- Home
- Conferences
- Conference Proceedings
- Conferences
Petroleum Geostatistics 2019
- Conference date: September 2-6, 2019
- Location: Florence, Italy
- Published: 02 September 2019
1 - 50 of 108 results
-
-
Using Deep Directional Resistivity for Model Selection and Uncertainty Reduction in the Edvard Grieg Depth Conversion
Authors O. Kolbjornsen, P. Dahle, M.D. Bjerke, B.A. Bakke and K.R. StraithSummaryWe consider the problem of depth conversion at the Edvard Grieg field. Measurements of deep directional resistivity suggest that the top surface on Edvard Grieg is much smoother than what is indicated by the interpretation of seismic reflectors. We investigate this problem by tools of standard depth conversion, by integrating measurements from deep directional resistivity into the standard kriging equations. We propose a statistical model which is able to reveal whether we should introduce a smoothing term for the time interpretations to improve the mapping of the top surface.
-
-
-
Inference of PluriGaussian Model Parameters in SPDE Framework
Authors D. Renard, N. Desassis and X. FreulonSummaryNew methodology for infering proportions and lithotype rule used for PluriGaussian Model
-
-
-
Parametric Covariance Estimation in Ensemble-based Data Assimilation
Authors J. Skauvold and J. EidsvikSummaryEnsemble-based data assimilation methods like the ensemble Kalman filter must estimate covariances between state variables and observed variables to update ensemble members. In high-dimensional, geostatistical estimation settings where the system state consists of spatial random fields, spurious entries in estimated covariance matrices can degrade the predictive performance of posterior ensembles. We propose to avoid spurious correlations by specifying a parametric form for the state covariance, and fitting this model to the forecast ensemble. The idea is demonstrated on a partially synthetic North Sea test case involving forward stratigraphic modeling.
-
-
-
To Drill Or Not To Drill? Mature North Sea Field Case Study
Authors L. Adamson, A. Kidd, T. Frenz and M. SmithSummaryThis paper shows the importance of uncertainty quantification for the mature small pool offshore field development. Mature small pool offshore field development is associated with high risks and high costs. Often projects in such fields are economically marginal therefore, the uncertainty quantification is very important to understand the full range of outcomes before making the ultimate decision on a given project. In this paper, we suggest an integrated workflow to account for a vast number of both static and dynamic uncertainties. To include the uncertainties into the project, we create an Uncertainty Matrix to group a huge number of uncertainties into a much manageable number of variables. In this work, we also address the challenges of capturing geological realism through facies modelling and propagating it whilst performing Uncertainty Studies. We demonstrate the application of the suggested workflow on a mature North Sea Brent Field with a limited data set. The subsequent results directly influence an infill well drilling decision on this field, which currently has two production wells and one injection well to date.
-
-
-
PluriGaussian Simulations with the Stochastic Partial Differential Equation (SPDE) Approach
Authors N. Desassis, D. Renard, M. Pereira and X. FreulonSummaryIn this work, the Stochastic Partial Differential Equation approach is used to model the underlying Gaussian random fields in the PluriGaussian models. This approach allows to perform conditional simulations with computational complexity nearly independent of the size of the data sets. Furthermore, by using non-homogeneous operators, this framework allows to handle varying anisotropies and model complex geological structures. The model is presented and the proposed simulation algorithm is described. The methodology is illustrated through two synthetic data sets.
-
-
-
Geostatistics: Necessary, but Far from Sufficient
Authors A.A. Curtis, E. Eslinger and S. NookalaSummaryAn explanation is given of both where and why there are several major steps in the reservoir characterisation and modelling process in which geostatistics are of little avail and for which other technologies must be used before geostatistics can then be invoked. A workflow is presented which overcomes one of the more intractable problems in reservoir characterisation: that of moving petrophysical properties, including saturation-dependent properties, from a fine scale to a coarser scale in the absence of suitable grids. Without a rigorous solution to this problem, the subsequent use of geostatistical algorithms to distribute what may be poor quality properties data is questionable. The solution, termed the CUSP workflow, uses a unique parametrisation based on Characteristic Length Variables (CLVs) which honour the principles of hydraulic similitude. A Bayesian-based Probabilistic Multivariate Clustering Analysis is used to carry out the Classification and Propagation of petrophysical properties based on the CLVs. The CUSP workflow is scale independent and has been implemented in readily available software. An example of the application of the workflow to move petrophysical properties from the core-plug scale to the wireline log scale is presented and an example for moving from the log scale to the geocell scale is provided.
-
-
-
Challenges and Solutions of Geostatistical Inversion for Reservoir Characterization of the Supergiant Lula Field
Authors E. Kneller, L. Teixeira, B. Hak, N. Martinho Cruz, T. Oliveira, J. Marcelo Cruz and R. Santos CunhaSummaryThe creation of reservoir model properties has become an art of bringing together hard and soft data, gathering ideas of geologists and geophysicists, constraining them with measured values in- and outside wells. Through lifecycle of the oil field the information coverage is growing - new wells are being drilled, new seismic acquisitions are performed, and new geological concepts are developed. The Brazilian pre-salt fields are no exception. However, these fields experience additional challenges, where the carbonates show significant lateral and vertical variability and the salt layer limits illumination and penetration of the seismic signal. In this paper, we investigate performance of three techniques on the Lula field: simulation, which "propagates" properties between wells; deterministic inversion, which transforms seismic amplitudes into elastic properties; and geostatistical inversion, which combines simulation and seismic-driven inversion. We demonstrate that geostatistical inversion brings together the best of both techniques and helps address the challenges of characterization of pre-salt carbonates.
-
-
-
Geostatistical Interpolation of Non-stationary Seismic Data
Authors F. Turco and L. AzevedoSummaryThe problem of sparsely collected seismic data is one of the main issues in reflection seismology, because most advanced data processing techniques require a dense and regular seismic data grid. We present a geostatistical seismic data interpolation technique based on sequential stochastic simulations with local structural anisotropies. This technique, contrary to conventional existing data-driven seismic interpolation approaches based on sparsity, prediction filters, or rank-reduction, predicts the value of seismic amplitudes at non-sampled locations by exploiting the statistics of the recorded amplitudes, which are used as experimental data for the geostatistical interpolation in the original data domain. Local mean and variance are computed on-the-fly to define intervals of the global conditional distribution function, from where amplitude values are stochastically simulated. The parameters to define subsets of experimental data from which mean and variance are calculated are given by local variogram models, which in turn are obtained from a local dip and azimuth estimation in the t-x-y domain. The geostatistical seismic data interpolation technique is applied to synthetic and real 2D and 3D datasets in both postand pre-stack domains. Besides being computationally cheaper than other methods, because the interpolation is carried out directly in the original data domain, the proposed technique provides a local quantitative analysis of the reliability of the interpolated seismic samples, which can be exploited in following processing steps.
-
-
-
Expecting the Unexpected: The Influence of Elastic Parameter Variance on Bayesian Facies Inversion
Authors C. Sanchis, R. Hauge and H. KjonsbergSummaryBayesian inversion is used for the prediction of lithology and fluids from AVO seismic data. We assume a multidimensional Gaussian rock physics prior model for the elastic parameters. In this study, we look at the role of the elastic parameters variance in the prior model and how it can impact facies predictions. When the facies classes contained in the prior model have different variance, this difference influences the inversion beyond just adding uncertainty to the seismic reflections. We examine the balance between the influence of this variance and the match with expected seismic data. Our results show that although the variance influence may lead to unexpected results in synthetic scenarios, it also helps to predict the facies configuration when the seismic data follows the prior distribution and forward model.
-
-
-
Geological Heterogeneous Effect on Fluid Flow and Solute Transport during Low Salinity Water Flooding
Authors H. Al-Ibadi, K. Stephen and E. MackaySummaryThis paper examines the impact of heterogeneity on Low Salinity Water flooding (LSWF) for a realistic field scale models. We examine various scenarios of permeability variations to cover a wide range of heterogeneity possibilities. Since heterogeneity is known to induce fingering and crossflow effects at the fine scale during conventional water flooding. We analyse these effects where the LSWF process is related to a change in wettability, to determine what should be captured, in terms of solute dispersion, in typical coarse scale simulation models.
-
-
-
Assisted History Matching of 4D Seismic Data - A Comparative Study
Authors K. Fossum and R.J. LorentzenSummaryIn this work we present two unique workflows for assisted history matching of seismic and production data, and demonstrate the methods on a real field case. Both workflows use an iterative ensemble smoother for the data assimilation, but differ in data representation and localization method. Further, publicly available seismic data are inverted for acoustic impedance using two different approaches. In addition, correlated data noise is estimated for the 4D attributes using different techniques. History matching results are presented for selected production and seismic data, and estimated parameters are shown for one layer in the model. Both workflows demonstrate that ensemble based iterative smoothers can successfully assimilate large amounts of correlated data. Despite methodological differences in the workflows, both methods are able to make significant improvements to the data match. The work demonstrates promising advances towards assisted assimilation of big data-sets for real field cases.
-
-
-
Stochastic Realizations of Gaussian Random Fields: Analysis and Comparison of Modeling Methods
Authors R. Gazizov, A. Bezrukov and B. FeoktistovSummaryHere a mathematical approach which can be used for comparison of three methods for modeling Gaussian random fields is developed. Namely, the known methods of Sequential Gaussian Simulation and Spectral Modeling as well as new method based of Fourier transform and spectral modeling random fields of Fourier coefficients are considered. We show that these methods give equivalent result when specific Gaussian fields are modeled. Also we discuss advantages and limitations of these methods, their applicability in practice problems, computational complexity and ways for their effective realizations.
-
-
-
A Statistical Workflow for Mud Weight Prediction and Improved Drilling Decisions
Authors J. Paglia and J. EidsvikSummaryWe study a drilling situation based on real data, where the high-level problem concerns mud weight prediction and a decision about casing in a section of a well plan. Sensitivity analysis is done to select the most relevant input parameters for the mud weight window. In doing so, we study how the uncertainties in the inputs affects uncertainties in the mud weight window. Our approach for this is based on distance-based generalized sensitivity analysis, and we discover that the pore pressure and unconfined compressive strength are the most important input parameters. Building on this insight, a statistical model is fitted for the mud weight window and the two main input parameters, keeping in mind their geostatistical trends and dependencies. Finally we use the fitted model to the decision situation concerning casing, in a trade-off between drilling risks and costs. We conduct value of information analysis to determine the optimal data gathering scheme at a given depth, for making better decisions about casing or not. In spite of being case specific, we aim to develop a workflow that could be applied in other drilling contexts.
-
-
-
Uncertainty Evaluation to Improve Geological Understanding for More Reliable Hydrocarbon Reserves Assessment: Case Study
Authors E. Kharyba, S. Frolov, M. Blagojevic, J. Kukavic, R. Yagfarov and A. BelanozhkaSummaryThe purpose of this project was to calculate oil reserves in order to make a decision on further workover activities. The main uncertainties faced were: permeability and porosity due to the lack of core from the target interval, PVT parameters due to the absence of downhole oil sample, 2D seismic profiles instead of 3D. In addition, there were risks connected to the proximity of OWC and faults possibly intersecting the wellbore.
-
-
-
Automatic Scenarios Extraction from Depth Uncertainty Evaluation
Authors P. Correia, J. Chautru, Y. Meric, F. Geffroy, H. Binet, P. Ruffo and L. BazzanaSummaryThe structurally lowest point in a hydrocarbon trap that can retain hydrocarbons is called a Spill Point and characterizing these locations over a depth horizon is a common approach in trap analysis. However, a horizon is an uncertain object typically produced through a time to depth conversion procedure which might involve several different variables like time, velocity, and fault position. Each of those variables brings its own uncertainty. By using geostatistical simulations, we produce different realizations of the depth horizons and further process them individually to determine the probability of presence of reservoirs and spill points associated to highly probable reservoirs. This paper presents a methodology to achieve such results including our analysis algorithm for trap and spill point characterization. By using a case-study we demonstrate that only proper characterization of all relevant realizations in the uncertainty space show us the possible scenarios, and their impact on traps volume.
-
-
-
Feedback Between Gravity and Viscous Forces in Two-phase Buckley-Leverett Flow in Randomly Heterogeneous Permeability Fields
Authors P. Alikhani, A. Guadagnini and F. InzoliSummaryData on hydrocarbon reservoir attributes (e.g., permeability, porosity) are only available at a set of sparse locations, thus resulting (at best) in an incomplete knowledge of spatial heterogeneity of the system. This lack of information propagates to uncertainty in our evaluations of reservoir performance and of the resulting oil recovery. We consider a two-phase flow setting taking place in a randomly heterogeneous (correlated) permeability field to assess the feedback between viscous and gravity forces in a numerical Monte Carlo context and finally characterize oil recovery estimates under uncertainty for a water flooding scenario. Our work leads to the following major conclusions:
- Uncertainty in the spatial distribution of permeability propagates to final oil recovery in a way that depends on the feedback between gravity and viscous forces driving the system.
- Uncertainty of final oil recovered (as rendered in terms of variance) is smallest for vertical flows, consistent with the observation that the gravity effect is largest in such scenarios and is dominant in controlling the flow dynamics.
- Uncertainty of final oil recovered tends to be higher when there is competition between the effects of gravity and viscous forces, the latter being influenced by the strength of the spatial variability of permeability.
-
-
-
An Integrated Approach to Uncertainty Management on the Example of Alexander Zhagrin Field
Authors A. Sidubaev, A. Melnikova, K. Grigoryev and S. TarasovSummaryThe formation of the exploration program is a complex process that requires an integrated approach. A successful exploration program is based on a multi-level analysis of all possible geological uncertainties, probabilistic assessment of reserves and resources, analysis of tornado charts, maps of variation coefficients, and calculation of the value of information. This work considers the consistent formation and execution of exploration works on the example of Alexander Zhagrin field, which allowed to start production in two years after discovery of oil field in autonomous conditions. The research region is located in the Khanty-Mansiisk autonomous districtof the Tyumen region. The main potentially productive formation is the river-dominated delta sediments of the Cretaceous complex represented by the stratum AS-9.
-
-
-
A Bayesian Approach to Uncertainty Quantification in Geophysical Basin Modeling
Authors A. Pradhan and T. MukerjiSummaryGeophysical basin modeling helps constrain the non-uniqueness of seismic velocity inversion methods by employing basin modeling to incorporate geo-history constraints into inversion. Traditionally, basin modeling is performed in a deterministic manner and thus does not facilitate uncertainty quantification. We present a Bayesian approach for propagation of basin modeling uncertainties into velocity models. Our methodology constitutes defining prior probability distributions on uncertain basin modeling parameters and likelihood models on basin modeling calibration data. Posterior realizations of basin models are generated by sampling the prior, performing Monte-Carlo basin simulations and evaluating the corresponding likelihood values. These posterior models are finally linked to velocity models by rock physics modeling. We demonstrate the applicability of our proposed workflow using a 2D real case study from Gulf of Mexico.
-
-
-
Advantage of Stochastic Facies Distribution Modeling for History Matching of Multi-stacked Highly-heterogeneous Field of Dnieper-Donetsk Basin
Authors A. Romi, O. Burachok, M.L. Nistor, C. Spyrou, Y. Seilov, O. Djuraev, S. Matkivskyi, D. Grytsai, O. Goryacheva and R. SoymaSummaryMost of the fields in the Basin of the current study are represented by multi-stacked thin reservoirs with total thickness up to 2 thousand meters containing oil, gas-condensate and dry gas with high lateral and vertical heterogeneity. The asset in this study is a mature gas field with more than 50 years of production history, that consists from 15 gas-bearing sands of variable gas composition, that are in commingled production through the slotted liner completions while some of the sands are not yet under development and therefore, shouldn't be considered in history matching and excluded from material balance P10 reserves calculation but rather in P50 and P90 resources.
This paper shows how the application of stochastic approach for facies modeling followed by petrophysical porosity, in the presence of non-resolutive 3D seismic could help to guide the property distribution and evaluate geological uncertainties. The next very important step in the applied workflow was flow-based ranking and selection of representative case based on connected (drained) volumes that helps to achieve history match for selected base case in the presence of additional high uncertainty in contact levels and quality of production data.
-
-
-
Incorporating Discrete Microfacies Sequences to Improve Permeability Estimation in Sandstone Reservoirs
More LessSummarySeveral cases have been conducted to address the permeability modeling and estimation, but all were not accurate because of the heteroscedasticity between data. Therefore, integrating the microfacies sequences into permeability modeling became a crucial to obtain accurate prediction and then improve the overall reservoir characterization. The discrete microfacies distribution leads to distinct regression lines given each microfacies type. Therefore, the Random Forest (RF) algorithm was considered in this paper for microfacies classification and Smooth Generalized Additive Modeling (sGAM) was considered for permeability modeling as a function of well logging data and the predicted discrete microfacies distribution. The well logging records that were incorporated into the microfacies classification and permeability modeling: SP, ILD and density porosity logs. These two approaches were adopted in a well in a sandstone reservoir, located in East Texas basin. The effectiveness of using RF and sGAM approaches was investigated by their performance to handle wide ranges of data given the five microfacies types. More specifically, the Random Forest Modeling was super accurate to predict the microfacies distribution at the missing intervals for the same well and other wells. Moreover, the sGAM resulted to obtain accurate modeling and prediction of permeability in high and low permeable intervals.
-
-
-
Extraction of Petrophysical Information and Formation Heterogeneity Estimation from Core Photographs by Clustering Algorithms
By D. EgorovSummaryCore data is the most reliable and representative source of information, however it is very expensive. It is the only thing that can be used for precise description of geological situation in a target formation including its depositional environment, petrophysical properties and oil saturation distribution, reservoir heterogeneity and compartmentalization degree induced by secondary processes. Despite the fact that core retrieving operation can increase well cost in few times only a little amount of core information is utilized during field development projects as a huge part of it is represented by qualitative description which is hard to be implemented into digital quantitative geological modeling process.
-
-
-
Amplitude Supported Prospects, Analysis and Predictive Models for Reducing Risk of Geological Success
Authors I. Tishchenko and I. MallinsonSummaryDirect Hydrocarbon Indicators (DHI) are commonly used for exploration prospects. Amplitudes as an independent source of information could be used as conditional probability within Bayes Theorem to assess risk of geological success. Following research is aiming to construct predictive model for estimating probability of hydrocarbons observing DHI, P(dhi|hc). In order to build such model, we used Rose & Associates DHI Interpretation and Risk Analysis Consortium database, which contains extensive descriptions of 336 drilled prospects, with known results, across various categories: Geology, Data Quality, Amplitude Characteristics and Pitfalls. Multiple Logistic Regression was used for predicting probability P(dhi|hc). Three methods were considered within the study: two data-driven models - stepwise regression and lasso shrinkage method plus the third one, a combination of data-and expertise- driven approach - stepwise regression plus manual addition of predictors to the model. All three models with key predictors are described and give similar accuracy of prediction − 77%. Performed data analysis and calculated models reveal several insights into R&A DHI Consortium database and amplitude prospects characterisation. The best method to create such models is probably a combination of data and expertise driven approaches, while selection of most appropriate model is a question of company's strategy.
-
-
-
Application of Long Short-term Memory Algorithm for Prediction of Shale Gas Production in Alberta
More LessSummaryShale resources are developed with high-density drilling and differ to predict future production rate with conventional methods such as decline curve analysis (DCA). We developed a prediction model for shale production using machine learning (ML) algorithm. Long short-term memory algorithm, which is suitable for training time-series data, is applied to more than 300 shale gas wells at Duvernay formation in Alberta. To increase performance of the prediction model, an additional feature, shut-in (SI) period, is introduced for input layer. SI information is easily extracted from production history and it is highly correlated with production rate. The trained model can give reliable and stable prediction of shale gas production rates. Especially, the two-feature model (production and SI period) shows better performance than the one-feature model (production only). Therefore, a domain knowledge for petroleum industry has an important role to build reliable ML model. The developed model can dramatically reduce analysis cost comparing with DCA, which requires expert's judgement for each shale well. Also, this concept can be extended to other conventional and unconventional production wells.
-
-
-
Machine Learning-based Approach for Automated Identification of Produced Water Types from Conventional and Unconventional Reservoirs
Authors P. Birkle, M. Zouch, M. Alzaqebah and M. AlwohaibiSummaryProduced water represents the most commonly recovered fluids in oil and gas operations. They are composed of natural fluids from deep groundwater systems (formation water) or artificial fluids from operations, or a mixture between both fluid components.
-
-
-
Application of a Serial Denoising Autoencoder for Geological Plausibility of a Channelized Reservoir in History Matching
More LessSummaryDenoising autoencoder (DAE) is utilized to preserve and improve geological reality and plausibility in a channelized reservoir model during history matching by ensemble smoother with multiple data assimilation (ES-MDA). As one of history matching methods, ES-MDA calibrates reservoir properties such as rock facies corresponding to production history. While ES-MDA modifies reservoir parameters, it recognizes them only as figures not honoring to geological features. Thus, conservation of geological characteristics during calibration of reservoir parameters is challenging in ES-MDA. DAE is trained to restore lost connectivity and pattern of an original geological concept and it is applied to posterior reservoir models after an assimilation by ES-MDA. ES-MDA combined with DAE shows not only geologically enhanced channel models but also well-matched production prediction.
-
-
-
Automatic Recognition of Lithological Units in Gas-bearing Shale Complex with Neural Networks (the Baltic Basin, Poland)
By K. BobekSummaryThe study was performed to check the possibility of automatic recognition of lithological units of shale complex and testing their heterogeneity in different parts of the Baltic Basin. For this purpose, two separate neural network models have been trained based on geophysical logs datasets: one for perspective units recognition, and second for tuffite interbeds detection. Obtained results have compared with units distinguished by sedimentologists in continuous core profiles. The performance of trained networks has been checked for two boreholes B-7 and B-1 not included in the training process. In case of the borehole B-7, the units predicted by the model match well to those distinguished manually in borehole core. In contrary, the score of model performance for the B-1 borehole was unsatisfactory. Significant differences in the lithological profile of this borehole were also recognized by sedimentologists. Due to the location of borehole B-1 in the marginal part of the basin, most of the properties of the studied units differ relevantly from recognized in other investigated boreholes. The performance of the network trained for tuffite interbeds detection is considered satisfactory. The results imply that the trained models might be applied in the boreholes where the core profiles are modest or non-existent.
-
-
-
Geomodeling Using Generative Adversarial Networks and a Database of Satellite Imagery of Modern River Deltas
Authors E. Nesvold and T. MukerjiSummarySeveral studies on deep generative models for use in geomodeling show encouraging results with binary training data. An important question is what type of training data to use, since realistic 3D geology with natural variability is difficult to create. The advent of multiple types of remote sensing data of subaerial and subaqueous sedimentary patterns provides new possibilities in this context. Here, we train a Wasserstein GAN using 20,000 multispectral satellite images of subsections of 40 modern river deltas. The generated output has three facies and a background facies, and all quantitative evaluation methods of the unconditional output show a close overlap between the model and training data distributions. Standard MCMC sampling conditional on soft and hard data works well as long as the likelihood model is balanced against the prior model. Transfer learning, i.e. fine-training a small subset of the network parameters on smaller dataset of interest, such as highly non-stationary images of river deltas with similar characteristics, also shows promising results.
-
-
-
Ensemble-based Kernel Learning to Handle Rock-physics-model Imperfection in Seismic History Matching: A Real Field Case Study
Authors X. Luo, R. Lorentzen and T. BhaktaSummaryModel imperfection is ubiquitous in geophysical data assimilation problems. A common approach to accounting for model errors is to treat them as random variables following presumed distributions. While such a treatment renders certain algorithmic convenience, its underpinning assumptions may often be invalid in practice. In this study, we adopt an alternative approach, and treat the characterization of model errors as a functional approximation problem, which can be solved using a generic machine learning method, such as kernel-based learning adopted here. To enable a seamless integration of kernel-based learning into ensemble data assimilation, we also develop an ensemble-based kernel learning approach. We show that an existing iterative ensemble smoother can be naturally employed as the learning algorithm, which thus inherits all the practical advantages of ensemble data assimilation algorithms, such as derivative-free, fast implementation, and allowing uncertainty quantification and applications to large scale problems. To demonstrate the efficacy of ensemble-based kernel learning, we apply it to handle model errors in a rock physics model used in history matching real 4D seismic data from the full Norne field. Our experiment results indicate that incorporating kernel-based model error correction into 4D seismic history matching helps improve the qualities of estimated reservoir models, and leads to better forecasts of production data.
-
-
-
3D Geological Image Synthesis from 2D Examples Using Generative Adversarial Networks
Authors G. Coiffier and P. RenardSummaryRecently, Generative Adversarial Networks (GAN) have been proposed as a potential alternative to Multipoint Statistics (MPS) to generate stochastic fields from a large set of training images. A difficulty for all the training image based techniques (including GAN and MPS) is to generate 3D fields when only 2D training data sets are available. In this paper, we introduce a novel approach called Dimension Augmenter GAN (DiAGAN) enabling GANs to generate 3D fields from 2D examples. The method is simple to implement and is based on the introduction of a random cut sampling step between the generator and the discriminator of a standard GAN. Numerical experiments show that the proposed approach provides an efficient solution to this long lasting problem.
-
-
-
Deep Stochastic Inversion
Authors L. Mosser, O. Dubrule and M.J. BluntSummaryNumerous geophysical tasks require the solution of ill-posed inverse problems where we seek to find a distribution of earth models that match observed data such as reflected acoustic waveforms or produced hydrocarbon volumes. We present a framework to create stochastic samples of posterior property distributions for ill-posed inverse problems using a gradient-based approach. The spatial distribution of petrophysical properties is created by a deep generative model and controlled by a set of latent variables. A generative adversarial network (GAN) is used to represent a prior distribution of geological models based on a training set of object-based models. We minimize the mismatch between observed ground-truth data and numerical forward-models of the generator output by first computing gradients of the objective function with respect to grid-block properties and using neural network backpropagation to obtain gradients with respect to the latent variables. Synthetic test cases of acoustic waveform inversion and reservoir history matching are presented. In seismic inversion, we use a Metropolis adjusted Langevin algorithm (MALA) to obtain posterior samples. For both synthetic cases, we show that deep generative models such as GANs can be combined in an end-to-end framework to obtain stochastic solutions to geophysical inverse problems.
-
-
-
Integrated Geo-modelling and Ensemble History Matching of Highly Faulted Turbiditic Reservoir Model
Authors V. Zaccardi, A. Abadpour, N. Haller, P. Berthet, D. Rappin and J. Grange-PraderasSummaryData assimilation methods and multi-realisation geomodelling workflow have been applied to history match a heavily faulted turbidic reservoir. With this approach it was possible to take into account a large number of uncertainties related to seismic data interpretation, geological modelling and dynamic data.
-
-
-
Ensemble History Matching Enhanced with Data Analytics - A Brown Field Study
Authors E. Tolstukhin, E. Barrela, A. Khrulenko, J. Halotel and V. DemyanovSummaryThis paper presents a methodology used in a subsurface uncertainty study for redevelopment of an oil field in the North Sea. A fractured chalk reservoir was depleted for more than 30 years with limited water injection. The uncertainty study aims to find an ensemble of geologically consistent scenarios that would honor production history. The scenarios then serve as input for the redevelopment concept selection, well placement and economic evaluation. The challenge in this study was that the field has long production history that must be respected. In addition, the uncertainty that may not be resolved by HM must be preserved in the scenarios in order to estimate all the risks and capture all the potential associated with the remaining oil pockets and future well targets. For the brown field, it is difficult to analyze all the information and utilize its full potential. In this work we use data analytics can improve efficiency of ensemble history matching by analyzing links between the static and dynamic model ensemble update: screening of the initial ensemble, model localization based on spatial analysis dynamic observations to the parameter update and identification of conflicts between groups of production observations that prevent balanced model update.
-
-
-
Updating MPS Facies Realizations Using the Ensemble Smoother with Multiple Data Assimilation
Authors A. Thenon, A. Abadpour and T. ChugunovaSummaryThis paper is presenting the results of a study focused on the update of facies models generated by multipoint statistics (MPS) within an ensemble history matching workflow. The tested parameterization consists in updating the uniform random numbers used by MPS to generate the facies realizations with an ensemble method as proposed in the work of Hu et al. (2012). The novelty of this study lies in use of the ensemble smoother with multiple data assimilation (ES-MDA) to update the random number realizations. Tests on synthetics cases show that increasing the iteration number within ES-MDA can significantly improve the quality of the history-match achieved by this approach. The workflow is briefly detailed in the first part of the paper. Then, we present and discuss the history-matching results for two synthetic cases: an inverted five-spot 2D model with channels and a more complex 3D model with dunes.
-
-
-
Improving Local History Match Using Machine Learning Generated Regions from Production Response and Geological Parameter Correlations
Authors T. Buckle, R. Hutton, V. Demyanov, D. Arnold, A. Antropov, E. Kharyba, M. Pilipenko and L. StulovSummaryWe present a data driven workflow to improve local history match quality by identifying model regions from correlation between production response and geological modelling parameters for use in an assisted history matching framework. This paper outlines the implementation and results from a large mature field case study. Regions are identified by calculating the partial correlation between individual well production misfits and uncertain geological modelling parameters across 500 models. Wells are then categorised into three groups based on their correlations: positive, negative and insignificant. A probabilistic neural network (PNN) is trained on the location of each well and its group. A map of regions can then be calculated using the PNN. The parameters used to define the region map are then varied separately in each region in an assisted history matching loop. In the full field case study, an 8.8% improvement in oil rate misfit within the positively correlated well group was achieved by regional modification of the net-to-gross multiplier, with no detrimental effect on the other groups match quality. This case study demonstrates the effective identification and utilisation of geologically and dynamically inferred regions which improve the local history match
-
-
-
Fluid Flow Consistent Geostatistical History Matching of an Onshore Reservoir
Authors E. Barrela, L. Azevedo, A. Soares and L. GuerreiroSummaryThis paper shows the application to a real field case of an iterative geostatistical history matching technique, integrating geological and engineering consistency. Current trends reflect a growing interest on developing workflows that simultaneously integrate petrophysical modeling with dynamic calibration of reservoir models to historical production data. Contrary to manual history matching techniques, where model perturbation often disregards geological or physical realism leading to poor production forecast, this example introduces geological consistency through geostatistical simulation and physical realism by using streamline regionalization, while holding the predictive capability of resulting petrophysical models. This is achieved by iteratively updating the reservoir static properties using stochastic sequential simulation and co-simulation, constrained to production data, while using streamline information for electing preponderant flow production regions of the model, focusing property perturbation. In order to capture the complex subsurface heterogeneities of the reservoir, petrophysical property realizations are obtained using the direct sequential simulation and co-simulation with multi-local distribution functions. The location and proportion of reservoir facies is also automatically updated throughout the iterative procedure, using Bayesian Classification. The technique was successfully applied to a real case study, located in North-East onshore Brazil, resulting in multiple history matched models that better reproduce historic data.
-
-
-
Revising the Method of Ensemble Randomized Maximum Likelihood
Authors P.N. Raanes, G. Evensen and A.S. StordalSummaryA popular (iterative ensemble smoother) method of history matching is simplified. An exact relationship between ensemble linearizations (linear regression) and adjoints (analytic derivatives) is established.
-
-
-
Theoretical and Practical Aspects of Stein Variational Gradient Descent with Applications to Data Assimilation
Authors A. Storksen Stordal, R.J. Moraes, P. Nima Raanes and G. EvensenSummaryWe have introduced and applied the Stein variational gradient descent (SVGD) algorithm to the reservoir history matching problem. The method has been extended with a Monte Carlo approximation of the gradient of the measurement function using the reproducible property of the reproducing kernel Hilbert space in order to avoid adjoint implementation. Furthermore we have implemented SVGD with p-kernels in order to extend the applicability to higher dimension than the standard Gaussian kernels used in the literature. The gradient approximation was validated both theoretically and on a toy model. Implementation on larger problems will be addressed in the future. We also showed using a reservoir model that the SVGD with p-kernels provided better estimates of the uncertainty than the standard Gaussian kernels. Further comparison with ensemble-based methods and other gradient-based methods is left for future research.
-
-
-
Well Logs Inversion Into Lithology Classes: Comparing Bayesian Inversion and Machine Learning
More LessSummaryLithology classification is a crucial challenge in geological research. Lithologies at thelocations without cores need to be predicted by using indirectly geophysics measurementssuch as well logs and seismic data. In this study, we use the spatial dependency of sedimentsand well logs data for inversion into lithologies by a kernel-based hidden Markov model(HMM) and a gated recurrent unit (GRU) model. We estimate model parameters from twotraining wells and predict lithologies from well log observations for a blind test well. Thepredictions are also compared to results from a deep neural network (DNN) model, whichassumes spatial independence. Results indicate that the lithology predictions supplied by theHMM and GRU models are more reliable than the ones from DNN in term of classificationaccuracy and make more senses in geological interpretation. Moreover, the HMM providesquantifications of the classification uncertainty.
-
-
-
Well Log Data Standardization, Imputation and Anomaly Detection Using Hidden Markov Models
SummaryThe main goal of this work is to develop a systematic approach to work with raw well log data. Toaccomplish this goal, we propose to fit a simple unsupervised generative model to the input data and au-tomate the preprocessing step using the generative model. This approach allows to detect the anomaliesin the data as the regions that the model struggles to explain (i.e., samples with extremely low like-lihood), infer approximations to the missing features using the Bayes rule and incorporate additional expert knowledge in the design of the model.
-
-
-
Seismic Estimation of Sub-resolution Reservoir Properties with Bayesian Evidential Learning: Application to an Offshore Delta Case
Authors A. Pradhan and T. MukerjiSummaryWe present a Bayesian framework that facilitates estimation of and uncertainty quantification of sub-resolution reservoir properties from seismic data. Our workflow exploits learning the direct relation between seismic data and reservoir properties by the evidential learning approach to efficiently estimate sub-resolution properties. The major focus of this paper is on a real case study of seismic characterization of reservoir layers significantly below the seismic resolution. We employ deep neural networks as summary statistics within Approximate Bayesian Computation to estimate posterior uncertainty of reservoir net-to-gross from 3D pre-stack seismic data without an explicit inversion.
-
-
-
Monte Carlo-based Framework for Quantification and Updating of Geological Model Uncertainty with Borehole Data
More LessSummaryUncertainty quantification is of importance for reservoir appraisals. In this work, we provide an automated method for uncertainty quantification of geological model using well borehole data for the reservoir appraisal. In our method, when new wells are drilled, multiple components of the geological model are updated jointly and automatically by means of a sequential decomposition following geological rules. During updating, we extend the direct forecasting method to perform such joint model uncertainty reduction. Our approach also enables updating geological model uncertainty without conventional model rebuilding, which significantly reduces the time-consumption. The application to a gas reservoir shows that, this proposed framework can efficiently update the geological model and reduce the prediction uncertainty of the gas storage volume jointly with all model variables.
-
-
-
Boolean Spectral Analysis in Categorical Reservoir Modelling
Authors N. Ismagilov, V. Borovitskiy, M. Lifshits and M. PlatonovaSummaryThe work introduces a new method for simulation of facies distribution for two categories based on Fourier analysis of Boolean functions. According this method, two categories of facies distributed along vertical wells are encoded as Boolean functions taking two values. The subsequent simulation process is divided into three consecutive steps. First, Boolean functions of well data are decomposed into a binary version of Fourier series. Then, decomposition coefficients are simulated over 2-dimensional area as stationary random fields. Finally, synthetic data in the interwell space is reconstructed as Fourier sum from simulated coefficients. The new method was implemented in an experimental software and tested on a case of real oil field.
-
-
-
Stochastic Seismic Inversion Based on a Fuzzy Model
Authors E. Kovalevskiy and M. VolkovaSummaryThe cause of the low efficiency of geostatistical seismic inversion based on sequential Gaussian simulation (SGS) is explained as follows. In spite of the non-stationary type of initial borehole impedance sections, SGS generates stationary realizations of the same vertical sections. This results in the stationary cubes of predicted impedance values in which all deterministic features are erased. This article proposes using impedance section realizations obtained from a fuzzy model rather than from SGS. The first accurately represent the local statistics of borehole impedance sections, which allows the resulting impedance cubes to clearly show the deterministic features of a geological object. The method is illustrated with an example of the inversion for a real seismic cube.
-
-
-
Integrated Res Characterization Tool to Construct High Resolution Geological Model in MR FM of DF Field
Authors A. AlShamali, N. Verma, R. Quttainah, A. Tiwary, G. Alawi and M. Al RaisiSummaryThis paper establishes the approach of finding a relationship between reservoir rock typing (RRT)-derived from core and well log data to generate full field continuous RRT models that can be used to predict RRT in undrilled locations and predicting blind wells In the present study, a total of 16 wells were analyzed uses a unique technology that integrates all data using artificial intelligence where a neural network is trained and tested using existing data. Multiple realizations are created, analyzed and validated through blind well testing.The RRT prediction was carried out using 'Ipsom' module in TECHLOG and the module is based on supervised neural network technology.This module using the core RRT as desired log and well logs as an input curves for the neural network training, RRT in the cored intervals were used as training set to obtain the neural network engine which will be used to predict the un-cored intervals and wells.Once the RRT were defined, the attempt was made to generate permeability and saturation height function for each of these RRTs. For permeability computation, FZI (Flow Zone Indicator) and Simple Geometric Regression Type Equations were tested. Having compared both methods of regression, it was observed that there was a minimal difference between both methods using FZI or using a simple regression type in permeability prediction. However, FZI gave a slightly better result when compared with the simple regression method within the range of data in the field.
-
-
-
Raising the Bar: Electrofacies as a Framework for Improving the Practice of Geomodeling
By D. GarnerSummaryA key impact on reservoir studies is a rigorous strategy around facies for modeling. The industry practices across small to large companies are highly variable regarding generating facies logs. Geomodeling workflows and geostatistical algorithms treat the facies log variable as hard conditioning information. Facies logs in practice have errors and carry petrophysical inconsistencies, real quality issues, which are not head-on addressed by the time they are used in a geomodeling workflow. Establishing electrofacies modeling best practices in the petroleum industry can help improve the preparation of facies logs for modeling and improve the fidelity of many geomodeling processes. This material presents basic theory, practical considerations, and example results from up to four different fields, depending on poster size. Further discussion is intended to further illustrate benefits of the use of electrofacies and help mature the understanding of the workflows which are not widely used.
-
-
-
Statistical Characterisation of Fluvial Sand Bodies: Implications for Complex Reservoir Models
Authors M. Franzel, S. Jones, I.H. Jermyn, M. Allen and K. McCaffreySummaryThe three-dimensional geometry of fluvial channel sand bodies has received considerably less attention than their internal sedimentology, despite the importance of sandstone body geometry for subsurface reservoir modelling. The aspect ratio (width/thickness, W:T) of fluvial channels is widely used to characterize their geometry. However, this does not provide a full characterization of fluvial sand body shape, since one W:T ratio can correspond to many different channel geometries. The resultant over- or underestimation of the cross-sectional area of a sand body can have significant implications for reservoir models and hydrocarbon volume predictions. There is thus a clear need for the generation of versatile, quantitative, and statistically robust models for sand body shape. The main aim of this research is to develop a new statistically-based approach that will provide quantitative data, derived from outcrop analogues, to fully constrain stochastic fluvial reservoir models. Here, we describe the construction of a new shape database and conduct a preliminary qualitative analysis in order to understand measurement and other uncertainties, and to explore the catalogue of shape configurations. A future quantitative analysis will develop a predictive model to enable forecasting of reservoir channel sand body geometries and shapes that can be built into existing reservoir models.
-
-
-
Integrated Reservoir Characterization and Multiscale Heterogeneity Modeling of Stacked Meander-belt Deposits, Lower Cretaceous McMurray Formation, Alberta
Authors S. Nejadi, J.A. Curkan, P.R. Durkin, S.M. Hubbard and I.D. GatesSummaryThe McMurray Formation is composed of large-scale fluvial meander-belt deposits that are highly heterogeneous. Repeated cut and fill events within the formation have led to a complex amalgam of stacked stratigraphic architectural elements. Lithological properties vary both laterally and vertically over short distances in the McMurray Formation. The youngest deposits of the reservoir studied at the Surmont site are well imaged using 3D seismic data; calibration with well-data enables construction of a particularly detailed reservoir model. The underlying deposits are characterized using wire line logs, core data, and stratigraphic dip analysis. For modeling purposes, internal stratigraphic architecture of both reservoir levels is mapped and distinct fluvial meander-belt architectural elements, including point bars, counter point bars, side bars and abandoned channel fills, are characterized as distinct zones. Each zone is characterized by distinct morphology, facies associations, petrophysical properties, and thus, reservoir potential. Deterministic geobody interpretations are implemented to guide geostatistical simulations; spatial distribution of facies are constrained to the mapped architectural elements. Constraining parameter estimations to deterministically interpret meander-belt architectural elements improves the predictive capability of the reservoir model. This modeling workflow preserves geological realism in models, allows spatial uncertainty to be captured adequately, and improves the ability to optimize development.
-
-
-
Statistical Characteristics of a Fractured Model from Seismic Data via Topological Analysis of Diffraction Images
Authors M. Protasov, T. Kchachkova, D. Kolukhin and Y. BazaikinSummaryA workflow for recovering fracture network characteristics from seismic data is considered. First, the presented discrete fracture modeling technique properly describes fracture models on the seismic scale. The key procedure of the workflow is 3D diffraction imaging based on the spectral decomposition of different combination s of selective images. Selective images are obtained by the prestack asymmetric migration procedure, while spectral decomposition occurs in the Fourier domain with respect to the spatial dip and the azimuth angles. At the final stage, we propose a topological analysis based on the construction of a merge tree from the obtained diffraction images. The results of the topological algorithm are modeling parameters for the discrete fractures. To analyze the effectiveness of the proposed workflow, a statistical comparison of the recovered parameters and true model parameters are provided. We use the Kolmogorov -Smirnov test for a statistical analysis of the fracture lengths, while the behavior of the Morisita index shows the statistical distribution of the modeled fracture corridors. Numerical examples with synthetic realistic models demonstrate a detailed, reliable reconstruction of the statistical characteristics of the fracture corridors.
-
-
-
The Effect of Fracture Clustering on Confined Fractured Zones: Numerical Modeling and Analyses
Authors A. Alali, K. Marfurt and N. NakataSummarySeismic reflection amplitude variation with offset and azimuth (AVOaz) provides a traditional technique to detect fracture in the subsurface and deduce their properties. AVOaz relies on the Effective Medium Theory (EMT) which treats the fractured formation as anisotropic medium. The underlying assumption of EMT is that the fractures are uniformly disturbed and sufficiently close such that only specular reflections from the boundaries of the fractures are observed. In contrast, for randomly spaced fractures clustering takes place and individual scattering occurs, and the assumptions of effective medium theory are violated.
-
-
-
Using Seismic Images for Scaling of Statistical Model of Discrete Fracture Networks
Authors D. Kolyukhin and M. ProtasovSummaryThe presented paper addresses the modeling and seismic imaging of fractured reservoirs. A three-dimensional statistical model of a discrete fracture network is developed. A flexible and efficient method to generate the random realizations of the statistical model for an arbitrary computational grid is suggested. The problem of scaling the developed fracture model using the analysis of seismic images for different grid steps is studied. Particular attention is paid to the models with a multifractal distribution of fractures.
-