Uncertainty Quantification on Foam Modeling: The Interplay of Relative Permeability and Implicit-texture Foam Parameters

Abstract

Efficient decision-making in foam-assisted applications, such as soil remediation and enhanced oil recovery, frequently relies on intricate models that are developed based on a selection of component models that describe the underlying physics of the phenomenon at hand. Modeling foam flow is challenging due to the complex interactions between foam properties, porous media characteristics, and flow dynamics, which results in significant uncertainties in model predictions. Previous studies on uncertainty in foam flow models have only analyzed foam properties and relative permeability separately, leading to limited reliability of the findings. This study aims to bridge the gap of integrating foam implicit-texture parametrization and relative permeability into an uncertainty quantification (UQ) framework to evaluate multi-phase foam flow simulations in porous media more comprehensively than previously available. A foam representation based on the CMG-STARS and a Corey relative permeability model are employed. Bayesian techniques and polynomial chaos expansion (PCE) are employed for inverse and forward UQ. These techniques enable the quantification of uncertainties and the identification of influential parameters within the model. An initial guess algorithm to represent prior beliefs objectively is introduced for the inverse uncertainty quantification step. An in-house foam displacement simulator, aided by a surrogate model, is employed in forward uncertainty quantification and sensitivity analysis. The research findings contribute to understanding and designing reliable foam flow simulations. Sensitivity analyses indicate that incremental strategies to fit parameters can produce inaccurate predictions. Additionally, the article discusses how inaccurately estimated parameters can lead to underestimation or overestimation of foam performance in simulations.