Conclusions

IF THE vapour transport rate is such that the normal interfacial velocity induced by diffusion cannot be ignored then the interdiffusion term should be included in the energy equation. The latter inclusion necessitates a composition-dependent mixture specific heat capacity if energy is to be conserved. A single parameter, the mass transfer rate factor, $\varPhi$, determines whether these three phenomena should be included or neglected.

When $\varPhi$ is small, the governing equations may be considerably simplified, becoming more like those of the analogous single fluid heat transfer problem. The compositional contribution to buoyancy and the energy transport due to the evaporation and condensation of the vapour should be retained in the low mass transfer rate limit. Indeed, the latter may even then be the dominant mode of energy transport.

Definitions for the normalized vapour and energy transfer rates (Sherwood and Nusselt numbers) have been proposed which differ from those which would be derived by considering low mass transfer rates from the outset. It appears that these definitions account for much of the dependence of the overall transport rates on the mass transfer rate factor. This means that mean Sherwood or Nusselt numbers obtained (from experiments, or numerical or analytical solutions) at low mass transfer rates may be applicable at much higher mass transfer rates. This fortuitous result has been previously reported for the Stefan diffusion tube. It does not apply to local fluxes.

Except possibly in the viscous limit, vapour transport in enclosures is likely to be strongly three-dimensional unless the enclosure is very tall and narrow. The latter restriction implies a small combined Grashof number. Future calculations of the onset of multicellular convection in tall cuboids should account for the finite span of the cavity, particularly if this does not exceed 1.7 times the breadth.