I'm pretty sure stls/regen.stl is the recovery core you're taking about tho I can't open it in a online stl viewer: https://drive.google.com/drive/folders/1py2YwmwBEcvmdw18SKwx...

In the photos it seems to be a bunch of nested single-perimeter cylinders that are joined at a few points to maintain spacing. Easy enough to model, but I agree the documentation is horrible and there's no way to contribute.

Commercial units are not comparable because they're way more expensive despite being so simple

You can get a commercial unit for under $400 USD. Search for "Pioneer 50 ERV", which claims 97% efficiency or look on aliexpress for units under $200.

That said, I find these "regenerative" heat exchangers too limiting as they generally only work for a single room/space.

My price is for a pair, remember, not one. And those pioneer things don't work very well, poor flow and efficiency unfortunately. They claim high efficiency but it's only at the lowest flow rates and not the average for the cycle, only at the beginning of the cycle. There is one company that keeps emailing me trying to sell me units wholesale, I keep telling them to send me actual test data and then I might buy some (for testing), and they never do.

Prusaslicer opens it, and it is the right size and shape to be the core, but there's no internal geometry in that STL, at least not how Prusaslicer renders it.

I'd be interested in seeing a diagram of how the air flows through it, if such a thing is available.

Edited to add: There are instructions in the WM12 manual to use your infill settings to make the old version of the core. Page 15 of the manual states there is a python script in the source files to generate the new core, but it only works for their particular printer. I wasn't able to find the script.

Thanks for mentioning that. It seems the PDF and the google docs "manual" have diverged a bit. (at least in page numbers).

I was curious about materials for the core. I know that they're supposed to exchange humidity as well as heat. I know PLA will absorb and release water but I would guess it wouldn't transfer enough to be very efficient. Though I would be happy to be wrong.

I assume the Core in my Panasonic whisper comfort was made out of something more permeable than "simply" extruded plastic. (would love to know more if someone has details).

The MPLA (not pla) does not absorb the water vapor, but it condenses on the plastic and then evaporates. There is also the option of sorbent, which grabs the water vapor and then releases it with each cycle, before condensation occurs, getting higher efficiency.

The material in the whispercomfort is IIRC a fiberglass paper. It can only transfer liquid water, not solid water, and not vapor, effectively. The water condenses, whets the paper, then evaporates out the other side. It's a reasonable approach but does not work well in very low temperatures, and also the condensation process implies certain limits to efficiency, same as a non sorbent coated regenerative heat exchanger. The water doesn't start to condense until 100% RH, so it is 100% when it leaves the exchanger, while the air coming in is lower than that, usually. Thus water is lost.

Very interesting! Do you have a method of adding sorbent to a 3d print?

I wonder if it'd be worth having three TW4 modules with sorbent on only one of them. That way you can control humidity better by choosing 2 of the 3 to use at a time. Eg after a long shower you might wanna discard water vapor for an hour

What's the significance of the model numbers 4 and 12?

Yeah that would probably work, it's just a matter of cost. A lot of people balk at $700 CAD, the eventual price is supposed to be $1300 for a pair, you do that and you are looking at a lot of $ for the flow and energy recovered.

The numbers are the number of versions before it stabilized, so 12 tries for the window mount, 4 for the through wall. It took a while.

There is the tw4, which is made to be put in a wall, and there is the WM12, which goes in the window. The main focus is the TW4. There are instructions in the manual for making an ERV core. It is not trivial.

I understand that the addition of desiccant material is the core aspect of what makes this an ERV. I don't actually see any clear explanation of how the desiccant material is added to the printed part. While the creator (open_erv2) mentions that sorbent/desiccant can be used to handle moisture ("If you have sorbent, it gets grabbed out of the air before it can condense"), they don't specify how it's incorporated into the design.

Is it added mid-print? After printing? Is it difficult to add?

I invented a process using some environmentally friendly solvents, grinding the sorbent and so on. It is not diy friendly unfortunately so there is little point in sharing the recipe.

Interesting. I'll paste it below.

Note on printing the regenerator/heat exchanger:

The latest and greatest heat exchanger is produced directly with python script generated gcode specific to the printer I use and cannot be practically produced diy, unfortunately. However the old model can be, and the STL is included for that, in the source repository. To do this, simply use Cura, load th STL in, put it in the center of the build plate, and set it to do “lines” infill with about 2.5 mm on center (between centers of the lines) spacing and 0.45 mm width, no top layer and no bottom layer (set them to zero). Check the preview and it should show you a structure which is much like grid infill, parallel channels which are square in cross section, with the outer wall. Tape can be applied over the nubs on the side to fit in an oversized pipe, or they can be sanded if the pipe is too small. You could also use grid infill, but the roads tend to have problems where they intersect. When the nozzle goes over one road, it wipes the plastic off, and not enough is deposited on the lee side. I don’t know how to solve this in Cura without using lines infill. If you could make it so the nozzle went in alternate directions each layer that would probably solve it well enough.