|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|
Decisions in E&P ventures are affected by Bias, Blindness, and Illusions (BBI) which permeate our analyses, interpretations and decisions. This one-day course examines the influence of these cognitive pitfalls and presents techniques that can be used to mitigate their impact. Bias refers to errors in thinking whereby interpretations and judgments are drawn in an illogical fashion. Blindness is the condition where we fail to see an unexpected event in plain sight. Illusions refer to misleading beliefs based on a false impression of reality. All three can lead to poor decisions regarding which work to undertake, what issues to focus on, and whether to forge ahead or walk away from a project. Strategic thinking and planning are key elements in an organisation’s journey to maximise value to shareholders, customers, and employees. Through this workshop, attendees will go through the different processes involved in strategic planning including the elements of organisational SWOT, business scenario and options development, elaboration of strategic options and communication to stakeholders. Examples are provided including corporate, business unit and department case studies. This seminar will teach participants how to identify, evaluate, and quantify risk and uncertainty in everyday oil and gas economic situations. It reviews the development of pragmatic tools, methods, and understandings for professionals that are applicable to companies of all sizes. The seminar also briefly reviews statistics, the relationship between risk and return, and hedging and future markets.
To drive progress in the field of data science, the authors propose 10 challenge areas for the research community to pursue. Because data science is broad, with methods drawing from computer science, statistics, and other disciplines, these challenge areas speak to the breadth of issues. Incorporating imagination into AI agents has long been an elusive goal of researchers in the space. Imagine AI programs that are able not only to learn new tasks but also to plan and reason about the future. Data scientists working with large data sets or in high-performance computational environments may find these programming languages essential to extracting data quickly and effortlessly.
The service companies plan to co-market an emerging well control system that can integrate with established managed-pressure-drilling components to enhance well construction safety and efficiency. The next step is to move toward optimization, then automation. An intelligent drilling optimization application performs as an adaptive autodriller. In the Marcellus Shale, ROP improved 61% and 39% and drilling performance, measured as hours on bottom, improved 25%. A real-time deep-learning model is proposed to classify the volume of cuttings from a shale shaker on an offshore drilling rig by analyzing the real-time monitoring video stream.
Maximizing stimulated natural and hydraulic fracture network is one of the primary hydraulic fracturing concerns for economic production from a horizontal shale gas well. Geomechanical facies and preexisting fractures in each stage are identified based on similarities in formation characteristics to optimize the locations of perforation clusters. This often requires analyzing large volumes of drilling, Logging While Drilling (LWD) and Measurement While Drilling (MWD) data. In this paper, we develop a methodology that calculates the mechanical specific energy (MSE) using real-time drill string acceleration signals directly from its definition. High resolution vibration signals have been collected using a tri-axial accerlometer, which was an auxiliary tool included in acoustic borehole imager. This technique provides a cost-efficient solution for engineered completion design. Furthermore, we adopt deep Convolutional Neural Network (CNN) with signal processing to build a data pipeline that effectively extracts patterns from dynamic acceleration signals for rock lateral MSE classification. First, we apply discrete wavelet transform and Short-Time Fourier Transform (STFT) for signal denoising and pattern recognition. Then we construct an image dataset using multi-scale image fusion at pixel level from 3 sensor channels, including axial, lateral acceleration spectrograms and zero-padded revolutions per minute (RPM). The resulted RGB image dataset includes 4,000 images of 5 MSE ranges with various rock strength conditions. Our results demonstrate that the proposed deep learning model can achieve more than 90% classification accuracy. The deep learning results, as a reference source, were applied in selected Marcellus Shale Energy and Environmental Lab (MSEEL) wells engineered completion located in the Marcellus shale gas site.
Sinha, Saurabh (University of Oklahoma, Los Alamos National Laboratory) | Pires De Lima, Rafael (Geological Survey of Brazil) | Lin, Youzuo (Los Alamos National Laboratory) | Y. Sun, Alexander (Bureau of Economic Geology - University of Texas at Austin) | Symon, Neill (Los Alamos National Laboratory) | Pawar, Rajesh (Los Alamos National Laboratory) | Guthrie, George (Los Alamos National Laboratory)
Due to international commitments on carbon capture and storage (CCS), an increase in CCS projects is expected in the near future. Saline aquifers and depleted hydrocarbon reservoirs with good seals and located in tectonically stable zones make an excellent storage formation option for geological carbon sequestration. However, stored carbon dioxide (CO2) takes a long time to convert into diagenetically stable form. Hence, ensuring the CO2 does not leak from these reservoirs in this time period is the key to any successful CCS project. Numerous methods are developed over the past couple of decades to identify the leaks which utilizes various types of geophysical, geochemical and engineering data. We demonstrate the automated leakage detection in CCS projects using pressure data obtained from Cranfield reservoir, Mississippi, USA. Our dataset consists of CO2 injection rates and pressure monitoring data obtained from a pressure pulse test. We first demonstrate the differences between the pressure pulse signal in case of a baseline pulse test and a pulse test with an artificially induced leak onsite. We then use machine learning techniques to automatically differentiate between the two tests. The results indicate that even simple deep learning architectures such as multi-layer feedforward network (MFNN) can identify a leak using pressure data and can be used to raise an early warning flag.
Evaluating the potential of the unconventional resources is a key for the development of this type of reservoirs. The currently adopted models for the well production forecast including decline curve analysis often fail to capture the complexity of flow performance by over-simplifying it and cannot produce reliable results due to the operational problems and most importantly the inadequate production history. In this study, a deep learning approach is developed to predict the long-term well performance based on a moderate duration of production data.
A data-driven procedure was implemented based on deep neural networks for flowrate predication using multivariate inputs. The production forecast was formulated as a time series regression problem where multiple inputs including tubing-head pressure and bottom-hole temperature are used as the input of a reverse model that estimates flow rate. Different recurrent neural networks (RNNs) including Long Short Term Memory, Gated Recurrent Units, and Bidirectional Recurrent Neural Networks were tested in this study to select the most time-efficient and accurate model of production forecasting.
The method presented in this paper provided a time efficient process which learned multi-domain sequence and was used to forecast production in unconventional resources. The developed deep learning networks did not require any feature handcrafting and could learn directly form the raw data. Reconstructed and predicted flowrates using deep learning was also used to estimate missing flowrate history. The study showed that deep neural networks have great capability to tolerate noise and optimize computation when multivariate input is used.
The technique can also be applied to other type of forecasting problems of prediction of pressure and rate in conventional reservoirs, prediction rate from temperature, and multi-well production forecasting.
Ashok, Pradeepkumar (The University of Texas at Austin) | Vashisht, Prabal (The University of Texas at Austin) | Kong, Hyeok (The University of Texas at Austin) | Witt-Doerring, Ysabel (The University of Texas at Austin) | Chu, Jian (The University of Texas at Austin) | Yan, Zeyu (The University of Texas at Austin) | van Oort, Eric (The University of Texas at Austin) | Behounek, Michael (Apache Corporation)
IADC dull bit grading is the current industry standard to assess the condition of a drill bit when it comes out of the hole. It is intended to capture the impact of drilling issues (e.g. drilling abrasive hard rock, drilling dysfunctions) on the bit and to improve future bit selection. However, the grading process is manual and subjective, making the bit grading outcome an inconsistent and unreliable metric. Recent advances in image processing and deep learning allow for bit grading to become more consistent and automated. Such a process is described in this paper.
The dataset used in this project consisted of multiple images (taken from different perspectives in a random manner) of used drill bits from 13 bit runs across multiple wells. As a preliminary step in developing the approach, only PDC bits were considered in this project. The first task was to identify all the cutters on a drill bit image using Convolutional Neural Networks (CNN). The CNN approach was chosen since it has shown remarkable success in solving the problem of object detection and classification in other fields. Next, the amount of damage to each cutter was quantified using image processing techniques. Finally, from information gathered in the previous steps, a holistic damage assessment of the drill bit was made.
The trained CNN was able to detect the cutters in an image to a high degree of accuracy. The accuracy of cutter detection was further improved through the use of heuristics that predict potential locations of cutters based on blade location and shape. The identification of unique cutters from a group of images of the same bit proved more challenging. Since the images could not be appropriately stitched together, each image was graded independently, and a holistic assessment of the bit was made by aggregation of the individual assessments. Additionally, not all of the cutters identified could be positively identified as damaged or not. For example, if the perspective that was available was at a right angle to the cutter's face, it is inherently not possible to quantify the damage. The computer-generated assessment of the bit was validated with collaborative assessments made by multiple human operators.
This paper presents a novel approach to bit damage classification that removes the subjective bias that comes with human evaluations. The application of deep learning techniques to cutter identification, damage detection and quantification is unique and has the potential to significantly improve bit design, selection, and thus, drilling efficiency.
Linking depositional properties and post-depositional diagenetic modifications of a rock with its petrophysical attributes remains a greatest challenge for carbonate rock characterization, formation evaluation and petrophysical rock typing. Generally, characterization of carbonate rock facies is labor intensive which requires an experienced geologist to interpret and integrate core, petrographic thin-sections and borehole image logs. In this approach, the carbonate lithofacies are identified with an emphasis on the diagenetic features, such as grain packing, micritization, cementation and dolomitization as well as diagenetic/karstic dissolution, and related connected or partial connected interparticle pores, intraparticle pores, separate and oversized vugs and micrite micro-porosity, etc. Here, we focused on developing deep learning based technique for automatizing manual facies identification process, a powerful tool to provide consistent and faster turnaround interpretations of geological facies for applications such as petrophysical parameter prediction.
In this paper, an architecture for unsupervised multi-class semantic segmentation of carbonate facies that incorporates deep U-Net based architecture is presented. The advantages of using such a network comes from adding skip connections which allows better flow of information in the network. This in return ensures comparable performances along with better feature representation for semantic segmentation tasks. Although many machine learning techniques have been previously applied for facies image analysis automation, the foundation is always the effectiveness of segmentation of multiple overlapping objects in the image. In case of carbonate rocks, diagenesis multiplies the heterogeneity complication. Therefore, in order to deal with this heterogeneity of carbonates we focused on unsupervised approaches because supervised learning methods can become very impractical due to the daunting task of manual feature labeling.
Multiple experiments are conducted on representative images of three types of carbonate facies (grainstone, rudstone, and packstone) to evaluate the performance of our segmentation algorithm and provide quantitative metrics useful for geological and petrophysical applications. Additionally, the segmentation algorithm is also used to detect primary resistive features from resistivity based borehole images. The consistent segmentation results have proved both the effectiveness and validity of the algorithm.
Navratil, Jiri (IBM T.J. Watson Research Center) | De Paola, Giorgio (Repsol) | Kollias, Georgos (IBM T.J. Watson Research Center) | Nadukandi, Prashanth (Repsol) | Codas, Andres (IBM's Brazil Research Laboratory) | Ibanez-Llano, Cristina (Repsol)
Despite considerable progress in the development of rapid evaluation methods for physics-based reservoir model simulators there still exists a significant gap in acceleration and accuracy needed to enable complex optimization methods, including Monte Carlo and Reinforcement Learning. The latter techniques bear a great potential to improve existing workflows and create new ones for a variety of applications, including field development planning. Building on latest developments in modern deep learning technology, this paper describes an end-to-end deep surrogate model capable of modeling field and individual-well production rates given arbitrary sequences of actions (schedules) including varying well lo-cations, controls and completions. We focus on generalization properties of the surrogate model which is trained given a certain number of simulations. We study its spatial and time interpolation and extrapolation properties using the SPE9 case, followed by a validation on a large-scale real field. Our results indicate that the surrogate model achieves acceleration rates of about 15000x and 40000x for the SPE9 and the real field, respectively, incurring relative error ranging between 2% and 4% in the interpolation case, and between 5% and 12% in the various spacial and time extrapolation cases. These results provide concrete measures of the efficacy of the deep surrogate model as an enabling technology for the development of optimization techniques previously out of reach due to computational complexity.