Salinity effects on fines migration in aquifers: Stochastic model and its upscaling

Abstract

Colloidal-suspension-nano flows with varying ionic strength are widely present in nature and industry. The variation of brine salinity, which highly affects electrostatic particle-rock interaction, triggers fines detachment and consequent rock alteration. The microscale models for fines detachment at the pore-particle and at rock-reservoir scales are widely used to predict core and field behaviour under fines migration, while the relationship between those models hasn't been established. This includes the lack of upscaling and downscaling procedures, which prevents the determination of micro-scale parameters from lab corefloods or well production histories and calculation of large-scale model functions from SEM and microfluid tests. In this work for the first time, we derive the rock-scale detachment model as expressed as a function of maximum attached concentration versus brine salinity (maximum retention function MRF) from particle-scale torque balance of attaching and detaching forces. Reflecting micro heterogeneity of the pore space and attached particles, we consider the mutual probabilistic distributions of geometric and electrostatic coefficients from the torque balance of detaching drag and attaching electrostatic DLVO forces exerting the particle. This determines the cumulative distribution of critical salinity, which is calculated from the torque balance and the mutual distribution of microscale parameters and defines the salinity-dependent MRF. This upscaling procedure is performed by the Monte Carlo algorithm for MRF calculation. The corresponding downscaling comprises tuning the mean and variance values for some micro-scale parameters from the MRF. These algorithms are used to treat three coreflood data sets with varying salinity, determine the MRF, and calculate mean values of lever arm and aspect ratios; the match is high, and the obtained microscale parameters are within the common intervals. The upscaling technique developed allows for sensitivity analysis of detachment with respect to microscale parameters and velocity. We also developed the upscaling procedure for MRF recalculation from dependency of one flow parameter to another. The velocity-dependent MRF was recalculated from three coreflood-based salinity-dependent MRFs, yielding lab-based prediction of well behaviour for water injectors and producers.