[BramT]

Dear all,

For my master thesis I'm running an optimization on a conventional wing using multiple FFD boxes:

1: FFD box enveloping the entire wing, used to change the sweep angle.
2: FFD box enveloping the shock wave area on the upper surface of the inboard wing.
3: FFD box enveloping the shock wave area on the lower surface of the outboard wing.

Last night I noticed some discontinuities between the inboard and outboard boxes (at their junction) and decided to change from having two separate boxes to have a single box enveloping the area of the upper surface that I am interested in (also for other reasons). See pictures below for clarity:

The old two boxes (deg_i = 9, deg_j = 19, deg_k = 1, for both):


The new box (deg_i = 19, deg_j = 19, deg_k = 1):


Before the change, running SU2_DOT for 400 variables spread over the two boxes would take 400 seconds (possibly less). After the change, running SU2_DOT for 210 variables takes at least 2100 seconds, if not more.. it's a ridiculous increase of computation time for no apparent reason (you would expect a decrease considering the change in number of variables!).

Does anybody know what causes this increase in computation time and possibly a way to avoid it? Your help will be greatly appreciated!

Regards,
BramT