- Home
- Conferences
- Conference Proceedings
- Conferences
Petroleum Geostatistics 2019
- Conference date: September 2-6, 2019
- Location: Florence, Italy
- Published: 02 September 2019
81 - 100 of 108 results
-
-
A Simulation Analysis of CO2 Capture and Underground Storage Monitoring in Smeaheia
Authors S. Anyosa, S. Bunting, J. Eidsvik and A. RomdhaneSummaryThe emissions of CO2 are an environmental problem and one possible solution is its capture and conduct underground storage (CSS). However, there is potential risk of leakage, and to aid in this challenge we propose to use statistical modeling for efficient monitoring and classification of sealing and leakage scenarios in the Smeaheia aquifer, in Norway. In this work, the approach is based on geostatistical simulations of the CO2 plume in the aquifer and on generating synthetic seismic data, for both leak and seal scenarios. The knowledge of estimated leakage probabilities allows better monitoring schemes and early leakage detection over time, which can be designed to support the decision making process on CO2 CSS projects.
-
-
-
Statistical Description of Weak Interface Properties and their Impact on Hydraulic Fracture Height
Authors C. Hammerquist, Y. Aimene, J. Nairn and A. OuenesSummaryInterfaces are recognized to a be major in-situ layered reservoir realities that are challenging the oil and gas industry. Their impact on the hydraulic fracture vertical growth is well established and yet their properties are difficult to assess with the current limited data. The purpose of this analysis is to estimate the hydraulic fracture vertical growth in layered rock while focusing on the heterogeneous interfaces properties. The model combines deterministic tools based on the Material Point Method (MPM) geomechanical model which includes interface modeling tools with a stochastic description of the interfaces in the geomechanical model. Combining this modeling capability with the Monte Carlo simulations, makes it possible to propagate the uncertainty of the interface properties to resulting fracture heights. The model shows a large difference of fracture heights in the presence of interfaces. The analysis demonstrated the ability of the model to capture the range of uncertainties in hydraulic fracture heights that can be input in probabilistic frac design and analysis software.
-
-
-
Application of Deep Learning in Reservoir Simulation
Authors S. Ghassemzadeh, M. Gonzalez Perdomo and M. HaghighSummaryReservoir simulation plays a vital role as oil and gas companies rely on them in the development of new fields. Therefore, a reliable and fast reservoir simulation is a crucial instrument to explore more scenarios and optimize the production. In each simulation, the reservoir is divided into millions of cells, and rock and fluid attributes are assigned to these cells. Then, based on these attributes, flow equations are solved with time-consuming numerical methods. Given the recent progress in machine learning, the possibility of using deep learning in reservoir simulation has been investigated in this paper. In the new approach, fluid flow equations are solved using a deep learning-based simulator instead of time-consuming mathematical approaches. In this paper, we studied 1D Oil Reservoir and 2D Gas Reservoir. Data sets generated using the numerical models were used to create the developed simulators. We used two metrics to evaluate our models: Mean Absolute Percentage Error (MAPE) and correlation coefficient (R2). Given the low value of these matrics (MAPE < 15.1%, R2 >0.84 for 1D and MAPE < 0.84%, R2 ≈ 1 for 2D), the results confirmed that the deep learning approach is reasonably accurate and trustworthy when compared with mathematically derived models.
-
-
-
Inverse Modeling with Deep Neural Network and K-medoids Clustering Under Uncertain Geological Scenarios
More LessSummaryThis paper presents an inverse modeling based on deep-neural-network, of which scheme integrates data-encoding with stacked autoencoder and k-medoids clustering to select the adequate geo-models for the supervised-training dataset in the presence of uncertain geological scenarios. The reliable geological scenarios are essential at the successful history matching as well as the accurate forecasting but the limited data obstruct the consideration of well-production-performances. The developed method reduces the errors of matching and forecasting profiles as workflow stages, and results out the reliable plausible geo-models satisfying different well-oil-rates. K-medoids clustering screens error-prone geo-models implementing flow-response-based distances. The results show that deep-neural-network can be applicable as a robust history-matching tool under multiple geological interpretations.
-
-
-
Geostatistical Seismic Shale Rock Physics AVA Inversion
Authors M. Cyz and L. AzevedoSummaryThe main goal of reservoir characterization is the description of the subsurface rock properties (i.e. porosity, volume of minerals and fluid saturations). This is commonly done in a sequential, two-step, approach: elastic properties are inferred from seismic inversion, which are then used to compute rock properties by applying calibrated rock physics models. However, this sequential procedure may lead to biased predictions as the uncertainties may not be propagated through the entire process. To overcome these limitations, here we propose the inference of shale rock properties directly from seismic data using a geostatistical direct shale rock physics AVA inversion. The purpose of the proposed geostatistical direct shale rock physics AVA inversion is to extract the properties included in the composition of a shale volume, such as brittleness, TOC and porosity from the seismic reflection data. The proposed method is applied to a real dataset from a Lower Paleozoic shale reservoir in Northern Poland.
-
-
-
Features of Factor Models in Seismic
More LessSummarySome seismic models that are close to classical models of analysis of variance are considered. They allow you to analyze and identify statistically significant variations of factors, which are important in data processing, as well as for solving inverse seismic problems. The focus is on the properties of these models, which distinguish them from the classical models. These properties are determined by the structure of observations characteristic of real seismic data, and the kinds of directions, allocated in the forming of factor models. As a result, new particularities appear in the model parameter estimation problem. In particular, the ambiguity increases, and to eliminate it, methods using truncation of observations and the formation of some additional conditions based on an analysis of the internal interaction between factors are proposed.
-
-
-
Statistical Properties of Multiplicative Factor Models
More LessSummaryThe results of the study, based on the study of the probabilistic characteristics of a random variable which appear after logarithmic of spectra of the path intervals, are presented. They make it possible to understand the properties of both the methods of factor decomposition and the estimates of the target parameters obtained or derived from them. In particular, such an analysis allows conclusions to be drawn about the quality of the primary observations and the degree of approximation of the highlighted signal component, as some regular element present in the observed wave field. Thus, the issue of the possibility of applying a particular model in processing the available experimental data can be resolved. As a rule, the distinguished regular element carries the basic information used later on at the level of interpretational models.
-
-
-
Iterative Approach of Gravity and Magnetic Inversion through Geostatistics
Authors A. Volkova and V. MerkulovSummaryIn this work was demonstrated iterative approach of potential fields (gravity and magnetics) inverse problem solution with the aim of correct accounting of sedimentary section. The approach consist of simple model creation and further potential fields forward solution and comparison with the base case and after that model modification. Different iterations of simple model construction were tested on the “ground truth” detailed model with the main features of the West Siberia Palaeozoic deposits. The objective of the research is to minimize effect of sedimentary layers on the gravity and magnetic fields by modeling with geostatistical approach (stochastic simulation with trends) and explain how to use this approach according to real data.
-
-
-
Study on the Fine Prediction of Ediacaran Fractured-vuggy Karst Reservoir
More LessSummaryFractured-vuggy carbonate reservoir has strong anisotropy and complex fracture-vuggy distribution. The quantitative description of dissolved pore and fracture is the key to reservoir prediction. The elastic parameter pairs of P-wave impedance and the P-S wave velocity ratio can be utilized to better remove the siliceous layers with the low P-wave impedance and low P-S wave velocity ratio, to identify the low P-wave impedance and the comparatively lower P-S wave velocity ratio, in order to reduce the ambiguities of the reservoir prediction. The curvature and texture attribute profiles have significant differences in response to different reservoir types, and their characteristics are mainly manifested as the texture attributes with a good connectivity and a large scale dissolution hole response and as the volume curvature attributes with the responses to faults and micro-cracks. Threshold fusion method is used to realize the attribute fusion, which can realize the spatial distribution of fracture and cave carving.
-
-
-
Seismic Attenuation in Two-Scale Porous Fractured Media — A Numerical Study
Authors V. Lisitsa, M. Novikov, Y. Bazaikin and D. KolyukhinSummaryWe present a numerical study of seismic wave propagation in fluid-filled fractured-porous media. We consider models of fractured media with two typical scales. The first one is the scale of a single fracture, the second one is the scale of the percolating fracture clusters. We generated the models with different percolation length, suggested and approach to characterize the geometrical properties of the clusters and then performed simulation of seismic wave propagation. According to the results of the simulation, we observe strong dependence of seismic attenuation on the length of the percolating clusters due to both the fracture-to-background wave-induced fluid flows and due to the scattering. Whereas fracture-to-fracture flows are not distinguishable at acoustic frequency band.
-
-
-
Seismic Tools to Mitigate the Challenges of Thin Tight Carbonate Reservoir: A Case Study
Authors S.K. Bhukta, E. Al-Shehri, S.K. Singh, P.K. Nath, A.S. Al-Ajmi, B. Khan and A. NajemSummaryTo identify a thin tight carbonate reservoir facies is one of the most challenging exploration task due to its spatial variation in terms of depostional settings, tectonics and diagenesis. The gross depositional environment plays a crucial role for insitu carbonate reservoir facies. However, the reservoir facies preservation depends on subsequent carbonate diagenesis. Though, the degree of diagenesis sometimes enhances the porosity but occasionally it ceases the porosity. However, the usage of the conventional seismic data analysis as well as state of the art tools like quantitative seismic inversion based reservoir characterization, geostatistical approach of waveform classification and the advent of the new machine learning tools like probabilistic fault likelihood, thin likelihood abetted to encompass the spatial variation, to identify the presence and hetrogeneity of the reservoir facies. Here, we have utilised these seismic tools through an integrated approach with other geological, geophysical and drilling data for futher hydrocarbon exploration and delineation.
-
-
-
Geostatistical Analysis of Seismic Data for Regional Modeling of the Broom Creek Formation, North Dakota, USA
Authors A. Livers-Douglas, M. Burton-Kelly, B. Oster and W. PeckSummaryThe Energy & Environmental Research Center is investigating the feasibility of safely and permanently storing at least 50 million tonnes of CO2 in North Dakota, United States. A regional geologic model of the injection target was created: the eolian sandstones of the Permian Broom Creek Formation. This study demonstrates how seismic data, covering a subset of the overall model region, were integrated using both multiple-point statistics (MPS) and variogram analysis. Seismic geobody interpretation enabled MPS training image development to define a lithofacies distribution, which was then used to constrain petrophysical property distributions. Alternatively, a seismic porosity inversion volume was used to calculate variograms, which were then applied in property distributions throughout the greater region. The mean and standard deviation of the porosity distributions were nearly identical in both, but porosity in the MPS case was bimodal (attributed to the facies model) versus a unimodal distribution in the variogram analysis case. These results do not indicate one approach is altogether better than the other, but geologic characteristics and control point density may make one approach more suitable. Relative agreement between the methods indicates the biggest overall benefit to a project occurs simply in having seismic data to inform model construction.
-
-
-
Modeling Study of the Unconventional Bakken Formation for Gas Injection EOR
Authors L. Jin, T. Jiang, N. Dotzenrod, S. Patil, R. Klenner, J. Sorensen and N. BosshartSummaryThe application of horizontal well and hydraulic fracturing technologies makes it profitable to produce a significant quantity of oil and gas from the extremely tight Bakken Formation. Typically, these fractured horizontal wells produce 5%-15% of original oil in place in the primary depletion stage, leaving a significant volume of oil in the reservoir. However, enhanced oil recovery (EOR) techniques have shown promise and are critical to the future development of the Bakken Formation. A series of modeling and simulation activities have been conducted in this study aimed at effectively modeling and simulating the production/EOR processes in complex fractured unconventional reservoirs. A geologic model and three simulation models with different scales were developed to investigate the flow behavior in the tight reservoirs and predict the gas injection EOR performance. Multiple interacting continua (MINC) and embedded discrete fracture model (EDFM) approaches were used to construct the fracture-matrix grid blocks in the models. The MINC method captured the early rapid transient flow but encountered numerical challenges in the gas injection simulation. The EDFM approach provided an effective way to improve simulation efficiency by reducing numerical failures in the computational process, which is critical for unconventional reservoir simulation efforts.
-
-
-
New Insights Into the Spatial Distribution of Complex Carbonate Channels Using Geostatistical Approach: A Case Study
Authors A. Al-Ali, K. Stephen and A. ShamsSummaryRecently seismic inversion method and geostatistical tools has been widely used in reservoir modelling workflows due to its excellent ability to capture the complex geobodies. In this study, the objective of this work is to characterize the spatial distribution of the Mishrif carbonate in the West Qurna Oil Field using seismic inversion results, well log data, rock physics model. Identification of the spatial distribution of channel fairway and lithology are keys for constructing Mishrif reservoir model, which have a great impact on the development of the most prolific reservoir in the field Mishrif reservoir.
-
-
-
Geostatistical Filtering of Noisy Seismic Data Using Stochastic Partial Differential Equations (SPDE)
Authors M. Pereira, C. Magneron and N. DesassisSummaryAn innovative geostatistical filtering approach is presented in this paper. It is based on Stochastic Partial Differential Equations (SPDE) and the idea is to solve kriging equations with the finite element method which requires the subdivision of a whole domain into simpler parts. This approach enables to deal with local variographic parameters while using a unique neighborhood even on large datasets. It opens the door to the operational processing of the most complex noise issues on seismic data. Post-stack and pre-stack. The methodology is described in details and two case studies are presented.
-
-
-
Joint Facies/Elastic Inference in Waveform Inversion
By J. GunningSummarySeismic AVO inversion for elastic parameters jointly with litho-fluid categories from migrated seismic data is now an established technique. Compared to conventional techniques based on adding smooth background models to impedance inversions, it has several advantages, including the ability to constrain elastic parameters to welllog-data distributions, and ability to directly predict fluids. These methods rely on the migrated amplitudes being inverted being faithfully scaled to reflectivity, and are vulnerable to the presence of non--primary seismic energy which is not modelled by conventional Born-style imaging, such as mode conversions or multiples. In any deconvolutional style inversion such wave energy adds to the noise rather than the signal. We show that it is possible to perform joint elastic/facies inversion on raw shot records, using a full-wave modelling operation in the likelihood of a hierarchical Bayesian inversion, with optimisation performed using the expectation-maximisation algorithm. Such full wave techniques can in principle model all wave modes, and should theoretically have higher S/N ratios than their AVO equivalents. Illustrative examples in high--contrast lithologies using joint acoustic/facies full wave inversion show that cleaner inversions are produced in this regime compared to convolutional methods based on traditional imaging.
-
-
-
Adaptive Ensemble-based Petrophysical Inversion for Seismically Constrained Static Model Building
Authors R. Moyen, R. Porjesz, P. Roy, R. Sablit, R. Alamer and F. AbdulazizSummarySeismic inversion produces a limited number of elastic variables (up to 3) however, the subsurface model is often described using a much larger number of variables such as porosity, clay content, fluids content, pressure etc. Through the use of a Petro-Elastic Model (PEM), it is possible to link the petrophysical properties to the elastic ones, but this forward model is not easily reversible as a given combination of elastic attributes (P -Impedance, Vp/Vs ratio...) can result from many possible combinations of petrophysical properties. Our adaptive ensemble optimization approach addresses this issue by sampling the solution space of this non-linear non-convex quadratic inverse problem through an ensemble-based model. A prior ensemble constructed from a prior model of petrophysical properties is used to sample the uncertainty of the parameters before entering the inversion process. Each petrophysical sample of the ensemble is then updated to reduce the mismatch between the elastic response given by the PEM and the elastic attributes. This update is given by a Gauss-Newton like approach where the first derivative matrix is adaptively estimated from sub-ensembles of petrophysical parameters and their corresponding forward model responses. The final ensemble provides an estimation of the uncertainty on the petrophysical parameters after the inversion process. We apply this technique on an on-shore clastic gas field in Pakistan, as part of an integrated multi-disciplinary workflow to obtain a robust, high-resolution static model integrating geology, sedimentology, petrophysics and seismic data. Stochastic modelling techniques are used to create three scenarios of varying levels of seismic influence, for a more rigorous uncertainty analysis.
-
-
-
Seismic Waveform Inversion of Elastic Properties Using an Iterative Ensemble Kalman Smoother
Authors M. Gineste and J. EidsvikSummaryProbabilistic inversion of subsurface elastic properties using seismic reflection data is considered. The methodology makes use of data partitioning as a divide-and-conquer strategy, while the conditioning to data makes use of an iterative ensemble Kalman smoother. Augmenting the ensemble Kalman framework with an variational approach is found suitable when conditioning on larger sets of seismic waveform data. The methodology is exemplified using a synthetic case for the inversion of acoustic- and shear velocity and density.
-
-
-
Self-updating Local Distributions in Geostatistical Seismic Inversion
Authors L. Azevedo, J. Narciso, R. Nunes and A. SoaresSummaryNumerical three-dimensional elastic models are a central piece of information in the geo-modelling workflow as they are often used to predict the spatial distribution of the reservoir properties such as porosity, volume of minerals and fluid saturations. Geostatistical seismic inversion methods have increasing its importance within this context due to their ability to infer high-resolution models while assessing uncertainties related to the spatial distribution of the inverted properties. Iterative geostatistical seismic inversion methods use stochastic sequential simulation and co-simulation as the perturbation technique of the model parameter space and a global optimizer based on cross-over genetic algorithms to ensure the convergence of the method. We propose a new alternative approach for model perturbation based on the concept of self-updating the local distributions of the elastic property. Iteratively, local probability distribution functions of the elastic property of interest are built and updated at seismic samples within the inversion grid based on the local mismatch between observed and synthetic seismic. This method avoids local fast convergence at early steps of the iterative procedure and allows assessment of local uncertainties at the seismic sample scale. The method was implemented in geostatistical acoustic inversion and applied to a non-stationary synthetic and a real case example with a blind well test.
-
-
-
Bayesian Rock Physics Inversion for CO2 Storage Monitoring
Authors B. Dupuy, P. Nordmann, A. Romdhane and P. EliassonSummaryWe present a two-step inversion workflow for quantitative CO2 monitoring. In the first inversion step, we carry out CSEM inversion and seismic FWI with uncertainty assessment. The uncertainty is propagated in the second step (rock physics inversion) using a Bayesian formulation. We show sensitivity tests and case study at Sleipner to highlight the importance of uncertainty estimation in the full workflow.
-