So I just looked into this and the following is true
1) When we do an offset surface, we offset all of the nodes in the source surface and we retain all of the original triangulation, however we don't lock up the original triangulation by putting internal 3D breaklines around all of the triangle faces so that they cannot be easily changed - so while you do not add any data to the offset surface you are in good shape, but as soon as you add additional data to the offset surface a retriangulation happens and that can give quite different results to the source surface (especially if it was modeled by a different system or the same system but with different model settings).
2) When you do an offset surface and use a clipping boundary, by default the offset surface will be different to the source, because the clipping boundary breaks the triangles in the source surface along its path and those extra nodes have to be triangulated into the new offset surface - so the offset surface is different in the triangles that connect to the clipping boundary line points that were generated. There is not much that can be done to avoid that failing creating 3D Breaklines around all of the triangles in the source surface and then breaking those using the Clipping line and retriangulating in the changed data - that will always give a slightly different result because the data is fundamentally different.
3) We will look at this next week to see what we can do to make that more robust - because if you add just one point it will retriangulate all of the data and without the triangle side breaklines to stop it or having the knowledge that was used to constrain the source design, TBC just triangulates the way that it triangulates all of the nodes without the breaklines which results in a different surface model entirely.
I suggest that we will add the ability to add in 3D Internal Breaklines around all triangle sides, and that if using a clipping boundary we try to hold as many of the source triangle faces as possible and allow it to just add the extra triangulation needed around the edges to incorporate the new data.
While we are at it, I have been meaning to add the ability to define several clipping boundaries, and for each one be able to define a different offset (eg different topsoil strip depths), so that we can create a new surface with multiple areas offset to remove different depths of materials - does that make sense to people?
Alan