Distributed Acoustic Sensing (DAS), a rapidly evolving fiber‐optic based technology for permanent in‐well and geophysical monitoring, was used to record VSP data in two field trials Shell assets onshore Canada USA. Useful velocity were gathered from the entire length of well wellhead TD (up 4 km), which compared with geophone recording sonic log data. Walk‐away yield images that are nearly equivalent conventional borehole geophones terms signal noise ratio resolution. Permanently installed...

Last year we introduced Distributed Acoustic Sensing (DAS) as a new method for VSP acquisition. In DAS, an optical fiber along the borehole serves seismic receiver array. The is interrogated by special device at surface, sometimes referred to lightbox. Improvements interrogator unit over past helped DAS data approach utility of conventional data. We show results from various field trials that demonstrate usability typical applications such checkshots, imaging and time-lapse monitoring....

Deep neural networks (DNNs) are becoming the core components of many applications running on edge devices, especially for real time image-based analysis. Increasingly, multi-faced knowledge is extracted via executing multiple DNNs inference models, e.g., identifying objects, faces, and genders from images. The response times multi-DNN highly affect users' quality experience safety as well. Different exhibit diversified resource requirements execution patterns across layers networks, which...

Federated Learning (FL) is a popular deep learning approach that prevents centralizing large amounts of data, and instead relies on clients update global model using their local datasets. Classical FL algorithms use central federator that, for each training round, waits all to send updates before aggregating them. In practical deployments, might have different computing powers network capabilities, which lead slow become performance bottlenecks. Previous works suggested deadline round so the...

Distributed Acoustic Sensing (DAS), a rapidly evolving fiber-optic based technology for permanent in-well and geophysical monitoring, has been used to record VSP data in multiple Shell assets onshore Canada, USA, Europe, the Middle East. Useful velocity were gathered from entire length of well wellhead TD (up 4 km), which compared with geophone recordings sonic log data. Walk-away yield images that are nearly equivalent conventional borehole geophones terms signal noise ratio resolution....

Distributed Acoustic Sensing (DAS) is a novel technology for seismic data acquisition, particularly suitable VSP. It break-through low-cost, on-demand, monitoring of reservoirs, both onshore and offshore. We will briefly explain how DAS works, then, demonstrate its usability typical VSP applications such as checkshots, imaging, time-lapse monitoring. show examples from around the world, discuss an enabler full-field with 3D

Deep neural networks (DNNs) are becoming the core components of many applications running on edge devices, especially for real time image-based analysis. Increasingly, multi-faced knowledge is extracted by executing multiple DNNs inference models, e.g., identifying objects, faces, and genders from images. It paramount importance to guarantee low response times such multi-DNN executions as it affects not only users quality experience but also safety. The challenge, largely unaddressed state...

M035 Summary TOMOGRAPHIC INVERSION OF FOCUSING OPERATORS B.E. COX and D.J. VERSCHUUR Delft University of Technology Lab. Acoustic Imaging Sound Control PO Box 5046 2600 GA The Netherlands Common Focus Point (CFP) method has proven to be an appropriate tool for velocity model estimation. CFP estimation that is described in this paper based on a separated inversion which first the focusing operators are updated with data-driven procedure after estimated. propagation velocities focus point...

Federated learning (FL) enables a set of geographically distributed clients to collectively train model through server. Classically, the training process is synchronous, but can be made asynchronous maintain its speed in presence slow and heterogeneous networks. The vast majority Byzantine fault-tolerant FL systems however rely on synchronous process. Our solution one first Byzantine-resilient algorithms that does not require an auxiliary server dataset delayed by stragglers, which are...

In time-lapse seismic studies involving compacting reservoirs, vertical time-shifts are frequently observed and prove useful for mapping out the areal distribution of reservoir compaction (Hatchell Bourne, 2005). addition to shifts, some work has shown that horizontal or lateral shifts also present in data (e.g., Hall et al., 2002; Hall, 2006; Hale, 2007) these large, directed outward from reservoir, carry information different shifts. Although methods extracting have been described, origin...

Down-hole seismic monitoring in the form of time-lapse 3D VSP has emerged as a technically viable alternative areas where reservoir via conventional 4D recorded on surface is not expected to be successful due low SNR or insufficient repeatability. But even after technical case proven for – often major and protracted task larger hurdle remains make technique affordable field-wide scale. We discuss how emerging technology Distributed Acoustic Sensing (DAS) may used clear this both onshore offshore.

We present a method to assess the technical and economic pros cons of various marine seismic monitoring techniques. The results are displayed in Value vs. Cost diagram guide business decisions on type frequency for specifics reservoir planned recovery method. estimation benefit is most useful part exercise, as it forces geophysicists engineers think about uses 4D that may be taken based data. illustrate methodology with an example inspired by pre-salt carbonate fields offshore Brazil....

We present an overview of new low-cost seismic monitoring technologies that may be applied in deepwater Brazil, particular the prolific pre-salt province. start with estimate "size prize" such a program deliver field undergoing WAG recovery, followed by survey geophysical and geodetic areal have been deployed or are under development other areas, including ocean bottom nodes cables, small sources, high resolution 4D, DAS VSP, seabed subsidence. discuss their advantages, limitations,...

The execution of deep neural network (DNN) inference jobs on edge devices has become increasingly popular. Multiple such models can concurrently analyse the on-device data, e.g. images, to extract valuable insights. Prior art focuses low-power accelerators, compressed architectures, and specialized frameworks reduce time single which are resource constrained. However, it is little known how different scheduling policies further improve runtime performance multi-inference without additional...

Federated learning (FL) systems enable multiple clients to train a machine model iteratively through synchronously exchanging the intermediate weights with single server. The scalability of such FL can be limited by two factors: server idle time due synchronous communication and risk becoming bottleneck. In this paper, we propose new architecture, our knowledge, first multi-server system that is entirely asynchronous, therefore addresses these limitations simultaneously. Our solution keeps...

The training of diffusion-based models for image generation is predominantly controlled by a select few Big Tech companies, raising concerns about privacy, copyright, and data authority due to their lack transparency regarding data. To ad-dress this issue, we propose federated diffusion model scheme that enables the independent collaborative without exposing local Our approach adapts Federated Averaging (FedAvg) algorithm train Denoising Diffusion Model (DDPM). Through novel utilization...

Summary Unconventional reservoirs are generally developed using hydraulic fracturing. Having a good understanding of the fracture characteristics helps in optimally and efficiently developing reservoir. Microseismic monitoring has proven to be valuable technique monitor fracturing operations. During treatment fluid is injected reservoir cracks form, which results occurrence microseismic events. The interpretation this events can lead better characteristics. used assess parameters like...

3D VSP has long been viewed as conceptually attractive for illuminating targets under complex overburden. However, the wide-spread use of hindered by cost and risk deploying geophones in a borehole, limited availability accessible wells. These hurdles are largely removed when acquiring downhole seismic with Distributed Acoustic Sensing (DAS) instead geophones. Over past several years Shell acquired more than 20 DAS VSP-s around world to test performance further its development. Early on we...

Seismic applications of Distributed Acoustic Sensing (DAS) have developed significantly in the last several years and more examples high quality DAS VSP products are being published literature, including 3D 4D images (Mateeva et al., 2014; Chalenski 2016; Hance 2016). Nonetheless, limits technology not been reached there is still room for further development. In which direction should these developments go? What that could bring largest value to energy industry? this abstract, we present our...

PreviousNext No AccessSEG Technical Program Expanded Abstracts 2003Combined CFP based global and local velocity analysis applied to North Sea dataAuthors: B. E. CoxP. L. A. WinthaegenD. J. VerschuurB. CoxDelft University of Technology, Lab. Acoustic Imaging Sound Control, P. WinthaegenNetherlands Institute Applied Geoscience TNO, D. VerschuurDelft Technologyhttps://doi.org/10.1190/1.1817770 SectionsAboutPDF/ePub ToolsAdd favoritesDownload CitationsTrack CitationsPermissions...

D-32 TOMOGRAPHICINVERSIONOF 3DFOCUSING OPERATORSUSING DATA-DRIVEN PARAMETERIZATION B.E.COX andD.J. VERSCHUUR DelftUniversityof Technology Lab. of Acoustic Imagingand SoundControl POBox 5046 2600GADelft The Netherlands Summary In the Common Focal Point (CFP) technology estimation focusing operators is separated fromtheestimationofthevelocitymodel. Focusingoperatorsareverysuitableforvelocitymodel because they represent one way traveltimes from reflection points to many surface locations which...