During the boundary layer separation process described by Van Dommelen & Shen [241], the upper part of the boundary layer is `ejected' away from the wall. It might be thought that this singular process would be reflected in the net forces experienced by the cylinder.
In fact, in fluid dynamics there is a close relationship between
boundary layer separation and drag forces. For a nonseparated
boundary layer such as exists at, say, , boundary layer
scalings predict a
drag coefficient that vanishes for high Reynolds numbers.
Since that is in contradiction with the finite values of the drag
observed experimentally on cylinders under steady conditions,
it is known as D'Alembert's paradox. For steady flows, the drag
is due to boundary layer separation which causes the boundary layer vorticity
to move far from the wall, changing the overall flow field and
inducing adverse pressure forces.
In the unsteady case the situation is different. While previously it
had been believed that drag cannot directly be predicted from the
standard first order boundary layer theory, (hence from the Van Dommelen
& Shen singularity), Van Dommelen & Shankar [229]
showed that this is
in fact possible. Ignoring the details of their derivations, their
final conclusion was that the leading order drag is not affected by
the initial unsteady separation!
Although the separation does introduce pressure forces,
the adverse and favorable forces cancel, leaving the net force
initially unaffected.
This prediction seems in good agreement with our results. Our computed
drag shows no sign of singular behavior until well after , when
the rolled-up vortex forms. Even then, in spite of the ejection of vorticity,
the drag goes down, rather than up.
There is a practical implication of the above conclusion that the Van Dommelen and Shen singularity does not affect the net force initially. It is that the control of unsteady separation in two-dimensional flows based on monitoring the net forces may not be very effective, since the separation has occurred before the forces change.