PNNL

Abstract

Constitutive relations between relative permeability ( ), fluid saturation (S), and capillary pressure ( ) determine to a large extent the distribution of brine and supercritical CO2 (scCO2) during subsurface injection operations. Published numerical multiphase simulations for brine- scCO2 systems so far have primarily used four models. For the relations, either the Brooks-Corey (BC) or Van Genuchten (VG) equations are used. The relations are based on the Mualem, Burdine, or Corey equations without the consideration of experimental data. Recently, two additional models have been proposed where the relations are obtained by fitting to experimental data using either and endpoint power law or a modified Corey approach. The six models were tested using data from four well-characterized sandstones for two radial injection test cases. The results show that for all sandstones, the VG-Mualem model predicts plumes that are considerably larger than for the other models due to the overestimation of the gas relative permeability. On the other hand, the predicted plume sizes are the smallest for the VG-Corey model due to the underestimation of the aqueous phase relative permeability, making it more difficult to inject the scCO2. Of the four models that do not use fits to the experimental relative permeability data, the hybrid model with Mualem and Corey relative permeabilities provide the best fits to the experimental data and always produce intermediate results. The model with the endpoint power law resulted in very low, uniform saturation outside the dry-out zone for the Tuscaloosa sandstone, as the result of a rapidly declining aqueous phase relative permeability. The results show that development of future hysteretic models should not be based on the VG-Mualem and VG-Corey models.