Pressure profiling, flow profiling, and a new rule of thirds

Need help with equipment usage or want to share your latest discovery?
User avatar
shadowfax
Posts: 3545
Joined: 19 years ago

#1: Post by shadowfax »

As some of you know, I've been interested in pressure profiling for a long time. As I enter my 5th year of involvement and experimentation with pressure profiling, I wanted to revive the discussion by trying to shed some new light on what some other people have been saying. Like I said when I started out investigating pressure profiling for myself, I don't think I have any really original insight to bring to the discussion, but I hope to illuminate the ideas of others and investigate them for myself and anyone else who is interested. With that aside...

Dynamics of Pressure Profiling: a New Rule of Thirds?

A year or two ago, I had the honor of spending some time with Scott Guglielmino of La Marzocco. We discussed pressure profiling espresso machines, their potential value, and what pressure profile(s) work best. One of the things Scott mentioned during our discussion was that they had observed that the biggest factor in determinining how a shot would flow in the last 2/3 of the shot was the pressure profile of the first 1/3 of the shot (roughly). Perhaps counter-intuitively, a gentle pressure ramp tends to yield a faster flow rate later in the shot, whereas a swift pressure ramp tends to yield a tightly restricted flow during the last 2/3 of the shot.

The practical upshot of this "rule of thirds," as it were, is that changing the shape of the initial pressure ramp can allow you to vary grind size dramatically while maintaining a desired brew ratio (and the texture/strength that accompany it), rather than using manipulations of grind size, dose, and shot time to achieve your desired brew ratio. With an exceptionally slow initial pressure ramp, you can grind significantly finer, all other things being equal, than you would with a conventional espresso machine with a fixed pump and flow restrictor. The idea, in this case, would be to make it simpler to dial in very light-roasted coffees as espresso-potentially opening up coffees to be extracted as espresso that really wouldn't be viable at all on a conventional machine.

Before I get into that though, how does flow enter these dynamics?

Pressure vs. Flow profiling: Pressure Profiling at 0 Bars

Some of you have no doubt read about Slayer espresso machines, and Jason Prefontaine's contention that flow profiling is critically different from pressure profiling (if not, check the thread above for Jason's very approachable explanation). In that thread, the OP (John) hit the nail on the head with the question:
bostonbuzz wrote:I'm missing something, otherwise flow profiling and pressure profiling are two ways of talking about the same thing... (except for the speed of simply filling the chamber with water at 0 bar ((once it's full it's simply slowly raising in pressure))).
Once a shot has stabilized to a positive pressure, pressure profiling and flow profiling are the same process with different variables held steady, sort of like the difference between shutter priority and aperture priority in photography. But in light of the "new rule of thirds," which demonstrates the impact of the initial pressure ramp, what about the flow rate before there's any appreciable pressure to ramp? That is, the part of the shot when water is trickling onto the puck but pressure isn't yet being applied at all because the space in the tube and grouphead above the puck and downstream of the 3-way valve hasn't been filled with water. At this stage, there is water flow, but no pressure. In this context, pressure profiling has no meaning, but flow profiling certainly does.

In all espresso machines I've seen that support pressure profiling by pump delivery variation (namely, the La Marzocco Strada EP), a fixed flow restrictor is used. In the case of La Marzoccos, this is 0.6 mm. Now, consider Jason's fire hose vs. garden hose analogy. You can put 3 bars or 9 bars of upstream pressure against such a flow restrictor, and it doesn't change the flow-rate against 0 bars of downstream pressure too substantially. I would say by not much more than 50-70%. With a variable flow restrictor, you can tweak flow rate in a much more dramatic way, even without manipulating brew pressure.

Given that the first third of the shot is so critical to the way a shot develops later on, is it possible or likely that the ability to "flow profile" at the earliest stages of extraction is much more important than being able to manipulate pressure or flow later in the shot? Is it possible that the ability to change the flow at the beginning stages of the shot is considerably more important than shaping pressure and flow later on?

As I mentioned earlier this year, I have pre-ordered a Slayer 1-group. I have reasons far outside the scope of the discussion at hand for buying one. If I wanted to duplicate its abilities, or even expand on them, adding a needle valve to my current machine would be an easy (or not) option.

But what's the real difference?

So this is the topic that has occupied my interest for some months. I've observed with my own equipment that dramatically reducing line pressure (to around 1 bar, vs. 2-3 bars typical line pressure) and turning my pump entirely off for the first 20s of brewing (and then ramping to 9 bars and extracting for another 20-40s), I can pull a shot with the same brew ratio that I'd get with a conventional pressure profile while grinding 3-4 notches (on a K10) finer for the same coffee. As you can imagine, this makes a huge difference in the shot's flavor. I've found so far that this works very well for dialing in very light roasted coffees to get maximum sweetness, balanced acidity, and thicker/heavier body; but does not lend itself to darker roasts where a fine grind is not desirable.

When I first played with pressure profiling, I focused almost entirely on the profile from when I achieved line pressure (at the time, around 3 bars) onwards, and in particular in declining the brew pressure later in the shot, like a lever machine. The differences I observed in blind testing this method were noticeable (and generally an improvement), but subtle. I'd like to revisit such rigorous testing again soon, focusing on profiling the beginning stages of the espresso extraction, rather than the end.

That may be some ways off, but I wanted to begin a public discussion of the topic with an eye toward collecting observations and suggestions from other forum members. I know a number of you on this forum also have access to equipment that lets you explore these variables, and I'd love to read your observations around this question. I'd especially love to read about your experiences in dialing in espresso in this fashion, if possible.
Nicholas Lundgaard

Advertisement
User avatar
another_jim
Team HB
Posts: 13871
Joined: 19 years ago

#2: Post by another_jim »

Thanks for this post. It's a real opener at least for me

It explains the Strega's "punch" (i.e. the need to grind finer or dose higher than on other machines). In the tank model, it takes about 12 seconds for the chamber to fill and the pressure to ramp up to 9 bar, which certainly qualifies as a long drawn out preinfusion. More generally, it's a beautifully simple explanation for the dose/grind variations one sees on all espresso machines.

I do have one mildly critical comment. Machines with a conventional brew heads, like the Elektra, Brasilia or Bezzera, that ramp up very rapidly, so that the dwell time is in the range of about 4 to 5 seconds are not useless for light roasted espresso. They simply requires very low doses and very low shot volumes. In other words, for the same basket, grind setting, and brew ratio, you may will need to dose much lower on these machines and make much shorter shots; but I don't see any reason why you can't still get a shot that tastes roughly the same. It may turn out that a pressure profiling is the only way to get properly extracted high dose shots; but that the grind setting, extraction, and brew ratios using old style Italian techniques (i.e. 6 to 7 grams in a single basket and a shot that weighs ten grams) are about the same.
Jim Schulman

User avatar
Marshall
Posts: 3444
Joined: 19 years ago

#3: Post by Marshall »

Thank you for the interesting post, Nicholas. I also talked to Scott and participated in their profiling comparison taste test at the L.A. Out of the Box event. I found the differences interesting and expected to see a lot of experimentation in the high-end coffee bars.

But, as far as I can tell from talking to many baristas using profilable LM's, there is zero interest in the shops in actually using these machines for profiling. The manometers on each brewhead might as well be hood ornaments, since the baristas are perfectly happy with letting the gicleurs do the job of a gradual ramp-up.

There is also talk in the barista Twitter-sphere about profiling as a fad that came and went. I'm afraid that, unless you can start some kind of campaign with the pros, profiling will mainly be a hobbyist tool. Of course that is fine for the hobbyists, but it pains me to see so much sophisticated hardware remain unused at retail.
Marshall
Los Angeles

OldmatefromOZ
Posts: 316
Joined: 10 years ago

#4: Post by OldmatefromOZ »

Just my 0.02AU cents, observations from a hobbyist point of view with Breville DB.

Lighter SO roasts as espresso (city to city+)
I have found that grinding significantly finer, so much so that at a straight to 9bar ramp just about chokes the machine.
Setting the pre infusion so that it sits around 3 bar, which it takes around 5 seconds to get to for total 10 - 12sec, at which time I can see even beading just starting on all holes of filter basket.
This allows for a much more stable shot from start to finish. There is more of a gradual progression toward blonding and I find that 50 to 55% brew ratio is achievable with loads of body and sweetness to balance the acidity / higher fruity notes.

Dropping the PI pressure back to 1 - 1.5bar, still around 10sec with the same dose / grind results in a more ristretto shot, intense syrup that can be quite interesting and pleasing with SOME lighter roasted coffees but I find it more suited to darker roasts (Full City) where there is rich caramels and chocolates and more body to be intensified.

Overall, I am addicted to having the ability to play with PI times and pressure(lead me to vintage manual home levers). I regularly experiment with turning the PI off on BDB and still can get some very nice shots, but they are less frequent and show up a few more visual cues of a less than optimal extraction. So in this regard i think it at least adds to being more forgiving to prep techniques adding to consistency. I fully understand that some machines with better pedigree may actually see less benefit from PI due to better group / shower screen design ect.

User avatar
shadowfax (original poster)
Posts: 3545
Joined: 19 years ago

#5: Post by shadowfax (original poster) »

another_jim wrote:More generally, it's a beautifully simple explanation for the dose/grind variations one sees on all espresso machines.
I've thought exactly the same thing in reflecting on this. Of course it's not the only factor, but I'd bet it's a first-order variable.

As for Elektras and a selection of other machines with near-instant 0-9bar pressure ramps working well with light-roasted coffees, where the compensation is made primarily in dose reduction (or should I say an increase in headspace?), it's a good point. But now that I am thinking about it, it makes me wonder if those machines are partly so sensitive to headspace because of their speedy pressure ramps.

It's an interesting observation, and may point to me being well off-base in my hypothesis, or at least require some nuance. It also makes me wonder how applying these brewing techniques to an Elektra Semiautomatica (for example) would change it.
Marshall wrote:As far as I can tell from talking to many baristas using profilable LM's, there is zero interest in the shops in actually using these machines for profiling. The manometers on each brewhead might as well be hood ornaments, since the baristas are perfectly happy with letting the gicleurs do the job of a gradual ramp-up.

There is also talk in the barista Twitter-sphere about profiling as a fad that came and went. I'm afraid that, unless you can start some kind of campaign with the pros, profiling will mainly be a hobbyist tool. Of course that is fine for the hobbyists, but it pains me to see so much sophisticated hardware remain unused at retail.
I agree. Even the few times I've seen a Slayer in use, I haven't found that baristas, even seasoned ones, have any experience adjusting its needle valve to tweak the machine's behavior. One thing that I suspect, at least for the Strada EP, is that for a tool to be of any use in a fast-paced environment, it needs to be adjustable quickly and incrementally (like a grind or dose change) in a relatively predictable manner. The Strada EP doesn't allow for this at all, and will likely relegate its power to lab and slow bar settings. Likewise with the Slayer, the needle valve adjustment is hidden away under the cup tray, with a single exception that I have seen. I suspect that pressure/flow profiling techniques won't really see widespread use in high-volume settings until they're as easy to tweak as grind setting and as widely understood.

Like you, I think that's fine for now. Maybe it is a fad that came and went, but trends move in circles, too-Perhaps it will be revived with a new angle.

Thanks for your thoughts, gentlemen.
Nicholas Lundgaard

User avatar
bostonbuzz
Posts: 1255
Joined: 13 years ago

#6: Post by bostonbuzz »

This is a great post distilling known information into a testable and probably theory that explains a lot of behavior and has quite a bit of intuitiveness. Thanks for alerting me about it.

I wonder what the role of the SPEED of the ramp up to the preinfusion pressure vs. simply the preinfusion pressure. I.e. slayer invests in a needle valve to adjust the speed of the ramp up to pressure X, whereas others are only concerned with pressure X. Maybe you can tell us when you own one! What is the delivery time on that anyway?

I feel the lack of professional enthusiasm. Two problems. First, the manual paddle machines are frustrating to profile with because of how they work (word of mouth from local coffee shops). The electronic profiling LM machines (where I drink my coffee) are set to a long preinfusion and a flat shot profile. They reported that LM had done tons of tests that resulted in the notion of a declining brew pressure as being not really that exciting. I don't know, but i suspect they have only played with preinfusion time rather than pressure (of course flow is off the table with a strada).

Dan may have something to be considered if I may quote him,
For what it's worth, I've noticed that excessively long preinfusion leads to "muddy" tasting shots (i.e., greater than 6 seconds).
(from this thread Preinfusion, channeling and some other rambles.)

I see that the BDB seems like a great machine for experimenting (except for the flow profiling aspect).
LMWDP #353

User avatar
shadowfax (original poster)
Posts: 3545
Joined: 19 years ago

#7: Post by shadowfax (original poster) »

bostonbuzz wrote:I wonder what the role of the SPEED of the ramp up to the preinfusion pressure vs. simply the preinfusion pressure. I.e. slayer invests in a needle valve to adjust the speed of the ramp up to pressure X, whereas others are only concerned with pressure X. Maybe you can tell us when you own one! What is the delivery time on that anyway?
With a jet (gicleur) at a set size and pump at a fixed speed (and pressure against the jet, as in the case of a rotary pump with a bypass valve), what we see is water being delivered through the jet at a roughly fixed rate (called the machine's water debit), delivering water into the grouphead and evacuating the air in the space above the puck. This air is gradually compressed and forced out through the puck. Before that happens, though, there is no appreciable pressure that registers on a pressure gauge downstream of the jet (e.g., the gauge on the top of the grouphead of a Strada MP) or on the coffee puck itself.

The combination of the pump's speed/upstream pressure and the size of the jet determine the time that this stage lasts. On an older La Marzocco Linea (a machine with a huge ~1.2mm jet) or Elektra A3 (a machine with a large 0.9 mm jet and very small volume of space between the 3-way valve and the top of the puck), this time is a mere second or two. On a machine like an E61 (which has a smaller jet and a big "gicleur chamber" inside its manual 3-way valve) or newer La Marzocco (they come with smaller 0.6mm jets now), the time it takes to fill this space is more significant.

As the water fills the space above the puck and the air it's displacing begins to really compress, the pressure on the coffee puck starts to rise. The water debit of the machine determines the rate at which this happens. A pressure profiling machine can manipulate the rate by dynamically adjusting changing the pump's speed (note: it's possible to pressure profile by adjusting a needle valve and achieve the same effect, but this technique isn't employed by any machine you can buy off-the-shelf).

The limitation that a "pressure profiling" system intrinsically has is that its feedback mechanism (pressure sensing) tells it nothing about this initial phase. To my knowledge, the Strada EP simply fixes the pump at a set speed for these first seconds (though it may scale the speed somewhat, depending on the position of the paddle). In either case, you don't see any feedback during this phase of the extraction, and from what I have seen, you don't really gain control of pressure with feedback on a Strada EP until 5 seconds or so after the shot starts. Likewise, I can manipulate this "0-bar phase" duration a little bit with my gear-pump powered shot brewer, but I can't get the time much over 7-8 seconds-I would need a needle valve or a smaller gicleur to do this.

A system that used flow rate as an input to a controller that actuated a needle valve would facilitate really fine-grained control of this duration. To my knowledge, this hasn't really been done, though as I alluded to in my post, member Sean Lennon is working on a major modification to his Synesso that will allow exactly this.

The Slayer takes a much more simplistic approach to manipulating the flow rate. The needle valve is adjustable, but is not intended to be changed dynamically during the shot. You can set the valve to deliver a specific flow rate when the paddle is in the middle position. This (theoretically) allows you to simply and repeatably set the duration of this "0-bar" phase in concert with the rate of the subsequent pressure ramp. It's not infinitely tune-able, but it's simple and easy to repeat. The Slayer also enters a "full flow rate/pressure" mode when the paddle is moved to the far left, where the water passes through both the needle valve and an 0.7mm jet. This is potentially important, because having a highly restricted valve for the entire shot can prevent you from ever attaining the pressure required to produce an actual espresso.

I will certainly report on how this works in practice when I receive mine. It is likely a couple of months away; they recently announced to us that they're beginning production in the next 2-3 weeks. Given my place in line, I will be thrilled if I receive it before the SCAA show this April.
bostonbuzz wrote:Dan may have something to be considered if I may quote him, "For what it's worth, I've noticed that excessively long preinfusion leads to "muddy" tasting shots (i.e., greater than 6 seconds)."
I've read this comment before, and it's a good connection to make (kudos for bringing it up). Actually, I've looked with interest at preinfusion in the past, and I had concluded previously that it was not such a good idea beyond taming an espresso machine with an overly aggressive pressure ramp (increasing the so-called forgiveness factor of the machine). I think that doing it at line pressure, with line pressure set around 3-4 bars, is a really major difference from what I outlined in my discussion above, where the pressure on the puck is held at or near 0-1 bars for an extended period of time. Jason Prefontaine calls this phase "pre-brewing" to differentiate it from preinfusion, which almost everyone assumes is done at line pressure. I think it's fair to say these are very different techniques, so while I'm somewhat reluctant to use the term Jason coined, I think it's important to keep the distinction in mind.
Nicholas Lundgaard

Advertisement
User avatar
another_jim
Team HB
Posts: 13871
Joined: 19 years ago

#8: Post by another_jim »

A lot of the taste work in preinfusion was done comparing E61 groups with faster dwell time groups. However, E61 groups also tend to start out significantly warmer than flooded LM groups. An extended preinfusion in a chamber at 96C to 97C may lead to very different results to one in a cooler chamber. I used to also got muddiness on E61s on excessive dwell shots; whereas this is not an issue on the long Strega preinfusions, where the group is a lot cooler.

In general; I think the time extended shots one does when using profiling machines may require lower group and brew boiler temperatures to be effective.
Jim Schulman

User avatar
aecletec
Posts: 1997
Joined: 13 years ago

#9: Post by aecletec »

Luke at LTD in Brisbane uses his EP in manual based on flow per shot - after dialing in grind he mentions that he changes pressure manually to different targets as staling sets in rather than changing grind. His shots are nearly always balanced in flavour profile.

User avatar
Possepat
Posts: 477
Joined: 12 years ago

#10: Post by Possepat »

Great thread. Thanks for sharing your thoughts everyone, this is what HB is all about.
"Do what you want, you're gonna do it anyways!" - My father

LMWDP #365

Post Reply