EXPERIMENTAL STUDIES OF VISCOUS INSTABILITY IN MISCIBLE DISPLACEMENTS (FLOW)

Available from UMI in association with The British Library.; This thesis presents the procedures and results from a series of laboratory experiments undertaken to study the relevant factors that control the displacement efficiency in a miscible process, particularly in unstable displacements. The main objective of this work was to provide quantitative experimental data that can be used to test and validate theoretical models of viscous flow in porous media.; Over 100 displacement experiments were performed, using completely miscible Newtonian fluid systems, based on brine and glycerol solutions, in 1D and 2D porous systems. The results show excellent reproducibility. In addition to in situ fingering patterns and recovery profiles, pressure profiles were measured within the 2D experimental systems (both 2D linear and areal five-spot floods). Only by using such pressure data is it possible to distinguish between certain types of averaged models of viscous fingering. This is because different models give very similar recovery profiles, but they make different predictions of the in situ mobility in the finger region. This work is the first that measures in situ mobilities in this way in a comprehensive range of viscous unstable floods (at M {dollar}approx{dollar} 4, 10 and 30) in the same packing material for different geometries. The results may now be used as "bench-mark" data for theoretical model development.; The experiments were modelled using both empirical models of viscous fingering and direct numerical simulation. The averaged type models included the methods of Koval (1963), Todd and Longstaff (1972), Fayers (1984) and Odeh and Cohen (1989). The results indicate that the Koval (1963) model underpredicts the effective mobility of the fluids in the finger zone, and does not provide adequate predictions of the displacement efficiency under certain experimental conditions. The corrections to the Koval (1963) model, suggested by Odeh and Cohen (1989), are in the right direction, but are not large enough to sufficiently improve the results. The Fayers (1984) model tends to overpredict the total effective mobility due to an unphysical shock front formation that appears in this model. The model which gave the best predictions, in all the experimental results for both recoveries and mobilities, was the model of Todd and Longstaff (1972). The direct numerical simulations also modelled the effective mobility of the fluids particularly well. The effluent concentration and recovery profiles were in good agreement with experimental observations for the direct numerical simulations.