Layer | Fill | Outline |
---|
Map layers
Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|
Fill | Stroke |
---|---|
Collaborating Authors
Results
Summary Using a combination of 3-D synthetic and field datasets, we motivate and demonstrate the effectiveness of a combined waveform inversion workflow using both single-frequency and broadband or time-domain inversion. We initially use single-frequency waveform inversion to isolate and invert the lowest temporal frequencies in the data, thereby updating the model with the longest spatial wavelengths first. We then follow these inversions with either broadband or time-domain waveform inversion to add high-frequency detail to the model. The advantage of this workflow is that by inverting for the lowest possible frequencies first, prior to the addition of higher frequencies, the chances of cycle-skipping are minimized. We motivate our workflow by examining gradient computations for single-frequency inversion, broadband discrete-frequency inversion, and time-domain inversions of synthetic data. We then present a data example illustrating the effectiveness of the workflow on field data from offshore Trinidad.
Summary Full Waveform Inversion (FWI), while now widely practiced industrially, remains expensive because of the number of iterations typically required by the descents-type methods employed. Direct calculation of pre-conditioners which approximate the inverse Hessian, has so far proved cost-ineffective or downright unfeasible. It has often been noted that the calculation of the gradient in FWI is formally identical to a Reverse-Time Migration (RTM) with a cross-correlation imaging condition. Inspired by this, we propose to precondition the FWI gradient by calculating an update direction analogous to a deconvolution imaging condition. That is, instead of applying a scaled correlation of the modeled source wave fields and the back-propagated residual wave fields, the FWI update direction is derived by de-convolving the back-propagated residual by the source. The gradient is therefore corrected by the source-side illumination on a shot-by-shot basis, which is similar to one factor in the application of the inverse Hessian. We find that this approach improves the convergence rate compared with the conventional gradient method.
- Europe > United Kingdom > North Sea > Central North Sea > Moray Firth > Moray Firth Basin > Moray Firth Basin > Witch Ground Graben > P.213 > Block 16/26a > Brae Field > Alba Field > Caran Sandstone Formation (0.99)
- Europe > United Kingdom > North Sea > Central North Sea > Moray Firth > Moray Firth Basin > Moray Firth Basin > Witch Ground Graben > P.213 > Block 16/26a > Brae Field > Alba Field > Alba Sandstone Formation (0.99)
- Europe > United Kingdom > North Sea > Central North Sea > Moray Firth > Moray Firth Basin > Fladen Ground Spur > Witch Ground Graben > P.213 > Block 16/26a > Brae Field > Alba Field > Caran Sandstone Formation (0.99)
- Europe > United Kingdom > North Sea > Central North Sea > Moray Firth > Moray Firth Basin > Fladen Ground Spur > Witch Ground Graben > P.213 > Block 16/26a > Brae Field > Alba Field > Alba Sandstone Formation (0.99)
Summary To make the l1-norm frequency-domain elastic full waveform inversion (FWI) less sensitive to initial guesses, we propose the weighting method that incorporates weighting functions to gradients at each frequency. The weighting function should be designed so that the final gradient can properly describe the differences between assumed (initial or inverted) and true models. In the l2-norm FWI, the backpropagated wavefields incited by source wavelet-deconvolved residuals, which can be easily obtained during the inversion process, can be used for the weighting function. However, in the l1-norm FWI, since the residuals normalized by their magnitudes are back-propagated to compute gradients, the normalized residuals do not reflect the differences between assumed and true models. For this reason, we approximate the residuals of the l2-norm objective function to define the weighting function for the l1-norm FWI. We demonstrate the weighting function for synthetic data with outliers obtained for the 2D section of the SEG/EAGE salt model. Numerical examples show that while the conventional l1-norm waveform inversion does not provide stable solutions for the salt model without good initial guesses, the weighting method gives inversion results comparable to true models even with the poorly estimated initial guesses.
Summary Geological features, such as faults, dikes and contacts, appear as lineaments in potential field data. Here we describe a new approach to automatically analyze sets of lineaments, (Total Horizontal Derivative maxima and zero contours of Tilt), derived from the same gravity data set. Coherency analysis โ ACLAS โ generates a set of consistent lineaments which more accurately identify the locations of geological features before determining their depth, density contrast and direction of throw. The automated method provides a "non-subjective" solution to identifying the portions of the Tilt zero contours that define structures. This improves the accuracy of edge/contact locations and the efficiency of structural mapping, especially for large data sets. The method could also be applied to combined (gravity and magnetic) data sets.
- Geophysics > Magnetic Surveying (1.00)
- Geophysics > Gravity Surveying (1.00)
Summary We analyze the effect of random encoding sequences on simultaneous-source full wavefield inversion (FWI) as a function of code length. The simultaneous-source method causes two artifacts in the FWI gradients compared to the gradients obtained in the conventional manner of stacking the gradients computed for each shot. One originates from the encoding of the seismic data by random sequences, and the other from the stacking of the encoded seismic data. We show that the effect of the code length on the former is more significant than that on the latter due to distortion of the frequency bands in the seismic data. The encoding sequences of ยฑ1 in Krebs et al. (2009) does not distort the frequency band of the data, and does not increase the computational cost compared to the longer encoding sequences, and so provides better performance when compared to longer encoding sequences.
Summary In this abstract, we discuss some fundamental issues of full waveform inversion (FWI). We believe that a successful waveform inversion should emphasize traveltime information and downplay the role of amplitude information. This is achieved by using an objective function that measures the cross-correlation between observed and synthetic data. Further, we propose that for FWI of reflection data in the data domain, migration and demigration processes are necessary to reduce the well-known cycle-skip problem.
- Geophysics > Seismic Surveying > Seismic Processing (0.95)
- Geophysics > Seismic Surveying > Seismic Modeling > Velocity Modeling > Seismic Inversion (0.85)
Summary An analysis of amplitude variation with offset (AVO) observations is applied in hydrate-bearing sands, free-gas- charged sands, and hydrate-over-gas sands. The elastic model parameters (Vp, Vs, and density) are obtained from well log measurements and a rock physics model. The study suggests that presence of gas hydrate and free gas affect the AVO of shallow unconsolidated sediments containing gas hydrate and free gas. Low-concentrated gas hydrate and low-concentrated gas hydrate overlying free gas have weak AVO behaviors while highly-concentrated gas hydrate and highly-concentrated gas hydrate overlying free gas have strong AVO behaviors. Both highly-concentrated gas hydrate and highly-concentrated gas hydrate overlying free gas are Class I AVO anomalies but the intercept of AVO is stronger negative for highly-concentrated gas hydrate overlying free gas. They may occur in different locations in the AVO intercept and gradient plane.
Pore-pressure prediction relies heavily on interpretation of seismic and well attributes such as velocity, resistivity, and density which capture porosity changes during shale compaction under vertical loading. Relationships such as Eaton (1975) were developed in the Gulf of Mexico using a relatively simple lithological mix of geologically young sandstones and shale mudrocks at relatively low temperatures. An alternative approach using data from the same region was more deterministicยฐalso with vertical stress (e.g., Hottman and Johnson, 1965; Bowers, 1995). Another approach using mean stress, developed by Harrold et al., (2000) used similar sand and shale sequences at relatively low temperatures from data in southeast Asia basins. All the above approaches can be shown to provide acceptable prediction of pore pressure in shale mudrocks in young, rapidly deposited siliciclastic sequences, such as the Baram Delta, Brunei (Tingay et al., 2009) and along the West African margin (Swarbrick et al., 2011). The results can be calibrated, with careful attention to evidence for lateral drainage or lateral transfer, using data from their associated reservoir. However, in higher-temperature environments (e.g., Malay Basin, Hoesni 2004) these methods fail to deliver predictions, which may be accurate enough for effective well planning and safe drilling. This paper reviews the above methods and how they can be modified for use at elevated temperatures (e.g., above 100ยฐC). However, the review also draws attention to the wide uncertainty inherent in conventional approaches to prediction in some regions, which may require an entirely different approach to pore-pressure prediction prior to drilling.
- North America > United States > Montana (0.28)
- North America > United States > North Dakota (0.28)
- Asia > Malaysia > South China Sea (0.25)
- Asia > Malaysia > Sarawak > South China Sea (0.24)
- Geology > Rock Type > Sedimentary Rock > Clastic Rock > Mudrock > Shale (1.00)
- Geology > Geological Subdiscipline > Geomechanics (1.00)
Abstract A review of several hundred Diagnostic Fracture Injection Tests (DFITs) from vertical and horizontal wellbores in a variety of lithologies (tight sandstone, siltstone and shale) within a spectrum of geological settings (passive margin, foreland, and active strike-slip/thrust basins) was conducted to determine potential controls on stimulation complexity determined from the DFIT Net Fracture Pressure (NFP). Not surprisingly large differences in NFP complexity exist within this diverse data set and the variability is best explained by grouping the data according to the tectonic setting of the basin. Tectonic setting is interpreted as the first order control on NFP complexity since increasingly complex tectonic and burial histories elevate stresses and create tectonic fractures that promote increasingly complicated interactions between induced hydraulic fractures and intrinsic rock fractures. As a result, rocks in the Gulf Coast passive margin basin (Haynesville, Bossier) which have relatively simple burial and tectonic histories exhibit the lowest NFP complexities whereas rocks in strike-slip/thrust basins with high present day tectonic stress and abundant tectonic fabric have the largest complexity. Rocks in foreland basins (Montney, Horn River, Cretaceous Deep Basin sandstones) have NFP complexity that is generally variable between passive margin and strike-slip/thrust basins. Within any particular tectonic setting, NFP complexity is controlled by a complicated interplay between the natural fracture intensity, net horizontal stress (NHS) and wellbore geometry. Increasingly stiff and brittle rocks are commonly increasingly naturally fractured and this favours greater NFP complexity, and DFITs from vertical wellbores generally exhibit lower NFP complexity as fracture initiation and growth is simpler from vertical wells than from horizontal wells. Relations between NFP complexity and NHS (closure โ pore pressure) are complicated by the degree to which tectonics has diminished and overprinted the pore pressure control on closure stress. In the Gulf Coast passive margin setting (eg.Haynesville shale) pore pressure is the dominant control on closure stress and NFP complexity is increased where pore pressures are lower, possibly due to frac geometry changes associated changing stress profiles. In more tectonically complex settings (foreland and strike-slip/thrust basins) pore pressure exerts less influence on closure while tectonic stresses increasingly influence regional NFP complexity.
- North America > Canada > British Columbia (1.00)
- North America > Canada > Alberta (1.00)
- North America > United States > Texas > Harris County > Houston (0.28)
- Geology > Structural Geology > Tectonics (1.00)
- Geology > Rock Type > Sedimentary Rock > Clastic Rock > Mudrock > Shale (1.00)
- Geology > Geological Subdiscipline > Geomechanics (1.00)
- Oceania > Australia > South Australia > Cooper Basin (0.99)
- Oceania > Australia > Queensland > Cooper Basin (0.99)
- North America > United States > Texas > Haynesville Shale Formation (0.99)
- (16 more...)
Abstract The Diagnostic Fracture Injection Test (DFIT) is a welltesting technique used to assess the fracture growth, fracture closure and permeability from reservoirs; recently, it has been applied successfully to unconventional reservoirs such as: coal seams, tight sands and shale gas. The conventional DFIT procedure shares similarities with the author's Step Rate Test (SRT) procedure; the author adapted the SRT for coal seams application in Australia more than 15 years ago, since then it has generated a data base of more than 700 tests. During this period, the SRT was primarily used to assess stress conditions in coal seam reservoirs. Because of the similarities between the DFIT and STR procedures, we decided to apply the DFIT analysis technique to the SRT data with the aim to assess the coal seam permeability; furthermore, compare these permeability results with Drill Stem Test (DST) and Injection Falloff (IFT) permeabilities of tests undertaken in the same coal seam. The permeability results from the three methods are not always similar and sometimes differ appreciably; this generates a dilemma for reservoir engineers because they need to choose a representative permeability for their production forecast and reserves recovery assessments. The explanation of these permeability differences relies on the interpretation of coal stress parameters provided by the DFIT analysis. Furthermore, the appropriate integration of the DFIT, DST and IFT analysis results with wellbore storage, wellbore calliper and leak-off type, will assist the engineers to choose the representative reservoir permeability for their production forecast and reserves estimation; for this reason, running these three tests at the early stages of the exploration project will assists substantially with the property valuations and the planning of future work.
- North America > Canada > Alberta (0.29)
- North America > United States > Colorado (0.28)
- Geology > Rock Type > Sedimentary Rock > Organic-Rich Rock > Coal (1.00)
- Geology > Geological Subdiscipline > Geomechanics (1.00)
- Geology > Rock Type > Sedimentary Rock > Clastic Rock > Mudrock > Shale (0.54)