I am trying to simulate the distortion of a (2-dimensional) casted F-shaped product after removing it from the mold (see fig.). The temperature and stress fields at the end of cooling within mold is propagated to the next step where the mold is removed using *Model Change command. Max temperature in the cast at the end of the previous step is around 200°C. Heat transfer is defined through *Sfilm as a surface film condition (Coeff. 0.05, Ambient temperature: 20°C). 

Problem:
I am getting temperatures far less than ambient temperature after cooling, and in some cases even negative. In the msg file I observed that, after mold removal, heat flux equations converges in all the iterations due to very low residuals. I am not able to understand why do we see this kind of spurious result. 

I have tried different heat transfer coeffiecients, different boundary condition (radiation), different time steps, different max. allowable temperature. But in all cases the temperature goes below ambient temperature.

I would really appreciate if someone can help me figure out this problem. Feel free to ask for more information to understand the problem if required.

Thanks in advance.

Abaqus ​​​​​​​