The Orca3D speed/power prediction for planing hulls is based on the well-known Savitsky method. The Savitsky method was developed for single-chined, prismatic forms (i.e., constant deadrise), within particular limits for length/beam ratio, deadrise angle, etc. The method is based on finding an equilibrium of forces and moments, using parametric estimates of dynamic lift and drag, coupled with hydrostatic and gravity forces.
Warped hulls, chine flats, bottom flats, rocker, etc. change the physics from the assumptions in the Savitsky method but, in general, the changes are not so dramatic that the method cannot still provide reasonable estimates. Stepped hulls, on the other hand, dramatically change the physics, and are outside of the bounds of a Savitsky analysis. Without a lot of historical experience benchmarking full-scale data for similar hulls, at best you might use it to determine if the combination of length, beam, deadrise, and displacement are reasonable. Presumably a stepped hull that ventilates properly would improve the performance over a non-stepped hull of similar proportions.
Hull geometries that don’t fall into the ranges used in the development of methods like Savitsky or parametric methods like Series 62, Holtrop, etc., are exactly why tools like Orca3D Marine CFD are so important. Because they take a first principles approach, they can analyze any shape and flow phenomena that you give them, and are not limited to a particular range of geometries.