Using Semi-Supervised Convolutional Neural Networks for Porosity Modeling Over a Fluvio-Deltaic Triassic Gas Field

2021 ◽  
Author(s):  
Haibin Di ◽  
Aria Abubakar
Keyword(s):  
Supervised Learning ◽  
Seismic Wave Propagation ◽  
Well Logs ◽  
Seismic Survey ◽  
Rock Properties ◽  
3D Seismic ◽  
Gas Field ◽  
Rock Property ◽  
Additional Information ◽  
Property Estimation

Abstract Robust estimation of rock properties, such as porosity and density, from geophysical data, i.e. seismic and well logs, is essential in the process of subsurface modeling and reservoir engineering workflows. Such properties are accurately measured in a well; however, due to high cost of drilling, such direct measurements are limited in amount and sparse in space within a study area. On the contrary, 3D seismic data illuminates the subsurface of the study area throughoutly by seismic wave propagation; however, the connection between seismic signals and rock properties is implicit and unknown, causing rock property estimation from seismic only to be a challenging task and of high uncertainty. An integration of 3D seismic with sparse wells is expected to eliminate such uncertainty and improve the accuracy of static reservoir property estimation. This paper investigates the application of a semi-supervised learning workflow to estimate porosity from a 3D seismic survey and 36 wells over a fluvio-deltaic Triasic gas field. The workflow is performed in various scenarios, including purely from seismic amplitude, incorporating a rough 6-layer deposition model as a constraint, and training with varying numbers of wells. Good match is observed between the machine prediction and the well logs, which verifies the capability of such semi-supervised learning in providing reliable seismic-well integration and delivering robust porosity modeling. It is concluded that the use of available additional information helps effectively constrain the learning process and thus leads to significantly improved lateral continuity and reduced artifacts in the machine learning prediction. The semi-supervised learning can be readily extended for estimating more properties and allows nearly one- click solution to obtain 3D rock property distribution in a study area where seismic and well data is available.

Estimating subsurface properties using a semi-supervised neural networks approach

Geophysics ◽  
2021 ◽  
pp. 1-38
Author(s):  
Haibin Di ◽  
Aria Abubakar
Keyword(s):  
Supervised Learning ◽  
Epistemic Uncertainty ◽  
Mapping Function ◽  
Linear Mapping ◽  
Well Logs ◽  
Seismic Survey ◽  
Rock Properties ◽  
Limited Bandwidth ◽  
Challenging Tasks ◽  
Noise Robust

Estimating static rock properties (e.g., density and porosity) from seismic and well logs is one of the essential but challenging tasks in subsurface interpretation and characterization. To compensate for the sparsity of well logs and the limited bandwidth of seismic data, a semi-supervised learning workflow is presented for efficiently integrating seismic and logs and simultaneously estimating multiple subsurface properties. It consists of two components: (1) unsupervised seismic feature engineering and (2) supervised seismic-well integration, each of which is implemented as a convolutional neural network (CNN). Compared to most of the existing methods, it advances in three aspects. First, it allows the use of local 3D seismic patterns for building an optimal non-linear mapping function with 1D logs, which is more noise robust and significantly improves the lateral consistency of machine prediction throughout the entire seismic survey. Second, it is capable of automatically bridging the gap of vertical resolution between seismic and well logs, which simplifies the workflow of data preparation, such as log upscaling. Additionally, it enables Monte Carlo (MC) dropout-based epistemic uncertainty analysis. The performance of the proposed solution is evaluated through two examples, relative acoustic impedance and porosity estimation in a synthetic PreSDM dataset of 36 pseudo wells and sonic and density estimation in the Groningen dataset of 375 wells. The good match between the machine predictions and the actual measurements demonstrates the capability of the proposed semi-supervised learning in providing reliable seismic and well integration and delivering robust estimation of subsurface properties, including those of a relatively weak physical link with seismic, such as density and porosity.

DEMETER HIGH RESOLUTION 3D SEISMIC SURVEY—REVITALISED DEVELOPMENT AND EXPLORATION ON THE NORTH WEST SHELF, AUSTRALIA

2006 ◽  
Vol 46 (1) ◽  
pp. 101 ◽  
Author(s):  
K.J. Bennett ◽  
M.R. Bussell
Keyword(s):  
High Resolution ◽  
Seismic Data ◽  
New Technologies ◽  
Seismic Survey ◽  
Future Application ◽  
3D Seismic ◽  
Gas Field ◽  
North West ◽  
The North ◽  
Exploration And Development

The newly acquired 3,590 km2 Demeter 3D high resolution seismic survey covers most of the North West Shelf Venture (NWSV) area; a prolific hydrocarbon province with ultimate recoverable reserves of greater than 30 Tcf gas and 1.5 billion bbls of oil and natural gas liquids. The exploration and development of this area has evolved in parallel with the advent of new technologies, maturing into the present phase of revitalised development and exploration based on the Demeter 3D.The NWSV is entering a period of growing gas market demand and infrastructure expansion, combined with a more diverse and mature supply portfolio of offshore fields. A sequence of satellite fields will require optimised development over the next 5–10 years, with a large number of wells to be drilled.The NWSV area is acknowledged to be a complex seismic environment that, until recently, was imaged by a patchwork of eight vintage (1981–98) 3D seismic surveys, each acquired with different parameters. With most of the clearly defined structural highs drilled, exploration success in recent years has been modest. This is due primarily to severe seismic multiple contamination masking the more subtle and deeper exploration prospects. The poor quality and low resolution of vintage seismic data has also impeded reservoir characterisation and sub-surface modelling. These sub-surface uncertainties, together with the large planned expenditure associated with forthcoming development, justified the need for the Demeter leading edge 3D seismic acquisition and processing techniques to underpin field development planning and reserves evaluations.The objective of the Demeter 3D survey was to re-image the NWSV area with a single acquisition and processing sequence to reduce multiple contamination and improve imaging of intra-reservoir architecture. Single source (133 nominal fold), shallow solid streamer acquisition combined with five stages of demultiple and detailed velocity analysis are considered key components of Demeter.The final Demeter volumes were delivered early 2005 and already some benefits of the higher resolution data have been realised, exemplified in the following:Successful drilling of development wells on the Wanaea, Lambert and Hermes oil fields and identification of further opportunities on Wanaea-Cossack and Lambert- Hermes;Dramatic improvements in seismic data quality observed at the giant Perseus gas field helping define seven development well locations;Considerably improved definition of fluvial channel architecture in the south of the Goodwyn gas field allowing for improved well placement and understanding of reservoir distribution;Identification of new exploration prospects and reevaluation of the existing prospect portfolio. Although the Demeter data set has given significant bandwidth needed for this revitalised phase of exploration and development, there remain areas that still suffer from poor seismic imaging, providing challenges for the future application of new technologies.

The Indefatigable Field, Blocks 49/18, 49/19, 49/23, 49/24, UK North Sea

2003 ◽  
Vol 20 (1) ◽  
pp. 741-747 ◽  
Author(s):  
C. W. McCrone ◽  
M. Gainski ◽  
P. J. Lumsden
Keyword(s):  
North Sea ◽  
Seismic Survey ◽  
3D Seismic ◽  
Gas Field ◽  
Recoverable Reserves ◽  
Phases Of Development ◽  
Four Blocks ◽  
Sandstone Formation ◽  
Well Data ◽  
The Uk

abstractIndefatigable is a mature dry gas field on the northeastern margin of the UK Southern North Sea Rotliegend Play fairway. The field was discovered, 49/18-1, by the Amoco operated group in 1966 and subsequent appraisal drilling established that the field extended over four blocks (i.e. 49/18, 49/19, 49/23 & 49/24). There have been several phases of development, initial production concentrated on the main horst block with first gas in 1971, followed by the west flank area in 1977/78. Then in 1987/88 the SW and SE Indefatigable satellite accumulations were brought on-stream.The Rotliegend Leman Sandstone Formation reservoir primarily consists of stacked aeolian dune sandstones (150-400 ft) of good reservoir quality (porosity 15%, permeability 100-1000 mD). However, the integration of the 1992/93 3D seismic survey, well data, reservoir pressure and production data has lead to a much more complex view of the field with 11 gas-water contacts and 15 reservoir compartments.This has resulted in an upward revision of the gas initially-in-place from 5.2 to 5.6 TCF and recoverable reserves from 4.4 to 4.7 TCF. Current work is focused on maximizing recovery from the various reservoir compartments and accessing this additional potential.

The Johnston Gas Field, Blocks 43/26a, 43/27a, UK Southern North Sea

2003 ◽  
Vol 20 (1) ◽  
pp. 749-759 ◽  
Author(s):  
David E. Lawton ◽  
Paul P. Roberson
Keyword(s):  
North Sea ◽  
Development Plan ◽  
Seismic Survey ◽  
3D Seismic ◽  
Seismic Attribute ◽  
Gas Field ◽  
Southern North Sea ◽  
North East ◽  
The North ◽  
The Uk

abstractThe Johnston Field is a dry gas accumulation located within blocks 43/26a and 43/27a of the UK Southern North Sea. The discovery well was drilled in 1990 and after the drilling of one appraisal well in 1991, a development plan was submitted and approved in 1993. Initially two development wells were drilled from a four slot sub-sea template, with commercial production commencing in October 1994. A further horizontal development well was added to the field in 1997.The field has a structural trap, fault bounded to the SW and dip-closed to the north, east and south. This field geometry has been established using high quality 3D seismic data, enhanced by seismic attribute analysis. The sandstone reservoir interval consists of the Early Permian, Lower Leman Sandstone Formation of the Upper Rotliegend Group. This reservoir consists of a series of interbedded aeolian dune, fluvial, and clastic sabkha lithofacies. The quality of the reservoir is variable and is principally controlled by the distribution of the various lithofacies. The top seal and fault bounding side seal are provided by the overlying clay stone of the Silverpit Shale Formation and the evaporite dominated Zechstein Supergroup.The field has been developed using a phased development plan, with the acquisition of a 3D seismic survey allowing for the optimized drilling of a high deliverability horizontal well.Current mapped gas initially-in-place estimates for the field are between 360 and 403 BCF, with an estimated recovery factor of between 60 and 75%.

scholarly journals The Seismic Interpretation for (X) Oil Field Depending on 3D Seismic Data of Nearby Oil Field, Southern Iraq

2021 ◽  
Vol 11 (4) ◽  
pp. 36-50
Author(s):  
Wessam Abdul Abbas Alhammod ◽  
Ban Talib Aljizani
Keyword(s):  
Seismic Data ◽  
Velocity Model ◽  
Oil Field ◽  
Seismic Survey ◽  
Rock Properties ◽  
3D Seismic ◽  
Water Contact ◽  
The Time Domain ◽  
Structural Closure ◽  
Southern Iraq

This research focused on using seismic data to review the structure of the (X) Oil Field, located 40 km SW of Basrah, Southern Iraq. The study utilises a 3D seismic survey conducted during 2011-2012, covering the (Y) Oil Field 2 km to the west, and with partial coverage across (X), to map the Top Zubair reflector. Seismic rock properties analysis was conducted on key (X) Oil Field wells and used to tie the Top Zubair reflector on (X) Oil Field. The reflector was mapped within the time domain using DecisionSpace Software, and then converted to depth using a velocity model. The depth structure map was then compared to the original oil water contact (OOWC) across the fields to understand the potential structural closure of the Top Zubair reservoir in both fields.

SEISMIC TO SIMULATION: INTEGRATED RESERVOIR MODEL FOR THE PATRICIA BALEEN GAS FIELD

2002 ◽  
Vol 42 (1) ◽  
pp. 83
Author(s):  
P. Fink ◽  
M. Adamson ◽  
F. Jamal ◽  
C. Stark
Keyword(s):  
Water Saturation ◽  
Subsequent Development ◽  
Seismic Survey ◽  
3D Seismic ◽  
Reservoir Model ◽  
Gas Field ◽  
Facies Model ◽  
Time Migration ◽  
Gas Fields ◽  
Elastic Impedance

The Patricia and Baleen offshore gas fields are located in the northeastern part of the Gippsland Basin in southeast Australia. Although discovered by two exploration wells almost a quarter of a century ago, the two gas fields only recently have again become the focus of appraisal and subsequent development activity through OMV’s acquisition of Cultus in 1999.After the drilling of a successful appraisal well in late 1999, a high resolution 3D seismic survey was acquired in early 2000. No further data acquisition will be undertaken. Special emphasis was therefore put on maximising the value of the 3D dataset by integrating the PreSTM (Pre. Stack Time Migration) seismic and several Elastic Impedance attributes with all other available subsurface data prior to building a sophisticated stochastic reservoir model for simulation.This paper describes how the integration of leading edge seismic technology with unconventional geological modelling was used to overcome a number of major challenges in order to build a coherent static reservoir model and constrain resource uncertainty given the limited amount of wireline and core data:A large proportion of the gas fields is strongly affected by seismic tuning which would introduce significant uncertainties on GRV and GWC estimations from seismic, if not accounted for properly. Likewise all seismic and to a somewhat lesser extent basic inversion based attributes used for reservoir property determination are strongly affected by this geophysical artefact: These challenges (and seismic pitfalls) were met by inverting the conventional 3D seismic for Pand S- wave impedances and generating a set of Elastic Impedance Cubes, difference cubes and LRM Cubes (standing for the elastic constants Lambda (λ), Rho (ρ) and Mhu (μ)), defining petroacoustic properties of the reservoir rocks. These cubes were tested for mathematical dependency and used for the conditioning of the facies and porosity models.The glauconitic Gurnard reservoir contains a high fraction of conductive minerals and is almost completely bioturbated. Conventional saturation estimations based on wireline-logs and conventional sequence stratigraphic facies description did not deliver a reliable picture: Instead a facies model based on ichnofabric analysis was built and constrained with data available at the three well locations. Saturation height functions were applied separately for each facies type. The Rho-Lambda (ρλ) cube was used to condition facies distribution away from the wells.More specifically, the results presented in the paper are:Elastic Impedance inversion provided vertical seismic resolution in the order of 4 m to 10 m, thereby allowing a more accurate seismic estimation of GRV and the GWC. Lamesf Constants were extracted from seismic in order to classify lithology.A realistic facies model was built utilizing the Rho- Lambda (ρλ) cube combined with ichnofabric analysis tied to permeability and water saturation distributions.Elastic Impedance Difference cubes were successfully calculated to eliminate tuning even further and condition the stochastic porosity model.Connected volume maps were used to optimise the production well pathsThe GIIP upside volume has been upgraded compared to that based on an earlier simplistic geological reservoir model used for simulation. A more realistic P10/P90 reserves range is now supported by a number of deterministic and stochastic reservoir models.

BEATING THE ODDS AT CASINO!—A SMALL AUSTRALIAN’S EXAMPLE OF RISK MANAGEMENT

2003 ◽  
Vol 43 (1) ◽  
pp. 85
Author(s):  
D.J. Poynton
Keyword(s):  
Risk Management ◽  
Financial Risk ◽  
Time Window ◽  
Weather Conditions ◽  
Seismic Survey ◽  
3D Seismic ◽  
Gas Field ◽  
Stakeholder Consultation ◽  
Operational Risks ◽  
Environmentally Sensitive

Strike Oil was a very small unlisted Australian company with a capitalisation of less than A$10 million when it decided to bid for block V98-4 (now VIC/P44) in the offshore Otway Basin in early 1999.Block V98-4 met Strike Oil’s gas strategy of pursuing opportunities in basins close to infrastructure and markets in the eastern states of Australia.Prior to making the bid Strike Oil identified the geological, financial and operational risks associated with exploring the permit, especially with regard to conducting a 3D seismic survey in the environmentally sensitive and sometimes hostile Bass Strait. This led to the implementation of, and adherence to, a comprehensive risk management plan.The geological risks were addressed by acquiring 3D seismic and conducting an analysis of the amplitudes and AVO responses associated with nearby gas discoveries and dry holes.Management of the financial risk centred firstly around not overbidding and secondly finding a farmee who could add value to the permit during both the exploration and exploitation phases.The operational risks were all associated with conducting the Casino 3D seismic survey. Local environmental considerations, particularly in relation to migratory whale species and the seasonal activities of local fishermen, meant there was only a six weeks’ time window available for unhindered operations. This window also coincided with the spring gale season, when weather conditions can stop marine operations.The use of experienced personnel, early stakeholder consultation, and the use of contingency plans, enabled Strike Oil to achieve its objectives under adverse conditions. The Casino 3D seismic survey, despite the odds, was completed on time, under budget, and with less than 7% infill, while still delivering high quality data.The farmout, the acquisition and processing of the 3D seismic data, and the discovery and appraisal of the Casino gas field were all achieved within 14 months.

Attribute selection in seismic facies classification: Application to a Gulf of Mexico 3D seismic survey and the Barnett Shale

2019 ◽  
Vol 7 (3) ◽  
pp. SE281-SE297 ◽  
Author(s):  
Yuji Kim ◽  
Robert Hardisty ◽  
Kurt J. Marfurt
Keyword(s):  
Gulf Of Mexico ◽  
Supervised Learning ◽  
Depositional Environments ◽  
Machine Learning Algorithms ◽  
Attribute Selection ◽  
Seismic Facies ◽  
Seismic Survey ◽  
Barnett Shale ◽  
3D Seismic ◽  
Facies Classification

Automated seismic facies classification using machine-learning algorithms is becoming more common in the geophysics industry. Seismic attributes are frequently used as input because they may express geologic patterns or depositional environments better than the original seismic amplitude. Selecting appropriate attributes becomes a crucial part of the seismic facies classification analysis. For unsupervised learning, principal component analysis can reduce the dimensions of the data while maintaining the highest variance possible. For supervised learning, the best attribute subset can be built by selecting input attributes that are relevant to the output class and avoiding using redundant attributes that are similar to each other. Multiple attributes are tested to classify salt diapirs, mass transport deposits (MTDs), and the conformal reflector “background” for a 3D seismic marine survey acquired on the northern Gulf of Mexico shelf. We have analyzed attribute-to-attribute correlation and the correlation between the input attributes to the output classes to understand which attributes are relevant and which attributes are redundant. We found that amplitude and texture attribute families are able to differentiate salt, MTDs, and conformal reflectors. Our attribute selection workflow is also applied to the Barnett Shale play to differentiate limestone and shale facies. Multivariate analysis using filter, wrapper, and embedded algorithms was used to rank attributes by importance, so then the best attribute subset for classification is chosen. We find that attribute selection algorithms for supervised learning not only reduce computational cost but also enhance the performance of the classification.

Benefits of Ultra-Dense 3D Spatial Sampling for Seismic Processing and Interpretation

2021 ◽  
Author(s):  
Mohamed Mahgoub ◽  
Guillaume Cambois ◽  
James Cowell ◽  
Suaad Khoori
Keyword(s):  
Model Building ◽  
Velocity Model ◽  
Sensor Data ◽  
Seismic Survey ◽  
Rock Properties ◽  
3D Seismic ◽  
Low Frequencies ◽  
High Resolution Data ◽  
Structural Interpretation ◽  
Seismic Acquisition

Abstract The advances in seismic acquisition systems, especially onshore nodes, have made it possible to acquire ultra-dense 3D surveys at a reasonable cost. This new design enables accurate processing sequences that deliver higher resolution images of the subsurface. These images in turn lead to enhanced structural interpretation and better prediction of rock properties. In 2019, ADNOC and partners acquired an 81 square kilometer ultra-high density pilot survey onshore Abu Dhabi. The receivers were nimble nodes laid out on a 12.5x12.5m grid, which recorded continuously and stored the data on a memory chip. The sources were heavy vibrators sweeping the 2-110 Hz frequency range in 14 seconds on a 12.5x100m grid. 184 million traces per square kilometers did make such small area, the densest 3D seismic survey ever recorded. The single sensor data were expectedly very noisy and the unconstrained simultaneous shooting required elaborate deblending, but we managed these steps with existing tools. The dense 3D receiver grid actually enabled the use of interferometry-based ground-roll attenuation, a technique that is rarely used with conventional data due to inadequate sampling, but that resulted in increased signal-to-noise ratio. The data were migrated directly to depth using a velocity model derived after five iterations of tomographic inversion. The final image gathers were made of 18 reciprocal azimuths with 12.5m offset increment, resulting in 5,000 fold on a 6.25x6.25m grid. The main structural interpretation was achieved during the velocity model building stage. Key horizons were picked after the tomographic iterations and the velocity model was adjusted so that their depth matched the well markers. Anisotropic parameters were adjusted to maintain gather flatness and the new model was fed to the next iteration. This ultimately resulted in flat image gathers and horizons that tied to the wells. The final high-resolution data provided a much crisper image of the target clinoforms and faults. This resulted in a more detailed interpretation of the reservoirs. The data was subjected to pre-stack stratigraphic inversion. The availability of low frequency signal (down to 3 Hz) means that less well constraints are needed for the inversion. Preliminary results are particularly encouraging. Amplitude variations with azimuth have yet to be analyzed but data density bodes very well for the process. Ultra-dense 3D seismic acquisition is feasible and results in a step change in image quality. Structural and stratigraphic interpretation provided a more detailed image of faults and clinoforms. Stratigraphic inversion benefited from the low frequencies of the vibrator source and the increased spatial resolution.

PORT CAMPBELL REVIEWED: METHANE AND CHAMPAGNE

1995 ◽  
Vol 35 (1) ◽  
pp. 418 ◽  
Author(s):  
J.D. Foster ◽  
A.J. Hodgson
Keyword(s):  
Late Cretaceous ◽  
Seismic Survey ◽  
Data Sets ◽  
3D Seismic ◽  
Gas Field ◽  
Natural Gas Markets ◽  
The North ◽  
Industrial Markets ◽  
Western Victoria ◽  
Gas Fields

Gas fields in the Port Campbell Embayment cur­rently supply all the natural gas markets (non-LPG) in western Victoria as well as commercial quanti­ties of carbon dioxide (C02) to industrial markets. Initial discoveries made between 1979 and 1981 were brought on-stream in 1986 with production from the North Paaratte field. Another substantial discovery was made in 1988, the Iona gas field, followed by the Boggy Creek C02 field in 1991, then the My lor and Langley fields in 1994. Discovery of Mylor marked the first recovery of oil from the Late Cretaceous Waarre Formation. Extensive 2D seis­mic data sets have been recorded in the region since 1979, and the first 3D seismic survey in the Otway Basin was carried out in 1993 extending beyond the area of the initial discoveries. No data on the fields have been published for nearly a decade and little detail about the structural and stratigraphic geol­ogy of the Late Cretaceous in the area has been documented. Summaries of the fields are presented incorporating many insights gained from interpre­tation of the 3D seismic data and its verification by the 'rotary lie detector'.

Sign in / Sign up

Export Citation Format

Share Document