Hello,
In lecdon_telemac3d.f (the subroutine that reads keywords for Telemac-3D), you can see that indeed, if a k-epsilon model is chosen for horizontal or vertical, the other (vertical or horizontal) one must be the same (otherwise, it is changed).
I am surprised that you do not have such a message when coupling Telemac-3D with Sisyphe. When coupling these two codes, you have first a listing for Telemac-3D then a listing for Sisyphe. Can you read the full listing again?
By the way, for k-epsilon model in Telemac, you can first read Jean-Michel's book.
Regards,
Chi-Tuan