Hyperbolic Adv-Diff, Navier-Stokes 1D/2D #811
Conversation
Files for the 1d and 2d Advection-Diffusion equation and files for the 1d and 2d Navier-Stokes equations are included.
The functions to call the Advection-Diffusion equation and the Navier Stokes equations need to be exported in the Trixi.jl file.
In case of the 1d hyperbolic Navier-Stokes equations the value of the "du" variable needs to be multiplied by a precondition matrix. This multiplication is added to the "rhs!" function.
Is case of the 2d Navier-Stokes equations the value of the "du" variable needs to be multiplied by a precondition matrix. This multiplication is added to the "rhs!" function.
The function is necessary for the Navier-Stokes precondition matrix multiplication.
An initial guess is added to the example exp_nonperiodic
The added elixirs execute the changed or added examples for the hyperbolic Navier-Stokes equations
Exporting the new examples for the Advection-Diffusion and the Navier-Stokes equations. The function calc_viscous_shock_solution which runs a fortran function to calculate an initial solution is also exported.
There was a problem hiding this comment.
Overall very nice work @MagLuc ! This is a very nice set of interesting equation systems to add into the Trixi capabilities.
I have put together a first batch of comments and suggestions. My initial reaction is that you have constructed a lot of manufactured solution testcases. We should strive to remove unnecessary ones and "trim" it down to effectively test your new equation systems but avoid duplication within the elixirs. You should also try to move the initial_condition + source_term functions into the elixir (when possible) as motivated by #685 . Once we have settled on which tests to keep you should then add them to the automatic test scripts to make sure that we keep the code coverage.
We should also get some feedback from @gregorgassner , @sloede and @ranocha about the way to move forward with this.
|
|
||
| """ | ||
| Example "idlCFD" | ||
| This example is taken from the free book "I Do Like CFD". |
There was a problem hiding this comment.
Can you please be more specific here. What page number / equation number is it. I think it is from Section 7.3.1 but it would be good to update the docstring here
| periodicity=false) # set maximum capacity of tree data structure | ||
|
|
||
|
|
||
| initial_condition = initial_condition_idlCFD_nonperiodic |
There was a problem hiding this comment.
| initial_condition = initial_condition_idlCFD_nonperiodic | |
| initial_condition = initial_condition_boundary_layer |
I do not think that this name is appropriate. We should change it to something more descriptive/obvious for other users. Like including the phrase "boundary layer", then it is implicit that it is not a periodic testcase
| periodicity=false) # set maximum capacity of tree data structure | ||
|
|
||
|
|
||
| initial_condition = initial_condition_myexp_nonperiodic |
There was a problem hiding this comment.
| initial_condition = initial_condition_myexp_nonperiodic | |
| initial_condition = initial_condition_convergence_nonperiodic |
Here the name of this initial condition is also vague. I think it means that it is a manufactured solution involving exponentials but it should be adjusted
| ############################################################################### | ||
| # ODE solvers, callbacks etc. | ||
|
|
||
| # Create ODE problem with time span from 0.0 to 1.0 |
There was a problem hiding this comment.
| # Create ODE problem with time span from 0.0 to 1.0 | |
| # Create ODE problem with time span from 0.0 to 3.0 |
| periodicity=false) # set maximum capacity of tree data structure | ||
|
|
||
|
|
||
| initial_condition = initial_condition_mytest_nonperiodic |
There was a problem hiding this comment.
what does mytest mean in this context? What does this configuration test that elixir_hypadvectiondiff_myexp_nonperiodic.jl and elixir_hypadvectiondiff_idlCFD_nonperiodic.jldo not?
If possible we could maybe remove this testcase
| # λ_max_ll = (abs(v1_ll)+abs(v2_ll))/2 + sqrt(T_ll) + abs(mu_h_ll / (u1_ll * L)) | ||
| # λ_max_rr = (abs(v1_rr)+abs(v2_rr))/2 + sqrt(T_rr) + abs(mu_h_rr / (u1_rr * L)) |
| mu_v = 4/3 * mu # viscosity of stress | ||
| mu_h = gamma*mu/Pr # viscosity of heat flux | ||
|
|
||
| # λ_max = (abs(v1)+abs(v2))/2 + sqrt(T) + abs(mu_h / (rho * L)) |
Co-authored-by: Andrew Winters <[email protected]>
Co-authored-by: Andrew Winters <[email protected]>
Co-authored-by: Andrew Winters <[email protected]>
DanielDoehring
changed the title
In this fork the Advection-Diffusion and Navier-Stokes equations in one and two dimensions are added to Trixi. Both equations are implemented with hyperbolic formulations, which were developed by Hiroaki Nishikawa. Serveral examples to all the new equations are also added.