To my eye, it looks like you might be using too many, when you've got fairly good flow already, you only want to make suggestions to it, not knock it down on the floor and kick it around until it will confess to anything and do whatever you want.
Seeing the shape of the roof it's on also, it's possible that the point where separation would have been occuring was a lot further back than typical on a sedan. A useful thing to know is that around legal highway speeds on a typical size vehicle, airflow will deflect between 8 and 12 degrees off an edge, and will attach to about 1/2 an inch or so of surface. So you can cut yourself a 172 degree angle, with 1/2 inch arms, and go testing curves with it... if the curve touches anything other than the points of the arms, the inside of them, that's about where it's probably separating... The scale of the vortices generated by turbulators should be approximately the same as the length of the turbulation device, so to allow a full vortex to form and touch back on the surface behind it, the turbulators should be about one to two lengths ahead of the separation point. Too many lengths forward and you're making unnecessary surface suction drag.
This is just simplistically explained rule of thumb stuff though for narrow ranges of situations though, so don't go designing passenger jets with it
Linear Mode
