Comparisons of pyrolysis parameters between source rocks and their clay-sized fractions: Implication for source material of hydrocarbon generation

In natural environments, organic-clay interactions are strong and cause organo-clay composites (a combination between organic matter [OM] and clay minerals) to be one of the predominant forms for OM occurrence, and their interactions greatly influence the hydrocarbon (HC) generation of OM within source rocks. However, despite occurring in nature, dominating the OM occurrence, and having unique HC generation ways, organo-clay composites have rarely been investigated as stand-alone petroleum precursors. To improve this understanding, we have compared the Rock-Eval pyrolysis parameters derived from more than 100 source rocks and their corresponding <2 μm clay-sized fractions (representing organo-clay composites). The results show that all of the Rock-Eval pyrolysis parameters in bulk rocks are closely positively correlated with those in their clay-sized fractions, but in clay-sized fractions the quality of OM for HC generation is poorer, in that the pyrolysable organic carbon levels and hydrogen index values are lower, whereas the residual organic carbon levels are higher than those in bulk rocks. Being integrated with the effects of organic-clay interactions on OM occurrence and HC generation, our results suggest that organo-clay composites are stand-alone petroleum precursors for HC generation occurring in source rocks, even if the source rocks exist in great varieties in their attributes. Our source material for HC generation comprehensively integrates the original OM occurrence and HC generation behavior in natural environments, which differs from kerogen and is much closer to the actual source material of HC generation in source rocks, and it calls for further focus on organic-mineral interactions in studies of petroleum systems.

Semi-Supervised Salt Segmentation Using Mean Teacher

Detecting subsurface salt structures from seismic images is important for seismic structural analysis and subsurface modeling. Recently, deep learning has been successfully applied in solving salt segmentation problems. However, most of the studies focus on supervised salt segmentation and require numerous accurately labeled data, which is usually laborious and time-consuming to collect, especially for the geophysics community. In this paper, we propose a semi-supervised framework for salt segmentation, which requires only a small amount of labeled data. In our method, adopting the mean teacher method, we train two models sharing the same network architecture. The student model is optimized using a combination of supervised loss and unsupervised consistency loss, whereas the teacher model is the exponential moving average (EMA) of the student model. We introduce the unsupervised consistency loss to better extract information from unlabeled data by constraining the network to give consistent predictions for the input data and its perturbed version. We train and test our novel semi-supervised method on both synthetic and real datasets. Results demonstrate that our proposed semi-supervised salt segmentation method outperforms the supervised baseline when there is a lack of labeled training data.

Quantifying the sensitivity of seismic facies classification to seismic attribute selection: An explainable machine learning study

During the past two decades, geoscientists have used machine learning to produce a more quantitative reservoir characterization and to discover hidden patterns in their data. However, as the complexity of these models increase, the sensitivity of their results to the choice of the input data becomes more challenging. Measuring how the model uses the input data to perform either a classification or regression task provides an understanding of the data-to-geology relationships which indicates how confident we are in the prediction. To provide such insight, the ML community has developed Local Interpretable Model-agnostic Explanations (LIME), and SHapley Additive exPlanations (SHAP) tools. In this study, we train a random forest architecture using a suite of seismic attributes as input to differentiate between mass transport deposits (MTDs), salt, and conformal siliciclastic sediments in a Gulf of Mexico dataset. We apply SHAP to understand how the model uses the input seismic attributes to identify target seismic facies and examine in what manner variations in the input such as adding band-limited random noise or applying a Kuwahara filter impact the models’ predictions. During our global analysis, we find that the attribute importance is dynamic, and changes based on the quality of the seismic attributes and the seismic facies analyzed. For our data volume and target facies, attributes measuring changes in dip and energy show the largest importance for all cases in our sensitivity analysis. We note that to discriminate between the seismic facies, the ML architecture learns a “set of rules” in multi-attribute space and that overlap between MTDs, salt, and conformal sediments might exist based on the seismic attribute analyzed. Finally, using SHAP at a voxel-scale, we understand why certain areas of interest were misclassified by the algorithm and perform an in-context interpretation to analyze how changes in the geology impact the model’s predictions.

Fractal characteristics of the micropore throats in the shale oil reservoirs of the Chang 7 Member of the Yanchang Formation, Ordos Basin

With the development of the global shale oil and gas revolution, shale oil became an important replacement field to increase oil and gas reserves and production. The Chang 7 Member of the Yanchang Formation in the Ordos Basin was an important shale oil exploration series in China. To study the micropore-throat structure characteristics of the Chang 7 Member, we launched nuclear magnetic resonance (NMR) and high-pressure mercury injection (HPMI) experiments to analyze the pore-throat structure features of the Chang 7 reservoir, and we considered fractal theory to study the fractal characteristics. The NMR results indicated that the T2 spectral morphology of the Chang 7 reservoir could be characterized by three main patterns encompassing early and late peaks with different amplitudes: the type 1 reservoir contained mostly small pores and few large pores, and the porosities of the small and large pores range from 4.16% to 9.04% and 0.70% to 2.40%, respectively. The type 2 reservoir contained similar amounts of small and large pores, and the type 3 reservoir contained few small pores and mostly large pores, while the porosities of the small and large pores range from 1.81% to 2.74% and 3.32% to 5.64%, respectively. The pore-throat structure parameters were obviously affected by the pore size distribution, which in turn influenced the reservoir seepage characteristics of the reservoir. The micropore-throat structure of the reservoir exhibited obvious piecewise fractal characteristics and mainly included dichotomous and trilateral fractals. The type 1 reservoirs were dominated by dichotomous fractals, and these two fractal types were equally distributed in the type 2 and 3 reservoirs. The fractal dimension of the pore throats of different scales exhibited a negative correlation with the corresponding porosity, but no correlation was observed with the permeability, indicating that the size of the reservoir determined by pore throats imposed a strong controlling effect on their fractal characteristics.

Improvement of generalization capability of two-dimensional salt segmentation via iterative semi-supervised learning

Accurate interpretations of subsurface salts are vital to oil and gas exploration. Manually interpreting them from seismic depth images, however, is labor-intensive. Consequently, use of deep learning tools such as a convolutional neural network for automatic salt interpretation recently became popular. Because of poor generalization capabilities, interpreting salt boundaries using these tools is difficult when labeled data are available from one geological region and we like to make predictions for other nearby regions with varied geological features. At the same time, due to vast amount of the data involved and the associated computational complexities needed for training, such generalization is necessary for solving practical salt interpretation problems. In this work, we propose a semi-supervised training, which allows the predicted model to iteratively improve as more and more information is distilled from the unlabeled data into the model. In addition, by performing mixup between labeled and unlabeled data during training, we encourage the predicted models to linearly behave across training samples; thereby improving the generalization capability of the method. For each iteration, we use the model obtained from previous iteration to generate pseudo labels for the unlabeled data. This automated consecutive data distillation allows our model prediction to improve with iteration, without any need for human intervention. To demonstrate the effectiveness and efficiency, we apply the method on two-dimensional images extracted from a real three-dimensional seismic data volume. By comparing our predictions and fully supervised baseline predictions with those that were manually interpreted and we consider as “ground truth”, we find than the prediction quality our new method surpasses the baseline prediction. We therefore conclude that our new method is a viable tool for automated salt delineation from seismic depth images.

Stratigraphic Characterization of the Eagle Ford Shale to Identify the Best Landing Zone: A Semi-analytical Workflow

Placement of the horizontal well within the best landing zone is critical to maximize well productivity, thus identification of the best landing zone is important. This paper illustrates an integrated semi-analytical workflow to carry out the stratigraphic characterization of the Eagle Ford shale to identify the best landing zone. The objective of this work is twofold: 1) to establish a workflow for stratigraphic characterization and 2) to understand the local level variability in the well performance.To establish the workflow, we have used the production data, petrophysical information and regional reservoir property maps. As a first step of the workflow, we subdivided the Eagle Ford shale into nine smaller stratigraphic units using the wireline signatures and outcrop study. In the second step, we have used statistical methods such as linear regression, fuzzy groups and theory of granularity to capture the relationship between the geological parameters and the well performances. In this step, we identified volume of clay (Vclay), hydrocarbon filled porosity (HCFP) and total organic carbon (TOC) as key drivers of the well performance. In the third step, we characterized the nine smaller units and identified four stratigraphic units as good reservoirs with two being the best due to their low Vclay, high HCFP and high TOC content.Finally, we reviewed the well paths of four horizontal wells with respect to the best stratigraphic units. We observed that production behavior of these wells is possibly driven by their lateral placement. The better producing wells are placed within the middle of the best stratigraphic units whereas the poor wells are going out the best stratigraphic units. This investigation provides a case study that demonstrates the importance of integrating datasets to identify best landing zones and the suggested workflow can be applied to other areas and reservoirs to better identify targetable zones.

Seismic attribute analysis of a fault zone in the Thebe field, NW Shelf, Australia

Seismic data is one of the main ways to characterize faults in the subsurface. Faults are 3D entities and their internal structure play a key role in controlling fluid flow in the subsurface. We aim to characterize a geologically sound fault volume that could be used for subsurface model conditioning. We present an attribute analysis of a normal fault from a high resolution seismic dataset of the Thebe Field, offshore NW Australia. We merge together a series of common attributes for fault characterization: dip, semblance and tensor (DST), and we also introduce a new Total Horizontal Derivative (THD) attribute to define the edges of the fault zone. We apply a robust statistical analysis of the attributes and fault damage definition through the analysis of 2D profiles along interpreted horizons. Using the THD attribute, we interpret a smaller width of the fault zone and a more straightforward definition of the boundaries than from the DST cube. Following the extraction of this fault volume, we define two seismic facies that are correlated to lithologies extracted from our conceptual model. We observe a wider fault zone at larger throws, which corresponds also to syn-rift sequence, hence more complex internal fault damage. Our method provides volumes at adequate scale for reservoir modeling and could therefore be used as a proxy for property conditioning.

Spatial distribution prediction of pre-Sinian volcanic and metamorphic rocks in Xinhenan-Sandaoqiao area, Northern Tarim Basin

The main type of oil and gas reservoir in the Xinhenan-Sandaoqiao area is buried hills. The distribution pattern and scale of reservoirs are obviously controlled by pre-Sinian basement strata. However, the lithologic combination and spatial distribution pattern of pre-Sinian basement in this area are still unclear. In this paper, the spatial distribution of pre-Sinian basement volcanic and metamorphic rocks is studied by using the method of multifactor comprehensive analysis. Firstly, the lithology and lithologic combination of igneous and metamorphic rocks are determined according to cores and thin sections. Guided by the seismic reflection characteristics of different lithologic combinations, different lithologic combinations are identified on the profile by combining the seismic reflection characteristics of single well and multiwell. Secondly, using cluster analysis technology, three seismic attributes sensitive to lithology are selected from 10 attributes, crossplots of three seismic attribute values are constructed, and the distribution range of attribute values corresponding to different lithologic combinations is defined for plane lithologic identification. Finally, the plane lithology distribution of the surface layer of pre-Sinian basement is described by combining plane and profile. Six distribution types were identified: deep metamorphic bedrock area in Kuqu depression, dynamic mixed metamorphic rock and intermediate-acidic intrusive rock area, metamorphic bedrock in thrust napple slopes area, thermal contact metamorphic rock area, intermediate-acidic intrusive rock area, dynamic metamorphic rock area and gneiss area in faulted uplift core.

Machine Learning Assisted Interpretation: Integrated Fault Prediction Extraction Case Study from the Groningen Gas Field, Netherlands

Seismic fault detection using machine learning techniques, in particular the convolution neural network (CNN), is becoming a widely accepted practice in the field of seismic interpretation. Machine learning algorithms are trained to mimic the capabilities of an experienced interpreter by recognizing patterns within seismic data and classifying them. Regardless of the method of seismic fault detection, interpretation or extraction of 3D fault representations from edge evidence or fault probability volumes is routine. Extracted fault representations are important to the understanding of the subsurface geology and are a critical input to upstream workflows including structural framework definition, static reservoir and petroleum system modeling, and well planning and de-risking activities. Efforts to automate the detection and extraction of geological features from seismic data have evolved in line with advances in computer algorithms, hardware, and machine learning techniques. We have developed an assisted fault interpretation workflow for seismic fault detection and extraction, demonstrated through a case study from the Groningen gas field of the Upper Permian, Dutch Rotliegend; a heavily faulted, subsalt gas field located onshore, NE Netherlands. Supervised using interpreter-led labeling, we apply a 2D multi-CNN to detect faults within a 3D pre-stack depth migrated seismic dataset. After prediction, we apply a geometric evaluation of predicted faults, using a principal component analysis (PCA) to produce geometric attribute representations (strike azimuth and planarity) of the fault prediction. Strike azimuth and planarity attributes are used to validate and automatically extract consistent 3D fault geometries, providing geological context to the interpreter and input to dependent workflows more efficiently.