I created this fan duct to see if a 360 degree part cooling duct could work on the Prusa. I ended up getting pretty ok results. My new v2 model has much more airflow directed right at the end of the nozzle when compared with my v1 model, while still maintaining a mostly downward airflow pattern. This model is still experimental and in development.

The v1 model is based on my previous design for the MeCreator 2 fan duct, which worked so well I thought I could try it again. The space constraints made things very challenging; things like the PINDA probe and the cables for the heater block got in the way. The right angle in the airpath right after the fan does not help with building a good static pressue whithin the fan duct. I want to see if using the angled fan improves results but don't see it as a worthwile endeavor.

My version 2 model aims to address these issues by allowing a larger area for air to exit. This creates the issue of not having consistent air pressure throughout the duct, which turns out to be the case, but it doesn't make much of a difference because my v3 model which accounts for this ended up performing worse. My v3 model allowed much more airflow to the rear but in real world testing and flow simulation, i found the rear airflow to interfere more than help.

Version 1 was a bit of a mess, so I cleaned it up to create a template. I then used that template to create version 2 (a brand new mess). The template and both models can be found by following the Onshape link below. The document can be copied and modified to your hearts content. Let me know if you create a better duct design than I have.
https://cad.onshape.com/documents/a7e64a740cdc3aa4d71dcca2/w/8a8a5974a2c91f518ba4a4dd/e/72afdd29311311e2e6b82430

PS. The main picture (the one that looks significantly better than the others) was by nlnpereira. I stole it from him because his print was way better than anything I made, and also it is genuinely a great shot.

NOTE/ADDENDUM:

After extensive testing, I have to conclude my fan duct design is not better than the stock one overall, and my findings are quite interesting.

I have done a variety of test prints, and I only found one case where my fan performed better than stock. In my testing, my fan does better at small prints than the stock fan. This is not because of rear airflow, because my v3 model performed notably worse than my v2. The only reason my fan duct did better in this case is because the stock fan duct blows air over a much larger area while mine has it focused at the nozzle. While small prints benefit from this, bridges can turn out worse.

I actually have an idea for a new fan duct design (forgoing rear airflow) which should perform even better for small prints than my current one. As far as creating a duct for rear airflow, I think I need to give it up until I get a new idea. If someone more skilled than I would like to give it a try, my work can be found and copied off Onshape using the link in the original post.