I'm working on a project where part of a unit needs to rotate around a fixed point, and be supported on two fixed (i.e. not castor) wheels.
When experimenting with the wheels, it feels like fixing them at 90 degrees to the unit results in them 'scraping' across the floor somewhat, as the arc of travel is always slightly misaligned with the plane of their rotation. Further, it feels like holding them at a slight camber mitigates or even eliminates this effect. However, I have no way of confirming this.
Is this basic observation correct, and if so, is there a method for calculating the optimal camber (a and b) for wheels different distances (d and e) from the pivot point x?
(Note, if it helps/matters, the wheels have a more rounded profile than those shown in the diagram.)



