Could you go over the theory behind tuft testing and what exactly to look for?
Tufts are wool/yarn/ribbon lengths taped on & around the area of the car you want to study.
In attached airflow, the tufts will stream back relatively smoothly, maybe the ends flicking slightly. In turbulence, tufts will swirl around more dramatically, sometimes in directions apparently at random depending on the severity of the turbulence.
Also in attached flow, the tufts will stream in the direction of airflow, often "pointing" towards low pressure areas, particularly near transitions (e.g. rear quarter / "C" pillar area on a 3-box vehicle).
Do a google image search and you'll find some examples like this:
... showing attached flow ahead of and above the wheel well, and turbulence after the wheel well. (Wheel skirts, anyone?)
If you've done any sailing, you're already familiar with applied tuft testing, since "telltales" on the sails are used to trim the sheets and/or change heading to maintain laminar flow on the sail for maximum lift.
. I've spent many hours looking at sail telltales. In this pic, the upper pair (one on each side) shows the top portion of the sail is properly trimmed, and the lower set shows separated flow on the outside of the sail (the outside telltale is pointing straight down), meaning it's stalled or overtrimmed.
Tuft testing a Lexus LS 400 - note separation at the forward end of the trunk/boot lid (tuft is lifted in turbulence) and reattached at the trailing edge (tufts streaming back).
Tuft testing helps you see airflow, while coastdown & road testing quantifies its effects.
I suspect that tuft testing is so widely used because there is no real alternative. From the resident authority over at homebuiltairplanes.com:
Personally I've played with CFD a little but generally don't depend on the tool(s) for the design process. Even some of the better codes have sufficient limitations to make the results less than useful. Only when particular procedures are in place that assure that the modeling environment and initial assumptions are accurate, can the results be representative of the eventual product. This either takes a lot of experience with a particualr product or, a combination of that and a very pricey system.
Several years back I was in a meeting with a design engineer who specalizes in CFD for United Technologies (Pratt and Whitney). When the subject of CFD for up front design came up, he smirked and indicated that as far as usefuleness is concerned, in his industry the rather pretty pictures are commonly refered to as "clown puke". In other words, they are pretty, make for good presentations, impress the non-technical crowd but in reality, unless there is a verifiable baseline in place, the results shouldn't be depended upon for a high level of accuracy.
i.e. there is no substitute for getting out there and seeing what the air actually does around the object you are designing.