**Source**

**Conference**

- 2003 SEG Annual Meeting (1)
- 2009 SEG Annual Meeting (1)
- ISRM International Symposium on In-Situ Rock Stress (1)
- SEG International Exposition and Annual Meeting (2)
- SPE Annual Technical Conference and Exhibition (1)
- SPWLA 60th Annual Logging Symposium (1)
- The Seventeenth International Offshore and Polar Engineering Conference (1)
- Unconventional Resources Technology Conference (1)

**Publisher**

**Theme**

**Author**

- Alpay, Server Fatih (1)
- Cristea, Alexandrina (1)
- Ergündüz, Okan (1)
- Farquharson, Colin G. (1)
- Gao , Ke (1)
- Geng, Meixia (1)
- Haghighi, Arash Moaddel (1)
- Haghighi, Iman Moaddel (1)
- Harrison, John P. (1)
- Jayaram, Vikram (1)
- Kaplan, Sam T. (1)
- Kececioglu, Tayfun (1)
- Lu, Wenkai (1)
- Lubo, David (1)
- Lubo-Robles, David (1)
- Luo, Yi (1)
- Marfurt, Kurt J. (2)
- Minoura, Munehiko (1)
- Naito, Shigeru (1)
- Pavlakos, Paul (1)
- Peace, Alexander L. (1)
- Qian, Zhongping (1)
- Rojas, Pedro A. Romero (1)
- Sacchi, Mauricio D. (1)
- Ulrych, Tadeusz J. (1)
- Welford, J. Kim (1)
- Zhao, Bo (1)

**Concept Tag**

- acoustic impedance (1)
- acquisition footprint (1)
- adaptive multiple subtraction (1)
- adaptive subtraction (1)
- algorithm (1)
- approach (1)
- Artificial Intelligence (8)
- attenuation (1)
- automatic learning gaussian mixture model (1)
- basis (1)
- basis function (1)
- basis patch (1)
- blind source separation (1)
- budgell harbour (1)
- budgell harbour stock (1)
- Channel complex (1)
- coefficient (1)
- component (1)
- correlation (1)
- dataset (1)
- demultiple (1)
- Denote (1)
- density contrast model (1)
- difference (1)
- dispersion (1)
- distribution (1)
- drilling operation management (1)
- eigenvalue (1)
- eigenvalue decomposition (1)
- Engineering (1)
- entropy (1)
- equation (2)
- Euclidean dispersion (1)
- evaluation (1)
- expansion (1)
- facies (1)
- function (1)
- Gaussian (1)
- Gaussian mixture (1)
- Gaussian mixture model (1)
- generalised variance (1)
- geologic modeling (1)
- geological modeling (1)
- GMM (1)
- Gravity Gradiometry data (1)
- hindcast (1)
- hyvärinen (1)
- ICA (2)
**independent component (9)**- independent component analysis (6)
- international exposition (1)
- intrusion (1)
- inversion (1)
- inversion result (1)
- knowledge management (1)
- kurtosis (1)
- location (1)
- log analysis (1)
- long wavelength signal (1)
- machine learning (7)
- management and information (3)
- marine transportation (1)
- matrix (4)
- method (1)
- mixture (2)
- mixture model (1)
- Moki (1)
- Multivariate (1)
- multivariate scatter (1)
- mutual information (1)
- New Zealand (1)
- NMR (1)
- noise (1)
- ocean (1)
- oil saturation (1)
- optimization problem (1)
- permeability (1)
- probability (1)
- problem (2)
- process (1)
- purple seismic facies (1)
- real dataset (1)
- realization (1)
- red rectangle (1)
- reserves evaluation (1)
- reservoir (1)
- Reservoir Characterization (7)
- reservoir description and dynamics (4)
- reservoir geomechanics (1)
- reservoir simulation (1)
- Rock mechanics (1)
- seismic processing and interpretation (3)
- signal (2)
- source (2)
- Symposium (1)
- Taranaki Basin (1)
- Tzz (1)
- University (1)
- Upstream Oil & Gas (7)
- Wave (1)

**File Type**

Rojas, Pedro A. Romero (Weatherford International) | Cristea, Alexandrina (Weatherford International) | Pavlakos, Paul (Weatherford International) | Ergündüz, Okan (ARAR AS) | Kececioglu, Tayfun (ARAR AS) | Alpay, Server Fatih (ARAR AS)

**ABSTRACT**

Nuclear magnetic resonance wireline logging and data post-processing technologies are continuously evolving, making significant contributions to rock, fluid typing, formation evaluation and characterization of the near-wellbore zone. In heavy oil fields, however, nuclear magnetic resonance (NMR) logging is known to provide an underestimated permeability, poor reliable oil typing and thus poor oil saturation and viscosity determinations, especially when the evaluation is based only on the spectra of transverse magnetic relaxation times (T2) (one-dimension NMR) [Romero et al., 2009]. Several attempts have been made to improve NMR results, mostly with limited success [Fang et al., 2004], especially in separating the oil component from the contribution of other fluids to the T2 spectra. The main reason lies not necessarily in the selection of the data acquisition parameters and sequences for a single-frequency or multi-frequency tool, but in the way how the data is post-processed.

The present study refers to a well drilled through the Derdere formation, a limestone/dolomite heavy oil reservoir in Turkey. The NMR data was acquired in with a centralized, single-frequency wireline tool in a 6-in. borehole, drilled with water-based mud in a freshwater carbonate reservoir. The generated T2 log was analyzed in a traditional way to obtain the NMR total porosity and its partitions based on standard cutoff values. For the given 12 API oil gravity, reservoir temperature (76 °C) and gas-oil-ratio (GOR) the T2Oil peak appears around 170 ms, right from the T2 cutoff for limestones; therefore, no corrections were needed on the permeability calculated from the Timur-Coates and Schlumberger-Doll-Research (SDR) equations. In the present well, only a diffused separation between oil and free water could be observed on the T2 distribution log from field data.

In the broader concept of Artificial Intelligence, the newly proposed post-processing steps to obtain the oil saturation start by deconvolving the T2 spectra, using blind source separation (BSS) based on independent component analysis (ICA) [Romero, 2016; Romero Rojas et al., 2018]. Based on its T2 peak value —the expected T2Oil peak response— calculated from the prejob planner/simulator, the deconvolution results show that one specific independent component corresponds to the oil, from which the oil saturation was determined.

Results demonstrated the usefulness of NMR logging technology in the characterization and evaluation of this reservoir. Data post-processing based on BBS-ICA enable adequate differentiation between fluid components from T2 spectra. For the reasons above, NMR has been proposed for additional wells in the same field.

Artificial Intelligence, blind source separation, Engineering, evaluation, independent component, independent component analysis, log analysis, machine learning, matrix, NMR, oil saturation, permeability, reservoir, spectra, spwla 60, Symposium, th annual, University, Upstream Oil & Gas, well logging

Country:

- South America (0.94)
- Europe (0.94)
- North America > United States (0.69)
- Asia (0.69)

Oilfield Places:

- Asia > Middle East > Iraq > Kurdistan > Zagros Basin > Taq Taq Field > Upper Kometan Formation (0.99)
- Asia > Middle East > Iraq > Kurdistan > Zagros Basin > Taq Taq Field > Shiranish Formation (0.99)
- Asia > Middle East > Iraq > Kurdistan > Zagros Basin > Taq Taq Field > Qamchuqa Formation (0.99)
- Europe > Romania (0.97)

SPE Disciplines: Reservoir Description and Dynamics > Formation Evaluation & Management > Open hole/cased hole log analysis (1.00)

Summary Seismic attributes are powerful tools that allow interpreters to make a more comprehensive and precise seismic interpretation. In this paper, we apply an unsupervised multiattribute technique called Independent Component Analysis to reduce dimensionality and extract the most valuable information of multiple spectral magnitude components in order to make an unsupervised seismic facies classification of channel complexes located in the Moki A Formation, Taranaki Basin, New Zealand. Introduction Depending on the seismic attribute that we choose, different information can be extracted (Infante-Paez and Marfurt, 2017) from the seismic volume, thus, relying on only one attribute information lead to an incomplete seismic interpretation. For this reason, multi-attribute techniques such as Principal Component Analysis (PCA), Selforganizing Maps (SOM) are commonly used. Based on higher order statistics, Independent Component Analysis separates a multivariate signal into subcomponents which are independent of each other (Hyva rinen and Oja, 2000), thus extracting more valuable information than techniques such as Principal Component Analysis (PCA) which tends to mix geology.

acquisition footprint, Artificial Intelligence, Channel complex, facies, independent component, independent component analysis, machine learning, matrix, Moki, New Zealand, purple seismic facies, red rectangle, Reservoir Characterization, seismic facies, spectral, Taranaki Basin, unsupervised seismic facies classification, Upstream Oil & Gas

Oilfield Places:

- Oceania > New Zealand > Tasman Sea > Taranaki Basin > Moki Formation (0.99)
- Oceania > New Zealand > North Island > Taranaki Basin > Mt. Messenger Formation (0.89)

SPE Disciplines: Reservoir Description and Dynamics > Reservoir Characterization > Seismic processing and interpretation (1.00)

Technology: Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (0.96)

Geng, Meixia (Institute of Geophysics and Geomatics, China University of Geosciences, and Department of Earth Sciences, Memorial University of Newfoundland) | Welford, J. Kim (Department of Earth Sciences, Memorial University of Newfoundland) | Farquharson, Colin G. (Department of Earth Sciences, Memorial University of Newfoundland) | Peace, Alexander L. (Department of Earth Sciences, Memorial University of Newfoundland)

We present 3-D inversion results of gravity gradiometry data over the Budgell Harbour Stock (BHS) intrusion, in northern-central Newfoundland, Canada, obtained using a probabilistic inversion method. We examine multiple density contrast models obtained by inverting the single component Tzz and by jointly inverting five independent components. The inversion results show that

Presentation Date: Tuesday, October 16, 2018

Start Time: 1:50:00 PM

Location: 213B (Anaheim Convention Center)

Presentation Type: Oral

Artificial Intelligence, budgell harbour, budgell harbour stock, density contrast model, drilling operation management, geologic modeling, geological modeling, Gravity Gradiometry data, independent component, intrusion, inversion, inversion result, knowledge management, long wavelength signal, reserves evaluation, Reservoir Characterization, Tzz, Upstream Oil & Gas

Oilfield Places: North America > Canada > Newfoundland Offshore > Jeanne d'Arc Basin > Hibernia Field > Hibernia Formation (0.98)

SPE Disciplines:

**Abstract**

In situ stress is an important parameter in many aspects of rock mechanics and often displays significant variability. Among many statistics, dispersion, which denotes how scatter or spread a data group is, is an effective tool to quantify the amount of variability. One common measure of dispersion is the standard deviation. However, this is only applicable to scalar data, and a robust approach to calculate a scalar-valued measure of stress dispersion is still not clear. This is mainly because of the tensorial nature of stress, which renders classical statistics inapplicable.

As multivariate statistics are widely used to analyse stress tensor components, here we investigate the use of three widely used scalar-valued scatter measures in multivariate statistics- namely, generalised variance, total variation and effective variance - to help identify a suitable scalar-valued measure to characterise stress variability. We compare these scatter measures with Euclidean dispersion, a tensorial statistic defined by us previously.

We show how the multivariate scatter measures are linked to Euclidean dispersion, and through the use of synthetic stress data illustrate that, in particular circumstances, all these measures can effectively characterise the variability of stress data. However, we have identified that there is significant potential for the multivariate total variation to generate erroneous results, due to a lack of coordinate invariance in the formulation of this statistic. Therefore, to remain loyal to the tensorial nature of stress, and avoid potential misuse, we recommend using the tensor-related Euclidean dispersion for characterisation of in situ stress variability.

**Introduction**

In situ stress is an important parameter for a wide range of endeavours, including rock engineering design, hydraulic fracturing analysis, rock mass permeability and earthquake potential evaluation (Amadei & Stephansson, 1997; Zoback, 2010; Latham et al., 2013; Matsumoto et al., 2015). Because of the inherent complexity of fractured rock masses in terms of varying rock properties, the presence of discontinuities and unclear boundary conditions (Matsumoto et al., 2015), stress in rock often displays significant heterogeneity (Martin, 1990). For all rock engineering endeavours the characterisation and quantification of stress variability is therefore important.

Artificial Intelligence, dispersion, Euclidean dispersion, generalised variance, independent component, machine learning, Multivariate, multivariate scatter, Reservoir Characterization, reservoir geomechanics, Rock mechanics, situ stress, statistics, stress data, stress tensor, stress variability, tensor, tensor component, total variation, Upstream Oil & Gas, variability, variance

Country:

- Europe (0.70)
- North America (0.70)

SPE Disciplines: Reservoir Description and Dynamics > Reservoir Characterization > Reservoir geomechanics (1.00)

Technology: Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (0.90)

**Summary **

Tying detailed well log measurements to lower resolution but a really extensive 3D seismic data volumes is key to quantitative seismic interpretation. Ties using a poststack or prestack convolution model are routine, while supervised classification tying well data to seismic attributes using neural networks and geostatistics are also well established. However, unsupervised classification ties where the objective is to identify unknown patterns in the data is less well established. In this paper, we use an automatic learning Gaussian Mixture Model to statistically characterize the well logs, evaluate the probability distribution functions of different lithologies and then tie them to corresponding 3D seismic attribute volumes. We precondition our four-dimensional data by projecting onto two dimensions using Independent Component Analysis.

We apply this workflow to Diamond M Field within the Horseshoe Atoll Reef Complex, Scurry County, TX, and find the Gaussian Mixture Model is able to statistically characterize and resolve lithological variations seen in the logs. In particular, we are able to clearly distinguish between lithologies from six different wells in the region of interest. The final result is a probabilistic map that statistically measures the variability of the seismic lithologies from the well logs.

acoustic impedance, algorithm, Artificial Intelligence, automatic learning gaussian mixture model, correlation, Gaussian, Gaussian mixture, Gaussian mixture model, GMM, independent component, independent component analysis, machine learning, mixture model, Reservoir Characterization, Upstream Oil & Gas

Oilfield Places:

- North America > United States > Texas > Permian Basin > Midland Basin (0.99)
- North America > United States > Texas > Permian Basin > Diamond M Field (0.99)
- North America > United States > Texas > Permian Basin > Cisco Formation (0.99)
- (9 more...)

SPE Disciplines: Reservoir Description and Dynamics > Reservoir Characterization > Seismic processing and interpretation (1.00)

Technology: Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning (1.00)

We use sparse coding to construct a series expansion of data. Sparse coding gives a data-driven set of basis functions whose coefficients follow a sparse distribution (the coefficients are called the sparse code). We use sparse coding in two noise attenuation algorithms, one applicable to additive random noise, and another applicable to coherent noise (free-surface multiple removal). To illustrate, we extract and, then, filter a sparse code from noisy data. The filter is designed to remove the portion of the code that is more indicative of noise than signal. First, we attenuate additive random noise by applying a threshold to the sparse code. Next, we consider a normal-moveout corrected common midpoint gather corrupted by free-surface multiples, and construct a filter using the dominant wave-numbers in the sparse coding basis functions. This allows us to filter out flat events (the signal), leaving an estimate of the noise (multiples) that are, subsequently, subtracted from the original gather.

In this abstract, we use sparse coding (e.g. Hoyer, 1999) to find, simultaneously, the coefficients y j and basis pj of the expansion. That is, the basis are data-driven. The coefficients y j are called the sparse code. The bases of the sparse coding expansion are interesting, resembling, for example, the set of basis functions studied in the local Radon transform (Sacchi et al., 2004), and curvelet transform (e.g. Starck et al., 2002).

We use the sparse coding representation of x to attenuate noise, both random and coherent. When the noise is random, we follow the work of, for example, Hyvärinen et al. (1998), and shrink the sparse code y (Kaplan and Ulrych, 2005). For a coherent noise example, we use the often studied application free-surface multiple attenuation (e.g. Hampson, 1986). In particular, we apply sparse coding to a common midpoint gather after normal moveout correction, filtering the sparse code based on the dominant wave-numbers in the corresponding basis pj. This allows us to filter out the primary events from the gather, leaving only multiples which are, subsequently, subtracted from the original data.

For our second example, we introduce a new application for sparse coding, applying it to free-surface multiple attenuation. In particular, we consider a single common midpoint (CMP) data gather which has undergone normal moveout (NMO) correction. Due to NMO, the primary events are flat, but the multiples are under-corrected, and remain hyperbolic. The idea, then, is that the hyperbolic and linear events are captured by disjoint subsets of the sparse coding basis functions, and so the multiples can be isolated, and subsequently subtracted from the data.

To test this idea we consider the real-data example in Figure 2. In Figure 2b, we plot d(t,h), the NMO corrected CMP gather. The realizations of x are extracted from d(t,h) using the same windowing procedure described in the previous section. Then, we compute the sparse code y and basis P, plotting the lexicographic reordered columns of P (the basis patches) in Figure 2a. The ordering of the basis patches depends on the initialization of the optimization procedure.

Artificial Intelligence, attenuation, basis, basis function, basis patch, coefficient, entropy, equation, expansion, function, hyvärinen, independent component, independent component analysis, international exposition, machine learning, noise, optimization problem, realization, Reservoir Characterization, reservoir description and dynamics, seismic processing and interpretation, sparse, sparse code, Upstream Oil & Gas

SPE Disciplines: Reservoir Description and Dynamics > Reservoir Characterization > Seismic processing and interpretation (1.00)

Technology:

Decomposing linear mixtures or superpositions into their components is a problem occurring in many different branches of science, such as telecommunications, Seismology, and biomedical signal analysis. Blind source separation (BSS) in particular, deals with the case where neither the sources, nor the mixing matrix or process of

mixing are known, the only available data are the mixed signals. The standard approach to BSS is Independent Component Analysis (ICA).

In fact ICA is a statistical technique that represents a multidimensional random vector as a linear combination of nongaussian random variables -'independent components'- that are as independent as possible. In 3D seismic surveys involved in exploration operations, recorded time series in each dimension are taken to be independent in nature and behavior, which is a direct result of physical response of materials into which seismic waves penetrate. But as an observation one dimension is sometimes contaminated up to one fifth by information from another dimension, resulting an increase in SNR.

Here we have applied FAST-ica algorithm to a 3D seismic record sample to extract least dependent recordings for all three spatial dimensions. In order to test the reliability of the decomposition we used mutual information (MI) transfer between signals to confirm the result of outputs from FAST-ica algorithm as, least dependent components, leading to more accurate interpretations of petrophysical parameters.

approach, Artificial Intelligence, component, distribution, ICA, independent component, independent component analysis, machine learning, management and information, matrix, mixture, mutual information, problem, Reservoir Characterization, reservoir description and dynamics, seismic processing and interpretation, separation, signal, source, source signal, statistically independent, transfer, Upstream Oil & Gas, vector, voice

Sea-states are assumed to be separated into independent components. A stochastic model based on the Markov process that includes a seasonal model is applied to these independent components. The propagation of a sea-state in the ocean is represented by the cross-correlation between present and past sea-states. The correlation is provided by sea-states with a given time step. A comparison of the occurrence probability and cross-correlation of the sea-states between simulations using our proposed stochastic model and the hindcast data shows that our proposed stochastic model and simulation scheme are valid. We observe a high sea area propagating from the west to the east in our simulations, as observed in actual seas. INTRODUCTION Sea-states used in the design and operation of ships and marine systems are provided in the form of statistical information from ocean waves and winds. A sea-state is represented by several parameters. For example, significant wave height or mean wave period is a sea-state parameter. The long-term occurrence probabilities and cross-correlations of sea-state parameters are commonly provided as wave tables, such as the Global Wave Statistics and the Waves and Winds in the North Pacific Ocean, which are constructed from a number of accumulated observation data and from hindcast data. These wave tables are useful for evaluating the lifetime of ships (Fukuda, 1969, Nordenström, 1973, Naito et al., 1998, 2006). Oceangoing cargo ships are becoming more sophisticated and larger. This is a result of the optimization of sea transportation. A more accurate evaluation of the structural strength, hydrodynamic performance, and operating cost is naturally desired, because these ships have a large influence on several aspects of safety and economy. For example, it is desirable to compare fuel consumption and arrival time, depending on the route, season, operating criteria. In general, wave tables that cover wide sea areas are not suitable for this purpose, and historical sea-state data on the sea routes where these ships sail is required.

Artificial Intelligence, Denote, eigenvalue, eigenvalue decomposition, equation, hindcast, independent component, location, machine learning, management and information, marine transportation, matrix, ocean, probability, process, reservoir description and dynamics, reservoir simulation, respect, simulated swell, stochastic differential equation, stochastic model, swell, Wave

SPE Disciplines:

Technology:

Adaptive subtraction is a critical step for the widely-used prediction plus subtraction multiple attenuation technique. In this paper, a new adaptive subtraction algorithm based on independent component analysis (ICA), which exploits the maximum kurtosis, is presented. The method has been applied on several synthetic datasets generated by simple convolution and finite-difference model (FDM) technique, and very encourage results has been obtained.

adaptive multiple subtraction, adaptive subtraction, dataset, demultiple, difference, ICA, independent component, independent component analysis, kurtosis, management and information, method, mixture, problem, real dataset, Reservoir Characterization, reservoir description and dynamics, seismic processing and interpretation, signal, source, subtraction, subtraction scalar, technique

SPE Disciplines:

Thank you!