[Christ]

gascort -

I haven't even noticed this thread until today!

I'm impressed with your work, albeit quick, fast, and somewhat clean.

I do have a suggestion about your cooling issue, though - Even though your (several) round ducts had the same surface area as your square duct, they would not have had:
1 - the same volume
2 - the same inlet/outlet velocity.

You changed two key features by adding several round ducts instead of a single square duct, in that the surface area remained similar, but you added much more non-flow area because of the increased number of surfaces the flow has to contend with.

In fluid dynamics, the fluid moving through an object nearly (with caveats) matches the fluid movement around an object. That is to say - there is still a turbulent boundary layer, and skin associated drag, etc.

More tubes of a smaller type would increase skin drag of the fluid movement through them, which explains why (even though the surface area was the same) you overheated with the duct work you used.

Second effect - the length of the ductwork creates an increasingly thick boundary layer up to a given point. A cylinder which is open to airflow at one end, and restricted at the other, is actually more like a directed nozzle, although less dramatic.

IOW - You weren't getting the flow you needed.

Adding back the square unit you had of the same dimensions may be the best idea, as well as sealing your radiator to the core support and ensuring that your flow through the radiator isn't disrupted by something in the flow path, such as grille elements.

If there needs to be a grille screen, it should normally be at the absolute surface of the extension you're working on, with tapered ductwork of increasing area per depth.

The overheating you were experiencing was also most likely accompanied by hot and cold spots in the radiator. If you had a surface temperature thermometer (IR touchless type) you could quickly see that there would be pinpointed spots where your duct work opened to the radiator that were cooler than the rest of the surface. Those points clearly were not enough to cool the flow sufficiently.