Surface chemistry

[ThomasHowarth]

I want to add surface chemistry to LMeX, but I am wondering if people have some thoughts on the correct treatment for the boundary condition. For the moment, we would just look at non-EB boundaries until we have the option of inhomogeneous Neumann boundaries for EB (AMReX-Codes/amrex#4402).

The boundary condition would look something like
$$\rho D_{i}\nabla Y_{i} = \dot{\omega}_{i,\text{hetero}}$$(+ Soret (#391) ).

What are the thoughts on using a lagged source term for the diffusion solve vs. an iterative approach?

[baperry2]

Just want to make sure I'm understanding this. $\dot{\omega}_{i,hetero}$ is a function of $Y_i$ at the surface, so you need the BC in order to define $\nabla Y_i$ in order to compute the surface $Y_i$ from the cell-centered value? And then the surface reactions are accounted for through the diffusive flux at the boundary rather than including the heterogeneous source term in the RHS?

I don't have strong opinions on whether an iterative approach for this term or lagging it in the SDC would be better.

[ThomasHowarth]

That would be exactly the idea, yes - happy to take other suggestions, but the heterogeneous reactions would basically only be considered as part of the boundary condition and species would diffuse into/out of the domain via the reactive walls, and not integrated 'in-the-domain' per se.

Lagging with the SDC will be easier to immediately set up and be notably cheaper, but I think we'd have to do some verifications to make sure that it's giving the right answer.

[baperry2]

My guess is that a lagged SDC approach would be first order accurate in space with 1 iteration (just computing heterogeneous source terms computed based on the cell centered Y_i), second-order with 2, but require a large number of SDC iterations to fully converge.

Depending on the relative stiffness and coupling between gas phase and heterogeneous reactions, it may be worth it to just use the first-order approximation that the surface composition matches the cell-centered composition in order to allow for directly coupled integration of both reaction sets with CVODE.