An Interesting Phenomenon

When computing the derivative of ``coefficient of lift'' with respect to ``angle of attack'' for the front slat of a multi-part wing using the derivative code generated by ADIFOR we saw the following:

The derivative values run from -1e9 to 1e9.

We therefore attempted to verify this result using finite differences. Using an ostensibly reasonable step size of 1.0e-1, and then progressively less reasonable step sizes of 1.0e-4 and 1.0e-6, we generated the following disturbing series of finite-difference plots:

	The derivative values run from -3 to 4.
	The derivative values run from -3000 to 3000.
	The derivative values run from -2.5e5 to 1.5e4.

Since the analytic derivative is a limiting process, the evidence indicates that the computed analytic derivatives are the correct derivatives of the CFD process. We are not sure, why the coefficient of lift appears to be so sensitive to the angle of attack.