Pino wrote:Not to dwell on the headspace, but here is a picture from Dalla Corte for your viewing pleasure.

Thanks for sharing. If nothing else, the broken translations are hilarious to read!

Dalla Corte wrote:Normally the lock of the cup can be good but the test is of anonymous espresso.

Oh, the terror to too much headspace!

The images and corresponding descriptions are intriguing. I wonder how much is due to the actual headspace versus not having a dose that is well matched to the basket size. I had always figured with proper grind, all you really have to worry about with a smaller dose is a soupy puck...

AssafL wrote: So if we grind fine and Insufficinet PI the entirety of the 9bar will rest on the dry layer of the puck (we know it can flow through the wet part). So it becomes an hydraulic press on the dry part which then seals the flow.

Ok, so if "we know the water can flow through the wet part" (how do we know that?), why would the entirety of the pressure rest on the dry part? I do get the logic that if this were to happen that the hydraulic press would seal the flow. Although I think of this more on the macro level of voids between granules then I do sealing fissures but I'm ok with the idea of both happening, sure.

But why? Water came down quickly and wetted the top layer. Got it. The top layer is wet. Why is the "next" layer immediately under the wet layer not getting wet? I get that this is all happening fast and that the exposed surface of the puck is overwhelmed with Water with no PI. I think maybe it's as simple as that. Water hits the puck (ground fine for PI) at full force and perhaps due to my hypothesis of puck resistance, the water can't get through the puck at the rate the pump would like to push it and the whole puck is compressed under 530+ lbs of super-tamp. This then triggers your model of collapsed fissures and general puck malady. None of it good for extraction.

I know you're on to something with the choking business. I tried an experiment the other day and tried pulling a shot straight up and the result was awful in epic proportions because of how fine I'm grinding. It sneaks up on you pretty quick, I had no idea how much finer (5 full tick marks on the SJ) from where I used to be, really was. The shot started slow, and then it was obvious the puck just exploded and I had geysers squirting all over the place and this thin watery mess in the cup. It was beyond terrible.

Here's my compromised model of reality to explain your side (why you can't grind fine without PI) and my side (why you can't grind coarse with PI) I think you may find the simplicity of my argument appealing...

Take my whole puck resistance model, which generates a quantifiable relationship between flow rate into a puck and back pressure generated by it. You take a standard pour (no PI) and determine the puck resistance. When you introduce PI, my model explains why the shot pours super fast, and why you can grind much finer since the maximum instantaneous flow rate into the puck has been effectively cut in half. The result is higher EY and other goodness in the cup that you couldn't possibly have gotten without grinding super fine like you can with PI.

Take your model of fissures and the hydraulic press compressing the puck without PI. It explains why you can't go from grinding for gradual PI back to a standard shot because my awful geyser shot is what comes out due to initial choking followed by catastrophic channeling. Your model explains why you must grind coarser without PI.

You put both our models together and you get a continuum of flow vs pressure defined by the puck and how we meter our water supply coupled with a hard transition point that says "Nope. You can't grind this fine if you're going to flow this much. Not going to happen." I think we're both right. But maybe I'm wrong. That's cool, too. I'm just glad this discussion is happening because it's really helping build an understanding of how complex everything is in that darned portafilter!

We both whole-heartedly agree that a finer grind results in higher EY and potential extraction of more flavor, and it's tough to want to waste good coffee brewing it the old way after getting used to extracting this way. I should really get a lever... As for the boulders, yep. I'm with you.

I would be very interested to see what extended PI would do for a lesser grinder. I don't have a great grinder, but it's pretty darn consistent for what I'm after. The difficult part would be finding a grinder that is questionable enough to evaluate an increase in performance which also gets fine enough to extract with a long PI. I'm sure a Rocky would be fine, but I'm thinking about a plastic burr carrier ceramic conical like an encore at best and a Krupps or Cuisinart at worst. Like a truly entry level burr grinder... I found an encore for $9.99 at goodwill a few weeks ago. Can't understand why I didn't grab it...

Cheers!

- Jake

Ok, so if "we know the water can flow through the wet part" (how do we know that?), why would the entirety of the pressure rest on the dry part?

Because if you waited a bit longer and added a bit more water the puck would flow. So the top part has water and would flow.

The only difference between a fully PI (time and amount of water) and partially PI puck is that the bottom of the puck is dry.

Now apply pressure. The top part wants to flow. If the puck was thinner it would flow. But for the bottom part.

Why is the "next" layer immediately under the wet layer not getting wet?

That was the baffling part and why I think the fissures collapse. The higher the pressure the more they collapse. That is why mains pressure PI also works (perhaps a bit less as it probably compresses on some of the fissures and crack). That is also why Slayer works better. As it compresses even less.

I assume water does seep through the dry layer eventually. Fissures do exist - pressure cannot make them disappear - just smaller and take more time to fill. And slowly they expand back.

It may max out the safety timers of the pump. But eventually water would drip and pressure relieved. I've had 100 second pours - Awful.

I think maybe it's as simple as that. Water hits the puck (ground fine for PI) at full force and perhaps due to my hypothesis of puck resistance, the water can't get through the puck at the rate the pump would like to push it and the whole puck is compressed under 530+ lbs of super-tamp. This then triggers your model of collapsed fissures and general puck malady. None of it good for extraction.

if so, why would the puck allow flow when PI was complete and not before? There is always a wet layer at the top. I agree that the entirety of the puck is under the full pressure. But the water would flow through the PI portion if there was a way out (spritzers and gushers!).

Come to think of it, Spritzers always happen when you are close to a good pull. If you grind too fine, puck clogs completely. Grind too coarse, and you get a thick, watery pour. Perhaps spritzers happen when the sealed layer (the dry at the bottom) is sufficiently thin to allow a specific area to burst through. PI was insufficient. A second or two away from a perfect pour.

Here's my compromised model of reality to explain your side (why you can't grind fine without PI) and my side (why you can't grind coarse with PI) I think you may find the simplicity of my argument appealing...

Here is why this is a worthy discussion: If I am right, incomplete PI acts as a pressurized portafilter - that is until water penetrates the sealed layer and begins to flow. Grinding coarser reduces EY. Grinding finer pressurizes the coffee; reduces emulsification (it is stuck - no agitation). Since there is no sufficient resistance (as you call it) there is no way out! The puck has no uniform gradient of pressure, and therefore never uniformly extracts.
Incorrectly setting PI leaves you with awful coffee. An altogether convincing argument for levers (which do PI as a natural byproduct of their design).

The whole puck resistance model is obviously correct - but I don't think it means what you think it does... or at least I don't understand it does.

Why does a GS/3 without a better PI still work (albeit at a coarser grind)? because the gicleur still does PI. Just a smaller amount of water and quicker time. So a coarser ground puck does end up filled with water and flowing. As an example, grind coarser but updose too much - and you may seal even a coarser ground puck (bottom would be dry).

One other reason I think a long slow PI is better: think about your puck resistance model. Pressure drops linearly (1 bar per 1/9 of the depth of the puck) if the puck is uniform. If however, you ramp up the pressure too fast, the bottom is still half filled with water. Any air will collapse. So the density will increase - it may not seal - but it will take up more of the pressure. Percolation will not be uniform (and EY will suffer).

To summarize - I think it is a question of uniformity - uniform grind, uniform distribution and grooming, uniform puck, uniform slurry and eventually uniform distribution of pressure across the entirety of the puck. That, in turn, ensures uniformity of percolation.

PI is, in effect, responsible for the uniformity of the slurry.

Just dawned on me why pucks come out relatively "dry". I never really had a good explanation for it...

When the pressure is released, any compressed fissures expand back - pulling water into the fissures and crevices. The slurry instantly converts to a sponge. Rather cool.

This also means that not all crevices see water. Otherwise (if all crevices would be completely filled with water) - since water doesn't compress - the puck would not spring back and remain a slurry. Maybe there are sealed bubbles in the boulders. Maybe some get filled or clogged with fines...

Read this if you dare...

Of particular interest are sections 2.2 on cake filtration, 2.2.3 on compressible cake filtration, 2.3 on blocking filtration and 2.4 on depth filtration. For those not inclined to read a 33 page excerpt from a textbook, I put some notes below...

Cake filtration describes, with great accuracy, our coffee puck. The section spends a bit of time on the building of a cake by feeding a suspension through a filter medium (the bottom of our basket) but obviously, we are building our cake when we grind and tamp, so that part doesn't really translate well. This, however, does:

They go on to define cake resistance as a function of quite a few physical properties of the filtrate and the cake dimensions, including but not limited to particle size, porosity and fluid viscosity on top of the height and diameter of the cake. Furthermore, it is clear that cake filters invariably exibit blocking filtration (wherein the boulders in the cake attract, accumulate and block the fines), along with deep bed filtration. Both topics are analyzed in detail in the text.

Below is an excellent depiction of the balancing act between compressive stress that tends to choke flow and fluid pressure throughout the height of the puck taken from section 2.2.3 on compressible cakes:
In this image, higher pressure is on the left and falls to the pressure on the other side of the filter medium on the far right, designated p₀. Its easy to see how more compressible cakes have less variability in hydraulic pressure (lower ∆P) higher in the puck and the cost of higher compressive stress above the filter medium, which leads to the "pressurized portafilter syndrome". They speak of "supercompactible" filter cakes that exhibit uneven filter resistance:

For practical purposes it can be approximated that such ''supercompactible'' filter cakes with n > 1 yield the same flow rate independently of the pressure applied: increasing pressure only increases the compressed layer adjacent to the filter cloth...
...The filter resistance often is concentrated in a thin, compressed layer facing the filter medium, the rest of the cake being very porous and wet.

One intersting takeaway from this could be that simply by lowering the pressure we may be able to get away with finer grinds without über-long preinfusion, as there is clearly a compressive force threshold that causes choking.

I have much more reading and digesting to do in this article, and much of it doesn't correlate directly on account of us extracting solids from the cake, rather than using the cake to pull solids from our supply. What we are doing may be more akin to the process of "deliquoring", in which depth filters are cleansed of fines and recharged...

Cheers!

- Jake

Awesome find. Great reading I hope.

Few points:
1. As you say, we are cleaning the filter rather than filtering the coffee. The spent puck is a clean filter devoid of soluble and fines.
2. Compressible is sort of applicable. Quasi applicable. Coffee, as can be seen in the transparent basket, does compress. However, their graph assumes a compressible filter matter. Say polystyrene balls squished by the pressure. In our case the material is both compressible and porous (according to Illy). Hence the puck is not uniform.
Perhaps even the fully wetted puck compresses. But any air in the puck would compress substantially more.

I think the model is getting clearer. Is it not?

AssafL wrote:Compressible is sort of applicable. Quasi applicable. Coffee, as can be seen in the transparent basket, does compress. However, their graph assumes a compressible filter matter. Say polystyrene balls squished by the pressure. In our case the material is both compressible and porous (according to Illy). Hence the puck is not uniform.

They specifically mention porosity as a determining attribute of the cakes, which are made in filtration by the boulders in the solution building up against a medium (the bottom of our basket). As such, I think compressible in the terms provided is fully applicable. Lucky for us, the boulders have ample soluble compounds for us to extract. There are also plenty of fines that start out evenly distributed. Being that our boulders form the filter cake, we do have some filtering out of the fines taking place, and the above mentioned research has this process very well characterized.

AssafL wrote:Perhaps even the fully wetted puck compresses. But any air in the puck would compress substantially more.

I agree, but I would follow up with "absolutely, a wet puck compresses, but air in the puck would compress even more", explaining your phenominum of rapid expansion post-shot that results in a dry puck...

One different view point I believe I have from you (and countless others, I'm sure!) is that I am not convinced the puck is ever liquefied into a slurry. I believe the boulders that we tamp into place basically form the structure of a filter cake and the water flows through the cake, stripping away soluble compounds and washing fines further down the cake. I gather from reading your last handful of responses that you view the "wet" part of the puck as an agitated and emulsified slurry and that the "dry" parts are the only bits that are locked into place. This is why I'm so stuck on the resistance model. I view the puck as a relatively static thing. We can compress the entirety of the puck by overwhelming the surface of the puck with more water than it can process (no PI with too high of a puck resistance), which results in a tighter matrix of boulders and more places for fines to become trapped and dam the flow, but I believe a boulder located in quadrant 4 of layer 1 effectively stays put throughout the extraction. (With the obvious exception of the very top layer that gets disturbed when the water hits it.)

I think a saturated and flowing puck resists being choked because there is a moderate pressure gradient across each layer of the puck. If we take our convenient 9 layer puck, layer 1 has 9 bar pushing it down, but 8 bar pushing it up, since fluid pressure acts in all directions. Layer 2 has only 8 bar pushing down on it, but 7 bar pushing up. It also has 1 bar of compressive pressure from layer 1 pushing down on it. This compressive pressure is caused by the resistance of fluid flow through layer 1. This goes on and on until we get to layer 9 that has 1 bar of fluid pressure on top and effective none underneath, but 8 bar of compressive pressure on top of it from the 8 layers above. Each layer has 9 bar in total acting on it and a pressure drop of 1 bar from the top of the layer to the bottom.

This is in stark contrast to what happens if we hit layer 1 with 9 bar before there is 8 bar established underneath and supporting it. 9 bar in layer 1 with a dry cake underneath causes some serious issues...

AssafL wrote: I think the model is getting clearer. Is it not?

It is!

But no less complicated

I am cogitating on this a bit and a question popped up. We talk about compression and eventually expansion (when pressure is released).

Why do we need the headspace that Della Corte care so much about? Why does the wet coffee expand? Is it like a sponge that has some sort of inner tension that is allowed to unwind when wetted?

The concept of a puck locked between two screens (which is supported by the evidence: amount of gunk above the shower/dispersion screen) supports the filter rather than slurry* hypothesis.

With a pressurized puck I assume the top wet layer is a slurry as there is little pressure gradient on it.

BTW - the compressed air (collapsed crevices and fissures) dry area means that the sealing happens in the dry/wet interface (probably more on the dry side). Not in the first layer. (well, I assume that the first layer is at the top...).

* PS - Semantics - but I though a slurry was just a wet powder - like mud. Be it a filter or not - isn't it always a slurry (in this case the slurry is the filter, caught between two fine mesh screens?)?

AssafL wrote: Why do we need the headspace that Della Corte care so much about? Why does the wet coffee expand? Is it like a sponge that has some sort of inner tension that is allowed to unwind when wetted?

I am not sure but I think the puck wants to expand because we have first compressed it by tamping. Why does getting it wet make it want relax? Assuming slow preinfusion, I think wicking by capillary action draws water into the puck but fails to evacuate air trapped in the interstitial voids. The result is increased volume from a swollen puck.

I'm quietly nodding to the rest of your post

Have fun cogitating!

**edit**

And of course, the puck swells due to the gasses released during bloom, which doesn't happen until the puck gets wet...

**/edit**

The more I cogitate over PI the more convinced I am that it is behind the "preferences" people have towards basket sizes, settings, grinders, etc.

A single, double. triple all need specific volumes of water and given a particle size - specific amount of time for the water to permeate and saturate the puck. The key is to avoid leaving a dry section at the bottom (that will either seal the puck or spritz). The dry part will - at the very least - skew the extraction yield - or altogether ruin the pull.

Obviously grind size is important. Objectively, the grind size is really defined by the 9 or so bar pressing on the pre-infused puck. If one wants them to pull in X seconds (so that emulsification can happen) a 21 gram dose will have to be ground coarser than a 7 gram dose. But it also affects PI timing so that PI for a 7 gram will be longer/shorter (don't know which) than for the 21 gram.

Why not over-wet the puck? Can one overdo wetting the puck? At least one drawback I can imagine to letting the coffee sit in too much water is that all the solubles will dissolve and there would be nothing left to emulsify (think trying to pull an espresso from something similar to leftover french press grinds).

So the following observations may be true:
1. Fixed (or limited range) "volume" PI machines - all the e61, most gicleur'd pump machines, etc. will have an optimum dose (or a narrow range of dose) where the results will be optimal.
2. Levers and flow restricted machines (like Slayer and Paddle machines) will have a substantially wider range of doses that can be proeperly extracted.
3. Obviously, PP machines (and especially FP & PP machines) will accommodate the widest range - but at some ranges/settings the outcome would not be traditional espresso (does 6 bar count as espresso?).

One simply cannot treat different basket sizes as a "simple" parameter - the basket size and dose work in concert with grind setting, PI volume, PI time, & pump pressure, to ensure proper extraction (EY). It all has to change and the main reason is pre-infusion.

A case in point is the triple basket thread. If I ever find my triple (I tried it twice and decided it was bad - talk about hubris ) I may try it to see if indeed PI volume is (was) the culprit.

NB - IMHO, a partial PI (not fully pre-infused) is not acceptable as proper technique. Trying to rationalize as to why I think it is a rule - is that there is absolutely no reason that partial percolation (i.e. channeling at the dry puck segment) or reduced extraction (i.e. grinding coarser so that the puck doesn't seal) are in anyway desirable. Also, achieving consistency with a partially wetted puck is nearly impossible. Which part is wet?

It is therefore an abuse of an e61 machine to use it with a 21gr dose. Unless one fits a finer gicleur, modulates a volumetric pump or otherwise changes the working point to the system to dump a lot of water - slowly - on the puck - before the PI chamber pressurizes.

That is not to say that mistakes don't sometime come out tastier (espresso is all about myopically focusing on localized maxima in flavor). If one over-extracts due to a channeling in a dry area of the puck (insufficient PI) one should really dose less, or grind coarser and fully percolate at the preferred extraction.

NB2 - 58mm. Larger portafilters have more force applied to them so it is clear why home machines prefer smaller filters (smaller pumps and cheaper bayonets). But why do larger portafilters (usually) perform better? My guess is more uniform PI given the fingering flow you show above. In fact, having pulled many times on a la Pavoni (48mm basket), i'd reflect that fingering flow in the puck makes getting PI right difficult.

What a fantastic and epic thread! And to think, I started it all by simply existing!

In all seriousness, I'm extremely proud of my younger brother for taking up the challenge of creating a consistent and amazing espresso and for taking the time to share it the way he has, in the fashion he has. Sadly, his exposure to my burr grinder (which I still have, though the terrible krupps espresso machine is long gone) was such a tiny and insignificant thing to me, that I completely missed the opportunity to really engage with him on how amazing a thing extracting coffee from ground up, nearly burnt seeds is.

I'm hoping to find some inspiration and will be on the lookout for an affordable machine in the not too distant future.

Learning lots here, keep the discussion going all!