An analytical and experimental investigation of the thin-film evaporation process in inert gases

Thin-film evaporative cooling appears widely in nature and in industry applications due to its high cooling-efficiency, however, the theoretical limit to this heat removal is orders of magnitude higher than what has practically been achieved. This research presents a theoretical study of the rate-limiting factors in the evaporative heat transfer process (for films in either a pure vapour or a vapour-inert gas mixture), by correctly considering the out-of-equilibrium kinetic boundary conditions. Existing work assumes there is negligible heat transfer in the gas domain, or that the boundary has negligible temperature change across it, or is in equilibrium. By relaxing these assumptions, it is shown that as the Knudsen number (or Peclet number for inert gases) exceeds O(1) (which occurs as the film thickness decreases), the interfacial resistance is comparable to the resistance through the liquid phase. At such thicknesses, the stability and homogeneity of the film may be disrupted by the onset of macroscopic flow instabilities. To investigate this problem, and quantify the rate of evaporation, an experimental campaign was carried out using a novel setup capable of heating a thin film from below, while performing simultaneous optical diagnostics, and measurement of the heat flux, film thickness, surface temperature, and far-field temperature.The Benard-Marangoni instability was visualised in transparent mixtures through lensing effects, and was shown to exist in mixtures and single fluids even at very thin thicknesses.

Reconstruction of the transient heat flux distributions on the heating surface was achieved by solving a transient three-dimensional inverse heat transfer problem explicitly based on measured temperature distributions on the boundary using a novel dual-IR imaging system and an IR-opaque transparent-conduct-oxide coating on a highly IR-translucent sapphire substrate. Here, the accuracy and limitations of existing approaches (including the widely adopted first-order heat flux reconstruction algorithm, ignorance of limited measurement precision, and uncertainties on initial conditions and boundary conditions) are quantified. In addition, the impact of key numerical analysis parameters (including the type of numerical scheme, the order of expansion for heat flux reconstruction on the boundary and the meshing sizes) on the accuracy of the reconstructed heat flux distributions, under different heating conditions, are systematically quantified; enabling improvements to be identified.