As can be inferred from the literature survey given in the ‘‘Introduction’’ section, experimental studies on original replicas of rock fractures are very rare. Most of them used smooth-parallel plates or rarely artificial rough fracture surfaces created by systematically ordered intrusions. In one of the very rare studies, Radilla et al. (2013) prepared replicas of sandstone and granite fracture surfaces and conducted single and multiphase-flow experiments. Their multiphase-flow experiments were conducted as the steady state flow of water and gas. They visually showed that the phase distribution was controlled by the roughness, which was determined by the lithology of the rock. The sandstone cases (example of grainy rock) showed ‘‘thin channels’’ compared to granite (example of crystalline rock) due to a more rough structure caused by its grainy structure. They did not characterize the fracture surfaces quantitatively, rather they provided relative permeability data for steadystate flow.
In the present study, we used more variety of rock samples (a total of seven models from limestone, marble, and granite) and performed unsteady state (displacement) experiments. We also defined the roughness quantitatively as mentioned earlier, which enabled us to correlate multiphase characteristics to roughness numerically as done for the single phase experiments summarized in the previous sections.
Attention was paid to several specific issues as to the displacement process and the effects of the following parameters on the displacement process were investigated:
Fracture roughness.
Lithology.
Normal load (and fracture closure) on the samples.
Model type and material (wettability to some extent) used.
Also considered was the repeatability of the experiments. Four types of experiments were conducted for four different displaced-displacing phase scenarios: water–gas, gas–water, oil– water, and water–oil. The observations and results are presented in the next sections.