PS01 - Multi scale method for Micro-scale simulation: An effective approach to understand and quantify the impact of micro-scale heterogeneity for improved and enhanced oil recovery
Mayur Pal Dept. of Applied Mathematics, KTU, Lithuania
Abstract: A micro-scale heterogeneity quantification methodology is presented, which uses petrographic image analysis with multi scale simulation. Petrographic images of micro-porous carbonate rocks are converted into microscale models with help of an image analysis workflow, which helps in assigning specific porosity, permeability and petro-physical properties to the microscale model. Thin section images are discretized into classes of specific porosity. Each specific porosity class is then assigned a permeability using the Kozeny-Carman equation for permeability. Doing so, the thin section image is transformed into a 2D porosity and permeability grid preserving the true heterogeneity at microscale. A multi-scale flow simulation using mixed finite element formulation is then used to perform fast flow simulation on the reconstructed grid representing the microscale model. Although researcher have used pore-network or Lattice-Boltzmann modeling techniques to understand impact of fine scale heterogeneities on fluid flow in porous media, these methodologies are computationally very expensive and impractical for large scale application. Authors present a pragmatic and time-efficient approach for simulating flow experiments at the microscale providing valuable insights into the impact of petrographic components on flow characteristics.
PS02 - PREDICT: PeRmEability DIstributions of Clay-smeared faulTs
Lluís Saló-Salgado (*1,2), J. Steven Davis (3,4), Ruben Juanes (1,2,5) 1 - Department of Civil and Environmental Engineering, MIT, Cambridge, USA 2 - Earth Resources Laboratory, MIT 3 - ExxonMobil Upstream Integrated Solutions Company, Spring, Texas, USA 4 - Now at Stanford Center for Carbon Storage, Department of Energy Resources Engineering, Stanford University, Palo Alto, CA 5 - Department of Earth, Atmospheric and Planetary Sciences, MIT, Cambridge, USA
Abstract: We focus on estimating the intrinsic permeability of faults in siliciclastic sedimentary basins at shallow depths (< 3 km). We introduce PREDICT, a physics-based, stochastic algorithm that computes probability distributions for the directional components (perpendicular, strike-parallel and dip-parallel) of the fault-scale permeability tensor. To achieve this, PREDICT models the main shear zone at high-resolution, in 2D, and takes advantage of MRST for flow-based upscaling. The algorithm requires a set of input parameters that describe the faulted stratigraphy. From these inputs, samples for a set of numerical quantities including fault thickness and layer residual friction angle, critical shale smear factor, permeability and permeability anisotropy are drawn. This allows populating a high-resolution discretization of the fault core with clay smears and sand-based fault materials. The final step consists in permeability upscaling, which uses a combination of the MRST modules incomp, coarsegrid, upscaling and mpfa (depending on user choices). This process is repeated multiple times, each repetition representing one realization, until the full permeability distribution for each component is obtained. We outline the differences between PREDICT’s output and that of previous fault permeability algorithms, and we illustrate how to use PREDICT with a MRST installation as well as its application to reservoir- and basin-scale flow simulation models of geologic CO2 sequestration in offshore sedimentary environments.
PS03 - Uncertainty Analysis of SCAL Data
Omidreza Amrollahinasab, Siroos Azizmohammadi, Pit Arnold, Holger Ott Montanuniversität Leoben, Austria
Keywords: Special Core Analysis (SCAL), data interpretation, simultaneous history matching, uncertainty analysis, saturation functions Module dependencies:ad-core, ad-blackoil, ad-props
Standard analytical techniques, which are used to calculate saturation functions from SCAL experiments, are limited to their underlying assumptions. Here we perform uncertainty analysis and inverse modeling, which enables us to have a better interpretation of the physics behind the SCAL experiments. This includes the simultaneous interpretation of capillary pressure (P_C (S_W)) and relative permeability (k_r (S_W)) data, as both saturation functions and experimental data are highly coupled. In this presentation, we show an open-source tool developed based on MATLAB-MRST library for comprehensive interpretation of SCAL data. The tool runs forward simulations using MRST and then matches the simulation predictions and the experimental data by varying the saturation functions using the MATLAB optimization toolbox. Saturation functions are constructed and varied in a point-by-point fashion to overcome the limitations imposed by given parametrization of the models like Corey and LET. Core flooding and centrifuge data are matched simultaneously using a single set of saturation functions and summing up the total error in a single objective function. The uncertainties are then analyzed using the Parallel High-Performance Delayed-Rejection Adaptive Metropolis Markov Chain Monte Carlo method, which provides a sampling tool to explore the response surface, and the uncertainty ranges around the history matching results. The tool utilizes the high computational efficiency provided by MRST and combines it with the parallelization, optimization, and sampling capabilities of the libraries in MATLAB to run the simulations in a fast and efficient way.
PS04 - An Uncertainty Quantification Workflow for Naturally Fractured Reservoirs using Proxy Modelling based on Poro-mechanically Informed Flow Diagnostics Simulations
L. Gutierrez Sosa, S. Geiger, and F. Doster Heriot-Watt University
Keywords: uncertainty; reservoir numerical simulation; flow diagnostics; heterogeneity; poro-mechanics Module dependencies: diagnostics and in-house modules build in MRST
Abstract: Carrying out uncertainty quantification and robust optimisation workflows for naturally fractured reservoirs is very challenging because exploring and capturing the full range of geological and mechanical uncertainties requires a large number of numerical simulations and hence computationally intensive. Specifically, the integration of poro-mechanical effects in full-field reservoir simulation studies is still limited, mainly because of high computational cost. As a result, poro-mechanical effects are often ignored in uncertainty quantification and optimisation workflows, which may result in inadequate reservoir performance forecasts. Computationally efficient poro-mechanical screening methods are therefore important to identify if poro-mechanics could impact reservoir dynamics and identify individual models from a model ensemble for more detailed full-physics reservoir simulations.
We introduce a new methodology that extends traditional uncertainty quantification workflows, through the use of poro-mechanical informed flow diagnostics and proxy models. This approach provides first-order approximations of the complex interactions between poro-mechanics and hydrodynamics using existing steady-state dual-porosity flow diagnostics and coupled dual-continuum poro-mechanics. This computationally efficient calculations allow us to quickly quantify poro-mechanical impact on reservoir dynamics and further enable us to select representative models that capture the uncertainty quantified in a reservoir model ensemble. These representative models can then be used in further, computationally intensive full-physics coupled reservoir simulations. The proposed poro-mechanical screening hence provides an efficient complement to traditional reservoir simulation and uncertainty quantification workflows and enable us to assess a broader range of geological, petrophysical and mechanical uncertainties.
Using a series of case studies based on a fractured carbonate reservoir analogue, we demonstrate how (1) uncertainty quantification workflows can be improved by considering different hydrodynamical-poro-mechanical scenarios, (2) how bias in the uncertainty estimation can be reduced by executing thousands of Monte Carlo realisations using ANN-based proxy models, and (3) how cluster analysis can be performed to identify a suitable set of representative models from a large model ensemble without reducing uncertainty in reservoir performance predictions. The proposed framework has been implemented using the open-source MATLAB Reservoir Simulation Toolbox MRST and was linked to a commercial reservoir simulation package to carry out the experimental design, construct the proxy model, and perform the sensitivity and uncertainty analysis.
PS05 - Dual porosity-dual permeability model in MRST
Nikolai Andrianov Geological Survey of Denmark and Greenland (GEUS)
Abstract: Fine-scale discrete fracture simulations provide a natural means to model fluid flows in fractured reservoirs. However, an application of discrete fracture modeling on the field scale is challenging due to uncertainties in fractures’ properties, difficulties in creating conforming meshes, and the computational complexity of fluid flow simulations.
Upscaling of flows in fracture networks has been traditionally used in order to cope with the challenges highlighted above. One common approach is to consider the matrix and the fractures as two inter- penetrating continua, characterized by different rock properties. If the matrix permeability is significantly lower than the fracture permeability, the flow in the matrix can be neglected and the resulting model is termed the dual porosity (DP) model; otherwise the flow in the matrix must be modelled and the corresponding model is referred to as the dual porosity-dual permeability (DPDP) model.
The present contribution describes an extension of the available dual porosity MRST module to the dual porosity-dual permeability model. To this end, a new inherited class for the two-phase oil-water fluid model is implemented, which describes the flow equations for the matrix medium and the associated boundary conditions. Besides, a new transfer function is introduced which allows for specification of variable shape factors, precomputed from fine-scale discrete fracture simulations per grid block.
The performance of the DPDP model is benchmarked against the fine-scale solutions for a synthetic and realistic fracture test cases. In all cases, the DPDP solutions provide a good approximation to the reference fine-scale results. In terms of computational cost, the DPDP runs are significantly faster than the corresponding fine-scale simulations, with the speedup of up to 3 orders of magnitude for the realistic fracture geometry case.
PS06 - Numerical Investigation of the Polymer-Gel Treated and Fractured Core Samples During Improved Oil Recovery by Using Embedded Discrete Fracture Model of MRST
Onur Alp Kaya (*1), İsmail Durgut (1), Serhat Canbolat (2) 1 - Middle East Technical University 2 - Near East University
Abstract: The dynamics of fluid flow in the matrix and in fractures are significantly different since fractures drastically impact the fluid flow. Fractures are considered the main flow unit since they are highly permeable flow channels, and the matrix is considered the storage unit since it occupies most of the reservoir volume. This study aims to numerically model waterflooding experiments in artificially fractured and gel-treated core plugs by using the Embedded Discrete Fracture Model (EDFM) of MRST.
MRST, which provides the numerical solution, was used to model different core plugs representing three prominent cases: non-fractured, artificially fractured, and polymer gel treated core plugs. 2 PV water was injected into all core plugs, and a 3D dimensional fluid saturation profile during the injection and oil recovery vs. time plots were obtained. The model was validated by the standard Buckley-Leveret solution and EDFM was used for the fracture modeling.
MRST results were compared with the experimental results. For non-fractured and artificially fractured cases, results were obtained after 2 PV water injection, and for polymer gel treated core plugs, 2 PV more water was injected after the polymer gel operation. For an artificially fractured core plug, mean water saturation was measured as 63.87% and 71.7% experimentally and 65.33% and 72.53% with MRST before and after polymer gel treatment. Similarly, oil recovery increased from 33.33% to 46.66% experimentally and increased from 34.15% to 44.47% in MRST after polymer gel treatment. Both experiments and MRST simulations show that fractures lead to non-uniform oil swept, decreasing overall oil recovery. Polymer gel treatment decreases the fracture permeability and leads to uniform swept, thus increases the recovery. Finally, the effects of fracture aperture and permeability on the recovery were investigated with MRST. Fracture permeability directly affects recovery for high values when the aperture is constant, and aperture directly impacts recovery for the low values when permeability is constant.
This study shows that numerical results and experimental results are significantly close to each other for all cases. In addition, obtained results are supported by the available literature and can be useful in the future.
PS07 - Application of MRST to the Simulation of LNAPL Transport in Groundwater
Ehsan Ranaee (*1), Giovanni Porta (2), Monica Riva (2), Fabio Inzoli (1), Alberto Guadagnini (2) 1 - Dipartimento di Energia, Politecnico di Milano, Milano, Italy 2 - Dipartimento di Ingegneria Civile e Ambientale, Politecnico di Milano, Milano, Italy
Abstract: Several scenarios environmental and industrial concern such as, e.g., oil displacement across a reservoir or migration of a contaminant plume involve multiphase flow of fluids through a porous medium. In this context, computational tools specifically developed for petroleum engineering applications are potentially transferrable to simulating contamination processes taking place in soil-water systems. Since operational conditions (e.g., confinement, consolidation, solid surface properties, and operating pressures) can be different from those encountered in a deep reservoir, feasibility studies should aim at assessing such applications.
The Matlab Reservoir Simulation toolbox (MRST) can handle complex geologic structures and fluid behaviors, including multiphase-multicomponent flows. Here, we assess the ability of the MRST toolkit to simulate soil contamination by light non-aqueous phase contaminants (LNAPL). We did so upon relying on a compositional formulation of the governing equations employed for reservoir simulations.
Our key objectives are: (i) compare the MRST solution against that provided by other specialized software packages and (ii) discuss uncertainties arising from the parameterization of the system. We rest on problem n. 7 (hereafter termed base-case) of the TMVOC user guide (Pruess and Battistelli 2002) to assess the capability of MRST to characterize LNAPL transport in the subsurface. The case study analyzes a multicomponent LNAPL spill in the unsaturated zone, followed by redistribution of the plume within the unsaturated zone and along the water table. We set up a numerical simulation model of the contaminated site with the same initial and boundary conditions of the base-case benchmark and compare the ensuing solution against the TMVOC one. We analyze the temporal evolution of water, air, and LNAPL saturation locally (i.e., in a total of 1944 cells along three observation vertical cross-sections) throughout the computational domain, as well as the time of contamination at a set of locations at increasing distance from the spill. We then consider a set of uncertain model parameters (including parameters of the relative permeability and capillary pressure models employed) and quantify the impact of such uncertainties on MRST simulation outputs through a rigorous uncertainty quantification analysis.
Simulation of multiphase flow in fractured reservoirs still poses a challenge due to the different timescales of fluid flow in fractures and matrix. Common approaches to model fractures in reservoir simulators include the discrete fracture and matrix method (DFM), where the fractures are explicitly represented as lower dimensional elements in the computational mesh, and multi-continuum approaches (e.g., dual-porosity and dual-permeability models) where the behaviour of the fractures and matrix are integrated and treated as distinct continua. The latter requires models (bespoke “transfer functions”) that upscale the multiphase transfer between fracture and matrix. There are several formulations for transfer functions available in the literature, and they are often application-dependent. Here, we propose a unified framework for simulation of flow in fractured media. The framework makes no distinction between dual-continuum and DFM methods, treating fractures and one or more matrix domains as “flowing domains” and “virtual domains”. Transfer functions are reinterpreted as fluxes between cells of different domains, which enables us to create an abstraction that encompasses both methods and makes it easy to build hybridized models including different regions with different matrix/fracture interaction concepts. We present a series of cases to illustrate the main differences between both modelling approaches and the benefit of a flexible implementation.
PS09 - An MRST module for modelling phase behavior of complex fluids using ePC-SAFT EoS
Elyes Ahmed * and Xavier Raynaud SINTEF Digital
Keywords: ePC-SAFT EoS, phase equilibrium, salt precipitation, multiphase flash calculation
The Carbon Capture and Storage (CCS) of CO2 in geological reservoirs especially in saline aquifers is expected to play a key role in reducing CO2 emissions from fossil-fuel combustion and industrial production. In addition, the most widely accepted approach is the injection of CO2 as an enhancer for the recovery of oil (EOR) into reservoir. This operation may experience salt precipitation, which reduces porosity and impairs permeability of the reservoir in the vicinity of the wellbore, and can lead to reduction in injectivity. Compositional and multiphase simulation is often a requirement for reservoir simulation involving CO2 storage. In particular, phase equilibria based on flash calculations with an Equation of State (EOS) is the widely used approach to accurately predict fluid properties. In this work, we present an MRST implementation of the electrolyte Perturbed-Chain version of Statistical Association Fluid Theory (ePC-SAFT) Equation of State (EoS). The ePC-saft is known to be accurate for calculating fluid properties compared to a cubic EoS, such as the Peng- Robinson (PR) or Soave-Redlich-Kwong (SRK) EoS. It provides also a good predictability for derivative properties and for inverting densities. The features of the ePC-SAFT EoS are tested in the simulation of the phase behavior of complex fluids such as asphaltenic crude oil and salt precipitation in the context of CO2 storage.
Abstract: We present a recently developed framework for computing parameter sensitivities for models of the GenericBlackOilModel class. The typical application is model parameter tuning, i.e., to adjust model parameters such that model simulation output (well component rates and pressure) matches some reference output as close as possible. In this case the sensitivity of a given model parameter is the gradient of some mismatch-function with respect to the parameter. With the framework, any number of parameters can be set up, and the corresponding sensitivities for a given simulation are computed by a single adjoint run. The new ModelParameter class is used to set up parameters and includes options for parameter multipliers, parameter groupings and non-linear parameter scaling. The class contains default setup of some common parameters such as transmissibility, permeability, pore volume, well connection factors, relative permeability function scalers and initial state. Custom parameters can easily be set up as long as they appear directly in the model equations and are defined in the model structure, schedule or initial state. We illustrate the use of the code by two simple examples. In the first example we tune the parameters of an initially upscaled model to better match the output of the fine underlying model. In the second example, we utilize the code for optimization of well valve settings to maximize simulated net-present-value (NPV).
Abstract: Gmsh is a popular open source geometry modeling and mesh generator software. I will show how to use Gmsh to generate grids and how to import them into MRST. Gmsh can be found at https://gmsh.info.
Abstract: During CO2 injection into geological reservoirs, CO2 may flow through faults and fractures present in seals. CO2 dissolution can acidify the formation water and drive a range of mineral reactions such as silicate mineral dissolution or precipitation of the carbonates. The dissolution reactions may increase the permeability of the fracture networks in the seals and also reduce the capillary entry pressure, which can lead to the CO2 leakage. On the other hand, formation of new minerals may decrease the permeability of the fracture systems and increase the caprock integrity. To predict how the dissolution and precipitation reactions affect the permeability of the fracture network in fractured caprocks, we develop a model that can simulate the reactive transport processes in fracture networks. The reactive transport model is based on the Discrete fracture and matrix (DFM) model and is implemented in the MATLAB Reservoir Simulation Toolbox (MRST). The developed model is then used to simulate the reactive transport process when the CO2-acidified brine is injected into different fracture networks (including a fracture network from Svalbard) and triggers calcite dissolution and precipitation within the fractures.
Abstract: Several graphical methods have been developed to understand the stratigraphy observed in wells and assist experts in estimating rock quality, defining limits for barriers, baffles, and speed zones, and in particular, delineating hydraulic flow units. The idea behind GAWPS is to provide a module to MRST with routines that carry out graphical analyses, both qualitatively and quantitatively by using classical methods such as the Modified Lorenz Plot, Normal Probability Plot, and Normalized Cumulative RQI Plot, as well as a secondary method we have developed for zone classification through flow unit speed. We will give an overview of the current implementations and applications embedded with GAWPS to a series of well-known benchmark reservoir models (e.g. SPE10, UNISIM I, and NORNE).
Abstract: In this poster I briefly present how two-coupled fluid flow and geomechanics can be modeled in MRST using the ad-mechanics module, and demonstrate the basic functionality on two well-known examples: Terzaghi's problem and Mandel's problem.
PS15 - Ensemble simulations in MRST using the ensemble module
Håvard Heitlo Holm * and Øystein Klemetsdal SINTEF Digital
Keywords: ensembles, monte carlo, history matching, uncertainty quantification Module dependencies:ad-core, ad-blackoil, mrst-gui, ad-props, example-suite, incomp
Abstract: Simulation of subsurface flow is always subject to a high degree of uncertainty. Typically, we only have partial and noisy knowledge about the geology and state of the reservoir, meaning that the computational model, grid, and its parameters are uncertain. A common way to express this uncertainty is to run an ensemble of simulations using parameters sampled from a probability distribution that reflects the uncertainty in the reservoir model. By simulating the ensemble members independently, we can analyse the corresponding uncertainty in the future well responses, giving us richer knowledge of the reservoir compared to running only a single simulation. Ensembles have several applications, such as quantifying the uncertainty in production data through Monte Carlo simulations, calibrating the uncertain reservoir parameters through history matching, or doing optimization under uncertainty.
This poster presents the recent work in creating a module within MRST for doing ensemble simulations. We present the MRSTEnsemble class and show the different components it is built around. We also highlight some of the options you have when running ensembles, such as parallel execution both with and without the parallel toolbox installed. Finally, we give a quick demonstration to how to use the module for history matching.
Julia is a modern, dynamically typed programming language that claims to have the simple syntax of MATLAB or Python with the potential to reach the same efficiency as compiled languages like C++ or Fortran. We present a small case study of how Julia can be used to significantly accelerate MRST simulations through a compact AD simulator written in pure Julia with support for compressible, multiphase flow and multisegment wells. The extension can run MRST cases directly on CPU and GPU, with support for MRST grids and wells for both simulation input and 3D visualization.
In this poster, we present the implementation of the MPSA method in MRST. The method is a cell-centered, finite-volume method for linear elasticity. After a brief introduction of the method, we review the examples included in the module, which also cover coupling with generic MRST flow solvers.
