Renjun Wen

Renjun Wen

Calgary, Alberta, Canada
2K followers 500+ connections

Join to follow

About

I am a mathematical geologist by training. My passion is to work with like minded people…

Articles by Renjun

  • AttributeStudio 8.4 Released
    Aug 28, 2020

    AttributeStudio 8.4 Released

    Explainable Machine Learning Workflows for Quantitative Integration of Seismic Attributes and Well Data CALGARY…

Activity

Join now to see all activity

Experience

  • SubsurfaceAI Inc Graphic

    SubsurfaceAI Inc

    Calgary, Alberta, Canada

  • -

    Calgary, Alberta, Canada

  • -

  • -

  • -

    Trondheim Area, Norway

  • -

    Yumen, Gansu Province, China

Education

  • Norwegian University of Science and Technology (NTNU) Graphic

    Norwegian University of Science and Technology (NTNU)

    -

    Activities and Societies: Geoscience software for reservoir characterization. Process-oriented modeling of reservoir heterogeneity. Members of AAPG, CSEG, CSPG, EAGE, IAMG, SEG.

  • -

  • -

    Jianghan Petroleum Institute is now part of Yangzhi Univeristy

Publications

  • Seismic Meta-Attributes and the Illumination of the Internal Reservoir Architecture of a Deepwater Synthetic Channel Model

    AAPG Search and Discovery

    We applied workflows of seismic attribute analysis and facies classification to a synthetic 3-D seismic volume, which was generated from velocity and density volumes in a 3-D turbidite facies model. The turbidite facies model was simulated based on a process-oriented method and is able realistically to capture major features in a turbidite environment. Because we had a priori knowledge of the geologic structure in the synthetic post-stack seismic volume, we were able to examine the efficacy of…

    We applied workflows of seismic attribute analysis and facies classification to a synthetic 3-D seismic volume, which was generated from velocity and density volumes in a 3-D turbidite facies model. The turbidite facies model was simulated based on a process-oriented method and is able realistically to capture major features in a turbidite environment. Because we had a priori knowledge of the geologic structure in the synthetic post-stack seismic volume, we were able to examine the efficacy of different seismic attribute analysis methods in delineating turbidite facies in the deepwater system. Based on attribute-analysis results of this synthetic seismic volume, we investigated the use of new seismic meta-attribute calculations for the detection of internal reservoir characteristics within deepwater reservoirs. New meta-attributes were created for characterizing features such as: the lateral continuity of amplitude response, the lateral continuity of similarly thick beds, and multiple combinations of geometric and response, energy, and instantaneous attributes, which are coined “ponding attributes.” The proposed meta-attributes were also used as input to neural-network seismic facies classification, which resulted in a closer match to the “ground-truth” facies model than conventional attributes.

    See publication

  • Interactive Seismic Attribute Analysis for Reservoir Characterization

    In "Attributes: New Views on Seismic Imaging -Their Use in Exploration and Production": 31st Annual GCSSEPM Research Conference Proceedings, Vol. 31, 2011, p. 473-495.

    Being “interactive” is not only important for traditional seismic interpretation, which maps subsurface structural and stratigraphic features; it should also be an essential feature in the seismic attribute analysis for reservoir characterization and prediction. In this paper we present interactive workflows that integrate major steps in seismic attribute analysis, namely attribute calculation, visualization, calibration, classification, and prediction. Interactive analysis workflows are…

    Being “interactive” is not only important for traditional seismic interpretation, which maps subsurface structural and stratigraphic features; it should also be an essential feature in the seismic attribute analysis for reservoir characterization and prediction. In this paper we present interactive workflows that integrate major steps in seismic attribute analysis, namely attribute calculation, visualization, calibration, classification, and prediction. Interactive analysis workflows are demonstrated as more effective and powerful than traditional linear workflows.

    We introduce the “strata-grid” as the major object for analyzing attributes in the interactive attribute workflow. A strata-grid defines a target stratigraphic volume, based on stratigraphic relationships, that is proportional, top-conformable, or bottom-conformable. Seismic attribute features are better delineated on stratigraphic slices within a strata-grid, rather than through traditional time slices or simple horizon slices. The strata-grid makes the interaction with the data possible, not only for visualizing the data but also for quantitatively analyzing attributes for large 3D seismic surveys – without handling the whole data volume.

    See publication

  • 3D geologic modelling of channellized reservoirs: applications in seismic attribute facies classification

    First Break, Vol. 23, p. 71 – 78.

    Geological models are usually used qualitatively in seismic interpretation. This paper illustrates that quantitative representations of detailed geological models can significantly enhance seismic attribute interpretation through facies classification. When applying seismic attribute classification to reservoir facies mapping, one often faces such typical questions as: ■ Which attributes should be used as input to classification? ■ How many classes should be used in the unsupervised…

    Geological models are usually used qualitatively in seismic interpretation. This paper illustrates that quantitative representations of detailed geological models can significantly enhance seismic attribute interpretation through facies classification. When applying seismic attribute classification to reservoir facies mapping, one often faces such typical questions as: ■ Which attributes should be used as input to classification? ■ How many classes should be used in the unsupervised classification method? ■ How many levels of hierarchy should be selected in the hierarchical classification method? ■ Does the seismic facies correspond to the geological facies? ■ How can attribute-derived facies models be validated? There are no unique and easy answers to the above questions. In this study, we aim to create a more accurate representation of the reservoir by using 3D synthetic Earth models to guide seismic attribute classification. We consider a channellized reservoir for which seismic attribute analysis has proven to be very useful, but results can be difficult to interpret. The next section describes a 3D stratigraphic modelling approach for the channellized reservoir. The major channel components and parameterizations are illustrated with examples. This is followed by a summary of seismic attribute analysis and classification workflow applied to a synthetic seismic volume. Results of attribute classifications using a self-organized map (SOM) (Kohonen, 1989) and waveform correlation maps are compared in relation to different input attributes and classification parameters.

    See publication

  • Petrophysical characterization of a heterolithic tidal reservoir interval using a process-based modelling tool

    Petroleum Geoscience, Vol. 11, pp. 71-28

    Heterolithic lithofacies in the Jurassic Tilje Formation, offshore mid-Norway, consist of three components – sand, silt and mud intercalated at the centimetre scale – and are generally difficult to characterize petrophysically with core and wireline data. A near-wellbore model of the lower part of the Tilje Formation in the Heidrun Field is constructed to illustrate the application of these results to formation evaluation studies. The sedimentological model is developed by detailed…

    Heterolithic lithofacies in the Jurassic Tilje Formation, offshore mid-Norway, consist of three components – sand, silt and mud intercalated at the centimetre scale – and are generally difficult to characterize petrophysically with core and wireline data. A near-wellbore model of the lower part of the Tilje Formation in the Heidrun Field is constructed to illustrate the application of these results to formation evaluation studies. The sedimentological model is developed by detailed parameterization of a cored well interval and the petrophysical properties are based on core plug data, taking into account sampling bias and length scale. The variation in petrophysical properties as a function of sample volume is examined by calculating the representative elementary volume. The sensitivity of the representative permeability values to the contrast between the three components is studied and gives a better understanding of the flow behaviour of this system. These results are used to rescale the core plug data to a representative value and thereby quantify the uncertainty associated with the wireline-based estimates of porosity and horizontal permeability and to give an improved estimate of the kv/kh ratio.

    Other authors
    See publication

  • Vertical Permeability estimation in tidal deltaic reservoir systems

    Petroleum Geoscience, Vol. 11, pp. 29-36.

    A method for estimation of vertical permeability in heterolithic tidal deltaic sandstones is proposed. Three-dimensional, stochastic, process-based models of sedimentary bedding are used to give estimates for the effective permeability of heterolithic tidal sandstone units where heterogeneities in the sandstone and mudstone components are evaluated explicitly.

    Other authors
    See publication

  • Multi-scale Characterisation, and Modelling of Heterolithic Tidal Systems, Offshore Mid-Norway

    in "Advanced Reservoir Characterisation for the 21st Century" T. F. Hentz (ed), Proceedings of GCSSEPM Research Conference: December 1999, Houston, (1999) 193-204

    Other authors
    See publication

  • Three-Dimensional Simulation of Small Scale Heterogeneity in Tidal Deposits - A Process Based Stochastic Simulation Method.

    Proc. 4 th Annual Conf. Of the International Association for Mathematical Geology, A. Boccianti, G. Nardi and R. Potenza (Eds.,) Naples, (1998), p. 129-134.

    Small-scale heterogeneity at the bed-set scale (ca 1 mm to 1 m) represents an important
    target for reservoir modelling. In particular, tidal deposits, which are typified by bi-directional
    currents and deposition of alternating mud and sand layers, have a high degree of architectural
    complexity at the bed-set scale. To quantitatively assess the impact of these small-scale
    heterogeneities on reservoir performance, we need a method to realistically reproduce 3-D
    permeability…

    Small-scale heterogeneity at the bed-set scale (ca 1 mm to 1 m) represents an important
    target for reservoir modelling. In particular, tidal deposits, which are typified by bi-directional
    currents and deposition of alternating mud and sand layers, have a high degree of architectural
    complexity at the bed-set scale. To quantitatively assess the impact of these small-scale
    heterogeneities on reservoir performance, we need a method to realistically reproduce 3-D
    permeability distributions. In this paper, we present a process-based stochastic simulation
    method to simulate 3-D permeability distributions in tidal bedding structures such as flaser and
    wavy bedding. Our process-based stochastic simulation method differs from existing
    geostatistical methods in that the formation process of bedding structures has been included in
    the simulation algorithm. The 3-D bedding geometry and grain-size distribution can be
    realistically reproduced. The new method is different from previous process-response models
    because we use geostatistical terms to describe the variation of bedding surfaces. Variations of
    bedform migration and deposition speed, as well as permeability and porosity distributions
    along individual bedding surfaces are modelled. In our method, we model bedding surfaces by
    a 2-D sine function and a stochastic component described by 2-D Gausssian random functions.
    Four stages of deposition, migration and erosion processes have been included to model each
    tidal cycle of deposition. To generate the correct 3-D geometry, we need to model variations of
    bedform migration speed, direction and deposition rate at each stage. Because of the periodic
    nature of flow in tidal deposits, these variations are modelled by a sine function. In addition, a
    1-D Gaussian function is used to provide flexibility and introduce variability...

    Other authors

  • Uncertainty in fractal dimension estimated from power spectral and variograms

    Mathematical Geology, vol. 29, no. 6, p. 727-753

    The reliability of using fractal dimension (D) as a quantitative parameter to describe geological variables is dependent mainly on the accuracy of estimated D values from observed data. Two widely used methods for the estimation of fractal dimensions are based on fitting a fractal model to experimental variograms or power-spectra on a log-log plot. The purpose of this paper is to study the uncertainty in the fractal dimension estimated by these two methods. The results indicate that both…

    The reliability of using fractal dimension (D) as a quantitative parameter to describe geological variables is dependent mainly on the accuracy of estimated D values from observed data. Two widely used methods for the estimation of fractal dimensions are based on fitting a fractal model to experimental variograms or power-spectra on a log-log plot. The purpose of this paper is to study the uncertainty in the fractal dimension estimated by these two methods. The results indicate that both spectrum and variogram methods result in biased estimates of the D value. Fractal dimension calculated by these two methods for the same data will be different unless the bias is properly corrected. The spectral method results in overestimated D values. The variogram method has a critical fractal dimension, below which overestimation occurs and above which underestimation occurs. On the bases of 36,000 simulated realizations we propose empirical formulae to correct for biases in the spectral and variogram estimated fractal dimension. Pitfalls in estimating fractal dimension from data contaminated by white noise or data having several fractal components have been identified and illustrated by simulated examples

    Other authors
    • Richard Sinding-Larsen
    See publication

  • Computer enhanced GLORIA side-scan sonar images of the surface of the Mississippi Fan

    Atlas of Deep Water Environments: Architectural Style in Turbidites Systems, Chapman & Hall, London, p. 297-299.

    Other authors
    • Richard Sinding-Larsen
    • N. H., Kenyon
    See publication

  • Mapping oil seeps on the sea floor by GLORIA side-scan sonar Images  a case study from the Northern Gulf of Mexico

    Nonrenewable Resources, vol. 5, no. 3, p. 141-154.

    An oil seep site in the northern Gulf of Mexico is characterized by high backscattering levels on the GLORIA (Geological Long-Range Inclined Asdic) side-scan sonar images against a background of low backscattering. The high backscattering from the oil-seep area are most likely caused by a combination of small-scale roughness and porosity reduction due to the precipitation of CaCO3 formed during biodegradation of the oil-seep. Geostatistical methods have been applied to analyze side-scan images…

    An oil seep site in the northern Gulf of Mexico is characterized by high backscattering levels on the GLORIA (Geological Long-Range Inclined Asdic) side-scan sonar images against a background of low backscattering. The high backscattering from the oil-seep area are most likely caused by a combination of small-scale roughness and porosity reduction due to the precipitation of CaCO3 formed during biodegradation of the oil-seep. Geostatistical methods have been applied to analyze side-scan images from both oil seep and nonseep areas. The results show that GLORIA images from oil seep areas can be distinguished from nonseep areas in terms of local histograms, variograms, and textural patterns. Pixels from seep and nonseep areas cluster into distinct groups on a textural feature plot. GLORIA side-scan images could be used as a reconnaissance tool to delineate oil and gas seep sites on the seafloor and thereby, reduce the dry-hole risk of petroleum exploration in deep-water frontier areas.

    Other authors
    • Richard Sinding-Larsen
    See publication

  • Nonparametric geostatistical studies of fault parameters in the Snorre Area, North Sea

    in J. O. Aasen, E. Berg, A. T. Buller, O. Hjelmeland, J. Kleppe and Torsæter O. (eds.), North Sea Oil and Gas Reservoirs - III, Kluwer Academic Publishers, p. 167-172.

    Other authors
    • Sinding-Larsen
    See publication

  • Probability kriging of sub-seismic fault throws with multifractal distributions

    Dimitrakopoulos R. (ed.), Geostatistics for the Next Century, Kluwer Academic Publishers, p. 488-497

    Faults with throws less than the resolution limit of seismic data (ca.15 meters) are numerous in faulted hydrocarbon reservoirs, but are not observable on a reservoir scale. Estimation of sub-seismic fault throws from seismically observable fault throws can be made using various methods under different assumptions. The purpose of this paper is to evaluate the efficiency of the probability kriging approach in estimating multifractally distributed fault throws.
    A fractal distribution may be…

    Faults with throws less than the resolution limit of seismic data (ca.15 meters) are numerous in faulted hydrocarbon reservoirs, but are not observable on a reservoir scale. Estimation of sub-seismic fault throws from seismically observable fault throws can be made using various methods under different assumptions. The purpose of this paper is to evaluate the efficiency of the probability kriging approach in estimating multifractally distributed fault throws.
    A fractal distribution may be fitted to fault-throw data across many scales of magnitude (103to 103m) for faults observed on cores, seismic sections, and outcrops. Such a distribution however only characterizes the bulk property of fault throws in an area. As the fractal dimension of fault-throws are spatially dependant, a multifractal model therefore appears to be a more realistic description of fault throws.
    A synthetic data base was simulated, based on a multifractal model whose parameters were estimated from real data. Using samples from the synthetic data, we estimated the conditional probability of having a magnitude of sub-seismic fault-throws less than selected threshold values using probability kriging by taking into account both the fractal distribution and biased sampling of throws. The results indicated that a multifractal distribution of fault throws can be used as a theoretical basis for probability kriging in the estimation of the local distribution of fault-throws.

    Other authors
    • Richard Sinding-Larsen
    See publication

  • Image filtering by factorial kriging  sensitivity analysis and application to GLORIA side-scan sonar images

    Mathematical Geology, vol. 29, p. 433-468, no. 4.

    Factorial Kriging (FK) is a data- dependent spatial filtering method that can be used to remove both independent and correlated noise on geological images as well as to enhance lineaments for subsequent geological interpretation. The spatial variability of signal, noise, and lineaments, characterized by a variogram model, have been used explicitly in calculating FK filter coefficients that are equivalent to the kriging weighting coefficients. This is in contrast to the conventional spatial…

    Factorial Kriging (FK) is a data- dependent spatial filtering method that can be used to remove both independent and correlated noise on geological images as well as to enhance lineaments for subsequent geological interpretation. The spatial variability of signal, noise, and lineaments, characterized by a variogram model, have been used explicitly in calculating FK filter coefficients that are equivalent to the kriging weighting coefficients. This is in contrast to the conventional spatial filtering method by predefined, data-independent filters, such as Gaussian and Sobel filters. The geostatistically optimal FK filter coefficients, however, do not guarantee an optimal filtering effect, if filter geometry (size and shape) are not properly selected. The selection of filter geometry has been investigated by examining the sensitivity of the FK filter coefficients to changes in filter size as well as variogram characteristics, such as nugget effect, type, range of influence, and anisotropy. The efficiency of data-dependent FK filtering relative to data-independent spatial filters has been evaluated through simulated stochastic images by two examples. In the first example, both FK and data-independent filters are used to remove white noise in simulated images. FK filtering results in a less blurring effect than the data-independent fillers, even for a filter size as large as 9 × 9. In the second example, FK and data-independent filters are compared relative to the extraction of lineaments and components showing anisotropic variability. It was determined that square windows of the filter mask are effective only for removing Isotropie components or white noise. A nonsquare windows must be used if anisotropic components are to be filtered out. FK filtering for lineament enhancement is shown to be resistant to image noise, whereas data-independent filters are sensitive to the presence of noise.

    Other authors
    • Richard Sinding-Larsen
    See publication

  • Stochastic modeling and simulation of small faults for reservoir characterization by marked point process modeling and kriging

    Kluwer Academic Publishers

    Other authors
    • Richard Sinding-Larsen
    See publication

Languages

  • English

    -

  • Chinese

    -

Recommendations received

1 person has recommended Renjun

Join now to view

More activity by Renjun

View Renjun’s full profile

  • See who you know in common
  • Get introduced
  • Contact Renjun directly
Join to view full profile

Other similar profiles

Explore more posts

Explore collaborative articles

We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.

Explore More

Others named Renjun Wen

15 others named Renjun Wen are on LinkedIn

See others named Renjun Wen

Add new skills with these courses

See all courses