NASA now says that the drag on the left wing of Columbia is consistent with asymmetrical boundary layer transition, which it has seen on about a dozen prior occasions. The boundary layer is the part of the airflow where it changes from the freestream velocity somewhere above the surface to zero at the surface of vehicle. Boundary layers are either laminar or turbulent. In laminar boundary layers, the flow moves smoothly along lines which are essentially parallel with no mixing as you move away from the surface; in turbulent boundary layers, which are thicker and have more drag, there are eddy currents within the flow, so there is mixing and therefore increased heat transfer.
The P-51 Mustang famously had a laminar flow airfoil. For airplanes, the flow over the wing and body is typically laminar initially and then transitions to turbulent – and one of the factors that determines where that transition occurs is surface roughness. One part of wind tunnel testing is to make sure that the boundary layer transitions on the subscale model in the same location as it would on the full scale, full Reynolds number vehicle. Transition strip, which is a strip of high surface roughness, is placed in the location where this occurs so that aerodynamics measured will correspond to the flight vehicle. Apparently the flow over the space shuttle wing starts outs totally laminar, and then later transitions to mostly turbulent. What happened with Columbia (and other flights) was that one wing was rougher than the other, and thus that wing transitioned from laminar to turbulent much sooner, leading to higher temperatures and higher drag on that wing. The drag meant that the control rockets had to be used more to correct for the induced yaw, and in the past NASA’s worry was that the control rockets would run out of fuel before the shuttle landed. So could the asymmetric transition itself have led to to the loss of Columbia? NASA doesn’t think so, and I have to believe the engineers who designed it wouldn’t have let a predictable occurrence like that have caused a vehicle failure. Of course, combined with another failure, it could have been a contributing factor.
