I guess initial temperatures are not relevant for thermal steady-state simulation. It only influences the starting condition of solver if your solver even use it.
But at the end, final temperature distribution MUST NOT be influenced by that.
I assume you have no thermal constraints. Keep in mind fixing temperature by means of
- grid point temperatures,
- convection definitions with fixed environment temperatures or
- radiation with constant environment temperatures and
(- constant heat flux)
will generate a steady-state temperature distribution.
Best wishes, Michael