What drives vaiation in mass transport coefficients?

I've been modeling and developing models for antibody-antigen SPR experiments for a few years now, and this question has bugged me for a while.

So commonly when an experiment is modeled, there is a variation of transport coefficients calculated between interaction sites. This makes sense because Edwards (2001) established the mass transport model as an approximation to the fluid dynamics system, which as a system will be quite different when concentrations are varied.

But what I've seen in my own work is variation between replicates of the same experiments. That is that sometimes 100nM of antibody reacted against a certain amount of antigen binds with a roughly Langmuir association curve and other times the same antibody, in the same concentration binds to the same ligand in approximately the same density giving a curve showing mass transport limitations (one example i am looking at now has for one replicate K_m as 1.7 and for another 5.8E+11 RU M^-1s^-1).

I wonder particularly that there might be inhomogeneities in the layer that the antigen is bound that produce differences in antigen densities when the same concentration is put across all the interaction sites of the chip, also produce differences in the distribution of antigen in the matix of the chip- such that some interaction spots might have it deeper in the matrix or others might have it shallower.

Or that there might be an issue that comes about from the chaotic nature of micro fluidics.

I haven't been able to find any studies on this (or even on the distrobution of transport coefficients). I get the impression that as most of the software that experimentalists use doesn't show transport coefficients this is not a question people are interested in.

Does anyone have any thoughts?