Added critical time step for heat transfer

[pabloseleson]

No description provided.

[streeve]

Yes, I think we should combine these in one function, but in a new function separate from the Inputs constructor since we need the force model details to determine whether we're running heat transfer or not

[pabloseleson]

Sure, could you please make the change?

[pabloseleson]

We need to account for "thermal_subcycle_steps>1" for the critical time step calculation for heat transfer.

[pabloseleson]

We may need a new PR for this, but the following does not seem correct.

// Set approximate bulk modulus if not set by user
// Used in timestep estimation
if ( !inputs.contains( "bulk_modulus" ) )
{
if ( !inputs.contains( "elastic_modulus" ) )
throw std::runtime_error( "Must input either bulk_modulus or "
"elastic_modulus." );
double E = inputs["elastic_modulus"]["value"];
double nu = 0.25;
double K = E / ( 3 * ( 1 - 2 * nu ) );
inputs["bulk_modulus"]["value"] = K;
}

The reason is that it will compute bulk modulus K assuming nu = 0.25, which is only valid for bond-based PD (PMB). Normally, one would input two constants (except for bond-based which would be just one) from the following: elastic modulus, bulk modulus, Poisson's ratio, and shear modulus.

Note: if we only compute the bulk modulus for the timestep estimation, we should compute it in that function as temporary and not add it to the inputs.

[streeve]

Fixed as you suggested

[pabloseleson]

This is only correct for mechanics, which is:

Note here "c" is the PMB micromodulus.

For heat transfer, we have:

Note here "c" is the is the specific heat capacity, which we denote in the examples by "cp". So, for heat transfer, it should be something like:
double dt_crit = safety_factor * rho * cp / sum;

[streeve]

I just copied your implementation originally, thanks for the details