Repository logo

Browsing by Author "Stright, Lisa, advisor"

Now showing 1 - 6 of 6
  • Thumbnail Image
    ItemOpen Access
    Evaluating covariance-based geostatistical methods with bed-scale outcrop statistics conditioning for reproduction of intra-point bar facies architecture, Cretaceous Horseshoe Canyon Formation, Alberta, Canada
    (Colorado State University. Libraries, 2022) McCarthy, Andrew Louis, author; Stright, Lisa, advisor; Ronayne, Michael, committee member; Bailey, Ryan, committee member
    Geostatistical characterization of petroleum reservoirs typically suffers from problems of sparse data, and modelers often draw key parameters from analogous outcrop, numerical, and experimental studies to improve predictions. While quantitative information (bed-scale statistical distributions) from outcrop studies is available, translating the data from outcrop to models and generating geologically-realistic realizations with available geostatistical algorithms is often problematic. The overarching goal of this thesis is to test the capacity of covariance-based geostatistical methods to reproduce intra-point bar facies architecture while guiding those algorithms with bed-scale outcrop statistics from the Late Cretaceous Horseshoe Canyon Formation in southeastern Alberta. First, general facies architecture reproduction is tested with 2- and 3-facies synthetic and outcrop-based experiments with variable hard data, soft data weight, and soft data reliability. Next, 3-D sector models compare performance of different geostatistical simulation methods: sequential / co-sequential indicator, plurigaussian, and nested truncated gaussian. Findings show that despite integration of outcrop statistics, all conventional covariance-based geostatistical algorithms struggle to reproduce complex facies architecture that is observed in outcrop. Specifically, problems arise with: 1) low-proportion facies and 2) a weak statistical relationship between hard data (measured sections) and soft data (probability models). Nested modeling partially mitigates low-proportion issues and performs better as a result.
  • Thumbnail Image
    ItemOpen Access
    Evaluating the impact of deep-water channel architecture on the probability of correct facies classification using 3D synthetic seismic data
    (Colorado State University. Libraries, 2021) Langenkamp, Teresa Rose, author; Stright, Lisa, advisor; Harry, Dennis, committee member; Eykholt, Richard, committee member
    Modeling studies of bed-to geobody-scale architecture in deep-water channel deposits reveal that channel element stacking patterns and internal architecture strongly control connectivity. This architecture is critical to understanding hydrocarbon flow and recovery but is unresolvable in exploration-scale seismic-reflection profiles. Forward seismic reflectivity modeling of a digital outcrop models is commonly used to explore how depositional architecture is interpretable in a filtered seismic response. One limitation of forward seismic reflectivity modeling studies is that they often stop short of qualitatively assessing the link between underlying depositional architecture and seismic response. This study addresses the gap between qualitative interpretation and quantitative evaluation by calculating the prediction reliability of inverted seismic data. Specifically, this study uses synthetic 3D seismic modeling and inversion of a 3D outcrop model of deepwater channels in the Tres Pasos Formation of the Magallanes Basin of southern Chile. The model includes outcrop- (bed and geobody) to seismic- (reservoir to basin) scale architecture. The primary objective is to quantify where and when channel architecture is accurately predicted by seismic facies classification. Bayesian classification is used to test the probability of correct facies classification from P-impedance and if the classification results are dependent upon architectural styles (e.g., channel element stacking patterns). Model sensitivity variables include seismic frequency (ranging from 15 to 180 Hz) and deep versus shallow rock properties. Results show that prediction reliability increased for both channel element axis sandstone and mass transport deposits with increasing frequency. Deep reservoirs or faster seismic velocities more accurately predict facies than shallow reservoirs or slower seismic velocities due to the increasing contrast between sandstone and shale velocities. Channel axis sandstone is less easily interpreted where channel elements are vertically aggraded, reducing acoustic impedance contrasts with background shale. When channel elements are laterally stacked or disorganized, facies can be predicted from seismic attributes with a higher confidence, due to a strong contrast between channel element sandstone and background shale. This study highlights that architectural information strongly impacts 3D inverted seismic data and highlights conditions that either hinder or aid accurate interpretation from facies classification.
  • Thumbnail Image
    ItemOpen Access
    Evaluating the impact of hierarchical deep-water slope channel architecture on fluid flow behavior, Cretaceous Tres Pasos Formation, Chile
    (Colorado State University. Libraries, 2021) Ruetten, Andrew, author; Stright, Lisa, advisor; Ronayne, Michael, committee member; Bailey, Ryan, committee member
    Channelized deep-water reservoirs inherently contain sub-seismic scale heterogeneity, resulting in uncertainty when evaluating reservoir connectivity and flow patterns. Stratigraphic architectural features, including stacked channel elements, channel element fill, mass transport deposits (MTDs), and channel base drapes, can have a complex and significant impact on fluid flow pathways. While this detailed stratigraphic architecture can be difficult to capture at the development scale, it can be effectively modeled at the sector scale using high-resolution outcrop data. The characterization of flow behaviors and reservoir performance at this finer scale can then be used in the construction of lower-resolution development-scale simulations. This study uses a three-part sensitivity analysis to test how fluid flow behavior responds to channel element stacking patterns, net to gross ratio, channel base drape coverage, and MTD properties. First, simplified models are used to isolate key flow behaviors. Then, field data is incorporated from the seismic-scale Laguna Figueroa outcrop of the Cretaceous Tres Pasos Formation, Magallanes Basin, Chile to construct a deterministic outcrop model that incorporates realistic stacking patterns and architectural features, including MTDs. Finally, stochastic object-based methods are used to try to replicate the flow characteristics of the outcrop model using established geostatistical methods and limited data input. Fluid flow was simulated using a constant flux aquifer at the base of the model and three producing wells at the top, and the results of the three modeling methods were compared in an effort to elucidate key flow behaviors.
  • Thumbnail Image
    ItemOpen Access
    Hypothesis-based machine learning for deep-water channel systems
    (Colorado State University. Libraries, 2020) Vento, Noah Francis Ryoichi, author; Stright, Lisa, advisor; Ronayne, Michael, committee member; Anderson, Charles, committee member
    Machine learning algorithms are readily being incorporated into petroleum industry workflows for use in well-log correlation, prediction of rock properties, and seismic data interpretation. However, there is a clear disconnect between sedimentology and data analytics in these workflows because sedimentologic data is largely qualitative and descriptive. Sedimentology defines stratigraphic architecture and heterogeneity, which can greatly impact reservoir quality and connectivity and thus hydrocarbon recovery. Deep-water channel systems are an example where predicting reservoir architecture is critical to mitigating risk in hydrocarbon exploration. Deep-water reservoirs are characterized by spatial and temporal variations in channel body stacking patterns, which are difficult to predict with the paucity of borehole data and low quality seismic available in these remote locations. These stacking patterns have been shown to be a key variable that controls reservoir connectivity. In this study, the gap between sedimentology and data analytics is bridged using machine learning algorithms to predict stratigraphic architecture and heterogeneity in a deep-water slope channel system. The algorithms classify variables that capture channel stacking patterns (i.e., channel positions: axis, off-axis, and margin) from a database of outcrop statistics sourced from 68 stratigraphic measured sections from outcrops of the Upper Cretaceous Tres Pasos Formation at Laguna Figueroa in the Magallanes Basin, Chile. An initial hypothesis that channel position could be predicted from 1D descriptive sedimentologic data was tested with a series of machine learning algorithms and classification schemes. The results confirmed this hypothesis as complex algorithms (i.e., random forest, XGBoost, and neural networks) achieved accuracies above 80% while less complex algorithms (i.e., decision trees) achieved lower accuracies between 60%-70%. However, certain classes were difficult for the machine learning algorithms to classify, such as the transitional off-axis class. Additionally, an interpretive classification scheme performed better (by around 10%-20% in some cases) than a geometric scheme that was devised to remove interpretation bias. However, outcrop observations reveal that the interpretive classification scheme may be an over-simplified approach and that more heterogeneity likely exists in each class as revealed by the geometric scheme. A refined hypothesis was developed that a hierarchical machine learning approach could lend deeper insight into the heterogeneity within sedimentologic classes that are difficult for an interpreter to discern by observation alone. This hierarchical analysis revealed distinct sub-classes in the margin channel position that highlight variations in margin depositional style. The conceptual impact of these varying margin styles on fluid flow and connectivity is shown.
  • Thumbnail Image
    ItemOpen Access
    Modeling of channel stacking patterns controlled by near wellbore modeling
    (Colorado State University. Libraries, 2023) Escobar Arenas, Luis Carlos, author; Stright, Lisa, advisor; Ronayne, Michael, committee member; Barnes, Elizabeth, committee member
    Reservoir models of deep-water channels rely upon low-resolution but spatially extensive seismic data, high vertical resolution but spatially sparse well log data and geomodeling methods. The results cannot predict architecture below seismic resolution or between well logs. Usually, the data and interpretations that provide constraints for modeling workflows do not capture sub-seismic scale architecture. Therefore, standard modeling methods do not generate models that include details that can impact hydrocarbon flow and recovery. Constraining models to well and seismic data is problematic. Employing measured sections in the Tres Pasos Fm. (Magallanes Basin, Chile) is feasible to predict deep-water channel architecture, specifically channel stacking patterns with 1D information analogous to well data. This research performed near-wellbore modeling to generate multiple scenarios of channel stacking patterns constrained by machine learning-derived probabilities using (i) conditional Monte Carlo simulation with soft probabilities per channel element within the measured section choosing the highest probabilities for each element (ii) conditional Monte Carlo simulation of channel stacking, (iii) template-based modeling, (iv) forward modeling with Markov transition probabilities without matching to thickness and (v) conditional Monte Carlo simulation constrained to measured section thickness. Machine learning workflows generate channel position probabilities (i.e., axis, off-axis, margin) within a measured section given the interpreted top/bases of channel elements. These probabilities constitute the input for Monte Carlo simulations capturing channel element stacking patterns at the measured section locations. The most likely 2D channel stacking pattern scenarios defined channel centerline points, and volumes of the individual channel elements can be generated connecting them. Surface-based modeling offers a way to depict reservoirs of hydrocarbons, water or low-enthalpy geothermal systems in which small-scale heterogeneity needs to be captured explicitly by bounding surfaces because it impacts fluid flow, improving our forecasts of resource exploitation. Furthermore, predicting heterogeneity controlled by depositional architecture is critical for transport and storage capacity in CO2 reservoirs. The dataset provided and the advent of these flexible and accurate methods to depict the subsurface offer the opportunity to overcome the historical limitations of grid-based models and allow us to assess multi-scale architecture that controls fluid flow. This research aims to show the results of modeling deep-water channels, including a 1D identification of architectural positions and a 2D arrangement of channel stacking patterns.
  • Thumbnail Image
    ItemOpen Access
    Using RMS amplitudes from forward seismic-reflectivity modeling of channelized deep-water slope deposits to inform stratigraphic interpretation and sub-seismic scale architecture, Tres Pasos Formation, Magallanes Basin, Patagonia, Chile
    (Colorado State University. Libraries, 2018) Nielson, Adam, author; Stright, Lisa, advisor; Schutt, Derek, committee member; Sale, Thomas, committee member
    Deep-water slope channels outcropping in the Tres Pasos Formation of the Magallanes Basin in southern Chile are used as the foundation of a forward seismic-reflectivity modeling study to better inform stratigraphic interpretation. The multi-scale architecture of deep-water slope channels is often difficult to interpret from low resolution seismic-reflectivity surveys. Valuable insight can be gained from forward seismic-reflectivity modeling using multiple-scales of architecture as building blocks (i.e., channel elements stacking into channel complexes) to provide insight into subsurface interpretation. Forward seismic-reflectivity models of channel elements with sub-meter scale heterogeneity are interrogated for RMS amplitude and apparent thickness as a function of true stratigraphic thickness and net sand thickness. Relationships between interpreted variables from the forward models (RMS amplitude and apparent thickness) compared to measured variable from the input models (true stratigraphic thickness and net sand thickness) provide recognition criteria for interpreting building blocks in subsurface seismic-reflectivity data. This study shows that decreasing RMS amplitude for constant apparent thickness is primarily controlled by vertically juxtaposed facies between multiple stacked channel elements. Furthermore, laterally stepping and vertically aggrading channel elements increase confidence in stratigraphic interpretation whereas laterally migrating channel elements are harder to delineate. An increase in frequency tends to improve interpretation of net sand thickness for multiple channel elements informing interpretation of lateral facies changes. Results from this study also show that RMS amplitudes and apparent thickness show patterns to help differentiate channel element stacking configurations and can be tied back to the known model variables, true stratigraphic thickness and net sand thickness. However, interpretation of exploration scale data, specifically RMS amplitude and apparent thickness interpretations is complicated by interfering reflections at increased frequency, complicating the recognition of multiple channel elements within a channel complex set.