An Analysis of Water Shielding in Water-Wet Porous Media

Abstract

This paper presents simulation results for water shielding in strongly water-wet Berea rock. The analysis of experimental data indicates three oil phase fractions; flowing, trapped and dendritic (refers to isolated oil fraction which is in contact with flowing oil phase), which are different for different Berea core samples. Model calculations demonstrate that the noted differences are significant and that data measured from one sample cannot be used to predict displacement performance in other samples. Berea rock exhibits relatively small amounts of dentritic oil; therefore, formulating a dendritic oil fraction in model calculations is not as critical as correctly estimating the trapped fraction. The trapped oil saturation can be estimated from the oil primary drainage and imbibition relative permeability data. Model validation using only an oil trapping function is supported by good agreement between model and experimental oil recoveries over a wide range of water/solvent injection ratios for tertiary CO2 displacements of a West Texas oil from water-wet Berea rock. Also presented are the results of a parametric study to identify key variables and their parametric study to identify key variables and their effect on water shielding phenomena. Introduction Miscible gas flooding applications usually employ an injected gas which has a lower viscosity than reservoir oil. This leads to unfavorable mobility ratios and lower sweep efficiency which can be treated by the water-alternate-gas (WAG) process. WAG may lead to high mobile water saturation which can result in oil entrapment in water-wet porous media. Several authors have discussed this oil trapping phenomenon as it occurs in first-contact miscible (CM) as well as multiple-contact miscible (MCM) displacements in water-wet rocks. Investigators have proposed a trapping mechanism whereby water shields oil from being displaced by solvent, and verified experimentally that the trapping phenomenon occurs for the nonwetting phase. Thus, oil-wet porous media exhibit almost no oil trapping. Early attempts to model water shielding include simple equations or correlations, e.g., Raimondi and Torcaso and Shelton and Schneider estimated the trapped oil saturation from the relative permeability data under steady-state conditions. Chase and Todd incorporated a modified version of the Raimondi and Torcaso correlation into their finite difference miscible flood simulator to predict flood performance under unsteady-state conditions. Salter performance under unsteady-state conditions. Salter and Mohanty extended the efforts by Stalkup and introduced a sophisticated description of water shielding whereby each phase was characterized in terms of three fractions: flowing, dendritic and trapped. The flowing fraction was allowed to flow due to a convective flux, while both dendritic and trapped fractions were treated as immobile. The dendritic fraction was, however, allowed to exchange mass with the flowing fraction according to a first-order rate law. Salter and Mohanty used a modified Coats and Smith dispersion-capacitance model to history-match effluent tracer histories which were run under steady-state conditions. Later, Dai and Orr developed a one-dimensional, four-component (water and three hydrocarbon species), finite-difference model based on Salter and Mohanty's modeling work. Dai and Orr used their model to predict the performance of unsteady-state displacements. Since their model was closely patterned after the work of Sa lter and Mohanty, it reproduced Salter and Mohanty's e xperimental data well. P. 243

A Literature Analysis of the WAG Injectivity Abnormalities in the CO2 Process

“…For high WAG injection ratios in Berea water-wet rock, shielding dominates the displacement process and the type of miscible process doesn't matter whether first contact or multiple contact miscible. 91 The trapping function and oil and solvent mobility in Berea water-wet rock control water shielding. Water-blocking estimates for reservoirs should be based on measurements in reservoir cores using reservoir fluids.…”

“…Water-blocking estimates for reservoirs should be based on measurements in reservoir cores using reservoir fluids. 91 Mobile water does not change the mass transfer process by which miscibility develops, 87 but the wetting conditions do affect miscible displacement. Displacement by the nonwetting phase is affected by highly mobile wetting phase saturation, where displacement by the wetting phase is not significantly effected.…”

“…In secondary CO 2 floods, local mixing caused by high water saturations reduce recovery only slightly because high mobility CO 2 mixes with and displaces the oil before injected water arrives and creates significant dendritic and trapped saturation. 91 Additionally, in tertiary CO 2 floods oil recovery is slowed and reduced by restricted local mixing because high water saturations cause significant dendritic and trapped fractions throughout the flood.…”

Mobility Control Experience in the Joffre Viking Miscible CO2 Flood

“…Therefore, we expected that some form of CO 2 mobility control would be required to enhance CO 2 conformance in the reservoir. Water injection with CO 2 is the most common method used to improve CO 2 conformance; however, water injection can reduce oil recowry by shielding oil from CO 2 , 2, 3 The key to a successful flood is an optimum injection strategy that maximizes oil recovery with efficient use of available CO 2 , During operation of the project, we experimented with different injection strategies to maximize oil recovery efficiency, including continuous CO 2 injection, W ACO z injection, and simultaneous CO 2 /water injection. Laboratory and field experiments were conducted with surfactant to generate foams to control mobility.…”

Theoretical Study of Water Blocking in Miscible Flooding

“…The water-blocked oil saturation was 0.3; both values are in agreement with the range expected for Berea sandstone. 5 Diffusional extraction of hydrocarbons or of a dendritic oil fraction was not included in this simulation. We also assumed that the actual multiple-contact-miscible displacement in the flowing fraction is well-approximated by a fully miscible process.…”