Description and verification of a novel flow and transport model for silicate-gel emplacement

Highlights

• Novel approach for numerical simulation of rapid gelation process

• Calibration and validation of simulation results of gelation from laboratory experiments

• Density-dependent movement of viscosity-changing solute in heterogeneous porous media

• Concentration dependent gelation process and continued movement of non-gelated solute

Abstract

We present a novel approach for the numerical simulation of the gelation of silicate solutions under density-dependent flow conditions. The method utilizes an auxiliary, not density-dependent solute that is subject to a linear decay function to provide temporal information that is used to describe the viscosity change of the fluid. By comparing the modeling results to experimental data, we are able to simulate the behavior and the gelation process of the injected solute for three different compositions, including long-term stability of the gelated area, and non-gelation of low concentrations due to hydro-dynamic dispersion. This approach can also be used for other types of solutes with this gelling property and is useful in a variety of applications in geological, civil and environmental engineering.

Introduction

Mathematical models continue to play an enormously important role in the study of problems of groundwater contamination and remediation. In theoretical applications, they have contributed to a deep understanding of the key flow and mass transport processes, the interactions of those processes, and the key parameters controlling them. In practice, hydrogeologists are able to model the physical, chemical and biological processes leading to the development of plumes of dissolved contaminants and to estimate their future patterns of spreading. In problems of remediating contaminated groundwater, flow and transport codes are now routinely used in selecting the best and most cost-efficient remedy and in designing the actual remedial systems.

Over the past 40 years, since the first simple flow and transport codes were developed, the pace of progress has been impressive. Early codes capable of simulating coupled flow and physical transport with simple reactions have given way to more powerful simulation models able to accommodate variable density flow, e.g., FEFLOW (Diersch, 1981), SUTRA (Voss, 1984), MITSU3D (Ibaraki, 1998), TOUGH (Oldenburg and Pruess, 1995), ROCKFLOW (Kolditz et al., 1998), HydroGeoSphere (Graf and Therrien, 2005), GeoSys (Beinhorn et al., 2005), and D3F (Grillo et al., 2010). More recently, codes like TOUGHREACT (Xu et al., 1999), MIN3P (Mayer et al., 2001), PHT3D (Prommer et al., 2002), HYTEC (van der Lee et al., 2003), RetrasoCodeBright (Saaltink et al., 2005), OGS (Xie et al., 2006), CrunchFlow (Steefel, 2009), and UG (Musuuza et al., 2011) have provided capabilities to incorporate homogeneous and heterogeneous reactions within an equilibrium or kinetic framework.

Evolution in the complexity of models to a significant extent has been informed by the needs of practice in terms of addressing types of dissolved contaminants, and processes affecting them. In this paper, we describe a new modeling approach that originates from a need in remediation problems to consider a miscible treatment fluid with a capability of evolving to a gel. More specifically, an aqueous solution with an ordinary viscosity can experience at least a four order-of-magnitude increase in viscosity as gelation occurs.

This modeling study is part of a larger study aimed at using oxidants (e.g., KMnO₄) in remediating large, deep and dilute plumes of chlorinated solvents. The efficacy of slow release systems in passively degrading contaminants in situ has already been well established. For example, slow-release solids have been developed by mixing potassium permanganate with other matrix materials such as clay-rich slurries (Siegrist et al., 1999) or oxidation resistant waxy polymers (Lee and Schwartz, 2007, Lee et al., 2009, Ross et al., 2005). These materials can either be injected as a slurry into horizontal hydraulically-created fractures, into vertical trenches and boreholes (Murdoch et al., 1997), or installed in vertical boreholes as solid cylinder-shaped slow-release solids (Lee and Schwartz, 2007). Such slow-release systems also have constraints on their lifetime, installation depths and spacings (Lee et al., 2008).

Our basic approach is to overcome these constraints through the use of dense fluids and slow-release gels to deliver remediation chemicals to deeper plumes and less permeable units. This approach takes advantage of the unique flow and mixing properties of dense fluids, and the potential of engineered gels as a diffusion-controlled mechanism for releasing the treatment chemical. The essence of this new remediation approach is all about increasing the residence time of oxidants close to the site where they are injected, which will be determined by characteristics of the actual contaminated site.

Previously, we performed flow-tank experiments to explore the behavior of dense viscous solutions in fresh water, which provided some understanding of processes involved and the experimental data needed for validating a modeling approach (Solpuker et al., 2012). The goals of this paper here are to describe and demonstrate an approach for modeling this variable density flow/gelation problem. It also presents illustrative data from three flow-tank experiments to verify the modeling approach.