Hi Yasemin,

I just wanted to chime in here as I agree with David’s comments. I think the only detail I would add is that if your system currently has some negative airspace thicknesses, it may be in a pretty rough spot in the solution space and local optimization may stagnate without some modifications to your system. One approach would be to, as David shared, increase the MNCA weights (and potentially just manually increase the distance between the offending surfaces and let optimization keep them positive with the increased MNCA weights). Additionally, you may also want to see if you have any earlier iterations of your design that do not yet have these negative airspaces, and use those as an intermediate “starting point” with increased MNCA operand weights.

Lastly, you could also always try one of the global optimizer routines -- probably the Hammer optimization is what you’d want (as Global Search will provide several starting points and is meant to generate initial ideas that you iterate further upon). Hammer Optimization won’t be limited by the local merit function minima (which local optimization may be finding with the negative airspace solution), but rather it will try to expand its search the longer you let it run.

Thanks, and good luck!