Discussion Forum

Hi,

I am simulating the solid mechanic's module with the heat transfer module.

The simulation will be compared with an experiment of rubber self-heating under dynamic loading.

The following points were used:
1. 2d-axisymmetric study (we have a cylindrical sample).
2. hyperviscoelastic model (Mooney Rivlin two parameters, relaxation times, and energy factors were evaluated from experiments with the optimization module).
3. force-displacement areas and dissipated energy agree with the experiment.

According to previous threads, I used a BDF solver with intermediate steps and 5-time steps per cycle [ex. t_step = 1/(frequency*5)] and it seems fine.

I defined the heat source as stress*strain*frequency (harmonic heat source).

My problem is when solving both modules together to investigate the temperature, the results agree with the experiment approximately till the time equals 2500s (when the temperature reaches equilibrium). Instead of staying constant or oscillating around a constant value, it decreases with time. This decrement increases with the frequency. The used frequency are 10, 20, 30, 40, and 50Hz. The attached figure shows a comparison between the experiment and simulation.

I don't know if this is the regular case for a harmonic heat source, or do I have some problems with time-stepping or solver?? I tried MUMPS and Paradiso. Also, I tried using more the five steps per cycle to 20. but I got almost identical results.

Thanks in advance for your help.

Regards,