Extended Three-Phase Relative Permeability Formulation and its Application to the History-Matching of Multiple WAG Corefloods Under Mixed-Wet Conditions

摘要

The interpretation of multiple water-alternating-gas (WAG) experiments carried out with different injection strategies (first injected fluid, lengths of injected slugs) is necessary to improve our understanding of WAG mechanisms and to build a reasonably predictive three-phase relative permeability model for reservoir simulation. This paper presents an extension to the well-known three-phase hysteresis model of Larsen & Skauge and its application to interpret an extensive series of WAG experiments carried out under mixed- wet conditions, on the same low permeability sandstone core, at three different levels of gas-oil interfacial tension (IFT), namely 0.04 (low), 0.15 (intermediate) and 2.7 mN.m~(-1) (high). After presenting the main experimental results, we describe an extended formulation - which was implemented in our in-house research reservoir simulator (IHRRS) - where the conventional Land formalism for the calculation of imbibition scanning curves and gas trapping is replaced by a more general model allowing arbitrary trapping functions and accounting for the nature of the displacing fluid (oil, water, or a mixture of both). This new approach - which is thought to be necessary for accurate modeling of WAG processes in mixed-wet systems and/or when gas-water and gas-oil IFTs differ significantly - is used within a history-matching workflow to generate multiple sets of fitting parameters consistent with the experimental data. For each IFT, an experimental dataset consisting of a gasflood and two long slugs (> 1 PV) WAG corefloods starting with either gas or water was history-matched. Each individual WAG coreflood was first interpreted together with the corresponding gasflood data, and then a simultaneous history-matching of all the coreflood data was attempted. An acceptable match of the experiments could only be obtained by restricting the common set of fitting parameters and by introducing a trapping model tuned to the experimental data and accounting for a different amount of trapped gas in the presence of an oil bank (observed in some experiments). This work offers new insights on the methodology to follow and the physical aspects to be improved, such as the gas imbibition scanning curves under three-phase flow conditions, in order to build a more reliable three-phase hysteresis model, capable of predicting multiple injection strategies and applicable to different wettabilities and gas-oil IFTs.