[JuPa]

In conjugate heat transfer problems we're solving the energy equation in a solid domain, which may be in contact with a fluid domain where we solve all of the conservation equations.

In the solid part, since all we're doing is simulating the diffusion of heat through the solid I assume the the mesh requirements for the solid part does not have to be amazing (but still adhere to orthogonality and skewness requirements). Is this a valid assumption?

Which leads me to my main question:

At the solid - fluid interface (see attached) can I have a sudden jump in cell size from the solid (grey) to the fluid (blue) domains?

Please note I have not finished the final mesh - I will have more cells in the solid (grey) domain to capture the temperature gradients.

Thank you.

[ghorrocks]

Steady state heat flow is just diffusion, so is linear. In that case a very coarse mesh can resolve this OK. You might be OK with just a single element thick - but I would check on an analytical case first before committing to it.

But for transient heat transfer, the transient bit can put additional thermal gradients into the material. Whether this is important to you or not is dependent on how fast the transients are and how accurately you need to resolve them.

[Opaque]

Keep in mind that diffusion solution is linear only if there is no curvature of the isotherms. For example, steady state one dimensional radial diffusion has a logarithmic profile between inner and outer radius. The isotherms may be packed near the boundaries depending of the radii ratio in that case.