Computational Geosciences最新文献

The Edvard Grieg field is a highly complex and heterogeneous reservoir with an extensive fault structure and a mixture of sandstone, conglomerate, and shale. In this paper, we present a complete workflow for history matching the Edvard Grieg field using an ensemble smoother for Bayesian inference. An important aspect of the workflow is a methodology to check that the prior assumptions are suitable for assimilating the data, and procedures to verify that the posterior results are plausible and credible. We thoroughly describe several tools and visualization techniques for these purposes. Using these methods we show how to identify important parameters of the model. Furthermore, we utilize new compression methods for better handling large datasets. Simulating fluid flow and seismic response for reservoirs of this size and complexity requires high numerical resolution and accurate seismic models. We present a novel dual-model concept for a better representation of seismic data and attributes, that deploy different models for the underground depending on simulated properties. Results from history matching show that we can improve data matches for both production data and different seismic attributes. Updated parameters give new insight into the reservoir dynamics, and are calibrated to better represent water movement and pressure.

Traditional interpolations might cause smoothing effect on geochemical anomaly detection due to the moving weighted average properties. Since Multiple-Point Statistics (MPS) is a kind of stochastic simulation based on regional variables statistical patterns in a certain space, it can reduce the smoothing effect and quantify the element distribution uncertainties effectively. However, due to the insufficient Training Images (TIs) in geochemical exploration fields, simulation processes cannot be directly applied on the original data. Meanwhile, element spatial distribution patterns cannot be finely characterized under single scale, with uncertainty exists during the attribute information prediction in some regions. In addition, due to the stochastic properties, it is difficult to identify geochemical anomalous information accurately based on various simulation results. Therefore, a hybrid framework combined MPS and Local Singularity Analysis (LSA) are mainly introduced in this paper. Firstly, rasterization algorithms are used to construct geochemical TI to ensure the MPS simulation processes. Then, two-step simulation, including large-scale and small-scale simulation, is applied to finely represent the geochemical element distribution patterns. Based on various simulation results, LSA and information fusion are finally introduced to construct the probability map of geochemical anomalies. The stream sediment geochemical data was mainly used in this paper to verify the feasibility of proposed methods. Results show that comparing with the Kriging-based ones, smoothing effect of different geochemical anomalous fields is significantly reduced, which shows a closer spatial correlation with the known deposits according to the ROC curve analysis. Based on the anomaly identification results, some mineralization indices can be preliminarily determined to offer some theoretical supports for further mineral exploration.

In this paper, three-dimensional numerical algorithm is constructed to simulate the behavior of the Brinkman-Forchheimer flow and thermal fields. Numerical results of velocity, pressure and temperature are obtained by applying the efficient modified two-grid marker and cell (MAC) algorithm on staggered grids with the second-order backward difference formula (BDF2) time approximation. The modified-upwind idea is introduced to convective heat transfer equations for improving accuracy without any numerical oscillation. The second-order convergence rate can be achieved for pressure, velocity and temperature of considered three-dimensional model. Some numerical experiments are presented to illustrate the efficiency of algorithm. The numerical example with analytical solution is used to validate the effectiveness and accuracy of the algorithm by comparing with the results of traditional MAC algorithm. A time-dependent test is proposed to show a detailed sensitivity analysis to indicate the influence of parameters including the (varepsilon ), Forchheimer number, Brinkman number and thermal diffusivity on the physical properties of Brinkman-Forchheimer flow and heat transfer in porous media.

In reservoir simulation, it is important to understand the mechanical behaviour of fractured rocks and the effect of shear and tensile displacements of fractures on their aperture. Tensile opening directly enhances the fracture aperture, whereas shear of a preexisting rough-walled fracture creates aperture changes dependent on the local stress state. Since fracture dilatation increases reservoir permeability, both processes must be included in a realistic and consistent manner into the mechanical reservoir simulation model. Here, we use the extended finite volume method (XFVM) to conduct flow and geomechanics simulations. In XFVM, fractures are embedded in a poroelastic matrix and are modelled with discontinuous basis functions. On each fracture segment the tractions and compressive forces are calculated, and one extra degree of freedom is added for both the shear and tensile displacement. In this particular XFVM implementation we assume that linear elasticity and steady state fluid pressure adequately constrain the effective stress. In this paper, shear dilation is not calculated a posteriori, but it enters the equations such that aperture changes directly affect the stress state. This is accomplished by adding shear dilation to the displacement gradients and therefore ascertains a consistent representation in the stress-strain relations and force balances. We illustrate and discuss the influence of this extra term in two simple test cases and in a realistic layer-restricted two-dimensional fracture network subjected to plausible in situ stress and pore pressure conditions.

This study deals with the numerical analysis of several discontinuous Galerkin (DG) methods for the resolution of the Navier-Stokes equations with power law slip boundary condition. The physical context corresponding to this problem is the description of a flow when a position and the direction slip boundary condition is taken into consideration. The main goal in this work is to examine the solvability, convergence of several DG methods, and to discuss their practical resolution by means of fixed point iterative algorithm. Theoretical findings are backed up by solid computational results.

The particle filter is a data assimilation method based on importance sampling for state and parameter estimation. We apply a particle filter in two different quasi-static experiments with models of subsidence caused by a compacting reservoir. The first model considers uncorrelated model state variables and observations, with observed subsidence resulting from a single source of strain. In the second model, subsidence is a summation of subsidence contributions from multiple sources which causes spatial dependencies and correlations in the observed subsidence field. Assimilating these correlated subsidence fields may trigger weight collapse. With synthetic tests, we show in a model of subsidence with 50 independent state variables and spatially correlated subsidence a minimum of (varvec{10^{13}}) particles are required to have information in the posterior distribution identical to that in a model with 50 independent and spatially uncorrelated observations. Spatial correlations cause an information loss which can be quantified with mutual information. We illustrate how a stronger spatial correlation results in lower information content in the posterior and we empirically derive the required ensemble size for the importance sampling to remain effective. We furthermore illustrate how this loss of information is reflected in the log likelihood, and how this depends on the number of model state variables. Based on these empirical results, we propose criteria to evaluate the required ensemble size in data assimilation of spatially correlated observation fields.

Current parameterizations of the hydrodynamic forces on irregular particles consider some shape dependencies, but lack an explicit dependence on the orientation with respect to the flow. In this paper, we propose a new parameterization of the drag and lift forces acting on whole Limpet shells at arbitrary orientations with respect to the direction of flow through the linear regression of fluid forces against the velocity components in an object frame of reference. The fluid forces were estimated using boundary layer-resolving Reynolds-averaged Navier-Stokes (RANS) simulations. We verified the accuracy of the shear stress transport (SST) (k-omega ) turbulence model on flat plates with varying angles of attack, and we achieved coefficients of determination versus existing data of approximately 0.95 for both the drag and lift coefficients. From the linear regression of our simulated force data, we developed a model as a function of 3-dimensional orientations to predict the hydrodynamic forces acting on a Limpet shell with coefficients of determination of 0.80 for normal forces and 0.51 for longitudinal forces.

In this work, we study the calibration of the parameters of a hydrogeological watershed model by comparing a 1D+2D approach that combines unsaturated 1D columns and a saturated 2D model, with a full 3D approach. In a first step, a heterogeneous permeability field is estimated by an inversion procedure for each model (2D saturated and 3D unsaturated). The fields obtained are similar but the calculation time is obviously much higher in the case of the 3D model: the 2D model seems therefore sufficient and more efficient to evaluate permeability fields using piezometric measurements in the case of vertically homogeneous aquifers. The second step focuses on the calibration of the hydraulic parameters by adjusting the hydraulic heights either derived from a 1D+2D reference model at several fictitious points distributed over the entire domain, or measured in a dozen real piezometers. Both approaches provide a good fit to the piezometric measurements, but the parameter values differ significantly: the van Genuchten alpha coefficient is unrealistic in the 1D+2D approach, reflecting a poorer consideration of the modeling unsaturated zone, while the porosity value is higher in the 3D approach, which can probably be remedied by developing a more suitable cost function.

This paper revisits the previously developed NH-2LR (reduced two-layer non-hydrostatic) model. The governing equations and numerical schemes are written in terms of normalized variables, with two dimensionless parameters representing dispersion and non-linearity. By utilizing analytical solutions and laboratory experiments, this study aims to validate the numerical NH-2LR model and investigate the effects of dispersion and non-linearity on the resulting waves. The first validation employs the analytical solution of the linear and fully dispersive model of a landslide moving with constant velocity on a flat bottom. The second validation involves a landslide hump sliding over a constant beach slope. A closer look at the run-up height reveals that this case is non-dispersive. Furthermore, we found that the dispersion effect was evident from the beginning of the wave formation process. Finally, we compare our numerical results to experiments on submarine landslides on sloping beaches. We found that dispersion is essential in the early generation and propagation of waves in off-shore regions. Moreover, non-linearity significantly influences the maximum run-up of landslide-generated waves.