A novel higher resolution spectral volume method coupled with a control-volume distributed multi-Point flux approximation (CVD-MPFA) is presented on unstructured triangular grids for subsurface reservoir simulation. The flow equations involve an essentially hyperbolic convection equation coupled with an elliptic pressure equation resulting from Darcy's law together with mass conservation. The spectral volume (SV) method is a locally conservative, efficient high-order finite volume method for convective flow. In 2D geometry, the triangular cell is subdivided into sub-cells, and the average state variables in the sub-cells are used to reconstruct a high-order polynomial in the triangular cell. The focus here is on an efficient strategy for reconstruction of both a higher resolution approximation of the convective transport flux and Darcy-flux approximation on sub-cell interfaces. The strategy involves coupling of the SV method and reconstructed CVD-MPFA fluxes at the faces of the spectral volume, to obtain an efficient finer scale higher resolution finite-volume method which solves for both the saturation and pressure. A limiting procedure based on the Barth-Jespersen limiter is used to prevent non-physical oscillations on unstructured grids. The fine scale saturation/concentration field is then updated via the reconstructed finite volume approximation over the sub-cell control-volumes. The method is also coupled with a discrete fracture model. Performance comparisons are presented for tracer and two phase flow problems on 2D unstructured meshes including fractures. The results demonstrate that the spectral-volume method achieves further enhanced resolution of flow and fronts in addition to that of achieved by the standard higher resolution method over first order upwind, while improving upon efficiency.
Accurate numerical modeling of fluid transport is essential in reservoir management. Higher-order methods help to improve accuracy by reducing the numerical diffusion, which is common for all first order methods. In this paper, we present an implementation of a MUSCL-type second-order finite volume method and demonstrate its capabilities on 2D and 3D unstructured grids. This includes corner point grids that are typically used in reservoir modeling.
A second order finite volume method is compared to standard first order method in terms of accuracy, performance and an ability to handle nonlinearities. There are several ways to build a second order finite volume method. In this paper we choose an optimization-based strategy to compute the steepest possible linear reconstruction. At the same time, a steepness-limiting procedure is included in the optimization as constraint. This ensures that the steepest possible reconstruction that does not lead to oscillations is computed. As a result, sharper fronts compared to standard schemes are obtained.
The paper demonstrates the described method on several benchmark cases with emphasis on relevant for practical reservoir simulation test cases. In particular, we use Norne field open data set, which enables cross validation with other implementations. We test the method on the transport case, where an analytical solution is known, to verify convergence behavior and to isolate the errors. Furthermore, the performance of first- and second-order methods is compared on multiphase flow problems typical for improved oil recovery: solvent and CO2 injection. The second order method shows superior performance in terms of accuracy.
This paper verifies the desirable properties of higher order method for reservoir simulation. Moreover, all the described implementations are available in an open source reservoir simulator Open Porous Media (OPM). As a result, these methods are accessible for reservoir engineers and can be used with industry standard modeling setups.
Simulation of flow and transport in petroleum reservoirs involves solving coupled systems of advection diffusion-reaction equations with nonlinear flux functions, diffusion coefficients, and reactions/wells. It is important to develop numerical schemes that can approximate all three processes at once, and to high order, so that the physics can be well resolved. In this paper, we propose an approach based on high order, finite volume, implicit, Weighted Essentially NonOscillatory (iWENO) schemes. The resulting schemes are locally mass conservative and, being implicit, suited to systems of advection-diffusion reaction equations. Moreover, our approach gives unconditionally L-stable schemes for smooth solutions to the linear advection-diffusion-reaction equation in the sense of a von Neumann stability analysis. To illustrate our approach, we develop a third order iWENO scheme for the saturation equation of two-phase flow in porous media in two space dimensions. The keys to high order accuracy are to use WENO reconstruction in space (which handles shocks and steep fronts) combined with a two-stage Radau-IIA Runge-Kutta time integrator. The saturation is approximated by its averages over the mesh elements at the current time level and at two future time levels; therefore, the scheme uses two unknowns per grid block per variable, independent of the spatial dimension. This makes the scheme fairly computationally efficient, both because reconstructions make use of local information that can fit in cache memory, and because the global system has about as small a number of degrees of freedom as possible. The scheme is relatively simple to implement, high order accurate, maintains local mass conservation, applies to general computational meshes, and appears to be robust. Preliminary computational tests show the potential of the scheme to handle advection-diffusion-reaction processes on meshes of quadrilateral gridblocks, and to do so to high order accuracy using relatively long time steps. The new scheme can be viewed as a generalization of standard cell-centered finite volume (or finite difference) methods. It achieves high order in both space and time, and it incorporates WENO slope limiting.
The problem encountered by the operator in this oil field is that the reservoir, an oil-filled packstone, is thin and laterally discontinuous. Despite having collected a high-resolution, state-of-the-art 3D seismic survey with usable frequencies up to 138 Hz, and despite having generated seismic attribute volumes in order to assist interpretation, the operator was unable to generate an interpretation manually that matched the rock-type interpretation at the wells. Therefore, the decision was taken to supplement the human interpretation with a machine-learning methodology.
ur Rehman, Syed Raza (Qatar University) | Zahid, Alap Ali (Qatar University) | Hasan, Anwarul (Qatar University) | Hassan, Ibrahim (Texas A&M University at Qatar) | Rahman, Mohammad A. (Texas A&M University at Qatar) | Rushd, Sayeed (King Faisal University)
Horizontal drilling technology has shown to improve the production and cost-effectiveness of the well by generating multiple extraction points from a single vertical well. The efficiency of hole cleaning is reduced because of the solid-cuttings accumulation in the annulus in cases of extended-reach drilling. It is difficult to study the complex flow behavior in a drilling annulus using the existing visualization techniques. In this study, experiments were carried out in the multiphase flow-loop system consisting of a simulated drilling annulus using electrical resistance tomography (ERT) and a high-speed camera. Real-time tomographic images (quantitative visualization) of multiphase flow from ERT were compared to the actual photographs of the flow conditions in a drilling annulus. The quantitative analysis demonstrates that ERT has a wide potential application in studying the hole-cleaning issues in the drilling industry.
The objective was to leverage prestack and poststack seismic data in order to reconstruct 3D images of thin, discontinuous, oil-filled packstone pay facies of the Upper and Lower Wolfcamp formation (Sakmarian time: 293-296 Ma).
The well-to-seismic tie was carefully established using synthetic seismograms, which enabled the facies log to be properly associated with the corresponding seismic samples. The seismic data were all resampled from 2 ms to 0.5 ms in anticipation of being able to recover facies thicknesses on the order of 2 m. Six neural networks with diverse learning strategies were trained to recognize the nine facies classes in the high-resolution seismic stack: Instantaneous Frequency, Instantaneous Q Factor, Inversion (P-Impedance), Semblance, Dominant Frequency, Most Negative Curvature, and eight Angle Stacks, using a two-stage learning and voting process.
At the wells, the nine facies were reconstructed from seismic at a 97% accuracy rate. The bootstrap classification rate, a proxy for blind well testing, was over 80%, which indicates a high-quality modeling process. The pay facies was described with no false positives or false negatives. In the 3D seismic volume between the wells, the procedure produced a Most Likely Facies volume (unsmoothed and smoothed), and nine individual Facies Probability volumes. The pay facies was visualized in a 3D voxel visualization canvas using opacity, and also in a two-way time thickness map. The usable vertical and horizontal resolution was much greater than that of the original seismic. Based on these classification results, additional drilling locations were chosen to further target the oil-filled packstones.
The classification results were created by neural networks, which can be used as a substitute for traditional AVO, inversion and cross-plotting techniques for seismic reservoir characterization. The time need to create the Machine Learning results for this small dataset was on the order of ten minutes.
Due to their potential instabilities, deploying personnel onto icebergs to make direct in-situ measurement is hazardous. The preliminary results from an investigation into the usage of Unmanned Aerial Vehicles (UAV) for surveying and monitoring icebergs are presented. The project had four objectives: (i) acquisition of imagery for the generation of iceberg topside reconstructions using photogrammetry; (ii) development of a GPS tracking device and a deployment mechanism to place it onto an iceberg; (iii) development of a motion sensor to record the motion of an iceberg and a deployment mechanism to deliver it onto an iceberg; and (iv) iceberg draft measurements from a UAV-mounted ice penetrating radar.
The project has used both commercially available and custom-built UAVs. The sensor packages (cameras, tracking devices, accelerometers and ground penetrating radar) were commercial products that have been modified for this study and, when required, mountings and delivery mechanisms have been designed and manufactured to integrate the system together.
Fieldwork was performed during the 2017 iceberg season in a near-shore environment (Bonavista, Newfoundland and Labrador, Canada) aboard a survey vessel and, in 2018, from an operational supply vessel offshore Newfoundland and Labrador. The field campaigns were conducted in parallel with an iceberg profiling system that uses an integrated multibeam sonar and LiDAR system to generate composite (topside and subsurface) iceberg reconstructions. These reconstructions can be compared with the results obtained from the photogrammetry and the radar survey.
During the 2017 program, iceberg imagery for photogrammetry was acquired and GPS tracking devices were deployed onto icebergs and sea-ice. The longest iceberg track obtained was 21 days. For the 2018 campaign, further photogrammetric data was collected and ground penetrating radar surveys of icebergs were performed. The photogrammetry topside reconstructions and the draft estimates from the ground penetrating radar produced results comparable to measurements from the iceberg profiling system.
This project has explored the capability of UAVs to deliver sensor packages onto icebergs, and to take aerial measurements over and around them. They are an emerging technology that, although challenging to work with in the harsh North Atlantic environment, have proved useful.
Mandelli, Sara (Politecnico di Milano, Italy) | Borra, Federico (Politecnico di Milano, Italy) | Lipari, Vincenzo (Politecnico di Milano, Italy) | Bestagini, Paolo (Politecnico di Milano, Italy) | Sarti, Augusto (Politecnico di Milano, Italy) | Tubaro, Stefano (Politecnico di Milano, Italy)
A common issue of seismic data analysis consists in the lack of regular and densely sampled seismic traces. This problem is commonly tackled by rank optimization or statistical features learning algorithms, which allow interpolation and denoising of corrupted data. In this paper, we propose a completely novel approach for reconstructing missing traces of pre-stack seismic data, taking inspiration from computer vision and image processing latest developments. More specifically, we exploit a specific kind of convolutional neural networks known as convolutional autoencoder. We illustrate the advantages of using deep learning strategies with respect to state-of-the-art by comparing the achieved results over a well-known seismic dataset.
Presentation Date: Wednesday, October 17, 2018
Start Time: 1:50:00 PM
Location: 204C (Anaheim Convention Center)
Presentation Type: Oral
Although drawbacks exist, finite-difference simulations remain a popular method to compute solutions to the seismic wave equation. One such drawback, compared to other modeling methods such as the spectral element method, is that staggered finite-difference computations cannot simultaneously output stress and velocity constituents at the free surface. We present an algorithm to obtain the response of all wavefield constituents at the free surface, using two additional, localized, finite-difference simulations. The method is based on the representation theorem, and requires the presence of a homogeneous layer between the free surface and the virtual recording surface just below it. The method is demonstrated on a simple acoustic and elastic model. Artifacts upon timereversal currently obstruct a clean retrieval of first arrivals, and the removal of these artifacts is the subject of further research.
Presentation Date: Wednesday, October 17, 2018
Start Time: 8:30:00 AM
Location: 205A (Anaheim Convention Center)
Presentation Type: Oral
Modern seismic data surveys generate terabytes of data daily leading to a significant increase of the cost for storage and transmission. Therefore, it is desired to compress seismic data. In this work, we propose a model-based compression scheme to deal with the large data volume. First, each seismic trace is modeled as a superposition of multiple exponentially decaying sinusoidal waves (EDSWs). Each EDSW represents a model component and is defined by a set of parameters. Secondly, a parameter estimation algorithm for this model is proposed using Particle Swarm Optimization (PSO) technique. In the proposed algorithm, the parameters of each EDSW are estimated sequentially wave by wave. A suitable number of model components for each trace is determined according to the level of the residuals energy. The proposed model based compression scheme is then experimentally compared with the discrete Cosine transform (DCT) on a real seismic data. The proposed model based algorithm outperforms the DCT in term of compression ratio and reconstruction quality.
Presentation Date: Tuesday, October 16, 2018
Start Time: 1:50:00 PM
Location: Poster Station 20
Presentation Type: Poster