Reactive transport modeling in heterogeneous porous media with dynamic mesh optimization

Abstract

This paper presents a numerical simulator for solving compositional multiphase flow and reactive transport. The simulator was developed by effectively linking IC-FERST (Imperial College Finite Element Reservoir Simu-laTor) with PHREEQCRM. IC-FERST is a next-generation three-dimensional reservoir simulator based on the Double-Control-Volume Finite Element method and dynamic unstructured mesh optimization and is developed by the Imperial College London. PHREEQCRM is a state-of-the-art geochemical reaction package and is developed by the United States Geological Survey. We present a step-by-step framework on how the coupling is performed. The coupled code is called IC-FERST-REACT and is capable of simulating complex hydrogeological, biological, chemical, and mechanical processes occurring including processes occur during CO 2 geological sequestration, CO 2 enhanced oil recovery and geothermal systems among others. In this paper, we present our preliminary work as well as examples related to CO 2 geological sequestration. We performed the model coupling through developing an efficient application programming interface (API). IC-FERST-REACT inherits high-order methods and unstructured meshes with dynamic mesh optimization from IC-FERST. This reduces the computational cost by placing the mesh resolution where and when necessary and it can better capture flow instabilities if they occur. This can have a strong impact on reactive transport simulations which usually suffer from computational cost. From PHREEQCRM the code inherits the ability to efficiently model geochemical reactions. Benchmark examples are used to show

Towards Understanding Factors Affecting Arsenic, Chromium, and Vanadium Mobility in the Subsurface

“…Subsurface systems have been modeled using mechanistic approaches both as individual systems and as components of coupled systems [ 111 – 114 , 117 – 123 ]. The well-established nature of mechanistic modeling of subsurface systems suggests a mature state of knowledge that belies the challenges associated with the mechanistic description of certain complex systems, such as those of focus herein.…”

“…For example, mechanistic models are systems of partial differential algebraic equations that necessarily require initial and boundary conditions, and model parameters (e.g., material properties, reaction rates). When these conditions vary in space and time in such a way that specification of the necessary model inputs are stochastic, uncertainty in the resultant model is a natural consequence [ 114 , 117 , 123 – 132 ]. Such uncertainty in mechanistic models of subsurface systems is universally the case, but often ignored by purely deterministic modeling approaches.…”

“…Batch models are made up of a set ordinary differential-algebraic equations with the number of equations dependent upon the number of unique non-equilibrium species-phase combinations [ 134 , 141 , 147 ]. Given the totality of subsurface and solid-phase biochemical complexity, the number of equations in such a model can number in the dozens [ 83 , 118 , 123 , 141 , 143 , 146 , 152 ]. Despite the system complexity, success in modeling the time evolution of the system and the final equilibrium state can be expected for most systems.…”

“…Even if accurate macroscale model formulations are developed, further issues related to the establishment of accurate initial conditions for geogenic contaminant sources and complicated redox conditions would remain [ 126 , 152 , 169 , 171 , 184 ]. This would be exacerbated by long-standing issues of the identification of material parameters such as permeability and porosity that vary significantly in space but can only be observed coarsely [ 123 , 124 , 132 , 152 , 184 ]. These issues have been found to have a greater impact on species transport compared to simpler flow problems [ 118 , 124 , 126 , 129 , 130 , 152 , 155 , 171 , 177 ].…”

“…These issues have been found to have a greater impact on species transport compared to simpler flow problems [ 118 , 124 , 126 , 129 , 130 , 152 , 155 , 171 , 177 ]. This inevitably leads to models of subsurface systems that are best represented stochastically, reflecting existing uncertainty [ 123 , 126 , 132 , 152 , 177 , 184 ].…”