Adds a Boussinesq source term for boyancy

[timofeymukha]

Formulated as $\beta (T - T{ref}) g$, where $\beta$ defaults to $1/T{ref}$. Reads the following entries:

• scalar_field: The name of the scalar that drives the source term, defaults to "s".
• reference_value: The reference value of the scalar.
• g: The gravity vector.
• beta: The thermal expansion coefficient, defaults to the inverse of ref_value.

I modify the RBC example to use the source term. In the example, the forcing is just equal to $T$ in the $z$ direction, which is achieved by manipulating the constants in the source term. Also, it seems possible to increase the timestep by a factor of 100, @adperezm can you confirm?

[MartinKarp]

Have you run the full RBC e.g. to T=250? The time step can be really large in the beginning as you don't have any convection, but will explode after if its not low. I think the transition to turbulence is around T=30 in this case.

[timofeymukha]

Have you run the full RBC e.g. to T=250? The time step can be really large in the beginning as you don't have any convection, but will explode after if its not low. I think the transition to turbulence is around T=30 in this case.

I think I did, but I will double check.

[timofeymukha]

I can run up to 250, but I guess it does not look very turbulent. This is the velocity in y at t=250

[timofeymukha]

I seem to get the same on develop though. @adperezm could you comment on how the solution should look :-)?

[MartinKarp]

It should undergo a transition at t=30, at least that was the case when I set it up orignally ;) For me that still happens though, here is the velocity mag for me when I run on develop:

[timofeymukha]

@MartinKarp @adperezm So, it turns out RBC is tricky, with a big time step it just doesn't transition, and you seem to need 2e-3 to keep it alive once it does. So,, I removed the dt change for the case. It takes at least an hour to run the whole case, so I ran for 10s with develop and the new source term and the solution is identical. So I think we are good

[MartinKarp]

I guess, if the time step is too big the initial disturbances are probably just smoothed out rather quickly. Of course, one can increase the noise in the ic, but yeah there is a trade-off to be made.

[timofeymukha]

@MartinKarp Can we merge?

[MartinKarp]

Yeah, go for it

[timofeymukha]

Yeah, go for it

You need to approve :)

[njansson]

Yeah, go for it

You need to approve :)

@MartinKarp, go, no-go?