COMPARISONS OF POINT AND AVERAGE CAPILLARY PRESSURE - SATURATION FUNCTIONS FOR POROUS MEDIA

The relationship between the volume of water occupying pores in soil or rock and its energy state is called the capillary pressure–saturation function. This is an important hydrogeologic property needed for modeling multiphase flow and transport. Standard methods used to determine capillary pressure–saturation behavior produce volume averaged functions rather than point functions. Average functions can produce erroneous simulations in flow models. Analytical expressions permit extraction of point functions from average functions, and predictions of average functions from point function parameters. These concepts are discussed in Chapter I.

Chapter II compares average and point functions from centrifugation of Berea sandstone, glass beads, and Hanford sediments. The averaging method smoothed out the drainage process, yielding less steep capillary pressure-saturation functions compared to the point functions. Maximum deviations in saturation between the two methods ranged from 0.08 to 0.28 cm³ [cubic centimeters] cm^-3, and generally occurred at low suctions. These discrepancies can lead to inaccurate predictions of hydraulic properties.

Chapter III presents computer simulations using the STOMP numerical model to simulate CO₂ [carbon dioxide] injection into saline aquifers for geologic sequestration. Average function parameters were compared to point parameters. Simulations using point parameters have injection flow rates 1.36 times faster than the injection flow rates produced by simulations based on the average parameters. The cost of CO₂ injected at a gas pressure of 1.5 x 10⁷ Pa [Pascals] was $0.87 [0.87 dollars] per ton using the average parameter set and $0.65 per ton of CO₂ injected using the point parameter set.

Chapter IV evaluates analytical expressions that relate point functions to average functions. These expressions can extract point parameters from average functions, and estimate average functions from point parameters. Average functions for nine columns of Flint sand between 4.3 cm and 55.0 cm tall were predicted from point functions and compared to average functions measured using the hanging column method. Agreement was generally good. Predictions of point parameters from observed average retention data produced variable results compared to point parameters measured by neutron imaging. Chapter V summarizes results from the preceding three chapters and provides some suggestions for future research.