mentioned here and there that as burrs heat up they actually move further apart and gradually grind fractionally coarser, or that warmer grounds extract and flow more quickly, amongst a few other things.
(For those using timed dosing, there can be some related effects in relation to this too).
It seems ludicrous that no one knew what causes this, until now (or at least there was nothing publicly available about it).
You would imagine that the reason for such a significant effect that is so familiar throughout the coffee world would already be understood, and explained with matter-of-fact confidence.
So, I wanted to see whether I could corroborate and confirm this newly suggested reason for the effect, arrived at from the graph data from the grinder manufacturers that was referred to.
Maxwell has mentioned on their blog that he hopes to publish data on this soon, and doubtless when this happens he and the (incredibly) specialised team involved will be able to present a far more thorough and insightful examination of this topic than I am capable of.
But, at present, the graphs referred to in relation to this theory have not been released. So, by posting the graphs from my own samples and laser sizing analysis here, it’s exciting to be able help with progressing knowledge forward a little regarding this topic that cafe baristas encounter and deal with every day.
Nice to release some data, publicly, to help substantiate the theory, and further support its accuracy, and to provide another reference point for anyone researching the topic.
The samples below have all been run on the same volume based analysis model. As Maxwell points out, if you run the same sample using different models, you can see strikingly different representations of any particular sample (and I might run the same samples, or new ones, via different models sometime). And as I’ve mentioned here in the past, the whole topic of the way particles are measured by this sort of equipment is incredibly complicated/scientific. The data can be interpreted or represented in many different ways, is often not as straightforward or conclusive as you might imagine, and the results often lead to as many questions as answers.
However, regardless of all this, you will see that the model selected here amply demonstrates in the graphs the effect on super fine particles, and on the overall shape of the distribution curve, of different burr/grinding temperatures.
And it does so far more dramatically and conclusively than I had expected.
All on a modified Anfim Super Caimano 2nd revision with steel burrs, dual time presets.
All samples taken at the same grind setting.
All using the same coffee, a washed, espresso profile blend from James’ Gourmet Coffee (composed of SukeQuto/Antonio/Paraiso/Potrero), across two batches with the exact same components (which performed identically, and which were rested for very similar periods), taken between 25/11/15 and 12/12/15.
Start of day.
Cold burrs (after loading the clean grinder with beans, and grinding a few doses to suitably prime the throat and other areas without excessive grinding).
Ambient temp, 16℃.
Towards end of same day.
Ambient temp 21℃.
End of day.
Particularly warm/hot burrs.
I had another sample tested (Sample 3), the graph for which I don’t have here, taken at the end of the day, testing the effects of selecting a finer grind setting, which demonstrated the same pattern in fines reduction with warmer burrs.
The graphs show, amongst other things, this key feature in question, of a vast difference in the amount of fines between cold and hot samples, with the fines reducing, and the grind particle distribution becoming increasingly uniform, as the burrs heat up, which seems to largely and convincingly explain the primary reason for alteration in flow.
A curious feature, which I hope to look at further, is the variation in the peak size/s across the 3 samples shown, despite all being at the same grind setting. With warm sample 2 peaking at a smaller/finer micron size than cold sample 1, and warm/hot sample 4 peaking coarser. Which is inconsistent, and which warrants deeper investigation (or maybe simply a better understanding of the graphs than I have!).
What’s interesting is how we choose to manage the phenomenon, generally, and, now that we think we do know what causes this change in flow, whether people might choose to manage it differently in the light of this new information.
Similarly, there is not a great deal of specific detailed substance published about the precise methods and practices used by people to counteract and manage this effect.
There are many generalised references to the commonly adopted approach of gradually grinding finer throughout the day…
And, of course, the idea of rigidly following set recipes and the brew ratios which we discover when initially dialling in (or as provided by a roaster’s recipe) is often cited as standard practice.
For reasons I mentioned in my previous post, the commonplace practice of grinding finer and finer throughout the day (and that of ostensibly adhering to sacrosanct fixed brew recipes), is potentially called into question by this new data, given that, in the light of this information, at the same parameters/recipe later in the day, at a finer grind (as is common practice), the shots will not actually be quite the same as they were (and might in fact be very different), because the distribution curve of the particles in the shot will be a different shape (regardless of whether the grind setting is changed or not), and because the mode particle size might have been significantly altered away from where it was initially…
If the grind had been getting coarser in a simple way, as some people thought, and we were just recalibrating it, and the distribution curve, back to where it had been when cooler, going finer would make perfect sense. But we know now that this is not the case.
Although, in a certain sense, to be fair to those who thought the grind got coarser, it kind of does of course. One grind profile with less fines at an approximately similar mode size than another with more fines is sort of coarser, with a higher proportion of, or balance towards, the larger particles, relatively. But it is not coarser in the traditional sense, the burrs have not (we think) moved further apart (or even if there is movement that certainly does not seem to be the primary cause of the effect on flow at least), and the mode size has not become (significantly enough, or at all) larger, and this therefore presents us with a very different scenario.
Anyway, my results concur with the theories presented by CS originally, and the analysis they had access to.
But, when I say we now finally know what causes this effect, that’s not entirely true.
Yes, we now know quite definitively that flow increases in this way because of decreasing levels of fines in the 0-100 micron range. But, we don’t yet know what actually causes this decrease to happen.
Whether this is due to the hotter burrs actually grinding differently (with less fines) for some reason (changes to the metal caused by heat), or because the warmer beans as heated by the hot burrs (above and beyond the effect of any increase in ambient temperatures), whilst directly above and feeding into the burrs, break and fracture differently (with less fines), or to what extent ambient temperatures and humidity have a contributory effect in this, is, I think, still undecided. I think it would take someone a lot of time and well structured, in depth testing to ascertain this.
But nonetheless, the important (revelatory?) takeaway from this for the time being is this alteration in the distribution curve and specifically the levels of fines as shown in these graphs posted here; this evidence that there are less super-fines when the grinder burrs (or beans) are running hotter compared to when cold. A lot less. This alone gives us a lot to think about.