Add time-averaged joule heating auxkernel

[jvwilliams23]

I added code for time-averaged joule heating, and made modifications to the example.

The figure below shows some test results, with end_t_temp = ${fparse voltage_period*20} and varying delta_t_temp. The compute time results are below:

• Black line: 838 s
• Blue dots: 616 s
• Green dots: 75 s

[nmnobre]

Hey @jvwilliams23,

Do you think you could confirm this still produces the same results?
I refactored the code slightly, and can't see any difference, but still.

Regarding the post processors, I think the temperature definitely belongs here, not sure if the power should be moved to the viz branch, what do you reckon?

[jvwilliams23]

Hi @nmnobre

I will check this tomorrow. R.e. power post-processors; I am not sure - it is useful to check the convergence of time-averaged joule heating, and is strongly linked to how well or poorly our sims agree with Apollo. In that sense, it could fall within scope for main.
Maybe once we have a better interpretation of optimal averaging procedure (e.g. number of cycles needed), then it could be moved to viz in future. What do you think?

[nmnobre]

Hi @nmnobre

I will check this tomorrow. R.e. power post-processors; I am not sure - it is useful to check the convergence of time-averaged joule heating, and is strongly linked to how well or poorly our sims agree with Apollo. In that sense, it could fall within scope for main. Maybe once we have a better interpretation of optimal averaging procedure (e.g. number of cycles needed), then it could be moved to viz in future. What do you think?

Okay, sounds good, let's keep it as is for now. on second thought, I think using ElementIntegralVariablePostprocessor might make more sense for P.

I was also just convincing myself that the green dots above should be below the blue dots, but that does seem consistent with the fact we're using implicit time stepping, so I'm happy.