Just thought of an way to approximate wind that could work much better than addforceglobal. Heres what you should do:
1. Iterate troough all vertices of the model every frame
2. Apply a force in the opposite direction of the normal of the vertex(only if the normal faces the wind)at the vertices position.
Or alternate #2 (Don't know which )
2. Apply a force in the opposite direction of the bounce vector of the normal of the vertex and the winds direction(only if the normal faces the wind) at the vertices position

I think #2 should work...

That sounds very complex but i think it might work pretty well(No scientific proof whatsoever, just improvising...)

This would theoretically spin the modell with the wind, not just move it, but since it doesn't take into account the sizes of the faces(every vertex has an equally sized effect) it's very crude and i don't even know if it is mathematically correct or works at all, just throwing an idea...

~Phantom88~