Parametric Hero Me Dual 5015 Duct

Update 2020-04-19: Mediaman has been working on an updated version of the base for this (Gen 4), and I am reworking the design to work with that. I am also doing the 'lightweight' version with the two halves printed separately with no brace. I have a few ideas for optimising that, but I should have something soon. I will need to do airflow modelling for that to see how it is.

To print this, the fan support fins, the flat front face of the fan holders, and the horizontal brace bar should all lie on the build surface. I am going to try printing this tilted 90 degrees instead of 45, with the ends of the duct nozzles towards the build plate. This may help strengthen the mounts due to a better orientation of the layer lines, and as far as I can tell, use less support material as well. I encourage people to try it :-)
Support from build plate only should be enabled as well.

Update 2019-06-06:
I have improved my simulation model, and made some changes, including strengthening the brace and moving it up. V13, both with and without fins should work.

Update 2019-05-06:
I made some small positioning fixes, so this should now work with the Microswiss and E3D hotends. I also increased the size of the horizontal brace, which should stiffen things up a bit. The version I recommend is dual5015_V11_fins.stl. Looking at the airflow simulations for the no-fins version, it looks like it isn't really putting the air in the right place, so the fins really do help

Update 2019-04-24:
Mediaman noted that the duct seems to be too far forward relative to the nozzle, which was particularly noticeable with the E3D V6 and Volcano hot-ends. I took a look at the simulations and I think he is right, so I now have dual5015_V10.stl and dial5015_V9_fins.stl. Simulations are in progress. If anybody wants to see the simulations I am doing, you need a free SimScale account, then go to: https://www.simscale.com/projects/acwest/parametric_hero_me_without_fins-_surface_at_0-5mm/

Update 2019-04-23:
I just printed a version dual5015_V8.stl and dual5015_V8_fins.stl that has the fins, and slightly thicker walls, to make it more resistant to heat deformation, as the walls at the tip of the nozzle are very thin. It shouldn't affect airflow much...
I have also included a number of levelling tools of different thicknesses to make setting the height easier, as I have found that it is a pain to get right.

Update 2019-04-22:
After mounting this, I discovered that at least on the version of the base I am using, the nozzle was a bit too high, so I reworked the nozzle height by a bit. Dual5015_V6.stl has the height corrected...
I did discover, using an earlier version of the duct, that the angle of it brings it too close to the heat block, so I am testing a version that should improve this. Not a problem if you are using ABS, but I'm not...
I just did more simulations, and the V7 duct which keeps the nozzle further way from the hot-end has almost identical flow characteristics as the previous version. dual5015_V7.stl is the latest

Update later on 2019-04-21:
I spent much of the day working in SimScale to optimise the design a bit. The latest version is now: dual5015_V5.stl, I am currently printing it, but the simulations are working out really well with it. The screen-shots from many of my simulation runs are also uploaded.

Update 2019-04-21:
Due to some issues with the output produced by my code when importing into SimScale, I did a pretty much total rewrite of much of my polygon 3D spline extrusion code. It was a pain (who knew triangulation of an arbitrary polygon would be so finicky??) but I have something that seems to be much higher quality. I am just starting the next round of simulations, but I am posting my latest duct version now. It is fairly radically different from the earlier ones as a result of my earlier simulation work. The latest duct is sufficiently efficient that I run the fans at 30% to prevent excessive cooling of the hot end. I will be doing a version using both a single 5015 fan and a single stock fan later, but the dual fan setup is still worthwhile, as two fans running at 30% are quieter than one fan running at a higher rate.
My simulation testing has shown very little if any advantage to using the fins, so I recommend going with duct_nozzle_1_75.stl

Update: 2019-03-27:
No new files yet. I eventually got SimScale to produce results that look reasonable, and am working on some cleanup for my code so that the STL it generates is more compatible to the system. Once I have it cleaned up a little more, I will be posting my latest version, as well as the code

Update: 2019-03-15:
I've been cleaning up the airflow, and made some modifications to the nozzle design. The V4 STL are now the current version, I am going to see what I can do about modelling the actual airflow next. For now I am testing things the hard way, with temperature towers at various fan speeds to see what works best
Update: 2019-03-12:
When printing this, I decided that the ducts were too square, so I redid the cross-section polygons to make it a little rounder. It might even improve the airflow, and I think it looks better... the V2 STL files have the newer shapes

Original post:
I really like the design of the Hero Me system, particularly the way the base works. I started using the dual 5015 fan version of it, but noticed that I was not getting the performance I would expect. My usual solution to fixing it was to see what I thought needed fixing. Unfortunately, the original design is done in tinkercad, which is not very friendly to exporting it to a useful file format, and the STL it produces for this part are so non-manifold they aren't repairable, I have no idea how it even loads into Cura.
This resulted in a rather long process of measurement and coding to make an openSCAD program to generate a Hero Me compatible duct.
Eventually I will get my scad files into a condition where I can let other people see it, but for now, here is the product.
My primary goal in reworking this design was to make the mounting angle of the fans adjustable. I noticed that the air routing was somewhat non-optimal, and I theorised that if you tilt the fans forward, they will have a much cleaner internal path to the nozzle, and the intakes would be less obstructed by the base and other accessories. Whether it works or not is anyones guess.
In the process of writing thins, I made some generalised code libraries to projects a polygon along a spline path, which I used to generate all of the curves in both the interior and exterior of the duct. I ran into a lot of rendering issues and openSCAD bugs, as this is exercising it to it's limit, but the results seem good.
I haven't printed the duct yet, that is tomorrow's job.
There are two possible issues. First, I did not include the internal fins in the nozzle opening, but they may be necessary for even distribution of the air over the part. Second, with the fans tilted forward, the weight distribution is different. It is possible that I will have to switch to a base that has bolted mounts at the front and rear, we shall see.
One of the benefits of the openscad version of this is that it is now possible to easily visualise the internal air path, which was very helpful while I was attempting to optimise it.
Many thanks to mediaman, he has been most helpful. I believe the original design came from Kelokera, as well.