Drying and wetting of capillary porous materials : insights from imaging and physics-based modeling

Abstract

This habilitation thesis aims to study the convective drying of capillary porous materials on the fundamental level. It also seeks to cover some impor-tant aspects relevant to the wetting process and the two-phase flow without phase transitions (i.e. drainage and imbibition) in such materials. The stud-ies are conducted by means of various experimental techniques along with discrete and continuum modeling approaches. Optical shadowscopy visualization is employed to measure the character-istics of the drying process in single and dual Hele-Shaw cells. Both cell types consist of a single layer of mono-disperse spherical glass beads sandwiched between transparent glass plates. Contrary to single cells, dual cells are comprised of two connected regions filled with small and large, hydrophilic and hydrophobic, separate and sintered particles, respectively. All cells are saturated with distilled water and dried at room temperature. Based on two-dimensional optical images wet and dry regions are determined. The time evolution of these regions clearly show how capillary effects depend on the pore structure, solid surface wettability and medium heterogeneity. The fluid invasion and phase distributions in transparent micromodels are investigated in real time by optical photography. The micromodels are quasi-two-dimensional networks made of PDMS or silicon. The PDMS mi-cromodels are used for drying and two-phase flow experiments, whereas the silica networks are employed for wetting measurements. From these exper-iments, important physical effects (such as the capillary valve effect) are unveiled and then incorporated into pore network models. Agreements be-tween pore network simulations and micromodel measurements are better when the capillary valve effect is accounted for in the models. Drying experiments with random packings of mono-disperse spherical par-ticles are carried out using in situ X-ray microtomographic imaging. Packed beds are filled with separate or sintered glass beads and saturated with either distilled water or with a salt solution. The three-dimensional evolution of the liquid distribution over time is visualized via a series of snapshots from the onset of the drying process until the end. On this basis, almost all details of the structure of the liquid phase in the packing are revealed. It is observed that at low saturation the liquid phase remains connected to the packing surface through a network of liquid rings formed at contact points between particles. Capillary rings are approximated by annular objects and then in-corporated into a pore network model. Pore network simulations show that capillary rings can play an important role in drying as they can accelerate the drying by establishing saturated air conditions in emptied pores even if they are not hydraulically interconnected. In studies related to wetting in this thesis, the layer porosity and pore size distributions of dry fibrous materials are determined from high resolution X-ray images. Steady-state liquid distributions in these composite structures are also deduced from the images. Pore network models are constructed which are representative of such real materials. Using these models the dis-tribution of liquid among the layers is computed. Pore network simulations are found to be in good agreement with the measured results. Pore network models that combine the algorithms of wetting and drying are developed to simulate the time evolution of a liquid droplet deposited on a porous particle. These models are extended by transport and accumulation of a dissolved species. As a result of these extensions, the range of applica-tions of pore network models is widened. A discrete pore network model is now available that can predict the influence of process condition, prod-uct structure, as well as formulation property on the impregnation-drying process. In addition to gaining new insights on the physics of two-phase flow pro-cesses (with and without phase transitions) in porous materials by discrete pore network simulations, they can also provide necessary input to construct continuum models and to validate them. For the latter purpose, some routes for formally connecting the discrete and continuum modeling approaches are presented in this thesis. Pore network simulations play the role of suitable numerical experiments and they are exploited to predict a set of macroscopic parameters of the parametrized continuum models. These include classical parameters responsible for fluid transport (intrinsic permeability, relative liq-uid and gas permeabilities, vapor diffusivity) and local equilibrium properties (capillary pressure and desorption isotherm). New interpretations for phe-nomena relevant to porous media drying (such as the non-local equilibrium effect for the vapor phase) are also suggested. Moreover, the velocity profiles in the liquid phase during the drying of a porous medium are computed and analyzed. The instantaneous volume-averaged velocity field determined from pore network simulations leads to step velocity profiles. By contrast, a simple mass balance in the continuum framework results in a linear profile in vertical direction. Finally, the coupling between the external and internal mass transfer in drying porous media is revisited. This study is motivated by the fact that pore network simulations can provide unequaled information on the phase distribution at the porous medium surface. The evolution of liquid clusters and gas clusters at the porous medium surface over time is characterized by means of pore network simulations. The relative contributions of these clusters to the total mass flux at the surface are thus quantified. On this basis, the quality of an existing analytical model is assessed.