- Home
- Conferences
- Conference Proceedings
- Conferences
82nd EAGE Annual Conference & Exhibition
- Conference date: October 18-21, 2021
- Location: Amsterdam, The Netherlands
- Published: 18 October 2021
61 - 80 of 1137 results
-
-
The Performance of Viscoelastic Surfactant-Polymer Flood in Heavy-Oil Carbonate Reservoir — Simulation Study
Authors A. Zaitoun, M. Al-Foudari, K. Zeidani, S. Al-Otaibi, A. Al-Ghadhouri, G. Omonte Rossi, J. Bouillot and A. ZaitounSummaryThis paper describes the simulations performed to evaluate different scenarios of water flood, polymer flood and Viscoelastic Surfactant (VES) combined with polymer blend in a Middle Eastern carbonate reservoir. Compared to classical Alkaline-Surfactant-Polymer (ASP) EOR technology, VES-Polymer does not require heavy water processing and is thus more robust and easier to deploy on the field. The simulation study used coreflood data set obtained by a laboratory study presented in another paper and aimed at optimizing a field pilot.
The simulation was conducted with a pattern of three parallel horizontal wells; one central injector and two lateral producers. A well length of 2000 meters and spacing of 100 meters was found to be the best configuration for the pilot. For waterflood, the unfavorable mobility ratio induced early water channeling. Due to more favorable mobility ratio, polymer flood shows better performances in terms of incremental oil production and VES-Polymer flood further increases the oil production compared to polymer flood due to combination of IFT reduction and increase in sweep efficiency.
Both polymer flood and VES-Polymer flood can thus be considered as valuable EOR options in this type of reservoir conditions which have not been considered so far
-
-
-
Structural and Depositional Features Controlling Permeability on Carbonate Platforms
Authors R. Loza Espejel and T.M. AlvesSummaryNaturally fractured reservoirs present a challenge when determining the permeability associated with different types and sizes of fractures. Permeability in these reservoirs depends on the heterogeneity and connectivity of open fractures; although depositional and diagenetic features also play an important role. In this study, a multi-scale analysis of the Cariatiz Fringing Reef Unit in SE Spain is completed based on outcrop and LiDAR data. Seven different features were found to influence the permeability of the Cariatiz Fringing Reef Unit. Structural features comprise: (i) joints, (ii) veins, (iii) vertical fractures, (iv) fracture swarms, and (v) karsts. Two types of depositional features were also recognised at outcrop:(vi) vertical Porites and (vii) pseudo-bedding surfaces. All contribute to increased permeability, apart from calcite-filled veins that create barriers to fluid flow. The results of this study highlight the complexity of carbonate systems and the need to collect data at different scales of analysis to decrease uncertainties in reservoir models. The approach in this work is valid in hydrocarbon exploration and production, geothermal reservoir characterisation, environmental studies and carbon sequestration projects.
-
-
-
Deep-Marine Hyperpycnal Sandstones and Implications for Exceptional Reservoir Quality Preservation
Authors J. Cater, T. Gould and B. UrsinusSummaryDeep marine hyperpycnal sandstones form prolific hydrocarbon reservoirs but remain poorly understood despite over 40 years of research. Predicting their geometry, composition and reservoir quality requires a thorough knowledge of the processes that formed them and the effects of diagenesis in the presence of brackish depositional pore waters. Dissolution of unstable grains (e.g. feldspars and volcanic material) and replacement by kaolinite and chlorite/smectite occurs more readily in the presence of brackish, acidic pore fluids. This is enhanced locally as compaction drives fluids through the aquifer. Pore lining chlorite cements can help to prevent chemical compaction of quartz grains and impede later quartz overgrowths, helping to preserve reservoir quality at depth. Commonly in hyperpycnal deposits, remnant pore fluids are of low salinity, resulting in anomalous low salinity DST results (e.g. Agat, NOCS). The salinity of the pore fluids soon after deposition can be quantified by measuring the isotopic composition of early carbonate cements, which may form strata bound or nodular baffles to flow within the aquifer. The influence and mobility of low salinity pore fluids during the early diagenesis of deep marine hyperpycnal deposits is a key subject for future research.
-
-
-
Lithology Identification Based on Hidden Markov Model and Random Forest
More LessSummaryBy combining the hidden Markov model (HMM) and random forest (RF), a new approach is proposed for lithology identification. To extract more useful information from elastic parameters, the HMM is used to provide a new hidden feature. The hidden feature reveals the inner relationship of elastic parameters and this is important for machine learning. With the new hidden feature and elastic parameters, RF is adopted for lithology prediction. To guarantee the quality of the hidden feature, it is updated in a loop iteration. Both synthetic data and field data tests demonstrate that the proposed approach can improve prediction results.
-
-
-
Use of Elastic Forward Modeling to Remove Complex Coherent Noises
Authors J. Tang, C. Peng, M. O’Briain and C. ShihSummaryIn the deep-water Campeche Bay area of the southern Gulf of Mexico, there are many complex shallow salt bodies and carbonate rafts that generate a significant amount of coherent noise energies, collectively for all non-primary reflection energies, as a result of the high impedance contrast between these bodies and the surrounding sediments. These noise energies include surface-related salt-diffracted multiples, interbed/internal multiples, bounces between salt bodies, and other types of prismatic waves as well as converted shear waves. These coherent noises cause difficulties in interpreting base of salt and subsalt seismic events. Identifying and removing them is crucial for optimal seismic imaging of subsalt targets.
We propose a method to model these noises using a geological imaging model and elastic finite-difference forward modeling. The method requires that the shallow part of the geological imaging model be accurate. We first compute elastic synthetic data using the model. Then, we migrate the synthetic data to generate a noise model in the image domain and use this noise model to pattern-match with another image volume migrated using field data. In this way, we can identify noises in the field data and remove them adaptively to obtain a cleaner image of the recorded reflectivity.
-
-
-
Performance Evaluation of Machine Learning Algorithms in Predicting Dew Point Pressure of Gas Condensate Reservoirs
Authors P. Ikpeka, J. Ugwu, P. Russell and G. PillaiSummaryAccurate knowledge of the dew point pressure for a gas condensate reservoir is necessary for the design of a field development plan and timing for optimization of mitigation operations for resources management. This study explores the use of machine learning models in predicting the dew point pressure of gas condensate reservoirs. 535 experimental dew point pressure data-points with max temperature and pressure of 304F and 10500psi were used for this analysis. First, multiple linear regression (MLR) was used as a benchmark for comparing the performance of the machine learning models. Neural Networks (NN) [optimized for the number of neurons and hidden layers], Support Vector Machine (SVM) [using radial basis function kernel] and Decision Tree [Gradient boost Method (GBM) and XG Boost (XGB)] algorithms were then used in predicting the dew point pressure using gas composition, specific gravity, the molecular weight of the heavier component and compressibility factor as input parameters. The performances of these algorithms were analyzed using root mean square error (RMSE), absolute average relative deviation percentage (AARD %) and coefficient of determination (R2). This work concludes that for large data sets neural network is preferred but for smaller data sizes, SVM shows better performance
-
-
-
Enhancing Massive Land 3D Seismic Data Using Nonlinear Beamforming: Performance, Quality and Practical Trade-Offs
Authors I. Silvestrov, A. Bakulin, D. Nekluydov, K. Gadylshin and M. ProtasovSummaryModern land seismic data are typically acquired using high spatial trace density but single sensors or small source and receiver arrays. These datasets are challenging to process due to their massive size and rather low signal-to-noise ratio caused by scattered near surface noise. Prestack data enhancement becomes a critical step in processing flow. Nonlinear beamforming was proven very powerful for 3D land data. It requires computationally costly estimations of local coherency on dense spatial/temporal grids in 3D prestack data cubes and poses inevitable trade-off between performance of the algorithm and quality of the obtained results. In this work, we study different optimization schemes and discuss practical details required for applications of the algorithm to modern 3D land datasets with hundreds of terabytes of data.
-
-
-
Improved Ooil Recovery by Low Salinity Water Injection Simulation
By J. TrivediSummaryPrimary and secondary oil recovery techniques together can produce less than half of the original oil in place due to some restricting phenomena such as rock heterogeneity, capillary, and mobility ratio problems during these first two stages in oil reservoirs.( Kamranfar and Jamialahmadi 2014 ; Lei et al. 2016 ). The overall objective of implementation of any chemical EOR method including alkaline, surfactant, and polymer flooding is to decrease the residual oil saturation left within reservoir rock porous media after primary and secondary productions. Recently, LSWI (low salinity water injection) is one of the emerging IOR techniques for wettability alteration in both sandstone and carbonate reservoirs.
-
-
-
Deep Learning for Anisotropy Parameters Estimation in Oil/Gas Fractured Reservoirs
Authors G. Sabinin, T. Chichinina and V. TulchinskySummaryWe study the applicability of Deep Learning in solving the problem of estimating the fractured medium parameters, represented as anisotropy parameters of a transversely isotropic model (HTI), using synthetic seismic data. Normal and tangential weaknesses of fractures ∆_N and ∆_T, the Thomsen anisotropy parameters ε, δ, γ, the crack density and the aspect ratio (crack opening) are considered. We develop a neural network model for solving this problem. Trained on synthetic seismograms, it provides quite accurate results. The effectiveness of Deep Learning for the inverse problem is demonstrated. The prospects for the development of this method for more complex rock-physics models are outlined.
-
-
-
Prestack Data Enhancement with Phase Substitution and Phase Corrections Guided by Local Multidimensional Stacking
Authors A. Bakulin, D. Neklyudov and I. SilvestrovSummaryWe revisit enhancement with local stacking in the context of seismic data corrupted by near surface scattering. We discover that phase spectra derived from local stacking contains critical information that could be used as direct estimate of signal phase (phase substitution method) or as a guide (phase corrections method) to correct frequency-dependent distortions obstructing prestack data. Combining corrected phase with original amplitude spectrum, we arrive at much better estimate of enhanced data compared to conventional multi-dimensional local stacking. Specifically, we eliminate loss of higher frequencies and preserve original amplitudes, while making originally invisible reflections to become discernable and coherent for further processing. While we only present example of two possible methods, this discovery paves the way to plethora of new approaches finally enabling removing corrupting effects of complex scattering near surface beyond conventional surface-consistent processing.
-
-
-
(When) Do Earthquakes Respect Traffic Lights?
Authors S. Baisch, P. Carstens, R. Vörös and K. WittmannSummaryTraffic light systems (TLSs) for limiting the strength of induced seismicity are used in different energy technologies. We use physics-based numerical models for investigating under which circumstances TLSs may not provide a robust mitigation measure. For seismicity induced by fluid injection, TLS efficiency can be limited by trailing effects caused by post-injection pressure diffusion and stress concentrations at the periphery of previous seismic activity. Seismicity caused by gas production exhibits a ‘characteristic earthquake’ pattern where earthquakes with similar (maximum) magnitude occur in the course of reservoir depletion. The characteristic earthquakes reflect repeated slip of the same reservoir fault patches. The maximum earthquake magnitude of a sequence can occur without precursors. Although trailing effects do not occur in an idealized reservoir with infinite conductivity, the lack of precursory seismicity limits the robustness of a TLS.
-
-
-
Moving Toward Direct DNN-Based Enhancement of 3D Pre-Stack Seismic Data
Authors K. Gadylshin, A. Bakulin and I. SilvestrovSummaryPre-stack data enhancement with multidimensional stacking is indispensable part of modern data processing that very compute-intensive since multiple wavefront attributes need to be estimated on dense spatial/temporal grid. At the core of this demand are conventional local or global optimization techniques. We propose two alternative approaches based of artificial intelligence that can greatly reduce computational effort of estimation stage. First approach performs traditional computations on sparser grid and inpaints to dense grid using deep neural network (DNN) with partial convolution layers. Second approach is direct DNN-based attributes estimation from the pre-stack seismic data itself. Both methods incorporate multiparameter attributes by encoding them into RGB-images. On synthetic and real 3D data examples, we demonstrate, that application of these methods for seismic data enhancement using nonlinear beamforming can greatly speed up the computational time while maintaining similar quality of output data.
-
-
-
Blended Acquisition with Temporally Signatured/Modulated and Spatially Dispersed Source Array: Productivity Enhancement in a Pilot Survey
Authors T. Ishiyama and T. WeiSummaryRecently, we established a blended-acquisition method: temporally signatured and/or modulated and spatially dispersed source array, namely S-/M-DSA, that jointly uses various signaturing and/or modulation in the time dimension and dispersed source array in the space dimension. We have acquired the first pilot survey with S-/M-DSA onshore Abu Dhabi. In this paper, we introduce this pilot survey and the resulting acquisition productivity enhancement in the time dimension. Furthermore, we discuss how this method could enhance the acquisition productivity in the space dimension as well. These show that S-/M-DSA significantly enhances the acquisition productivity compared to conventional blending methods.
-
-
-
Mixed Phase Seismic Wavelet Estimation using the Bispectrum
By M. BekaraSummaryThe ability to estimate a mixed phase wavelet is a useful tool for processing and quality control in seismic imaging. The wavelet is estimated using higher order statistics of the data. In practice, these methods tend to show some instability issues when the wavelet length is increased. To improve the stability of the solution, this abstract proposes a new formulation of the wavelet estimation problem that constrains the solution to be a finite duration, phase-only compensation applied to a known base wavelet. The proposed solution works in the frequency domain and consists of three steps. First, the bispectrum of the data is deconvolved using the bispectrum of the base wavelet to increase its bandwidth. This helps to improve the sensitivity of third order statistics to phase information. Then, a phase-only wavelet is estimated from the deconvolved bispectrum using an iterative least-squares approach without phase unwrapping. Finally, the estimated phase-only wavelet is conditioned using a projection onto convex sets type algorithm to enforce the constraint of the finite time duration giving the user a control on the amount of phase deviation from the base wavelet. Test examples on synthetic and real data both show reliable results with robustness to noise contamination.
-
-
-
Least-Squares Kirchhoff Psdm with a Local Based Inversion Approach and Compensation for Limitations in Modeling.
Authors Ø. Korsmo, S. Crawley, C. Zhou, S. Lee, E. Klochikhina and N. CheminguiSummaryReliable seismic amplitudes are crucial for the estimation of rock properties. In conventional depth imaging, amplitudes and resolution will be influence by propagation effects in the imaging model. These limitations origin from the formulation of the migration operator, implemented as the adjoint rather than the inverse of modeling. Least-squares migration (LSM) tries to eliminate these effects and resolve the real reflectivity model.
In this study, we make use of a newly developed local calibrated image-domain Kirchhoff least-squares migration to deconvolve the system response from the depth migrated gathers. We demonstrate how the inversion de-blurs the image and adjusts the prestack amplitude response, following better the expected response from well synthetic. The method is demonstrated on a North Sea dataset from the Viking Graben area, covering the Verdandi/Lille Prinsen discovery.
-
-
-
Seismicity-Permeability Coupling in the Breaching and Sealing of Reservoirs and Caprocks
Authors D. Elsworth, Y. Fang, K. Im, C. Wang, T. Ishibashi, Y. Jia, E.C. Yildirim and F. ZhangSummaryThe presence of pre-existing faults and fractures in the upper crust contribute to induced seismicity as a result of fluid injection, in hydraulic fracturing, deep storage of CO2, and stimulation of EGS reservoirs. In all of these, either maintaining the low permeability and integrity of caprocks or in controlling the growth of permeability in initially very-low-permeability shales and geothermal reservoirs are key desires. We explore styles of permeability evolution using both experimental and computational methods to explore how fracture permeability changes in response to fracture/fault reactivation and investigate the roles of (1) mineralogy and (2) fracture roughness in conditioning response; together with (3) intrinsic controls of healing on the earthquake cycle and permeability evolution.
-
-
-
A Perfectly Matched Layer Technique for the Lattice Spring Model
More LessSummaryThe lattice spring model (LSM) combined with the velocity Verlet algorithm is a newly developed scheme for modeling elastic wave propagation in solid media. Unlike conventional wave equation based schemes, LSM is established on the basis of micro-mechanics of the subsurface media, which enjoys better dynamic characteristics of elastic systems. But LSM is still suffering the boundary reflections and little work has been reported on this topic. The focus of the present study was to develop a special form of absorbing boundary condition based on the perfectly matched layer (PML) concept for LSM. The PML formulation is tested using a homogeneous model and the Marmousi model. The perfectly matched layer concept appears to be very well suited for LSM.
-
-
-
Design of Non-Replicated Acquisition Geometries for Time-Lapse Measurements
Authors G. Blacquiere and G. BlacquiereSummaryTime-lapse, or 4D seismic is capable of satisfying the continuously increasing demand for high-quality subsurface images to reveal both static and dynamic elements during the field development. However, in practice, challenges of pursuing this strategy lie in different perspectives related to budgetary, operational and regulatory constraints. Seismic surveys, performed in a compressed manner in time and/or space, can provide high-quality seismic datasets in a cost-effective and productive manner. The processing of data acquired in this way usually requires decompression, e.g., deblending and data reconstruction. The decompression performance is of fundamental importance in determining the success of compressed measurements. Our decompression approach deals jointly with deblending and data reconstruction via a sparse inversion, coupled with constraints on causality and coherency. Additionally, we carry out the inversion simultaneously for all available vintages, sharing static information between them while extracting the dynamic changes. We use this inversion as the kernel of a survey-design scheme. We use artificial intelligence (convolutional neural network) to speed up the computations. In our experiment, using time-lapse data from the Troll field, the improvement of designing the acquisition geometry combined with the simultaneous inversion of all available vintages was 6 dB.
-
-
-
Application of Stochastic Method for Geomechanical Parameters under Uncertainty Quantification to Design Mud Window
Authors M.A. Ebrahimi, M.J. Ameri and M. AhmadiSummaryA novel methodology is developed to design a reliable safe mud window based on most updated geometrical uncertainty distribution. The developed approach allows to estimate the uncertainty ranges for geometrical parameters and their dependent parameters such as collapse pressure and fracture pressure. A trustable mud window design based on posterior probability reduces the risks of wellbore stability problems and less kick.
In this research, the Markov chain Monte Carlo simulation used to quantify the geomechanical uncertainties in order to make it more clear and trustable.
-
-
-
Seismic Quality Factor Estimation with a Simulated Annealing Approach: A Practical Example of the Sichuan Basin
More LessSummaryFluid movement and grain boundary friction are the two main factors responsible for the anelastic attenuation of seismic data. The quality factor Q quantifies the degree of anelastic attenuation and is commonly used in assisting the identification of gas reservoirs. We propose to employ the seismic reflections at near offset as referred seismic signals in the quality factor computation while the seismic reflections at medium and far offsets are regarded as target seismic signals. We then employ simulated annealing to simultaneously obtain the quality factor values of the targeted seismic signals. The proposed method is applied to both synthetic and real seismic data to demonstrate the validity and effectiveness. The application of SiChuan field data demonstrates that the estimated Q values using our method can be used as direct indicator the for gas reservoir.
-