Two‐phase (air‐water) flow experiments were conducted in horizontal artificial fractures. The fractures were between glass plates that were either smooth or artificially roughened by gluing a layer of glass beads to them. One smooth fracture with an aperture of 1 mm and three rough fractures, one with the two surfaces in contact and two without contact, were studied. For both types of fractures, the flow structures are similar to those observed in two‐phase flow in a pipe, with structures (bubbles, fingering bubbles, films, and drops) depending on the gas and liquid flow rates. The pressure gradients measured for different liquid and gas velocities were interpreted by three models. First, using Darcy's law leads to relative permeability curves similar to conventional ones for porous media. However, these curves depend not only on saturation but also on flow rates. This effect is caused by inertial forces which are not included in this approach. Second, the standard approach for two‐phase flow in pipes (Lockhart and Martinelli's equation) agrees with experimental results, at least for small pressure gradients. Finally, the best fit was obtained by treating the two phases as one homogeneous phase. All the properties are averaged, and the pressure drop is deduced from an empirical correlation between the two‐phase Reynolds number and the friction factor.

This paper describes an experimental and theoretical study of thermal convection in a sloping porous layer. The saturated layer is bounded by two parallel impermeable planes maintained at different temperatures. Several types of flows were observed: a unicellular movement and a juxtaposition of longitudinal coils or of polyhedral cells.A theoretical analysis has been made using the standard bases of the linear theory of stability and by taking into account some assumptions suggested by experimental observations. The critical conditions for the transition between unicellular and polycellular flows has been determined. For flow in longitudinal coils or with polyhedral cells the average heat transfer depends mainly on the filtration Rayleigh number and on the slope of the layer.The experimental study was made in a Rayleigh number range 0–800 and for various slopes (0–90°). For both the transition criterion and the heat transfer, a good fit was observed between the experimental and theoretical results. For maximum slope, i.e. 90°, a correlation which connects the Nusselt number with both the Rayleigh number and the vertical extent of the model is proposed.

No abstract