Playing with Pump Pressure: Parte Due

[AndyS]

For your reading pleasure, here are a couple more tests to measure the effect of varying pump pressure on the rate of espresso extraction. Last time the results suggested that increasing pump pressure above ~8 bar actually decreased the extraction rate, and I wanted to verify this result.

In the first experiment, I kept dose and grind constant and simply measured how much espresso I got in a 25 second extraction at various pressures. Once the pump came up to full pressure, the pressure was maintained as flat as possible. In particular, I was looking at the lower pressure realm, because some data in Illy's book suggested the the flow rate might go up around 5 bar. I saw no evidence of this:



In the second experiment, I kept the dose and grind constant and set the flowmeter to terminate the shot after 48 ml of water had been pumped into the boiler. This resulted in shots that were fairly uniform in volume (19.0-21.4 grams). To get the average flow rate, I divided the volume in grams by the seconds of flow (total seconds minus 8.5 seconds dwell time). This data suggests that the flow rate peaks somewhere around 8 bar:



Notes:
1. I've been saying for years that it's the pressure/volume relationship in the first few seconds of the extraction that is most critical in determining flow rate. I still believe this, so I kept the parameters in the initial 4 seconds of each extraction constant.
2. I tasted all the shots, especially in the second test, where similar volumes permitted realistic comparisons. It would be very easy to draw hare-brained conclusions from these, but I believe I'd have to run a lot more shots with a blind taste panel to get any definitive results.
3. I also have a hare-brained theory for why espresso seems to come out best around 8-9 bar, but I'll save it for when I can actually prove it (if ever).
4. I bought a new 1.5% pressure gauge to calibrate my setup. I believe the pressures in the graphs aren't perfectly accurate, but they're pretty close to what the cake is actually seeing.

[another_jim]

Wow! What a terrific set of measurements.

Well my old economist self is pleased. Seems the flow, ceterus paribus, is maximized at the 8 to 9 bar range; just where one would expect the curve to have its maximum given that this is the brewing range used for the last 50 years.

I also believe the dwell time is probably where all the interesting action is. The fines migrating downward are in a race against the particles swelling from water absorption and locking them in (Think Indiana Jones fines dodging flood waters and ever enlarging boulders). Presumably, the flow rate around 10 seconds after the flow starts (this is roughly where I decide to dump a gusher or dripper) will reflect how close the fines got to forming a barrier in that initial race.

But does this mean the fines are getting locked in higher up in the puck at 8.5 bar than at 5 or 12 bar?

[AndyS]

The fines migrating downward are in a race against the particles swelling from water absorption and locking them in (Think Indiana Jones fines dodging flood waters and ever enlarging boulders).

Couldn't have said it better myself.

Presumably, the flow rate around 10 seconds after the flow starts (this is roughly where I decide to dump a gusher or dripper) will reflect how close the fines got to forming a barrier in that initial race.

But does this mean the fines are getting locked in higher up in the puck at 8.5 bar than at 5 or 12 bar?

The fines seem to be an "x factor" whose migration is highly dependent upon the preinfusion.

As far as the 5 bar, 8.5 bar, 12 bar extraction rates go, they may be explainable through the interaction of two conventional, yet competing factors:
(1) higher pressures normally produce higher flow rates through a fixed resistance
(2) but the resistance isn't fixed, because higher pressures compact the cake, reducing its porosity, resulting in lower flow rates.

[Ken Fox]

I keep wanting to make an intelligent comment on this thread, whose initial post I read a number of hours ago. But it escapes me.

So I am reduced to saying, "thank you for your efforts, Andy," and don't interpret a lack of further commentary, on my part or on the part of others as disinterest. Rather, we lack the intellectual horsepower to say anything that gives the appearance of being an intelligent response.

ken

[AndyS]

don't interpret a lack of further commentary, on my part or on the part of others as disinterest.

Ken, thanks for your generous comments! But the problem with these Pump/Pressure posts of mine is that so far, they reside in the realm of theory only. I'm hoping that eventually I'll stumble upon something that'll actually be of use in making better espresso (see photo of actual espresso below). Until then, this is probably just of interest to geeks and nerds like me.

[mrgnomer]

Thanks Andy for the work!

I won't pretend to understand the theory completely but I think I get the gist of it. It's interesting to consider the actions of the fines with respect to blooming especially with pre infusion and the overall effect on puck density/resistance also relative to pump pressure.

From what I've read pulling a lever shot doesn't generate as high of a pressure as 8+bars. Less gets extracted and that would account for a smoother, maybe sweeter shot? That seems to be what your results are showing.

It's a difficult thing to measure, I think, when extraction/flow rates seem so sensitive to grind composition, dose, distribution and packing. What comes to mind is ensuring ideal conditions with some sort of imaging device coupled with some sort of high frequency shaker. The imaging device could scan the grind distribution to activate the shaker until a uniform distribution/density is achieved. An automatic tamper set for pressure could finish with a theoretically perfect tamp.

[AndyS]

From what I've read pulling a lever shot doesn't generate as high of a pressure as 8+bars. Less gets extracted and that would account for a smoother, maybe sweeter shot? That seems to be what your results are showing.

I don't own a lever machine, so I have no experience there. But if extracting less (or extracting at lower pressure) was the secret of smoother, sweeter shots, pump machines would have no problem doing that....

[cafeIKE]

Until then, this is probably just of interest to geeks and nerds like me.

There may be more geeks here than you can chuck a puck at.

You, Ken, Jim et al. are truly appreciated.

Many Thanks.

[erics]

Andy -

Thanks for the time you took in running these tests.

I have read the particular sections of Illy's book you mention and, to be honest, I still trying to figure out what he is trying to illustrate.

The illustration is reproduced here for discussion purposes:


Another idea would be to video a scale readout during the brew process and graph the data much as you did with the Silvia flat-liners you did previously.

[another_jim]

Another idea would be to video a scale readout during the brew process and graph the data much as you did with the Silvia flat-liners you did previously.

Ken has this really small 0.1 gram scale from Hongkong. You could put something like this under the cup, along with a timer, while you video the pour, to get a proper output volume estimate. You already have the flow meter for input volume; so perhaps there is some gold in looking at how the input and output curves relate. Perhaps there are differences in the amount of water the puck absorbs at different pressures.

One suggestion: at this point you aren't looking for the best shot, just the way the system behaves. A wider range of pressure (e.g 4 bar for a low, and 14 bar for a high) may give a better idea of the curve.

[AndyS]

I have read the particular sections of Illy's book you mention and, to be honest, I still trying to figure out what he is trying to illustrate.

Hi Eric:

Yup, we've all tried to figure out what the heck he was talking about. I explained my take on it in a previous thread.

[erics]

Greetings Andy, Jim, Ian, Ken & Kirk -

Just so there's no mistake as to what I see - when one does a poor man's integration of one of his (Petracco's) curves, you end up with ~ 53 ml of espresso. I did the 5 bar curve because it happed to be the easiest one to "pick out" of the grouping. What I don't understand is why he did all this in 12+ seconds and not, say, 25 seconds. I also don't understand the really high flowrates he illustrates but enough on that.

Pressure profiling is interesting to me because I firmly believe it is a methodology that will enable one to get the most from the least. Strictly intuitively speaking, preinfusion is good, tapering off the pressure in the last third of the shot to prolong the onset of "blonding" is good.

[another_jim]

I also don't understand the really high flowrates he illustrates but enough on that.

Does anyone know whether the peak of the pressure to flow curve remains at the same place for different grind settings? Perhaps the coarser the grind, the lower the flow peak (I can hardly imagine it being the other way around)

[AndyS]

when one does a poor man's integration of one of his (Petracco's) curves, you end up with ~ 53 ml of espresso. I did the 5 bar curve because it happed to be the easiest one to "pick out" of the grouping. What I don't understand is why he did all this in 12+ seconds and not, say, 25 seconds. I also don't understand the really high flowrates he illustrates but enough on that.

The 12 second part is easy. Petracco heard that double shots run 25 seconds. But in Italy they mostly pull singles, so he figured he had to cut the time in half.

But seriously, the graph is a mess. Why did he choose 3, 5, and 7 bar when he states in the same chapter that 9 bar works best? And does he really believe the flow chokes down to zero after second 12?

But some parts do make sense, I think. If you look at his flowrates as water input to the portafilter, not as espresso exiting the spouts, it's fairly realistic for a double. My double shots take about 30 ml water input before flow begins, and 15-20 ml after that.

It's possible you really would get graphs that look like these if you used no gicleur at all. The initial "slam" up to 7 bar might actual result in a lower flow rate than the gentler slam up to 5 bar. Maybe. It wouldn't be that hard to test, but what's the point?

[AndyS]

Ken has this really small 0.1 gram scale from Hongkong. You could put something like this under the cup, along with a timer, while you video the pour, to get a proper output volume estimate.

That may work, but there are two things to watch out for:
1. the cheap digital scales have a built in delay of 2-3 seconds
2. the kinetic energy of the falling espresso stream may give a false high reading (could be minimized by pulling into a shallow vessel that is as close as possible to the portafilter)

One suggestion: at this point you aren't looking for the best shot, just the way the system behaves. A wider range of pressure (e.g 4 bar for a low, and 14 bar for a high) may give a better idea of the curve.

Yes, I should do that soon.

[AndyS]

One suggestion: at this point you aren't looking for the best shot, just the way the system behaves. A wider range of pressure (e.g 4 bar for a low, and 14 bar for a high) may give a better idea of the curve.

Here are two series of shots run at a wider range of pressures. Each shot received a five second preinfusion at line pressure. The rest of the extraction (seconds 6-27) was performed with a rotary pump set to various pressures:



Notes:
(1) The absolute peak is somewhat fuzzy here; it's somewhere between 8 and 10 bar. This curve depends on a lot of factors which probably make a blanket statement like "the maximum amount of espresso is extracted at X.X bar" suspect.
(2) Speaking of which, I want to repeat this experiment with little or no preinfusion. I speculate that going immediately to pump pressure (with no pump delay) and with no gicleur (ie, no flow restriction) may yield results similar to the ones that Eric posted from Illy's book. In other words, peak flow may be found at a much lower pressure. Key word is "may."

[another_jim]

Here are two series of shots run at a wider range of pressures. Each shot received a five second preinfusion at line pressure. The rest of the extraction (seconds 6-27) was performed with a rotary pump set to various pressures:

Thanks for doing this. The results appear to be consistent for your machine, and perhaps inconsistent with Petraccho's results (hard to tell, with all the trade secret obfuscations in the book -- the graph doesn't deal with a regular extraction, but doesn't say exactly what it deals with).

The results are beginning to support a larger hypothesis of mine -- the Italians have done a lot of engineering, either in a scientific or a more seat-of-the-pants way (by this I mean just setting one variable at a time so it taste best with whatever else they are doing), and have more or less optimized their machines for the temperatures, pressures, doses, etc for which they are designed. This sort of fits in with the dosing experiments, what we are finding out about the way jets and HXs are tuned on the better machines, etc etc. Schomer's "mediocrity by design" piece is plainly wrong, at least for the better espresso machine and grinder companies. It may be possible to improve on what they have done, but probably not by simply changing the variables. The real problem of course is the same that obfuscates the Illy book in places -- In Italy, it seems absolutely everything is a trade secret; so they simply don't have anything like the information flow we have on this.

[AndyS]

the Italians have done a lot of engineering....It may be possible to improve on what they have done, but probably not by simply changing the variables. The real problem of course is the same that obfuscates the Illy book in places -- In Italy, it seems absolutely everything is a trade secret; so they simply don't have anything like the information flow we have on this.

It's a shame that nearly everything we've "discovered" is probably just a rediscovery of something that's been known in Italy for 30 or 40 years. But since they're not talking, we still have to sweat the details to figure things out.

[AndyS]

I guess many of you are sick of seeing a stream of graphs that don't answer the basic question, "How can I make better espresso?"

Me, too. But here's one last (maybe) graph. It compares shot-pulling with a rotary pump, with and without a gicleur (flow restrictor). Dose and grind were kept constant.

Nowadays most rotary pump machines (GS3, GB5, Synesso, etc) come with a gicleur that restricts the flow of water going to your portafilter so that the initial buildup of pressure takes 5-8 seconds rather than 1-2 seconds.

Interestingly, this initial slowdown makes the shot run far faster once full pressure is reached (see graph). It also means that baristas use a significantly finer grind on gicleur-equipped machines as compared to non-gicleur machines. How that affects the resulting shot is a mystery.

It's worth noting that gicleurs may be unnecessary on vibe-pump machines. Their slow buildup to full pressure mimics the action of a rotary with gicleur. Numerous studies, however, have shown that vibe pumps eventually cause hearing loss and may result in prolonged bouts of insanity.

There's also a shot on the graph that represents a non-gicleur shot that was preinfused with line pressure water (~3 bar) for five seconds before the pump came on. This had a similar effect to the gicleur shots.

Sometime soon when I get most of this preliminary stuff out of the way I'd like to compare the taste and texture of gicleur vs non-gicleur shots.



Just kidding about the insanity part...maybe. :-)

[Randy G.]

....Numerous studies, however, have shown that vibe pumps eventually cause hearing loss and may result in prolonged bouts of insanity.

If you can't back up these claims with graphs and some hard, factual data, you should not make such comments about me unless you have met me... err.... never mind.

Thanks for the hard work, Andy. You have been on the forefront of such research for a good, long time and a lot of us are enjoying better espresso because of it...