Brew pressure profiling update 2

Beginner and pro baristas share tips and tricks.
gscace

Postby gscace » Mar 11, 2007, 8:40 pm

Howdee:

Well I feel quite intimidated to post my paltry attempts at learning about pressure profiling after the Ken and Jim juggernaut published their freezing results. But here goes.

I had time this weekend to devote to coffee since Casey and Anneke are off visiting relatives. So two group of people came out to play with pressure profiling. On Saturday, Alan Munter, Sarah Munter, and Nicholas Cho showed up for a lunch of smoked salmon and asparagus, with coffee for dessert. Also at the house was my friend Dave Arnoff, who had just received a Brewtus as a gift, and was diving into espresso - once we fixed his machine. That's another story.

Saturday's testing was not blind, but involved everyone standing in part of my basement, in relatively close proximity to the espresso equipment. Fortunately everyone seemed acceptably clean and well bathed. Nick brought coffee with which he is very familiar - Counterculture Toscano. Nick led the way, pulling shots in which the extractions were divided into thirds by time, with the collected volume in each third contained in a separate shot glass. It was pretty interesting to taste what was coming out of the spouts. The thickest mouthfeel was in the beginning third, and also the sweetest taste. The last third was watery and bitter by comparison. It made sense to try to limit the amount of liquid extracted in the last third in comparison to the amounts extracted in the first and second thirds. I adjusted the pressure profile to accomplish this. The pressure profile was adjusted so that the maximum pressure was 140 psi at the pump, which corresponds to about 135 at the cake, under WBC flow conditions. At extraction's end, the pressure was 60 psi, just enough to keep flow going thru the cake. Nick was of the opinion that the pressure was too high, so I adjusted the profile downward in pressure, eventually settling on a value of 110 psi max. That's 7 2/3 bar under WBC flow conditions, but over 8 in no-flow equivalent pressure (what a dumbass unit, huh?). This made the shots thicker, but smaller in volume, well less than an ounce. This morning, I made a Nick-style shot and weighed it at 18g of syrup extracted from an 18 g cake. I have to say that Nick's shots ended up being very sweet, and I became a bit of a convert, until I tried to replicate them today. I haven't said much up til now about profiling vs. constant pressure. But we tried both, with Nick leaning toward profiling and thinking that there was enough to it that maybe I oughtta think about bringing the pump system to Portafilter 2007 in just over 2 weeks.

Today a different cast of malcontents showed up, including Steve Jones, the famous Eric S., and Kurt Heinrich. All are schooled in the art (just in case I need to invoke the "obvious to one schooled in the art" schtick later). The idea was that if Nick's pressure profile brought out the sweetness in Counterculture's Toscano, I should be able to brew shots of Toscano using profiling and not using profiling, nice volunteer tasters should be able to discern differences and the differences should demonstrate a preference for one pressure method over the other. Or not. During the tests one person acted as a runner between me and the folks out on the deck basking in the late winter sun (actually very pleasant). In this way the tasters wouldn't know whether or not pressure profiling was being used, and they could make disparaging comments out of earshot. The runner rotated after every few shots, so only two of the three got to taste any one shot. 11 shots were brewed. Shots 1 and 2 were not profiled, with both tasters preferring shot 2 to shot 1. Shot 3 was profiled, with both tasters preferring shot 3 over 2. Taster 1 called shot 3 slightly sweeter than 2 with sweetness lingering longer. Taster 2 wrote "very sweet on the tongue immediately, excellent mouthfeel. Shot 3 was his favorite of all shots tasted. Taster 1's notes on shot 4 (also profiled) seem to prefer shot 4 for a combination of bright sweetness and caramel. Taster 2 found it bright initially, but better tasting in a longer sip. My notes for shot 4 indicate that I made a grinder adjustment in the coarser direction since shot 3 was pretty short in volume. Taster 3 replaced taster 2 for shots 5-8, and taster 1 for shots 9-11. Shot 5 was profiled. Taster 3 reported "nice mouthfeel - delicious". Taster 1 said "slightly acidic - minimal sweetness" - Hard to know if taster 3 was calibrating his taste buds while taster 1 was getting tired. Shot 6 was brewed at constant pressure (no profile). Taster 1 noted "hint of bitterness on 1st sip. Slight bitterness on 2nd sip, overall slightly muddy". No preference was given between shot 6 or 5. Taster 3 clearly preferred shot 5 to 6. Shot 7 was not profiled. Taster 1 liked shot 7 as very clean, bright, hint of chocolate and very sweet. Taster 3 also liked this shot, comparing it favorably with shot 5, his first. He added that there was a touch of bitterness in the last part of the shot. Shot 8 was not profiled, and I ran out of Toscano, substituting some coffee that I had roasted a week ago. Both indicated preference for previous shots (there ya go about my roasting, huh?), with Taster 1 writing "very thick mouthfeel, but most unpleasant shot so far. Muddy flavor combo, nothing stood out." Taster 3 opined "no "nose". Ok mouthfeel - same as 6.

So far, shot 3 (profiled) stood out as particularly good. Taster 1 seemed to be able to discern the switch back to un-profiled in shot 6, with preference for shot 5. Taster 3 also preferred profiled shot 5 to unprofiled shot 6. Both tasters picked up that shot 7 was better than 6 even though both were unprofiled. My notes for 7 indicate that the height of my tamper in the basket indicated that I updosed it compared to my norm, and the extraction time was long at 32 seconds, compared to shot 5 at 26 secs and shot 6 at 24 seconds. Here updosing and resulting long extraction time seemed to trump everything else. Both tasters knew something was up on shot 8. Of course what was up was that I had changed coffee.

Taster 2 replaced taster 1 at this point. Shot 9 was profiled, and shots 10 and 11 were not. Taster 2 preferred shots 10 and 11 over shot 9, noting that shot 9 was much like 4, his least favorite, and that shot 10 was pleasantly bright with some sweetness and 11 was the mellowest of 9 -11. Both 10 and 11 had good mouthfeel. Taster 3 clearly preferred shots 10 and 11 over 9, with 10 and 11 both "possessing earthy taste, then sweet."

I possibly obfuscated the results for shots 9 - 11 by trying to do a better job of grinder control. I brought my Kony home from work, with the idea that I would set the Robur for pressure profiling, set the Kony for non-profiling, then halfway through the tests, switch grinders so that the Kony was grinding for profiling etc. In practice there was just too much stuff going on. First, only one grinder (the Robur) was filled with Toscano, so no Kony action could happen early in the tests. The opportunity for using both grinders presented itself once both were loaded with the same coffee (apres switch). Unfortunately, by now the tests had progressed fairly far, and I just introduced something new into the mix.

By now everyone's taste buds were pretty blown out. And the results were somewhat inconclusive in one sense, but conclusively showed that I am pretty inexperienced at this. I learned that there is a really limited number of shots that can be tasted before it just gets hard to taste. That means ya gotta keep it simple. I can't tell if shots 10 and 11 were preferred over 9 by virtue of grinder choice, or by not pressure profiling. By looking at folks' preferences it appears that a difference could be discerned, but then there was that shot 7, the updosed-non-profiled one that got good marks. That means that the level of benefit is of the same order as the noise between shots made by an amateur barista using good equipment and fresh coffee. That is to say that it's a real bitch trying to be consistent enough to discern the gain or lack of it.

Afterward, we celebrated our adventure with more coffee (not Canadian Club you lush). The reason we did this was because we sat around discussing the shot volumes that I had produced. For these tests we used the pressure profile that Nick seemed to like. In my opinion, Nick's profile automatically limited the amount of coffee volume because the pressure values were generally so low. I think everyone thought the shots were very small in volume compared to the amount of coffee used, so I reloaded the profile I had been using before Nick's arrival. This profile ramped to 140 psi, rather than 110, so it represented a substantial difference. With this profile, shots were produced with more volume by at least a factor of 2, with viscosity well-controlled at the end of the extraction. Visually, the shots were of satisfactory volume compared to what the tasters expected of an espresso made with 18 g of coffee - that is to say a double. With crema, the volume was at least a factor of 2 greater than produced by the Nick profile. The tasters liked them for their balanced taste and good mouthfeel, and so did I. But that's about the gist of it. We couldn't do any more testing if we wanted to.

My opinion so far: First, I think the idea of controlling volumetric flow rate during the extraction by varying pressure has merit because the extraction fractions that I tasted in shots that were parsed out into thirds tasted best in the first two thirds, and worse in the last third. Thanks Nick for demonstrating this to me on Saturday. Doing the evaluation is devilishly complicated. Eric rightfully points out that isolating a single variable is impossible when investigating pressure profiling. For example, not-profiling requires a coarser grind in order to keep the shot volume the same, so that means particle size is now different and we're comparing pressure profiling with fine particle size to non-profiling with coarser particle size. Which cart is driving which horse and by how much? My hat's off to people (Jim and Ken) who can pull off days of shot testing. It's hard as hell to be consistent. That being said, I'd like to do the tasting experiment again. I learned that one really has to keep it simple because it's pretty difficult to do the tasting. Thanks to Nick, Alan, Sarah, Steve, Eric, and Kurt for their help!

More when I get something worth spouting off about.

-Greg

User avatar
HB
Admin

Postby HB » Mar 11, 2007, 9:37 pm

Thanks Greg for the early results. I now am beginning to appreciate the complexity of testing for beyond gross differences; I think the discussion going forward will be as much about the means of comparison as the comparisons themselves. In the past we skipped this step, listening politely to assertion X or assertion Y without questioning the basis of the comparison. Ken and Jim's recent publications have made it difficult to skate the issue.

I am thinking about it though, especially in light of a grinder project on the horizon. I've learned the hard way that your taste buds burn out fast, so my leading "keep it simple, stupid" idea is to have two setups optimally dialed in at all times. For example, say you wish to compare two grinders. Dial in grinder X and grinder Y to the best of your abilities. Now dose / prepare two (discreetly marked) baskets out of the portafilter, ask one of your kids to randomize them. Now make and taste the two, ideally side by side if you've got a multi-group machine, otherwise in rapid sequence. Record simple 0...3 scores (same, slight preference, strong preference, mind-blowing preference). Over the course of several weeks, a pattern should emerge.

I don't know if such a lazy man's approach would withstand statistical scrutiny, but it's at least practical.
Dan Kehn

User avatar
another_jim
Team HB

Postby another_jim » Mar 11, 2007, 10:13 pm

A good piece of testing, Greg. The point at this stage is not to get anything "statistically significant" (every stats person despises the phrase anyway); but to get a feel for how the whole setup works, and the types of problems that need solving before formal testing can begin.

It looks like you've found that mouthfeel is the prime taste variable, and that pressure profiles based on flow rates is the control variable. That's a big step right there.

You've had the "learning experience" of how tough it is to analytically taste a lot of shots in a row.

I'm sorry I'm repeating myself, but maybe now with some testing under your belt, you'll appreciate my point: how much of the tasting can you replace with physical measurements? Again (tell me when I'm boring you), this has nothing to do with Ken's article or design, where all attempts to physically or organoleptically characterize the difference between fresh and frozen failed in the design stage, and which was therefore designed as a no difference test, eliminating all extraneous variables. You hope to do the exact opposite -- you want to characterize the physical difference in the cup between different pressure profiles and prove that these correlate to some organoleptic properties that people value.

Here's some suggestions:

On the profile end -- how tough would it be to measure flow rates, say in mL/sec and graph them over the course of the shot? If this data could be collected, I could run a time series regression of how the flow rate now correlates to what the pressures are doing up to now. This would aid in designing new profiles to achieve desired flow rates.

On the tasting end: The density, fat content, and solids content of the shots would be good candidates as correlates of mouthfeel. Is there any easy way of measuring these?

The payoff would be an equation relating pressure profiles to those physical variables in the cup which correlate to mouthfeel. Once you have that, designing a taste test to prove or disprove everything will be a piece of cake.
Jim Schulman

gscace

Postby gscace » Mar 11, 2007, 11:02 pm

another_jim wrote:A good piece of testing, Greg. The point at this stage is not to get anything "statistically significant" (every stats person despises the phrase anyway); but to get a feel for how the whole setup works, and the types of problems that need solving before formal testing can begin.

It looks like you've found that mouthfeel is the prime taste variable, and that pressure profiles based on flow rates is the control variable. That's a big step right there.

You've had the "learning experience" of how tough it is to analytically taste a lot of shots in a row.

I'm sorry I'm repeating myself, but maybe now with some testing under your belt, you'll appreciate my point: how much of the tasting can you replace with physical measurements? Again (tell me when I'm boring you), this has nothing to do with Ken's article or design, where all attempts to physically or organoleptically characterize the difference between fresh and frozen failed in the design stage, and which was therefore designed as a no difference test, eliminating all extraneous variables. You hope to do the exact opposite -- you want to characterize the physical difference in the cup between different pressure profiles and prove that these correlate to some organoleptic properties that people value.

Here's some suggestions:

On the profile end -- how tough would it be to measure flow rates, say in mL/sec and graph them over the course of the shot? If this data could be collected, I could run a time series regression of how the flow rate now correlates to what the pressures are doing up to now. This would aid in designing new profiles to achieve desired flow rates.

On the tasting end: The density, fat content, and solids content of the shots would be good candidates as correlates of mouthfeel. Is there any easy way of measuring these?

The payoff would be an equation relating pressure profiles to those physical variables in the cup which correlate to mouthfeel. Once you have that, designing a taste test to prove or disprove everything will be a piece of cake.



Ya know I gotta say that I was really trying hard not to be the measurements guy on this, and get down to what really matters in coffee, which is TASTE, and in the end I know that you're right on the quantification thang. One thing that struck me on Saturday was how easy it was to tune the flow rate of the various subsections of the extraction process by varying pressure. Nick was just changing out the cups every 11 seconds and I was looking at the levels and damn if we didn't get them all equal. I can easily suck them up in a syringe and measure that, so there is a quantifiable parameter. Alan Munter also works at the Institute of Slow Thinkers, in the Neutron Physics bunch. We were talking about how to quantify the chemical makeup of some of this stuff. I suspect that I can get samples analyzed without too much trouble, or I can find out what tests can detect changes in fat content. I did a few measurements of solids content. I'm having a few problems here because I have got to standardize on a shot volume that I think is a good compromise for mouthfeel and other sensations, and I'm concerned about the drying process. I dried coffee cakes in the oven after brewing, and the weighing results were time dependent, even after baking at 300 degrees for 1.5 hours.

But even though I found it somewhat difficult to wrap myself around the testing today, the aggregate of yesterday and today's work seems to point in a direction that feels reassuring to my engineering self. Also, thanks for pointing out the difference in quantifying the lack of detectable change, vs. characterization of differences. And you ain't boring me by no stretch of the imagination. I very much wanna get to the bottom of this, and I think that I can.

-Greg

User avatar
AndyS

Postby AndyS » Mar 11, 2007, 11:39 pm

gscace wrote:I'm concerned about the drying process. I dried coffee cakes in the oven after brewing, and the weighing results were time dependent, even after baking at 300 degrees for 1.5 hours.



Try carefully breaking them up and spreading them out on a small plate. They will reach an equilibrium weight much faster that way...assuming you didn't do this already.
-AndyS
VST refractometer/filter basket beta tester, no financial interest in the company

User avatar
another_jim
Team HB

Postby another_jim » replying to AndyS » Mar 11, 2007, 11:59 pm

My ditto on that. A complete cake takes about 3 to 4 hours, while a spread out one does nicely at 275 for 2 hours. One possible gottcha; I did this in the ultra-dry winter -- spring in Baltimore may make coffee drying a whole lot tougher.
Jim Schulman

User avatar
another_jim
Team HB

Postby another_jim » Mar 12, 2007, 12:08 am

gscace wrote:Ya know I gotta say that I was really trying hard not to be the measurements guy on this, and get down to what really matters in coffee, which is TASTE, and in the end I know that you're right on the quantification thang.



Espresso is a very lousy area to play out the science versus art game. When cupping coffee, everyone uses completely uniform roasts, grinds and brewing -- the only variable is the coffee, and your nose is a whole lot more accurate in figuring out how two different coffees vary than any GC/MS. With espresso one is playing with pressure, temperature, water/coffee proportions, particle sizes, times and flows -- for this stuff instruments beat the crap out of any nose. So even an artsy-fartsy type like me thinks it's best to characterize how the cups differ physically before investing a lot of tasting time.
Jim Schulman

User avatar
AndyS

Postby AndyS » Mar 12, 2007, 12:10 am

gscace wrote:The thickest mouthfeel was in the beginning third, and also the sweetest taste. The last third was watery and bitter by comparison. It made sense to try to limit the amount of liquid extracted in the last third in comparison to the amounts extracted in the first and second thirds. I adjusted the pressure profile to accomplish this.



Greg, we're all in this exploratory phase together.

I appreciate how complex this stuff can get, but I don't understand. If limiting the amount of liquid in the last third is your goal, you don't need the fancy profiling gear. Just stop the pump and enjoy the ristretto!
-AndyS
VST refractometer/filter basket beta tester, no financial interest in the company

User avatar
diab0lus

Postby diab0lus » Mar 12, 2007, 1:49 am

gscace wrote:I've also learned that there are pressure issues at play in all pump-driven espresso machines that affect orderly pressure ramping at the start of brewing and may or may not have a small effect on coffee taste. Relatively cold water is pumped to the brewing water reservoir (boiler or heat exchanger) in espresso machines without feedwater pre-heat. The water is heated within the reservoir, and the pressure within the reservoir increases due to thermal expansion of the liquid water. When the group solenoid is activated, the water pressure is instantly released and a small spurt of water flows onto the coffee. I've reduced the effect of this to a large degree, but I haven't eliminated it entirely. I think that the way to almost completely eliminate it might be to install a normally open solenoid-actuated valve inline with the pressure-relief line from the brew boiler to the pressure-relief valve in the drain box. I would connect the valve to the pump circuit, so that the valve would close when the pump is activated. Rather than adjust the pressure-relief valve to the recommended 12 bar, I would adjust it to crack open at 2 bars. When adjusted in this way, the boiler would maintain only 2 bars pressure when idling. The solenoid valve would close on pump actuation, allowing pressure to build to brewing pressure levels. At the end of the brewing cycle (pump de-activation), the solenoid valve would re-open, again limiting boiler pressure to a low level.



I copied this post from the first thread on this topic.

Last week I tee'd the line between the pump and the boiler on my Silvia and installed a Parker series 25 NC 1/4" solenoid valve followed by a Swagelok RL3, which relieves itself into the water reservoir. This is my take on preinfusion since I am not plumbed in [it works well too]. I switch on the valve at the same time I hit the brew switch, the pressure climbs to 3 bar over a few seconds, then I power off (close) the valve once the first drop appears [no bottomless pf, yet], pressure glides to 8.2-8.5 bar and the shot continues. Anyhow, after reading what you have posted above, it makes sense to open the valve before I hit the brew switch to adjust the boiler pressure down to preinfusion pressure, then hit the brew switch. This mod could serve two purposes. Your thoughts?
-Ryan

gscace

Postby gscace » Mar 12, 2007, 6:58 am

AndyS wrote:Greg, we're all in this exploratory phase together.

I appreciate how complex this stuff can get, but I don't understand. If limiting the amount of liquid in the last third is your goal, you don't need the fancy profiling gear. Just stop the pump and enjoy the ristretto!



Yeah.

But now that we've got this fancy schmancy thang, I had to make a cool profile that looks sort of like the Sydney Opera House.

-Greg