Instead of a full streamliner concept that fits over everything, the images above of adding wheel pants (ala Carbinite) or Orpheus-style housing would be interesting. (and fairly easy to integrate without damage and could be done remotely and shipped easier due to smaller sizes) The addition of these panels would increase surface area itself -- but the frontal aspect would remain about the same and the airflow would be cleaner than the existing exposed wheels.
Given the baseline testing of the belly tank and known data about everything ala speed/gearing/displacement, it would make an interesting scientific project without breaking the budget. Possibly three iterations or more of testing: (and feedback in regard to directional stability)
1) Simple Carbinite NACA 6600 profile derived wheel covers over all four wheels (or rear two only). Or other shapes?
2) Orpheus configuration of integrated front/rear cover.
3/4) Variants of 1/2 with additional appendages to cleanup airflow over suspension / axles / etc.
We had considered this in a super cheap way a few years ago with using some Morris minor fenders as "cheater guards" to simply change the class, and we even built some bolt on ones that didn't do anything for aero but did the class change only to cheekily mop up two open class records. We have yet to run them though.
This is something to consider as an exercise indeed even if it is for interest sake for all concerned and I am sure there would be much interest and discussion derived from it. I could pretty easily draw the NACA shape in 3D and cut it a mold using our 5 axis router at uni...
The thing is though your version 3/4 above is sort of what I am proposing anyway but hopefully in a more stylish manner than the accretion method described would produce (IMO).
The main issue in my mind is the use of ground effects or not. If that was the piece cleaning up airflow as you describe in 3/4 then I could make a couple of different types that gave proof of concept. If it doesn't work then there is no point building a whole new car based on this idea and better to go a different route.
There are a number of factors governing whether it will work or not in the manner we want it to work which will include which record we are going for (ie what top speed), the type and power of the engine, the final cD of the car including the ground effect device compared to not having it, all of which is not yet decided but each are important variables. What I want to do is see if there is a best set of variables given established records and run that package.
Superleggera you have me thinking... if we are to build a new chassis then a modular designed body might be the solution where it is designed with the ability to test all aspects of bodywork... it is a matter of defining joint lines and hard points (a contemporary auto manufacture methodology).
I still want to do some proof of concept stuff on the SOS chassis though...
rH+