ATS Solute Mass Fluxes

[Wyatt Tatge]

Hi Phong,

I've recently installed a new version of ATS master, after you guys had added the mass_flux evaluators. I've been playing around with them in my model and the regression tests and have been getting some weird results.

In the regression test surface_tracer_advection_diffusion_withsources.xml the diffusion flux is always 0, no matter what the diffusivity is set to. The advection fluxes would change with the change in diffusivity. 

I modified the column_drainage.xml regression test to make sure I was understanding how the mass_flux observations work. I used the bottom drainage face to look at the flux out of the domain. When doing this I seem to get diffusion numbers that don't make any sense. They are usually larger than the total tracer input. But if I ignore the diffusion flux and only use the advection flux to perform a mass balance it seems to work out. I use the cumulative advection flux + storage at the last time step and I am able to reproduce the starting amount on the surface. This advection flux is also negative even though I thought I set the positive direction out of the domain. 

I was hoping you guys could maybe shed some light on what I'm seeing and/or how to utilize these new features. 

Thanks

Wyatt

[Wyatt Tatge]

Hi,

I've continued to mess around with the regression tests, and I modified the column_drainage some more by splitting the column in half and adding a middle face to output water and mass fluxes. When I perform a mass balance for the water everything works out perfect for the whole, top, and bottom domains. When I do the same calculations for the mass fluxes they do not work out perfectly, as shown below. I'm not sure if this is an issue with how I've been performing the calculations, setting up the model outputs, or a bug with the solute mass fluxes. I am attaching the input file for the modified column_drainage.xml. 

Thanks

Wyatt

[Phong Le]

Hi Wyatt,

I looked at the surface_tracer_advection_diffusion_withsources.xml example. Initially, I thought the reason we saw zero diffusion flux was because the east face (1000 m) is too far away from the source (0-10 m). So, I placed another face closer to the source for observation. Now we can see that the diffusion mass flux is non-zero. However, I did see high concentration of tracers (break though curves) at the last cells of the domain, so diffusion should be happening there and diffusion mass flux should not be 0.

To isolate diffusion, I set the water_flux = 0, making transport entirely governed by diffusion. I observed a good breakthrough curve (mole_fraction of tracers 1 and 2) at the east face. The advection mass flux is 0 (expected), but the diffusion mass flux is also 0, which is incorrect. This means that diffusion is happening properly within the domain modeled by ATS, but the way we extract the diffusion mass flux may not be correct.

The diffusion mass flux is calculated the same way as the diffusion of water, but you mentioned that your mass balance for the water looks good. This will require some thoughts for me and Ethan to understand what might be causing this issue. 

We've created an issue to fix the bug here: https://github.com/amanzi/ats/issues/315

Thank you for checking the tests.

Phong