This is pretty interesting! And Thanks!!! My question is how do things look once one introduces 1) radial flow non-uniformity (faster in the center compared to the perimeter) and 2) changes in solubility coeficient and concentration gradient based on radial and axial temperature variation due to fluid flow non-uniformity. Is it feasible to do these studies here or does serious coin and time need to be spent?

-Greg

another_jim wrote:It looks like the exraction lines converge. So if you ran the shot, say, 90 seconds; all layers would be extracted almost equally.

Yes. There is also the caveat that the "extraction" of the first and subsequent layers only really counts when those solids have left the last layer.

That is to say that solids from the first layer that are in the last layer when you stop the shot have no bearing on the extraction that is in the cup. So, the effective extraction of each layer would end a second or so before the layer beneath it, which drives the layers towards equal extraction even earlier than it appears. Basically you can subtract a portion of the dwell time for each layer. If the last layer begins 6 seconds into the shot, then the first layer ends roughly 6 seconds before the end of the shot (offset by the amount of absorbed water).

This is what led me to abandon the idea of an inverted cone portafilter to increase the contact time at lower levels in the puck. It simply isn't necessary for top to bottom extraction.

gscace wrote: This is pretty interesting! And Thanks!!! My question is how do things look once one introduces 1) radial flow non-uniformity (faster in the center compared to the perimeter) and 2) changes in solubility coeficient and concentration gradient based on radial and axial temperature variation due to fluid flow non-uniformity. Is it feasible to do these studies here or does serious coin and time need to be spent?

-Greg

Greg, I can take a crack at it.

These are excel models for 1D computations. Temperature is easy enough to build into this model, as would an approximation of average viscosity using concentration and temperature as the driving forces.

Splitting it into a radial model with the average conditions tied back to the expected output of the 1D model also seems doable, but it will take a little time. Right now, I have a 29 by 15 matrix for each layer that is driving the model. The initial condition is fresh water added to a puck layer, which interacts with it and then moves to the first line in the second layer as the second line in the first layer receives another dose of fresh water. The whole thing cascades from here.

If each cell in these layer matrices represent an average value for the radial cross section of a basket, we need another 29 by X matrix for each value, with X being the resolution across the basket. 29 seems as good a number for X as any, as it gives us 1mm accuracy across the radius of a 58mm basket. This would give us half a radial cross section, with the average value tied to the 1D model, a zero-flow boundary layer and a velocity profile driving the rest of the shape being characteristically derived by temperature and viscosity. Seems doable, but CFD would be the preferred method.

Anyone here with CFD access/experience want to collaborate?

Cheers!

- Jake

I'm really intrigued by some of the ideas here about changing basket shape to compensate for changes in EY vs. depth in an espresso puck. I've done some 1-D modeling siilar to Jake's work that might be helpful. The model is described on a high level here:

https://quantitativecafe.com/2021/11/14 ... xtraction/

and implemented in a Jupyter Notebook here:

https://github.com/quantitativecafe/blo ... s-ey-model

I then measured TDS for salami shots using fine and coarse grinds of the same coffee, and adjusted the parameters of the numerical model to fit these results. With these parameters, I was able to take a closer look at the model output for each grind setting to get some idea of what might be happening inside the puck for the two shots. One result which has really helped me build intuition is this animation, showing changes in concentration of soluble coffee in the water and in the coffee grounds vs. vertical position in the puck vs. volume output:



Each frame shows the situation at a particular output volume for both shots, so with the coarse shot the clock runs slower, because the coarse shot reaches that volume much faster. Interestingly, plotted this way, it's actually quite difficult to tell the difference between the shots.

We also see that the concentration of soluble coffee in the grounds is higher at the bottom of the puck, as expected, because extraction is driven by concentration differences, and for most of the shot the water passing through this part of the puck is already quite saturated.

Finally, in either shot, after about 20 seconds the curve is quite a bit shallower than when it starts, so we're looking at more even extraction conditions.

One thing I have not yet done is to keep track of percent extraction at each moment in time for each "packet" of water passing through the puck, then use this to look at what percentage of the extracted coffee in the cup comes from early extraction vs. late extraction. Although the animation makes it look like the fine and coarse shots are very similar at a given output volume, I wonder if we would see a significant difference in the early/late extraction contributions between these two shots.

With all of this in mind, I'm trying to think if a different basket shape could be used to compensate for changes in EY vs. depth in the puck. The suggestion has been made that a conical basket could produce lower flow lower in the basket, increase contact time, and increase EY toward the bottom of the puck. But it feels somehow like this might be the wrong direction to go, since the concentration of dissolved solids in the water changes quite a lot with time--indeed, even the shape of the curve changes quite a lot--which suggests that a single basket shape could not compensate entirely for this effect.

One other thing that's been pointed out is that a very thin puck would have more even extraction than a thicker one. Talking with others previously, I've had the impression that this matches up with their experience, e.g., of using the same dose in a basket with a smaller diameter--if I remember correctly, the impression was that this gives a more rounded cup, but also less clarity.

I'm a bit late to the discussion...but...

gscace wrote:There is much more available energy to drive dissolution of coffee solids at the top compared to the bottom and in regions where fluid flow is higher, since the available energy for driving the process is directly related to water flow rate, specific heat and temperature as the water flows by the coffee.

I see that a lot of discussion that follows focuses on the flow rate part of that statement.
I wonder if we could reduce the temperature-related energy unevenness by preheating coffee and/or basket?

ADDED:
Also, I just looked at SWorks basket...wouldn't gradual increase of hole density towards the edges of the basket and shower screen improve flow evenness?

Those are good questions. WRT heating coffee I think that increased staling rate of heated coffee in air might affect itnegatively. An interesting experiment would be to heat ground coffee in a microwave and then dose it, brew it, look for a difference in extraction yield and look for a taste difference. With respect to baskets, the thermal conductivity of stainless steel is really poor. It transfers heat very poorly by conduction and especially to irregularly shaped particles which have minimal contact with basket walls. Regarding flow uniformity - there are ways to promote flow toward the perimeter and in my opinion they are worth investigating. Increasing hole density toward the perimeter is worth looking at.

-Greg

fiu-bździu wrote:I'm a bit late to the discussion...but...

I see that a lot of discussion that follows focuses on the flow rate part of that statement.
I wonder if we could reduce the temperature-related energy unevenness by preheating coffee and/or basket?

ADDED:
Also, I just looked at SWorks basket...wouldn't gradual increase of hole density towards the edges of the basket and shower screen improve flow evenness?

Seems like mischief makes the cheapest of the new generation baskets:
https://www.mischiefworkshop.com/produc ... c%8ch26mm/

Anyone know how many holes are in the Weber 20g Unibasket?