There was a similar question here: Approach for fluid cooled tube heat transfer
The easiest way would be to split the surface into segments and apply different temperatures. UTEMP subroutine would be the hardest approach but you could also try CFD (or actually CHT) as mentioned in the referenced thread.