In The Professional Barista's Handbook, Scott Rao makes an observation that I found fascinating:

I've often wondered why so many Italian baristi use dispensing temperatures in the 185-195F range, while many American baristi...use 198-204F. I think one part of the answer is most Italian baristi use 7 gram doses to yield 1 oz shots, whereas many Americans use 18-21 grams to yield 1 oz shots. Despite the differences in dispensing water temperatures, both systems result in similar average extraction temperatures. Why is that? Because the larger dose used by Americans absorbs more heat from the brewing water.

Scott sketches out a "thought experiment" in support of his theory. Well, my feeling is that a mediocre physical experiment trumps a good thought experiment any day of the week. So...I decided to do some testing.

Measuring "extraction temperature" is a little tricky, since it involves getting a thermocouple inside the portafilter, near the bottom, without screwing up the extraction. Since I don't have Dan Kehn's skill or patience at doing this, I figured that a low-mass bead thermocouple measuring the espresso temperature as it exited the portafilter might give a reasonably accurate "integration" of what's going on (temperature-wise) inside the portafilter. It wouldn't be absolutely accurate - only the thermocouple inside the pf could approach that. But I figured it might be accurate enough to discern some differences in extraction temperature, should they occur. Here's a closeup of the rig:



And here's one just before the flow starts:



The temperature readings are useless early on in the extraction, but once a good flow out the bottom of the basket is established, the thermocouple bead is covered in espresso and the readings are quite reproducible.

Here's a picture taken around 20 seconds into a typical extraction:



(to be continued in the next post)

To perform the experiment, I pulled a whole bunch of shots at a 200F dispensing temperature. I used 14 gram doses at various brew ratios and another set of 21 gram doses at various brew ratios. Extraction time was kept fixed at 30 sec. Past experiments have shown that my modified Silvia can dispense brew water at a very consistent temperature (probably +/- 0.5F), so whatever larger differences show up can be pretty confidently ascribed to the interaction of dose and brew ratio.



After looking at the data a bit, a few things became apparent:

1. At low brewing ratios ("long" shots), the exit temperature leveled off by 30 secs; it wasn't going to get any higher. The fact that it leveled off about 8F lower than the water inlet temperature is related at least in part to the method of measuring just below the filter basket.
2. Exit temperatures measured 25 seconds showed the strongest correlation with the brewing ratio.
3. Exit temperatures at 20 sec were strongly correlated with brewing ratio, but there was more variability than at 25 sec.
4. One can say that the "average extraction temperatures," meaning the significant temperature conditions through the "heart" of the extraction, are strongly correlated with brewing ratio.
5. All other things being equal, going from ristretto to normale to lungo increases the average extraction temperatures.
6. The difference in #5 above is dramatic: going from a 100% ristretto to a 50% normale raises the extraction temperature through the "heart" of the extraction by about 12F!
7. Since a few degrees F is considered to be a "tastable" difference with most coffees, the extraction temperature is a major factor in the differing flavor balance between ristrettos and normales.
8. Perhaps one's regular procedure when experimenting with longer shots should be to drop the inlet temperature, and vice versa.
9. In this experiment, it was not the dose per se but the brewing ratio that influenced the average brew temperature.

Thank you, Scott, for getting the ball rolling on this.

In thinking about this a little more, I believe that the graphs somewhat exaggerate the extraction temperature differences between ristretto and normale. That's because the real "integration" of extraction temperature is the mass of espresso extraction times the temperature at which it was extracted. And since ristretto shots are very slow to get going, their very low temperatures at the beginning have less of an impact on the average extraction temperature.

I feel the correlation between brew ratio and average brewing temperature is a strong one, but I don't have all the data now to nail down the relationship. Sorry....

Very interesting start... I've often wondered about this, but i figured the dry puck couldn't absorb that much heat to be a factor. Perhaps that's not the case. I've also thought about trying to bring the puck up to temp ahead of time, but I've always heard (although never personally tested) that heat kills coffee (i.e. the pursuit of a low temperature grinder). What are your thoughts on that? Feasible to preheat the puck? Or will it kill the coffee?

Thank you for sharing the data. It certainly agrees with my anecdotal experience: when my shots run fast, they are hot (as in my mouth gets burned if I drink them immediately), whereas I can comfortably drink slower flowing shots.

One point regarding your list of conclusions: it seems that while items 1-6 and 9 are factual observations about the data, 7 and 8 belong in the class of "ideas worthy of further evaluation" (which are inspired by the data). Does that seem like a reasonable characterization?

Navin

popeye wrote:I've often wondered about this, but i figured the dry puck couldn't absorb that much heat to be a factor. Perhaps that's not the case.

The dry coffee does absorb a lot of heat, lowering the outlet temperature a significant amount.

popeye wrote: I've also thought about trying to bring the puck up to temp ahead of time, but I've always heard (although never personally tested) that heat kills coffee (i.e. the pursuit of a low temperature grinder). What are your thoughts on that? Feasible to preheat the puck? Or will it kill the coffee?

I've never tested it, but many people feel that heat will negatively affect the puck. USBC competition standards penalize a barista who does not begin the extraction immediately upon inserting the portafilter into the machine.

PS, say hi to Olive Oyl for me.

Navin wrote:One point regarding your list of conclusions: it seems that while items 1-6 and 9 are factual observations about the data, 7 and 8 belong in the class of "ideas worthy of further evaluation" (which are inspired by the data). Does that seem like a reasonable characterization?

Yes, I think you are correct.

AndyS wrote:In thinking about this a little more, I believe that the graphs somewhat exaggerate the extraction temperature differences between ristretto and normale. That's because the real "integration" of extraction temperature is the mass of espresso extraction times the temperature at which it was extracted. And since ristretto shots are very slow to get going, their very low temperatures at the beginning have less of an impact on the average extraction temperature.

I feel the correlation between brew ratio and average brewing temperature is a strong one, but I don't have all the data now to nail down the relationship. Sorry....

Great work

As you say, the graph shows the potential differences at their most dramatic, rather than the actual, quotidian differences. The top of the puck extracts early in the shot, the bottom of the puck, late in the shot. So by the time each section of the puck is extracting, the water temperature has risen to close to the inflow temperature. Just as with the fines migration and wetting, physics makes the extraction temperatures somewhat self correcting.

Oddly enough, the self correction may be going too far. Ristrettos tend to be more bitter and over extracted than normales, so the added brew time appears to be more than counteracting the reduced flow and temperature.

another_jim wrote:Oddly enough, the self correction may be going too far. Ristrettos tend to be more bitter and over extracted than normales, so the added brew time appears to be more than counteracting the reduced flow and temperature.

Here's a couple of shots measured with a TC at the bottom of the puck inside the basket. Both shots were pulled to about the same volume. These shots were measured for a different investigation, but is interesting as it shows the exit temperature remains relatively flat once the hot water hits the bottom of the puck.

In the restricted shot, the coffee spends about twice as much time at temperature as the quick shot.

AndyS wrote:Well, my feeling is that a mediocre physical experiment trumps a good thought experiment any day of the week.

Welcome to the club! <VBG>

Now, if you want to raise the bar from "mediocre" to "good", try checking out the original aim of the experiment (that lower inflow temperatures and lower coffee volumes can give the same level of extraction and flavour IN THE CUP as higher inflow temperatures with higher coffee volumes) with Bob & Jim sucking shots. Forget the "Lungo, Normale, Ristretto" bull, give me milliliter shot volumes! Tasting notes! Grinds in, coffee solids out extraction ratios!

I suspect that you probably already have the data, but haven't thought through the presentation and conclusions yet. What we really need to know is if a 7g/88C/25 sec/30ml single tastes the same as a 14g/93C/25 sec/30ml single, due to cooling and flow factors in the puck.

Alan

(And yeah, I enjoy being a PITA, almost as much as Barry.)