Doing exactly in the same manner that with rectangular footings should be safe enough, counting with the loads really acting (segment of a circle). More conservative of course would be dimensioning for the maximum cantilever in the direction.

If you want more exact values you can model the footing with thick plates or even solid brick elements and see what happens, both overkill.

I need to know the segment you mentioned,is circular or ... tangent to a circle with D distance of the circular column on it?

A segment, that is a sector minus the included triangle.

I don't find any example in a book, but all say at d of face, hence at above.

If your application is critical better do a detailed analysis, then you may average a bit the stresses along any plane you consider.

Really I never practice footings where shear controls so if too adjusted to the strength I would likely proceed the FEM way.

I would proceed as ISHVAAAG has outlined in defining the critical section (distance 'd') away from the column.  If the column is round use the equivalent square to determine where 'd' must be measured from.  Use the segment of the cirle outside of this critical section to determine the shear.  This will be conservative for all locations or variations.

If there are problems, then increase the footing thickness.  Concrete is cheap compared to design time and worry.

I would not, employ a FEM as it would take entirely too long to develop (even in a commercial package) for what amounts to an approximation which is what we're doing via the aforementioned analysis.