Drag force seems wrong in Cycling Shoe simulation
 

I have done many versions of this simulation with many shoe design changes and at different AOA (if you can call it that with a shoe).

Qualitatively, the results look good (and make sense when different versions are compared to eachother) when viewed in ParaView - see attached image. But the drag force seems off, especially when looking at the differences in drag force between designs.

For instance, with the geometry in the case file linked below, the shoe has two round closure mechanisms on it called BOA's. Specialized has publicized wind tunnel results between two pairs of shoes, one with these BOA's and one totally smooth. They claim a 35 second advantage over a 40km time trial. This back calculates to roughly 35g reduction of drag, depending on some assumptions. When I duplicate this test in OpenFOAM, when the shoe is level as in the attached case, removing the BOA's only yields a 5g reduction of drag per pair of shoes (though admittedly at 22.4 mph, not 30mph as in the Specialized test 5g still seems like too little) . Several other design changes yield very small differences, when I would expect them to be even larger than removing the BOA's.

Can anyone look over the case files and suggest what might be going wrong, if anything? I'm relatively new to CFD so I can easily be missing something important.

Thank you so much for your time.


Case files & geometry:

https://drive.google.com/file/d/19wj...ew?usp=sharing