Early- and late-time prediction of counter-current spontaneous imbibition, scaling analysis and estimation of the capillary diffusion coefficient

TL;DR

This study develops a scaled model for 1D counter-current spontaneous imbibition, enabling accurate prediction of recovery profiles and diffusion coefficients across diverse conditions without solving complex equations.

Contribution

The paper introduces a normalized scaling approach for imbibition, linking recovery behavior to a single coefficient, and provides a method to predict recovery profiles and diffusion coefficients efficiently.

Findings

Scaled recovery follows a square root time dependence at early stages.

The correlation model predicts recovery with R^2=0.9989 across 5500 cases.

Parameters describing the diffusion process are linked to recovery behavior.

Abstract

Solutions are investigated for 1D linear counter-current spontaneous imbibition (COUSI). The diffusion problem is scaled to depend only on a normalized coefficient {\Lambda}_n (S_n ) with mean 1 and no other parameters. A dataset of 5500 functions {\Lambda}_n was generated using combinations of (mixed-wet and strongly water-wet) relative permeabilities, capillary pressure and mobility ratios. Since the possible variation in {\Lambda}_n appears limited (mean 1, positive, zero at S_n=0, one maximum) the generated functions span most relevant cases. The scaled diffusion equation was solved for all 5500 cases and recovery profiles were analyzed in terms of time scales and early- and late time behavior. Scaled recovery falls exactly on the square root curve RF=T_n^0.5 at early time. The scaled time T_n=t/{\tau}T_ch accounts for system length L and magnitude D of the unscaled diffusion coefficient via {\tau}=L^2/D, and T_ch accounts for {\Lambda}_n. Scaled recovery was characterized by RF_tr (highest recovery reached as T_n^0.5) and lr, a parameter controlling the decline in imbibition rate afterwards. This correlation described the 5500 recovery curves with mean R^2=0.9989. RF_tr was 0.05 to 0.2 units higher than recovery when water reached the no-flow boundary. The shape of {\Lambda}_n was quantified by three fractions z_(a,b). The parameters describing {\Lambda}_n and recovery were correlated which permitted to (1) accurately predict full recovery profiles (without solving the diffusion equation); (2) predict diffusion coefficients explaining experimental recovery; (3) explain the combined impact of interactions between wettability / saturation functions, viscosities and other input on early- and late time recovery behavior.