ABSTRACT We compare two iterative approaches for depth-migration velocity analysis by image-wave continuation in CIGs. One implementation aims at updating the whole model in each iteration, whereas the other updates it one layer at a time, from top to bottom. Previous investigations found that the first iteration of the whole-model approach works acceptably well in media with lateral velocity variations, but convergence is slow and not guaranteed. We test both implementations on synthetic data from a rather simple (but representative) model which simulates two sedimentary layers between a water layer at the top and a salt layer at the bottom. Our numerical examples demonstrate that the layer-stripping approach helps to improve convergence. Since this is a rather inexpensive procedure which starts at no a-priori knowledge of the medium, it can be used to build initial models for more sophisticated inversion techniques. Presentation Date: Tuesday, September 17, 2019 Session Start Time: 8:30 AM Presentation Start Time: 8:55 AM Location: 217B Presentation Type: Oral

Summary Common-Reflector-Surface processing (CRS) generally improves the signal-to-noise ratio of a data set. In addition, it is the best available algorithm to regularize data in any domain such as shot-receiver or azimuth/offset domain. In seismic data processing, there are multiple prestack migration algorithms around which may require input data optimized in both, data quality and spatial sampling depending on the selected migration approach. CRS processing can help in both, improving the overall input data quality as well as its sampling in order to reduce migration artefacts and thus improve the seismic image. This is demonstrated at both, a Kirchhoff migration and a Reverse Time Migration (RTM). Kirchhoff migration usually operates in the offset domain and for that reason shows less artefacts if a fully offset regularized data set is input. In this particular example, the data was separated into six azimuth classes which were independently offset regularized by CRS in order to better image faults according to their orientation. The azimuth/offset regularized data was then prestack migrated for each azimuth class. An RTM requires shot gather input. Here, we use a shot-receiver regularization to both, improve the signal-to-noise ratio, and suppress migration artefacts at locations with decreased shot-receiver coverage. Introduction Due to limitations in seismic data acquisition, either inaccessibility or related to no-permit-areas, data gaps are a common problem especially for land seismic data. A mitigation of such data gaps requires sophisticated algorithms to reconstruct as much information as possible from the available measurement (Fig. 1). In CRS processing, wave field attributes are calculated from the measured data, which describe reflections in detail. Reconstruction or correction of seismic events according to these attributes improves the amplitude of reflections over random noise and enhances reflector continuity. As a consequence, it is possible to fill small data gaps with the information of wave field propagation recorded in their vicinity (Fig. 1). Traces can therefore be relocated and a regularization of data is possible.

ABSTRACT Reverse time migration (RTM) is a very expensive depth imaging algorithm, particularly at high frequencies. The cost of RTM scales to the 4 power of the relative frequency. As such, the increase of the bandwidth for RTM may result in prohibitive costs, also reducing the ability to handle accurately the elastic properties, or to gather output such as surface offset gathers. Herein, we propose a novel method to efficiently cut the cost of RTM without sacrificing image quality, hence enabling the use of most advanced features in production environment. This new approach involves using a depth-variant migration strategy, with spectrum balancing. Presentation Date: Tuesday, October 16, 2018 Start Time: 1:50:00 PM Location: 207A (Anaheim Convention Center) Presentation Type: Oral

Summary For most of the present multiple migration methods need to separate multiples from the original data firstly, which is a time-consuming and high-cost job. So the joint migration method of primaries and surface-related multiples by replacing the source function with the recorded data together with a synthetic wavelet and use the recorded data as receiver wavefield was proposed, which can avoid the necessity of multiple separation. However, an incorrect wavelet will harmfully affect the accuracy of migration result, besides, serious crosstalk will be caused by the cross-correlation of unrelated events during the migration. In this paper, we propose a more reasonable method for joint migration, mainly by introducing two creative strategies to solve the problems proposed above. Firstly, we suggest to extract an accurate wavelet from the original data, then use the extracted wavelet together with the recorded data as the source wavefield. Secondly, the stereographic imaging condition is introduced to suppress the crosstalk artifacts in the image result. According to the numerical examples, the feasibility and effectiveness of this method is verified. Introduction For a long period, only primary reflections are considered as signal in the typical migration strategy. While other events including multiples are regarded as noise, which need to be suppressed as much as possible before migration. However, multiples are also real response reflected from the reflecting interface under the surface, and if properly migrated, they can provide wider illumination, higher fold, and better imaging result of the subsurface than primaries (Berkhout, 1994; Guitton, 2002; Shan, 2003; Liu, 2011). At present, most commonly used multiple imaging techniques are performed by altering the boundary conditions, replacing the impulsive source wavelet with the recorded data and using the multiples as the input receiver data instead of primary reflections, then with Kirchhoff migration (Reiter, 1991) or wave-equation migration (Berkhout, 1994; Liu, 2011) multiples can be imaged to the position of an equivalent primary. But the implementation of these techniques need to separate multiples from the recorded data, which is a time-consuming and high-cost job, besides, these techniques cannot effectively deal with the combination of the separated primary image and multiple image.

ABSTRACT Prestack RTM is a powerful tool to image complex structures. Based on computing numerical solutions to a two-way wave equation, it does not suffer from dip limitations like one-way downward continuation techniques, thus enabling overturned reflections to be imaged. As well as correctly handling multi-pathing, RTM has the potential to image internal multiples when the boundaries responsible for generating the multiples are present in the model. In this paper, we address normalization of RTM images, an important step in improving amplitudes for input to inversion. We also introduce a new hybrid normalization scheme. We use wide-azimuth (WAZ) data acquisition in the Gulf of Mexico data to demonstrate the impact of the various RTM normalization schemes.

ABSTRACT Based on the full two-way wave equation for wavefield extrapolation, reverse time migration (RTM) is considered as a powerful imaging technique that avoids the approximation of wave equation, without dip and extreme lateral variation of velocity limitation. However, this algorithm suffers from very expensive computational costs and high storage requirement. In this study, we introduce curvilinear coordinate system such that the depth of the Cartesian coordinate was converted to vertical time domain to overcome oversampling problems at higher speed regions, and derive the 1st order velocity-stress seismic wave equation in vertical time domain. This feature has a more profound effect on the areas with large speed differences, especially in the mid-deep high-speed areas. Besides, we deduce the difference formula of the approximate source wavefield using the optimization operator, and reconstructs the differential-order layer wavefield in the calculation area by storing the optimization operator of each point around each time slice, and uses this optimization operator boundary storage strategy to improve 3D RTM algorithm in vertical time domain. Our RTM algorithm is success in accurately imaging of complex structures and reducing 68.4% of the storage and 35% of the computation times. Presentation Date: Tuesday, October 16, 2018 Start Time: 1:50:00 PM Location: 207A (Anaheim Convention Center) Presentation Type: Oral

SUMMARY In this paper we introduce a new algorithm for seismic imaging based on the flow out of Gaussian wave packets. We follow the standard strategy of decomposing data into wave packets and flowing them out along rays to approximate the downward wavefield extrapolation, and finally applying an imaging condition. We revisit each computational step to gain efficiency. Furthermore, we develop procedure for seismic data decomposition in order to obtain highly sparse representations with Gaussian wave packets. As a result we get fast algorithm heavily exploiting sparse data representation and analytic description of Gaussian wave packets. We tests our algorithm on synthetic example of migrating common-shot gather. INTRODUCTION Gaussian beams were used for modeling seismic wave propagation since early 80s (Popov, 1982; C? erveny´, 2001), while Gaussian wave packets introduced at the same time (Babich and Ulin, 1984) were not as popular in practical computations. The practical implementation of prestack migration with Gaussian beams was proposed by (Hill, 2001). Migration based on Gaussian packets was discussed in Bucha (2009); ?Z ´a?cek (2004); Klime?s (2004) with a detailed discussion of the related issues. Recently, different versions of so called “beam migration” are have become popular in the seismic industry. The main strategy is to perform directional analysis of data and “steer” beams accordingly into the subsurface following the detected directions. It was noticed that curvelet frames can be used for sparse representation of data and migration operators (Douma and De Hoop, 2007; Chauris and Nguyen, 2008). In this paper we use Gaussian wave packets to represent seismic data, and use the fact that they are described by explicit analytic formulas that are invariant under many operations, e.g., translation, scaling, rotation, multiplication, convolution and Fourier transformation. We address some new approaches to implementing the seismic migration operator based on the flow-out of Gaussian wave packets. Following Bucha (2009), we will discuss the main steps of the migration strategy for 2D common-shot gathers: • data decomposition into Gaussian wave packets; • flow-out of wave-packets into the subsurface; • imaging condition (cross-correlation of wave packets). Data decomposition - sparse representation So far we have implemented a 2D decomposition that can be used for getting sparse representation of 2D seismic data sets with rather few wave packets. Sparse data representations are important for reducing the computational cost for the whole migration procedure, i.e., reducing the number of rays to be traced and the cost of applying imaging condition. In addition, we essentially get a high quality analysis of data directionality that can be used in many different ways apart from migration: detecting slopes, data regularization, as a part of event picking etc. Flow-out of wave packets Wave-packet flow-out is the most restrictive step of our migration strategy. It was noticed by many authors that a smooth migration velocity model should be used for propagation (flowout) of Gaussian wave packets. Even using a smooth model one gets in trouble trying to propagate Gaussian packets for a long distance.

Summary We compare two different approaches to input data regularization for azimuth-limited Kirchhoff prestack migration, namely azimuth sectoring and prestack interpolation via band-limited Fourier reconstruction. We evaluate the performance of both approaches as they are employed in a P-wave fracture detection flow using real and synthetic data. The prestack interpolation gave excellent results in both the real and synthetic cases; moreover we found that a careful implementation of azimuth sectoring also gave good results. By contrast, a naïve implementation of azimuth sectoring produced an unacceptable level of artifacts in the output fracture attribute volumes and we conclude that careful data regularization prior to prestack migration is a critical step in minimizing artifacts in post-migration fracture analysis. Introduction P-wave fracture detection tools are enjoying widespread use as interest in unconventional reservoir development continues to surge. P-wave fracture analysis exploits azimuthal variations in stacking velocity (“VVAZ”) and/or amplitude-versus-offset (“AVAZ”). In theory both AVAZ and VVAZ analyses should be performed in the migrated domain, but unfortunately 3D prestack time migration (PSTM) of wide-azimuth land data is known to be sensitive to the effects of irregular and/or sparse spatial sampling. For example, in unstructured data regimes where identification of subtle anomalies requires artifact-free imaging, stacks after PSTM may suffer from more sampling-induced migration noise than their counterparts produced after the relatively simple process of stack plus poststack migration. Unfortunately the stringency of the sampling requirements for avoiding migration artifacts is further heightened if no stacking is performed after PSTM simply because we lose the natural power of the stacking process in eliminating migration noise. This effect was recently explored by Hunt et al., (2008), who sought to minimize migration artifacts on PSTM image gathers generated by common offset migration in order to improve the quality of post-migration AVO inversion. As in the AVO case, post-migration fracture detection analysis is also performed on migrated image gathers (although the associated migration operates on offset-and-azimuthlimited, rather than offset limited, data subvolumes ), so we expect a similar sensitivity to the effects of irregular sampling, and correspondingly, the strong possibility that excessive levels of migration noise may completely negate the theoretical advantages associated with operating in the migrated domain. The traditional approach to forming the offset-and-azimuthlimited data subvolumes appropriate for input to PSTM in migrated-domain fracture analysis is to perform azimuth (and offset) sectoring (Lynn et al., 1996). When carefully implemented, this approach provides a degree of implicit data regularization which can help mitigate some of the migration artifacts related to the imperfect sampling. In this paper we consider prestack interpolation as an alternative approach for minimizing this migration noise. Specifically, we use a multidimensional Fourier reconstruction technique to synthesize regularly sampled common azimuth and offset data subvolumes which are in turn used as input to PSTM. It is worth noting that common offset vector gathering (also known as “ offset vector tiling”) (Cary (1999); Vermeer (2002)) has recently emerged as another useful approach for forming these offset-and-azimuth-limited data subvolumes (Calvert et al. (2008). Although this abstract focuses on comparing azimuth sectoring to prestack interpolation, examples from COV gathering will be also be included in the oral presentation.

SUMMARY The imaging condition used in reverse-time migration requires that the source wavefield (computed via a forward recursion) and the receiver wavefield (computed via a backwards recursion) must be made available at the same time in an implementation of the algorithm. Several strategies to organize the calculation can be employed, differing in balance between memory and computation. This paper describes and compares these different approaches, and argues that strategies favoring computational complexity over memory (to the point where disk i/o can be avoided) are attractive for 3D prestack migrations. An example of 3D reverse-time migration applied to wide-azimuth data from the Gulf of Mexico is presented to support the claim. INTRODUCTION Reverse-Time Migration (RTM) was introduced in the late 1970''s (Hemon, 1978) but despite showing promising imaging capabilities (Baysal et al., 1983; Whitmore, 1983; McMechan, 1983; Loewenthal and Mufti, 1983), it was not used in practice due to its stringent requirements, both in terms of computation and memory. Until recently, RTM was therefore largely confined to 2D and/or post-stack imaging but computer technology has now reached the point where 3D prestack RTM isfeasible (Yoon et al., 2003, 2004; Bednar and Bednar, 2006; Farmer et al., 2006; Guitton et al., 2007; Jones et al., 2007). The core of the algorithm is the crosscorrelation of two wavefields at the same time level, one computed by stepping forward in time, the other computed by stepping backwards in time. The forward recursion is usually carried out first, in which case the entire time history must be made available during the backwards recursion in order to compute the imaging condition. Several options can be used to arrange the calculation, differing in the amount of memory and computation requirements. The next section describes the realization of a particular approach to RTM based on leapfrog time stepping for a scalar field (pressure). The following section compares the various strategies which can be used to organize the computations. Finally, an example of 3D RTM applied to wide-azimuth data from the Gulf of Mexico is presented. SIMULATION AND REVERSE-TIME MIGRATION The forward problem involves marching this scheme for n = 0,1, . . . ,N -1. The simulation of synthetic seismic data dn is related to pn by a sampling operator Sn which extracts time samples of the wavefield at receiver positions, at the time sample rate of the output data traces. The implementation used for the examples shown below does not assume any relation between the computation grid and the acquisition geometry, or between the simulation time step and the sample rate of the seismic traces. It uses bilinear interpolation for the spatial variables and cubic interpolation in time. Similarly, the source field is adjoint-interpolated onto the computational grid. The Laplacian is approximated using an eighthorder centered difference scheme with optimized coefficients (Ye and Chu, 2005; Etgen, 2007) and Perfectly Matched Layers (PML) absorbing boundary conditions (Cohen, 2001) are used to simulate unbounded domain wave propagation.

Summary VSP-derived images can suffer from severe migration artifacts where the migration "smiles" resulting from the wave-equation and ray-based migration approaches are not be canceled out completely. These artifacts primarily result from limited acquisition aperture and limited receiver array used in the data acquisition. This can be especially important with sediments below a complicated salt structure where the desired data contains not only signals of interest but strong undesired diffractions and other arrivals caused by the sharp velocity contrast. In this work, we demonstrate how to improve migration image quality by decomposing wavefields into local beams in the migration process and optimizing the decomposed local beams using a generalized imaging condition. Tests on synthetic and field data indicate that this migration scheme is capable of attenuating migration artifacts and yields the image with less noise. However, some care should be taken when dealing with high angle structure such as faults and salt flanks where more beams will be required. Introduction The success of prestack depth migration for sediment imaging strongly depends on not only on the accuracy of the velocity model, but also on sufficient coverage of the illuminated subsurface. Over the years, in order to highlight the weak subsalt reflections in seismic data, much effort has been made in developing various migration approaches for surface seismic data in different domains. These include the fast beam (Gao et al, 2006), in which they decimate the data before generating the desired beams in order to reduce the processing cost, boost the subsalt signals, and attenuate the noise at the potential risk of losing some useful signal in data set. Luo et al., (2004) and Wu et al. (2006) developed the beamlet based migration method in order to provide a more accurate wavefield propagator for strong contrast velocity models by using beamlet transform to eliminate the numerical anisotropy and noise from the FD operator. Instead of a global FFT for space domain, they make use of a local cosine or a Gabor-Daubechies (G-D) transform to decompose the wavefield into local wavefields in both space and wavenumber domain which is then applied to the amplitude compensation and directional illumination analysis. To attenuate migration smiles, Luth et al. (2005) and Takahashi (1995) proposed using polarization information from multi-component data, or slowness information from single component data, to derive the wave propagation direction and restrict the Kirchhoff ray-based migration operator by locally mapping the recorded amplitude into the depth domain. This technique may encounter difficulty for subsalt sediment imaging below complicated salt structure. In this work, we focus on attenuating the migration noise in VSP image caused by limited receivers and migration aperture by applying local beam migration. In the first part, we will show what kinds of waves are recorded in our data besides primary signals and what causes such complicated characteristics. In the second part, we describe the basic idea of using local beam migration to attenuate migration artifacts. Examples will be presented and finally some conclusions will be given.

ABSTRACT Interbed multiples cause artifacts in subsurface images because they are incorrectly migrated when using standard primary-based migration algorithms. While surface-related multiple elimination has been well established as a standard step in seismic processing, the usage of interbed multiple attenuation (IMA) remains low due to various practical challenges, such as inaccuracy in prediction, difficulty in subtraction, and prohibitively high compute cost. We propose a workflow with a few strategies to improve the applicability of IMA. Using a 3D field data example, we demonstrate the effectiveness of our workflow in predicting and attenuating interbed multiples. Presentation Date: Tuesday, September 17, 2019 Session Start Time: 1:50 PM Presentation Start Time: 3:05 PM Location: 304B Presentation Type: Oral

Abstract Storing carbon dioxide (CO2) underground is considered the most effective way for long-term safe and low-cost CO2 sequestration. This recent application requires long-term wellbore integrity. A leaking wellbore annulus can be a pathway for CO2 migration into unplanned zones (other formations, adjacent reservoir zones, and other areas) leading to economic loss, reduction of CO2 storage efficiency, and potential compromise of the field for storage. This CO2 leakage through the annulus may occur much more rapidly than geologic leakage through the formation rock. The possibility of such leaks raises considerable concern about the long-term wellbore isolation and the durability of hydrated cement that is used to isolate the annulus across the producing/injection intervals in CO2-related wells. With the lack of industry standard practices dealing with wellbore isolation for the time scale of geological storage, a methodology to mitigate the associated risks is required. This requirement led to the need and development of a laboratory qualification of resistant cements and the long-term modeling of cement-sheath integrity. This article presents the results of a comprehensive study on the degradation of cement in simulating the interaction of the set cement with injected supercritical CO2 under downhole conditions. The methodology and the equipment are described for testing conventional Portland cement and measuring the evolution of its alteration process with time under CO2 conditions. Experimental details and analytical methods are discussed. Data relating cement-strength loss and CO2 penetration in Portland cement are presented. The evolution of cement chemistry and porosity with time is highlighted by scanning electron microscopy analyses, back-scattered electron images, and Hg-porosimetry measurements. A first fluid-flow-geochemistry modeling for Portland cement is proposed. The results are compared to equivalent studies on a new CO2-resistant material; the comparison shows significant promise for this new material. This CO2-resistant material will enable the hydrocarbon production industry to store the burnt residue over the long term in a safer and more responsible manner. Introduction Storing carbon dioxide (CO2) underground is considered the most effective way for long-term safe and low-cost CO2 sequestration[1,2]. There are three main types of geological reservoirs[3] with capacity sufficient to store captured CO2: depleted oil and gas reservoirs, deep saline aquifer reservoirs, unminable coal beds. The reservoirs need to be at a depth greater than 800 m so that the CO2 is in a supercritical state at a temperature and a pressure above its critical point (31.6°C, 7.3 MPa). These pressure and temperature conditions allow storing CO2 in a relatively small volume. The ideal storage site would involve high pressure but at the lowest possible temperature to be in the most dense properties of the supercritical CO2. For these reasons, regions with low geothermal gradients are preferable. Piping CO2 emissions for underground injection is not a novel concept and is already often used for the purposes of enhanced oil and gas recovery[4,5,6]. However, this application of CO2 injection is not intended for long-term storage, which is a more recent concept that needs a long-term wellbore integrity strategy to be developed. Indeed, the major risk associated by the public with CO2 injection is a well failure, which may result in escape of CO2 that will migrate upwards. The likelihood of a sudden escape of all CO2 stored in an underground reservoir is extremely small. The main risks are: CO2 and CH4 leakage, seismicity and ground movement (subsidence or uplift). Failure of the cement, in the injection interval or beyond it, may create preferential channels for CO2 migration back to the surface. This may occur on a much faster time scale than geological leakage. It is hence important to explain that wellbore integrity will ensure that CO2 stays underground for several hundred years and beyond.

ABSTRACT As a latest branch of ray theory, seismic beam method has certain unique advantages, such as amplitude regularity at caustic and shadow zones and tube ray paths with a narrow skin depth, all of which make it popular in seismic depth imaging. But like Kirchhoff migration, beam-based continuation operator also maps the reflected waves to the travel-time ellipse isochrones, which produces lots of imaging noise and swing artifacts, especially for low SNR or sparsely acquired data. In order to improve the robustness of beam migration, we proposed an effective data-driven optimization strategy in this paper, which could suppress migration artifacts and highlight the dominant reflections. For convenience, the classic Gaussian beam migration is taken as a typical representative of beam migration family to demonstrate how our algorithm is designed. The main idea of our method is that through analyzing the coherence attributes of common-receiver and common-shot pre-stack data, we firstly obtained the instantaneous emergence angles at sources positions or beam centers, which is based on the local plane-wave assumption. Then using this information, beam-front curvature radius and the first Fresnel radius, we constructed a quality control factor and applied it to beam migration frame to suppress imaging noise and improve imaging quality. Typical numerical example and the field data processing result proved the feasibility and adaptability of our method. Presentation Date: Wednesday, October 19, 2016 Start Time: 1:30:00 PM Location: 171/173 Presentation Type: ORAL

Introduction Summary This paper is being presented at a special session honoring J.H.F. Gardner who was my mentor. Here I discuss a phenomenon which we studied together in 1992. At that time we noticed that migration amplitudes are highly affected by the acquisition geometries, and that migration introduces very significant amplitude distortions to the data. We suggested ways to solve these problems during the migration. Since then, I have often received data for AVO inversion which was naively migrated and contaminated by this artifact. Re-migration was not an option in many of these cases. Here I suggest a practical solution which can be applied after migration and before AVO. It is a very heuristic approach, but is practical for many cases where re-migration is not available. Amplitude preservation is the most important element in seismic data processing and migration, when it comes to AVO and inversion. Users who perform amplitude inversion and extract rock properties from their seismic data expect that the data they use is “true amplitude”. At the same time the developers of migration programs emphasis the “amplitude preservation” qualities of their programs. The reality is often very different. In many cases we notice that the data used for inversion does not look like “true amplitude” or “amplitude preserved” even though the processors usually insist that they used the best available amplitude preserving migration. I can testify that 90% of the datasets that I have received for AVO inversion did not look at all like true amplitude data. The problems are mainly associated with acquisition geometries and conventional practices in pre-stack migration programs (both Kirchhoff and wave equation migration). In Kirchhoff migration, for example, people may spend much effort to compute and apply the amplitude preserving weights (Belkin determinant, 1985) which have a second order effect on the resulting amplitudes, but completely ignore a first order effect of the acquisition geometries. The issue was discussed by several authors, including Albertin, et. al, 1999, Cooper et.al, 2009 and others who suggested various solutions to be implemented as part of the migration process. In this paper I will discuss some practical “post-mortem” solutions to balance datasets which had already been migrated, but whose amplitudes were distorted by the migration program. I will concentrate mainly on one aspect - the artifacts caused by 3-D acquisition in common offset pre-stack migrations, which is probably the most popular migration strategy in the context of AVO and inversion. To better understand the problem, let us look at some examples. Figure 1 display migrated gathers from various 3-D seismic surveys acquired at different parts of the world. Notice that there are similarities in the general AVO of these gathers, all of them show high amplitudes at center offsets compared with the near and far offsets. We addressed this artifact before in a number of papers (Canning & Gardner, 1994, 1996). The problem can be intuitively explained using a simple synthetic example. A spiral diagram shows the geometry of one shot which exhibits wide azimuth distribution.

There have long been attempts to apply Full Waveform Inversion (FWI) to land data sets because it has the potential to build higher-resolution velocity models compared to traditional ray-based model building techniches. However, several difficulties hinder the successful application of FWI to land data: cycle skipping, unreliable amplitudes, as well as source and receiver signature estimation. To address these issues, we have developed three new FWI methods: the Dual Frequency Phase Difference (DFPD) Laplace Fourier domain FWI, Surface Offset Gather Flattening (SOGF) Wave Equation Migration Velocity Analysis (WEMVA), and Seismic Reflection Slope FWI. The Dual Frequency Phase Difference (DFPD) Laplace-Fourier domain FWI greatly simplifies the input data and promises the global minimum, and thus, is ideal for initial velocity model building. The Surface Offset Gather Flattening (SOGF) WEMVA method aims at maximizing the flatness of surface offset gather, which is a better objective than maximizing the focus of subsurface time/spatial lagged gather in terms of handling models with large lateral change. The Seismic Reflection Slope FWI method is capable of identifying small velocity anomalies such as gas clouds and faults, thereby producing a higher-resolution velocity model. Three model data examples are used demonstrate the capability of our methods: first we use a simple two-layer model to demonstrate that the Dual-Frequency Phase Difference (DFPD) Laplace-Fourier domain FWI is indeed immune to cycle skipping; second we use a steep boundary model to demonstrate that our Surface Offset Gather Flattening WEMVA is efficient at handling models with large lateral velocity change; finally we use the BP 2004 benchmark model to demonstrate the Seismic Reflection Slope FWI method is capable of identifying small velocity anomalies. We also note that these three methods can be combined together to form a FWI based model building strategy that produces high-resolution velocity model for land data set. Two field data examples are used to illustrate the effectiveness of this model building strategy. First, we applied the strategy to a 2D field data set acquired in Inner Mongolia. Two starting models were used, one is a simple model converted from time domain RMS velocity, and the other is from the traditional ray based model building approach. For both cases, our strategy was able to greatly improve the velocity model and the migration image. The migration image starting from the traditional ray based model had a superior focus and better followed the geological setting, indicating that our FWI strategy is compatible with traditional method and that, in order to get the best possible result, one would use our strategy subsequent to utilizing the traditional method. We then demonstrated this by applying our refined workflow to a 3D field data set acquired in the Pre-Caspian Basin, Kazakhstan, producing a velocity model with an improved migration image, containing fine details such as faults and channels.

Brownfield residual oil zones (ROZ) may benefit from specific strategies to maximize production. We evaluated several strategies for producing from the Seminole ROZ. This ROZ lies below the main pay zone (MPZ) of the field. Such brownfield ROZs occur in the Permian Basin and elsewhere, formed by the action of regional aquifers over geologic time. CO 2 can be injected into these zones to enhance oil recovery and carbon storage. Since brownfield ROZs are hydraulically connected to the MPZs, development sequences and schemes should influence oil production, CO 2 storage, and net present value (NPV). We conducted economic assessments of various CO 2 injection/production schemes in the Seminole stacked ROZ-MPZ reservoir based on flow simulations. First, we constructed a high-resolution geocellular model from a seismic survey, wireline logs and core data. To calibrate the geological model and constrain the interface between the ROZ and the MPZ, we performed a comprehensive production-pressure history matching of primary depletion and secondary waterflooding. After this, we conducted flow simulations of water alternating gas (WAG) injection into the reservoir while considering several injection/productions schemes (e.g., switching injection from the MPZ to the ROZ, commingled production). For each scheme, various WAG ratios (i.e., reservoir volume ratio between injected water and CO 2 ) were tested to find the maximum oil production and maximum CO 2 storage. We assessed the economic results for each WAG ratio case on NPV. The results from simulating various injection/production schemes showed that simultaneous CO 2 injection into the MPZ and ROZ favors oil production. If instead, CO 2 is injected into the MPZ and ROZ, then into the ROZ alone, this leads to increased CO 2 storage. Storage performance is influenced by the interplay between the crossflow from the MPZ to ROZ and WAG ratios. As the WAG ratio increases, the amount of CO 2 stored decreases more for commingled injection cases than for separated ROZ injection cases. Also, the WAG ratio leading to maximum oil production does not necessarily yield the largest NPV, because of the complicated interactions among CO 2 consumption, reservoir heterogeneity, and oil recovery. Brownfield ROZs are common below San Andres reservoirs in the Permian Basin, and they can be exploited to increase oilfields’ NPV and carbon storage potential. Our case study on the Seminole MPZ-ROZ is an analog for other similar reservoirs. We demonstrate that development sequences and WAG ratios influence the performance of CO 2 EOR and storage. Thus, this work provides valuable insights into the further optimization of brownfield ROZ development and helps operators to plan flexible storage goals for stacked ROZ-MPZ reservoirs.

ABSTRACT Gas hydrate accumulations in continental shelf sediments are considered a promising resource for future gas supply by several non-COST countries (e.g. USA, Japan, China, India, South Korea, and Taiwan). In 2013, the Research Consortium for Methane Hydrate Resources in Japan (MH21) produced gas during a successful offshore field test. In Europe, as elsewhere, demand for natural gas is continuously increasing. This COST Action is designed to integrate the expertise of a large number of European research groups and industrial players to promote the development of multidisciplinary knowledge on the potential of gas hydrates as an economically feasible and environmentally sound energy resource. In particular, MIGRATE aims to determine the European potential inventory of exploitable gas hydrates, to assess current technologies for their production, and to evaluate the associated risks. National efforts will be coordinated through Working Groups focusing on 1) resource assessment, 2) exploration, production, and monitoring technologies, 3) environmental challenges, 4) integration, public perception, and dissemination. Study areas will span the European continental margins, including the Black Sea, the Nordic Seas, the Mediterranean Sea and the Atlantic Ocean. INTRODUCTION Gas hydrates accumulating in continental margin sediments are considered as promising energy resource. Numerous countries around the world (e.g. Japan, South Korea, USA, China, Taiwan, India, New Zealand) are investing in hydrate R & D to explore their coasts and national waters, constrain the resource potential, and develop technologies for gas production from gas hydrates. Several production tests conducted both onshore and offshore have proven that gas can be produced from these unconventional natural gas reservoirs. Natural gas from indigenous gas hydrate deposits should play an important role in the future European energy system. It could i) enhance the security of energy supply, ii) contribute to the reduction of CO2 emissions by replacing coal, and iii) complement renewable energies and stabilize the power grid by proving electricity during low-wind and/or low-light periods. Ultimately, gas hydrates could replace Europe's conventional gas reserves that will be depleted within the next decades and mitigate the growing dependence of Europe on natural gas imports.

This paper presents typical multi-fractured horizontal well failures together with potential mitigation strategies. The discussion is based on information compiled over the last decade from operations in the United States, Argentina and Canada. The analysis focuses mainly on well integrity and well accessibility events associated with production casing, production liner and tubing strings. A comprehensive failure database covering over 150 cases from operations in shale plays was created. Failure here is defined as any event that led to either compromised well integrity or loss of well accessibility. The database was populated with verified cases from various operators and product origins, as well as various sizes and types. Causes were grouped in ten different categories, resulting in conclusions for the following well stages: strings’ installation, stimulation, and production. The analysis allows us to estimate the failure occurrence per stage and to infer the inherent failure risk for the different phases of a well's life cycle. Moreover, the obtained overall failures ranking shows which categories require more stringent attention in order to avoid problems throughout the life of a well. The database analysis also revealed distinctive failure patterns for the different stages. These deliver important lessons on how to select tubular goods, design the installation, or even operate the well in order to diminish the failure risk. This work represents an important effort to quantify the metrics and impact of the different well failures during the installation, stimulation and production stages throughout the past decade. The conclusions were obtained by using information from different operators, regions and product types, which means lessons learned can be considered as a valuable reference regardless of the particular contexts. Introduction Multi Fractured Horizontal Wells (MFHWs) exhibit unique challenges for tubulars and connections. Loads and demands and, therefore, requested performance may change radically among different well life stages (i.e. well construction - casing installation, stimulation and production).

Summary Leikoupo is an important gas-bearing formation of the Longnusi structure in the Central Sichuan Basin. The reservoir rock is of Middle Triassic age and is composed of dolomitic limestone with an overlying unit of halite and anhydrite. The underlying Lower Triassic Jialingjiang formation also contains a halite and evaporite unit. As such, Leikopu reservoir unit has been subjected to complex folding by the compressional action of the over- and underlying halite-anhydrite(or evaporite) formations. The complexity in the geometry of the reservoir unit has been successfully delineated by the seismic images derived from 3-D prestack time and depth migrations. The unique features of the Longnusi structure are described by low-dip thrust faults and blind thrust faults combined with subtle but intensive faulting and folding. In this paper, aside from inferring the structural style from the 3-D seismic image, we also discuss the genesis of the Longnusi structure. Introduction Figure 1 shows a sketch of the location of the area of investigation. A comparison of the seismic images derived from 2-D poststack, 3-D poststack, and 3-D prestack time migrations is shown in Figure 2. Note that the topreservoir event is ambiguously defined in the image from 2-D poststack time migration, whereas the best interpretation is achieved by the image from 3-D prestack time migration. The results of the interpretation have been combined with the geologic and borehole data to generate prospects in the area. The Leikopu halite unit above the dolomitic limestone reservoir level has a variable thickness between 0-115 m. Additionally, there exists undulations in the geometry of the reservoir unit itself and lateral velocity variations caused by the differences in lithology within the Leikopu formational sub-units. As such, 2-D and 3-D poststack migration strategies (Yilmaz, 2001) for seismic imaging of such a subtly complex strata have provided but only limited resolution (Figure 2). Instead, 3-D prestack time migration and, ultimately 3-D prestack depth migration (Figure 3), has provided the desired image quality for a detailed structural interpretation. The time structure map derived from the interpreation of the top-reservoir event in the image volume from 3-D prestack time migration exhibits subtle, but complex folding with intensive faulting (Figure 4). Delineation of the Structural Style within the Leikoupo Reservoir Unit The well-log data indicate that there exist two separate strata of halite within the Leikoupo formation (Figure 5). The thickness of the lower halite unit does not vary significantly, while that of the upper halite unit changes significantly in the lateral direction. The two units of halite are interbedded with a limestone strata and a thinly bedded anhydrite strata with a total thickness that varies between 27-30 m. Nevertheless, at some well locations, three cascaded Leikoupo halite strata have been encountered. This indicates the strong influence of thrust faulting. A pea-shaped tuff horizon marker present at two well locations further ascertains the presence of an intensive thrust fault system within the Leikoupo reservoir unit. The seismic image volume also exhibit the intricate complexity of the folding and faulting that are inferred from the well data.