- Home
- Conferences
- Conference Proceedings
- Conferences
77th EAGE Conference and Exhibition 2015
- Conference date: June 1-4, 2015
- Location: Madrid, Spain
- Published: 01 June 2015
941 - 960 of 980 results
-
-
Full-waveform Inversion of Surface Waves - Numerical Sensitivity Analysis
Authors I. Silvestrov, A. Bakulin, M. Dmitriev, K. Gadylshin, P. Golikov, D. Neklyudov and V. TcheverdaSummaryWe present the results of analysis of two full-waveform inversion formulations using surface waves. We explicitly construct the approximate Hessian matrix and study its properties using a truncated SVD approach. This analysis allows us to understand characteristics of the inversion algorithm such as the expected shape of the recovered solution, uncertainty of the results and the expected properties of local minimization algorithms being applied to this problem. This information should be useful for further development of the inversion strategy and for interpretation of the inversion results.
-
-
-
An Elastic Full-waveform Inversion Method Based on Wavefield Separation
More LessSummaryOwing to its potential to provide more subsurface model parameters with high accuracy, elastic full waveform inversion (FWI) becomes more attractive than acoustic FWI. However, the coupling of different parameters and wave modes results in more local minima and serious nonlinearity. To improve inversion accuracy, we propose a new elastic FWI method based on the wavenumber-domain wavefield vector decomposition scheme. First, the gradients expressed by separated P- or S-wave are derived. Then, a three-stage inversion strategy is introduced. The nonlinearity and non-uniqueness are mitigated by using the P- and S-wave separation and our hierarchical steps. A synthetic example reveals that our new method can obtain better inversion results than the conventional FWI method.
-
-
-
Spectral Analysis of Post-imaging Seismic Data
Authors A. Khalil, H. Hoeber, B. Deschizeaux and S. CampbellSummaryA migrated seismic image represents the spatially variable reflectivity of the medium where migration effectively rotates the wavelet to be normal to the imaged reflectors. While this is the general case, it is often disregarded, and one-dimensional spectral analysis of the vertical coordinate is commonly used. We show that spectral processing of post-imaging data in the direction normal to the reflectors provides accurate results for steep and complex structures. We introduce the concept of Orthogonal-Image-Gathers (OIGs) which facilitate this approach while providing a platform for handling spatially variable spectral distortions due to the velocity field and other medium-related properties.
-
-
-
Generalized S-transform Based Time-frequency Domain Spectral Modeling Deconvolution
Authors S. Wei, O.Y. Yonglin, Z. QingCai, H. JiaQiang and S. YayingSummaryIn the conventional deconvolution techniques, the reflection coefficients were assumed random white noise and the wavelet was unique. In order to get rid of the limitation, we proposed a new method which was called time-frequency domain Spectral Modeling Deconvolution (SMD). By using the generalized S-transform, the seismic data was transformed to time-frequency domain, the spectral of wavelet which was changed with time and frequency was simulated with the method of spectral modeling. The purpose of our process was to enhance the resolution of seismic data, thus, the spectral of wavelets should be eliminated, the remaining were spectral of reflection coefficients, after perform inverse S-transform, the time domain seismic data were obtained again with a more balanced spectrum and broader frequency band.
-
-
-
Double Sparsity Dictionary for Seismic Noise Attenuation
More LessSummaryIn this paper, we propose a double sparsity dictionary (DSD) for seismic data in order to combine the benefits of both analytic and learning-based sparsity-promoting transforms. We provide two models to learn the DSD: the synthesis model and the analysis model. The synthesis model learns DSD in the image domain, and the analysis model learns DSD in the model domain. As a tutorial, we give an example of the analysis model and propose to use the seislet transform and data-drive tight frame (DDTF) as the base transform and adaptive dictionary in the DSD framework. DDTF obtains an extra structure regularization when learning dictionaries, while the seislet transform obtains a compensation of the transformation error caused by slope dependency. The given DSD can provide a sparser representation than the individual transform and dictionary and therefore can help achieve better performance in denoising applications. Field data example confirms a superior denoising performance of the proposed approach.
-
-
-
Filtering Traces Detection Using Hough Transform in ST-TFPF
More LessSummaryIn recent years, spatiotemporal time-frequency peak filtering (ST-TFPF) has been proposed and successfully applied to seismic random noise suppression. It improves the shortcomings of conventional TFPF and processes seismic data along radial or quadratic traces in spatiotemporal domain. The matching degree of filtering trace and reflection events directly influence the denoising result of ST-TFPF.
Therefore, filtering trace selection has vital significance and plays an important role in ST-TFPF.
Nevertheless, in its existing models, there is hardly any effective selection method. The trace parameters were either fixed or selected by some immature algorithms, which consequently produce inaccurate or error filtering traces. Thus, this paper presents a novel filtering trace selection approach using Hough transform (HT), which designs filtering traces according to the distribution morphology of seismic events. The experimental results prove its strong accuracy and applicability in the aspect of filtering trace selection.
-
-
-
A Study on the Stationarity of the Random Noise in Seismic Prospecting
More LessSummaryIn the denoising process of seismic prospecting, random noise is often assumed to be stationary. However, this is not always appropriate for real seismic data. Here, using the passive noise records collected according to the requirements of the actual seismic prospecting, we use the statistical testing methods to investigate the stationarity of land-seismic-prospecting random noise. The results show that the noise can not be considered as a stationary stochastic process. By analyzing the behaviors of the dataset, we obtain that the non-stationary noise always have more energy in high frequency bands, and the energy distributions of the non-stationary noise are disordered in high frequencies. Thus, the stationarity of the random noise should be improved by attenuating the energy in high frequencies. We use an experiment to prove the correctness of our results. Based on the generating mechanism of the random noise, we also give a reasonable explanation. The findings have implications for future noise reduction and signal detection algorithms.
-
-
-
Ground Roll Attenuation Using Variational Mode Decomposition
More LessSummaryGround roll is the main interference wave in seismic data, which is characterized by low frequency and high amplitude. Common techniques for ground roll attenuation include frequency filtering, windowed frequency filtering, f-k filtering, and so on. In this paper, a novel time-frequency analysis method is presented based on the variational mode decomposition (VMD), which aims at depicting the non-stationary characteristics of seismic data. The proposed approach can provide more preferable time-frequency decomposition than the complete ensemble empirical mode decomposition (CEEMD). Synthetic examples show that the VMD can exactly capture the variable frequency components compared with CEEMD. The field data examples demonstrate that the VMD can be employed to effectively attenuate ground roll in seismic shot record, therefore render this technique is promising for seismic processing.
-
-
-
A Lifted l1/l2 Constraint for Sparse Blind Deconvolution
Authors E. Esser, T. Lin, R. Wang and F.J. HerrmannSummaryWe propose a modification to a sparsity constraint based on the ratio of l1 and l2 norms for solving blind seismic deconvolution problems in which the data consist of linear convolutions of different sparse reflectivities with the same source wavelet. We also extend the approach to the Estimation of Primaries by Sparse Inversion (EPSI) model, which includes surface related multiples. Minimizing the ratio of l1 and l2 norms has been previously shown to promote sparsity in a variety of applications including blind deconvolution. Most existing implementations are heuristic or require smoothing the l1/l2 penalty. Lifted versions of l1/l2 constraints have also been proposed but are challenging to implement. Inspired by the lifting approach, we propose to split the sparse signals into positive and negative components and apply an l1/l2 constraint to the difference, thereby obtaining a constraint that is easy to implement without smoothing the l1 or l2 norms. We show that a method of multipliers implementation of the resulting model can recover source wavelets that are not necessarily minimum phase and approximately reconstruct the sparse reflectivities. Numerical experiments demonstrate robustness to the initialization as well as to noise in the data.
-
-
-
Time-varying Wavelet Estimation by Hybrid Method
More LessSummarySeismic wavelet estimation is an important part of seismic data processing and interpretation, whose preciseness is directly related to the results of deconvolution and inversion. The methods for seismic wavelet estimation can be classified into two basic types: deterministic and statistical. By combining the two types of methods, spectral coherence method ( Walden & White, 1998 ) of deterministic method and skewness attribute method ( Fomel & Van der Baan, 2014 ) of statistical method, the amplitude and phase of the time-varying wavelet are estimated separately. The skewness attribute is used to estimate time-varying phase of propagating wavelet instead of locally observed wavelet. Phase-only corrections can then be applied by means of a time-varying phase rotation. Alternatively, amplitude and phase deconvolution can be achieved to enhance the resolution. We illustrate the method on both synthetic and real data examples.
-
-
-
Time Lapse Seismic to Assess Blowout Effects - A Case Study from Upper Assam Basin, India
Authors S. Chakraborty, S. Maji, Y. Ranjeet Singh, P.K. Paul and R. DasguptaSummaryThe following study utilizes 3D seismic data from 1995 and 2008 to investigate blow-out related seismic amplitude changes in the Eocene reservoirs of the field. Differences in reflection amplitudes and other attributes in these data vintages are used to interpret and constrain the spatial distribution of reservoir properties and probe into insight of the depletion process due to blow-out. 4D seismic data is considered as a relative measure of change in the reservoir and it has to be suitably calibrated and normalized. NRMS data is used to remove more anomalous measurements. NRMS amplitudes can be interpreted as a fractional change in amplitude relative to the original. Production profiles for several wells and seismic based attributes align towards the results predicted through NRMS amplitude differences. This study normalizes the two seismic surveys acquired in two vintages and interprets the amplitude differences, as well as correlating the seismic volumes thereby identifying the regions which may be affected by the blowout. On the process of updating the reservoir by calibrating 4D data of different vintages has given a better understanding of changes to the reservoir and also reduces risk in forecasting, leading to better decisions about reservoir maintenance and infill well targeting.
-
-
-
Volumetric Bounds on Subsurface Fluid Substitution Using 4D Seismic Time-shifts with an Application at Sleipner
Authors P. Bergmann and A. ChadwickSummaryA method for volumetric estimation of subsurface fluid substitution is presented that relies on the analysis of 4D seismic time-shifts. Since time-shifts cannot resolve for fluid saturation and layer thickness simultaneously without additional constraints, mass estimates are derived from the complete set of possible fluid saturations and layer thicknesses. The method considers velocity-saturation relationships that range from uniform saturation to patchy saturation. Based on a generalized velocity-saturation relationship that is parameterized by the degree of patchiness, explicit upper and lower fluid mass bounds are provided. We show that the inherent ambiguity between fluid saturation and layer thickness has a severe impact on the convergence of these mass bounds. That is, roughly linear velocity-saturation relationships with patchy saturation tend to provide significantly better accuracy in a mass interpretation than the strongly non-linear velocity-saturation relationships associated with homogeneous saturation. The method is validated at the Sleipner storage site, where injected fluid masses are known. Moreover, a linear relationship between 4D time-shifts and injected mass is observed, suggesting that the evolving patterns of fluid saturation and fluid mixing in the CO2 plume at Sleipner have remained roughly constant with time.
-
-
-
Time-lapse Seismic Monitoring of Overburden Compaction due to Hydraulic Fracturing in the Montney Shale
By I. ViñalSummaryRecent studies have shown convincing evidence that time-lapse changes in seismic data occur not only within the reservoir interval but also in the overburden. Observations that production at the reservoir level and subsequent decrease in pore pressure lead to modifications in the stress field and variations in the overburden have been documented ( Hatchell et al., 2003 ; Hudson et al., 2007 ). The study of the opposite case, that is, the analysis of the effect in the overburden of an increase in pore pressure in the reservoir, has not been so well documented yet and is the focus of this work.
The analysis is performed using multicomponent data corresponding to three seismic surveys acquired to evaluate the hydraulic stimulations of two horizontal wells in the Montney Shale.
The hypothesis is that the increase in the reservoir pressure due to the hydraulic well treatment might produce overburden compaction, leading to an increase in stresses that would be translated into an increase in the seismic velocities and therefore, into detectable time-lapse time shifts above the reservoir interval. The expectation for the compacted areas is to be correlated with zones of anisotropy in the reservoir. The results obtained from this study show evidence of this correlation.
-
-
-
Coupled Geomechanical-fluid Flow Impacts on Time-lapse Seismic for Waterflooding and Thermal Recoveries
Authors A. Shamsa, L.R. Lines and M. PaydayeshSummary4D seismic quantitative interpretation remains a challenging process because simultaneous changes in several dynamic properties can make the estimation of these properties ambiguous from 4D seismic. In this study, a coupled geomechanical-flow simulation has been utilized to investigate the individual and combined impact of fluid saturation, pressure, temperature and induced stress on compressional wave velocity in water flooding and thermal recoveries.
The time-lapse response is modelled based on coupled geomechanical-flow simulation. The analysis reveals specific characteristics of thermally induced stress for an enhanced recovery such as steam injection in heavy oil. Our study suggests that thermal stress can be substantial during the steam injection and should be considered in addition to pressure induced stress in the reservoir. Injected steam significantly reduces the velocity due to low velocity nature of steam. On the other hand steam injection adds to total stress which will act like an additional load due to the expansion of the reservoir rock and resistance of overburden and side burdens. The examinations performed in this paper demonstrate the crucial need to do 4D seismic interpretation based on the coupled analysis of fluid flow and geomechanics.
-
-
-
Time-lapse Seismic Monitoring for Reservoir Waterflooding Recovery at Nigeria Deep Sea Field
Authors T.E. Fan, X.W. Zhang, G.Y. Hu, H.L. Zhang and N. TianSummaryThe paper studies the technology of the time-lapse seismic monitoring and the prediction of remaining oil distribution on water-driven reservoir. We first make the rock physical analysis through Gassmann’s fluid substitution, and find Ip2-2.15Is2 as the most fluid sensitive parameter. And then multicube pre-stack Ip2-2.15Is2 inversion and quantitative interpretation for time-lapse seismic data are evolved, especially in ways that integrating 4D seismic data with other types of reservoir data such as well logs, static reservoir data and production data. Finally, the application of above technologies shows that robust 4D anomalies clearly identify the injector preferential pathways and water front movement within the main producing interval, meanwhile the remaining oil distribution is well predicted.
-
-
-
Calibration of Rock Stress-sensitivity Using 4D Seismic Data
Authors H. Amini and C. MacBethSummaryRock stress-sensitivity is a key aspect of 4D seismic studies, particularly when discriminating the effects of pressure and saturation change. Typically, laboratory measurements on core samples are used to address the rock stress-sensitivity in petro-elastic models. Such measurements are subject to uncertainties and their applicability to the in-situ field-scale reservoir response is questionable. Here, as an alternative to laboratory measurements, a technique is presented to calibrate the rock stress-sensitivity based on the 4D seismic data alone. After years of injection, the variation of the 4D signal in the fully-flooded zone in the vicinity of the injectors can be solely attributed to pore pressure variations. The proposed method utilises comparison of the observed with synthetic 4D seismic response in this fully water flooded zone for a range of rock stress-sensitivity models. The results of the modelling around two injectors, for a deep-water turbidite reservoir in the West of Shetland on the UKCS, suggest higher rock stress-sensitivity compared to the stress-sensitivity determined from laboratory measurements.
-
-
-
Intra-survey Pressure Variations - Implications for 4D Seismic Interpretation
Authors V. Omofoma and C. MacBethSummaryDuring the time taken for seismic data to be acquired, reservoir pressure and saturation may fluctuate as a consequence of field production and operational procedures. This has consequences for the quantitative analysis of 4D seismic data and particularly for understanding of the pressure signal that diffuses rapidly into the reservoir over a time-scale of hours or less. A modelling study using actual acquisition data (permanent seabed sensors and also towed streamers), reveals that the signature of pressure variations in the pre-stack domain is complex, and thus the resultant post-stack image is not representative of the true reservoir mechanisms that caused the pressure changes. This is of particular concern when trying to accurately resolve small pressure changes away from wells with post-stack data. It appears however that larger signals closer to the well may still be detected adequately. Our results have implications for post-stack quantitative 4D seismic analysis, as well as processing and acquisition workflows for detailed seismic time-lapse studies.
-
-
-
Joining Two Probabilistic Approaches - 4D Seismic Data and Simulation Models
Authors A. Davolio and D.J. SchiozerSummaryIncreasingly, reservoir uncertainties are considered in geophysical and petroleum engineering problems through the use of probabilistic approaches. The increase of computer power has allowed the choice to consider several scenarios instead of a deterministic reservoir model and a deterministic estimation of a seismic attribute. Concerning the integration between reservoir simulation models and 4D seismic data (4DS), probabilistic approaches seems to be a reasonable choice due to the high level of uncertainties associated to the generation of simulation models and interpretation of 4DS. However, if the two data are treated within probabilistic approaches it is necessary do develop a procedure to integrate them. This work proposes a methodology to integrate two probabilistic approaches, namely, several simulation models yielded from a probabilistic well-history matching and several estimations of pressure and saturation resulted from a probabilistic procedure applied to 4D seismic data. The methodology proposes a classification based on the comparison between the histograms generated for the two dataset. The final result is a “Map of Classes” that provides a detailed view of the mismatch between the two dataset, which is a rich source of information to be used in history matching procedures.
-
-
-
Multi-Azimuth Variable-depth Streamer Acquisition and Processing - Martin Linge Case Study
Authors J.C. Morante-Gout, J.P. Mascomere, L. Janot and G. MikkelsenSummaryThe Martin Linge Field (ex-HILD) has a long history, being until June 2012 the largest un-developed gas discovery in the Northern North Sea. The field development has been delayed due to the complex structural setting of the Brent reservoirs and the poor seismic imagery. Two realizations contributed to unlock the development decision, the first one was the results of the 2005–2008 OBC seismic acquisition/ processing project that showed a step-change in the quality of the seismic image and the second one was the drilling in 2010 of the appraisal well 30/4-D-1H in the main Brent structure called Martin Linge East. Nevertheless, imaging has remained an issue, mainly with regards of faulting at Brent level, as proved by the last appraisal well, where an up to that point unknown fault was encountered. Subsequently, the decision was made to test the promising broadband variable-depth streamer technique in order to improve the structural understanding of the area prior to the drilling of development wells. This paper will illustrate how multi-azimuth broadband technology through acquisition, pre-processing, velocity model building including full waveform inversion and PreSDM imaging techniques can improve the seismic imaging to reduce uncertainties in a complex field development such as Martin Linge.
-
-
-
Wide-azimuth, Broadband, and High-density 3D Seismic Survey in Eastern Edge of the Pre-Caspian Basin
Authors B.Q. Zhang, S.T. Jin, R.J. Zhang, T.J. Zeng, J. Xiao and R. LiSummaryThe reef-flat facies reservoirs in carbonate are dominantly developed along the Eastern Edge of the Pre-Caspian Basin. Due to the influences of salt dome and strong heterogeneity, the pre-salt targets, stratigraphy-controlled traps, cannot be well imaged with conventional source, geometry, and processing methods. The narrow-azimuth and low-fold geometry cannot provide high S/N data and cannot image the salt-domes and the pre-salt reef-beach bodies distinctively; the lack of low frequencies and the loss of high frequencies severely affect the resolution of seismic data and inversion quality. The broadband vibrator which enhances low frequency from 1.5Hz has been developed and deployed in eastern edge of the Pre-Caspian basin, Kazakhstan in 2013. A Wide-azimuth, Broadband, and High-density (WBH) data volume was acquired. Following the acquisition, a WBH processing flow has been set up, the imaging accuracy of the salt dome boundaries and pre-salt reflections and resolution have been greatly improved. The geological problems of this block have been solved satisfyingly, and good results have been achieved.
-