Something still looks weird to me… I can’t reproduce your result starting with plane intersection.
Why would a full ellipse need to be revolved ‘0 To 2*Pi’?
Seems like 180 degrees (‘0 To Pi’) should be enough?

P.S. OK, I got it to work by using YZ for intersection which yields one half ellipse curve, which can be rotated ‘2*Pi’ (360 degrees) to give a surface that works fine with Project and other tests.

It’s strange because it works in R5… maybe it is not the Project to Brep component but something to do with the revolve surface component but that just revolves an ellipse through 0 to pi.

Attached is the full gh def up to the project component and below shows that the curves I want to project lie within the Brep so should intersect (and indeed do work in R5)

The original surface in the first post was created by rotating a full ellipse through 0 to pi radians not, as you asked, through 0 to 2*pi. You can see that in the def I posted above… the D Domain input for the RevSrf is 0 to pi

Thanks for the idea to use a half ellipse. I had misunderstood some of your reply about
intersecting the internalised surface but I see what you were doing now.

I have no idea why…

I guess the lesson here is do not use a closed curve, rotated 0 to pi for RevSrf.

I seem to have a vague memory of a discussion about this either here or on the old forum. Does it create a bad seam in the surface?