Hi, i'm checking this issue which was in our to-do list.
RailClone works slightly different from Max modifiers, regarding accumulated transformations (or modifiers).
In Max, the geometry is transformed by each modifier and flows through them. So, if you apply two "Material" modifiers, the geometry IDs are modified in sequence twice.
Instead, RailClone just checks the changes for each operator and computes a "final" transformation, which is applied to the segment only once time.
This method is more efficient, and well suitable for our instancing engine.
But as side effect, Material operators are overwritten when used in sequence, because the IDs of each operator are not computed at all.
Although i know that's not very intuitive, we prefer to keep it as now. An alternative method would be complex to implement.