Advances in digital technologies have the potential to enhance model predictive capability and redefine its boundaries at various scale. Digital oil with accurate representation of atomistic components is a powerful tool to analyze both macroscopic properties and microscopic phenomena of crude oil under any thermodynamic conditions. Digital oil model presented in this paper is the key input in molecular chemistry modeling for designing chemical enhanced oil recovery formulation. Hence, it is constructed based on a fit-for purpose strategy focusing in oil components that have large contribution to microemulsion stability. Complete crude oil composition could comprise over 100,000 components. Lengthy simulation time is required to simulate all crude oil components which is impratical, despite the challenges to identify all crude oil components experimentally. Therefore, we established a practical experimental strategy to identify key crude oil components and constructed the digital oil model based on surrogate components. The surrogate components are representative molecules of the volatiles, saturates, aromatics and resins. Two-dimensional digital oil model, with aromaticity on one axis, and the size of the molecules on the other axis was constructed. We developed algorithm to integrate nuclear magnetic resonance response with architecture of the molecular structure. A group contribution method was implemented to ensure reliable representation of the molecular structure. We constructed the digital oil models for a field in Malaysia Basin. We validated the physical properties of the digital oil model with properties measured from experiment, predicted from molecular dynamics simulation and calculated from quantitative property-property relationship method. Good agreement was obtained from the validation, with less than 5% and 13% variance in crude density and Equivalent Alkane Carbon Number respectively, indicating that the molecular characteristic of the digital oil model was captured correctly. We adopted the digital oil model in molecular chemistry modeling to gain insights into microemulsion formation in chemical enhanced oil recovery formulation design. Digital oil is a robust tool to make predictions when information cannot be extracted from experimental data alone. It can be extended for engineering applications involving processing, safety, hazard, and environmental considerations.

The Daman marginal field is a prolific gas-producing clastic field with highly unconsolidated Paleo-Miocene sandstone formations and a wide variety of lithologies across multistack sand layers. As such, high-rate water packs (HRWPs) are the ideal completion method in many Mumbai fields. Because multistack reservoirs require good zonal isolation, and to prevent crossflow between reservoirs with different pressure regimes, multistack sand exclusion (MSSE) methodology was selected for primary well completions with minimum rig time and a high degree of treatment placement accuracy. From an operational standpoint, exploiting these layers using this method means more control points can be achieved across these heterogeneous layers, and the MSSE completion is ideal for multiple applications in a shorter period, helping sustain sand-circumscribed gas production from these unconsolidated layers. During the design phase, grain-size distributions and core study defined the sand range from generally clean, coarse, and sorted to poorly sorted, with high-fines content and clay rich. To address the unique challenges of deep offshore operations, formation technical difficulties, high-stakes economics, and the significant untapped potential from these Daman sands, the MSSE approach was designed and implemented in this field. Historically, for multistack wells, an HRWP is performed zone by zone whereby the process of sump packer installation, perforation run, deburr run, screen assembly installation, and pumping is repeated for each zone. In Well A, the MSSE system was applied without any repetition and all in one phase. All layers were perforated and positively isolated. Each interval was individually opened for the HRWP treatment using a low-friction low-residue carrier fluid. Using a high-packing-factor proppant at a higher rate, the well was treated sequentially from the bottom of the interval to the top. Many marginal fields in this basin have become uneconomical because of the high cost and complexity of sand control methodology. Therefore, reducing costs and time becomes vital to help ensure economic viability, as well as achieving significant operational efficiencies. Additionally, reducing near-wellbore (NWB) mechanical skin and ensuring good productivity from the reservoir are among the major solutions when implementing an MSSE completion. The methodology adopted significantly helped reduce expenditures by standardizing completion design, simplifying the core complexity, and enhancing overall reliability and operational efficiency. The optimized engineering workflow was fit for purpose, rather than the conventional “cookie-cutter” method to address sanding propensity in this field. This paper discusses the cutting-edge MSSE completion systems that focused on downhole completion and modifications for pumping operations. Additionally, the paper reviews challenges addressed during this campaign, workflow adapted, detailed strategy success factors, and positive results obtained during evaluation. This has helped reduce potential risks and improve reliability and performance, which can act as best practices and can be applied within similar fields.

Oil and Gas operations are now being "datafied." Datafication in the oil industry refers to systematically extracting data from the various oilfield activities that are naturally occurring. Successful digital transformation hinges critically on an organization's ability to extract value from data. Extracting and analyzing data is getting harder as the volume, variety, and velocity of data continues to increase. Analytics can help us make better decisions, only if we can trust the integrity of the data going into the system. As digital technology continues to play a pivotal role in the oil industry, the role of reliable data and analytics has never been more consequential. This paper is an empirical analysis of how Artificial Intelligence (AI), big data and analytics has redefined oil and gas operations. It takes a deep dive into various AI and analytics technologies reshaping the industry, specifically as it relates to exploration and production operations, as well as other sectors of the industry. Several illustrative examples of transformative technologies reshaping the oil and gas value chain along with their innovative applications in real-time decision making are highlighted. It also describes the significant challenges that AI presents in the oil industry including algorithmic bias, cybersecurity, and trust. With digital transformation poised to re-invent the oil & gas industry, the paper also discusses energy transition, and makes some bold predictions about the oil industry of the future and the role of AI in that future. Big data lays the foundation for the broad adoption and application of artificial intelligence. Analytics and AI are going to be very powerful tools for making predictions with a precision that was previously impossible. Analysis of some of the AI and analytics tools studied shows that there is a huge gap between the people who use the data and the metadata. AI is as good as the ecosystem that supports it. Trusting AI and feeling confident with its decisions starts with trustworthy data. The data needs to be clean, accurate, devoid of bias, and protected. As the relationship between man and machine continues to evolve, and organizations continue to rely on data analytics to provide decision support services, it is imperative that we safeguard against making important technical and management decisions based on invalid or biased data and algorithm. The variegated outcomes observed from some of the AI and analytics tools studied in this research shows that, when it comes to adopting AI and analytics, the worm remains buried in the apple.

Fluid properties are key factors for predicting single well productivity, well test interpretation and oilfield recovery prediction, which directly affect the success of ODP program design. The most accurate and direct method of acquisition is underground sampling. However, not every well has samples due to technical reasons such as excessive well deviation or high cost during the exploration stage. Therefore, analogies or empirical formulas have to be adopted to carry out research in many cases. But a large number of oilfield developments have shown that the errors caused by these methods are very large. Therefore, how to quickly and accurately obtain fluid physical properties is of great significance. In recent years, with the development and improvement of artificial intelligence or machine learning algorithms, their applications in the oilfield have become more and more extensive. This paper proposed a method for predicting crude oil physical properties based on machine learning algorithms. This method uses PVT data from nearly 100 wells in Bohai Oilfield. 75% of the data is used for training and learning to obtain the prediction model, and the remaining 25% is used for testing. Practice shows that the prediction results of the machine learning algorithm are very close to the actual data, with a very small error. Finally, this method was used to apply the preliminary plan design of the BZ29 oilfield which is a new oilfield. Especially for the unsampled sand bodies, the fluid physical properties prediction was carried out. It also compares the influence of the analogy method on the scheme, which provides potential and risk analysis for scheme design. This method will be applied in more oil fields in the Bohai Sea in the future and has important promotion value.

Well real-time flow rate is one of the most important production parameters in oilfield and accurate flow rate information is crucial for production monitoring and optimization. With the wide application of permanent downhole gauge (PDG), the high-frequency and large volume of downhole temperature and pressure make applying of deep learning technique to predict flow rate possible. Flow rate of production well is predicted with long short-term memory (LSTM) network using downhole temperature and pressure production data. The specific parameters of LSTM neural network are given, as well as the methods of data preprocessing and neural network training. The developed model has been validated with two production wells in the Volve Oilfield, North Sea. The field application demonstrates that the deep learning is applicable for flow rate prediction in oilfields. LSTM has the better performance of flow rate prediction than other five machine learning methods, including support vector machine (SVM), linear regression, tree, and Gaussian process regression. The LSTM with a dropout layer has a better performance than a standard LSTM network. The optimal numbers of LSTM layers and hidden units can be adjusted to obtain the best prediction results, but more LSTM layers and hidden units lead to more time of training and prediction, and LSTM model might be unstable and cannot converge. Compared with only downhole pressure or temperature data used as input parameters, flow rate prediction with both of downhole pressure and temperature used as input parameters has the higher prediction accuracy.

The ultimate goal of oilfield development is to maximize the investment benefits. The reservoir performance prediction is directly related to oilfield investment and management. The traditional strategy based on numerical simulation has been widely used with the disadvantages of long run time and much information needed. It is necessary to form a fast and convenient method for the oil production prediction, especially for layered reservoir. A new method is proposed to predict the development indexes of multi-layer reservoirs based on the injection-production data. The new method maintains the objectivity of the data and demonstrates the superiority of the intelligent algorithm. The layered reservoir is regarded as a series of single layer reservoirs on the vertical direction. Considering the starting pressure gradient of non-Newtonian fluid flow and the variation of water content in the oil production index, the injection-production response model for single-layer reservoirs is established. Based on that, a composite model for the multi-layer reservoir is established. For model solution, particle swarm optimization is applied for optimization of the new model. A heterogeneous multi-layer model was established for validation of the new method. The results obtained from the new proposed model are in consistent with the numerical simulation results. It saves a lot of computing time with the incorporation of the artificial intelligence methods. It showed that this technique is valid and effective to predict oil performance in layered reservoir. These examples showed that the application of big data and artificial intelligence method is of great significance, which not only shortens the working time, but also obtains relatively higher accuracy. Based on the objective data of the oil field and the artificial intelligence algorithm, the prediction of oil field development data can be realized. This technique has been used in nearly 100 wells of Bohai oilfields. The results showed in this paper reveals that it is possible to estimate the production performance of the water flooding reservoirs.

The tracer-based production logging technology can be used to obtain the well production data continuously for several years without the need for risky well interventions and expensive equipment. The paper examines the case of placing polymer-coated tracers dopped proppant in a horizontal well with ten multi-stage frac intervals and using two different tracers dopped proppant codes for two frac ports (the first and the last ones) to identify the performance of the far and near zones of a hydraulic fracture. Upon the completion of the hydraulic fracturing operations, the collected reservoir fluid samples were studied in the laboratory. Chemical tracers contained in the samples were detected by flow cytofluorometry using custom-tailored machine learning-based software. The studies helped identify the productivity of each frac port, calculate the contribution of each port in percentage points, and also evaluate the productivity of the near and far hydraulic fracture zones in the first and the last intervals. The analysis provided data on the exact content of oil and water in the production profile for each frac interval. The results of tracer-based logging in the well in question revealed that the interval productivity is changing in the course of several months of surveillance. The most productive ports and those showing increasing oil flow rate were identified during quantitative analysis. The use of tracer dopped proppant with different codes within one multi-stage frac interval enabled detecting a peak release of chemical tracers from the far fracture zone in the initial periods of well operation followed by a consistent smoothing of the far and near zones’ production profiles. Laboratory analysis of reservoir fluid samples and hydraulic fracturing simulations proved the uniform distribution of proppant across the entire reservoir pay zone and laid the foundation for further research required to better understand the fracture geometry and reduce uncertainties in production optimization operations.

Subsurface engineers pivot on surveillance of reservoir performance for future production rates and plan the optimization strategies at earliest. There are some techniques preferred for unconventional reservoirs such as numerical simulation and decline curve analysis (DCA) for production forecasting, but the uncertainty of uneconomical well test data often occurs in unconventional resources. Moreover, reservoir engineers can also hit a tailback in optimizing and tuning the model. Further, for DCA this approach is only appropriate for well/reservoir that are under boundary dominant flow regime, whereas fracture dominant flow regime is often observed for a longer period in unconventional hydraulically fractured reservoirs. Therefore, to resolve this issue, oil & gas industry (O&G) can adopt AI (Artificial Intelligence) based Algorithms for production forecasting. This paper presents a data-driven algorithm, known as Artificial Neural Networks (ANNs), along with time series forecasting that is a well-known statistical technique. Machine learning model trained by a past well performance data such as tubing head pressure (THP), flowing bottom-hole pressure can predict future production rates. This can be an efficient technique for subsurface engineers to monitor and optimize well performance. Time series neural networks were used for training the model at top and bottom node of the well with variating pressures in the past. After training and validation, the model predicted a target parameter that was gas rate. ANNs are inspired by biological neurons that are present in human brain, a powerful computing tool to make decisions after fueling itself with data. Moreover, prediction (t+1) nonlinear automated regression is preferred for accurate step ahead. Production rates and constraints of unconventional reservoirs were used to train the model. In our results, the NN based model gave the co-efficient of determination (R 2 ) of 0.996 that shows nearly an exact precision. Furthermore, the values generated from NN Model and Arp's decline curve calculations were plotted for validation and it turned out that ANN can accurately predict the parameters. The Neural Network model is a novel approach for production forecasting, of unconventional reservoirs and help engineers in corporate decision making. This approach can mitigate the need of uneconomical well test operations and further provide confidence to production engineers in terms of data and result expectations.

Adequate implementation of probabilistic dynamic modelling workflow allows to explore full range of solutions and helps in better decision making. Probabilistic modelling workflows comprise of a series of steps. This paper closely scrutinizes three key steps of a probabilistic workflow – sensitivity analysis, experimental design and model selection for decision making. There are different methods and processes practiced in the industry for these key steps. A comparative study of commonly available and practiced techniques is performed, and their pros and cons are demonstrated using open available Brugge field model. In the first part the commonly used tornado/one variable at a time (OVAT) is compared against other global sensitivity techniques suchas ANOVA (Analysis of Variance) and multi-regression analysis. The difference among the techniques is explained and its application is demonstrated through field example. It is shown that the global sensitivity analysis is more robust and provides better results but at a higher computational cost. Next experimental design techniques are reviewed with focus on the optimum experimental design. Key concepts of experimental design such as aliasing, resolution and confounding pattern are explained. It is shown that an appropriate experimental design can be used to avoid confounding of key parameters. The confounding patterns have been demonstrated for factorial design experiments. The concept is shown with the example of Brugge model. Model selection after history matching process is a challenging task. In the last part of the paper two different multi-dimensional visualization techniques PCA (Principal Component Analysis) and t-SNE (t-Stochastic Neighbour Embedding) are explained and utilized to visualize the ensemble of history matched models. These techniques in combination with k-mean clustering has been used to select representative models for prediction. The overall uncertainty captured by the representative models is compared to the full ensemble of history matched models. It is shown that clustering along with visualization provide a robust framework for model selection. The paper is intended to provide a good understanding of some of the key steps of the probabilistic modelling workflow. The technical concept behind each technique is briefly described but the main focus is on the practical aspects and implemtation of these techniques.

Optimizing multiple assets under uncertain techno-economic conditions and tight government policies is challenging. Operator needs to establish flexible Plan of Development (POD)s and put priority in developing multiple fields. The complexity of production and the profit margin should be simultaneously evaluated. In this work, we present a new workflow to perform such a rigorous optimization under uncertainty using the case study of PHE ONWJ, Indonesia. We begin the workflow by identifying the uncertain parameters and their prior distributions. We classify the parameters into three main groups: operations-related (geological complexity, reserves, current recovery, surface facilities, and technologies), company-policies-related (future exploration plan, margin of profit, and oil/gas price), and government-related (taxes, incentives, and fiscal policies). A unique indexing technique is developed to allow numerical quantification and adapt with dynamic input. We then start the optimization process by constructing time-dependent surrogate model through training with Monte Carlo sampling. We then perform optimization under uncertainty with multiple scenarios. The objective function is the overall Net Present Value (NPV) obtained by developing multiple fields. This work emphasizes the importance of the use of time-dependent surrogate approach to account risk in the optimization process. The approach revises the prior distribution with narrow-variance distribution to make reliable decision. The Global Sensitivity Analysis (GSA) with Sobol decomposition on the posterior distribution and surrogate provides parameters’ ranking and list of heavy hitters. The first output from this workflow is the narrow-variance posterior distribution. This result helps to locate the sweet spots. By analyzing them, operator can address specific sectors, which are critical to the NPV. PHE ONWJ, as the biggest operator in Indonesia, has geologically scattered assets, therefore, this first output is essential. The second output is the list of heavy hitters from GSA. This list is a tool to cluster promising fields for future development and prioritize their development based on the impact towards NPV. Since all risks are carried by the operator under the current Gross Split Contract, this result is advantageous for decision-making process. We introduce a new approach to perform time-dependent, multi-asset optimization under uncertainty. This new workflow is impactful for operators to create robust decision after considering the associated risks.

Predicting permeability in low-medium quality reservoirs (> 10 md to <100mD) is important in brownfields since many of them can still produce hydrocarbons. Developing an approach relating geologic properties to permeability prediction can increase field reserves and extend producing life. The common practice of predicting permeability includes linear regressions of core-porosities vs. core-permeabilities applying different lithofacies. However, these methods discount data scattering around regression-lines. This paper describes an innovative-technique for permeability prediction that combines rock-types, flow-zone-indicator (FZI), and machine-learning techniques (ML). FZI is a reservoir-flow-unit that controls hydraulic fluid-flow and is influenced by pore-geometry resulting from diagenetic-processes. In reservoirs, pore-geometry usually is heterogenous due to mineral-composition, rock-texture, cementation, and compaction. Thus,the commonly used permeability equation of Kozeny-Carman (KC) equation still can be used but it needs to be modified for better connecting FZI to hydraulic-flow-units. The modified KC equation incorporates heterogeneous poregeometry as a non-linear-function of porosity by adding cementation-exponent (m) into the equation, where the original KC equation assumes m is equal to one. The semi-log cross-plot between Reservoir-Quality-Index (RQI) vs. PHIZ*Por (m-1) (or FZI m ) from the modified KC equation can be constructed using rock-type class. The ML approach was applied to predict FZI groups using 4 standard-logs: gamma-ray, resistivity, density, and neutron-porosity. Cross-plots of RQI vs. PHIZ (conventional FZI) can be compared to RQI vs. PHIZ*Por (m-1) (modified FZI model) usingdata from 11cored wells in oil field offshore Malaysia. The modified FZI model shows less data clustering compared to the conventional FZI model, shown by higher R 2 coefficient correlation accuracy. The proposed modified FZI model shows narrower permeability range at low porosity which is a good indication of more accurate hydraulic-flow-unit interpretation. When applying the original and modified FZI models, each lithofacies may occur in more than one hydraulic-flow-unit due to pore-geometry difference within the same lithofacies. Furthermore, the hydraulic-flow-unit generated by the modified FZI model is more sensitive to total porosity when comparing to original FZI model. Each generated hydraulic-flow-unit has better correlation to total porosity and with less scattered permeability at the same porosity. The permeability calculated by modified FZI model was then verified with core permeability showing an excellent overall match. On the ML technique, the "Random Forests" technique will be utilized due to recognized as one of the most recent ML algorithm(s) developed as an innovative technique based on both classifications and regression trees techniques. The Random Forests technique has shown its great accuracy on predictive exactness for these challenge permeability estimations. The prediction quality was benchmark by R 2 value of > 0.9 for all crossplots (porosity, permeability, and water saturation) when comparing to routine core analysis lab measurements.

G oilfield in China is a dual medium oilfield dominated by carbonate reservoir, where faults and fractures are well developed. After G oilfield was put into production, water cut has rised quickly and oil production decreased rapidly. The heterogeneity of fracture development is very strong. Because of unclear understanding of the fracture distribution, oilfield development plan can not accurately identify risks and potentials accurately. As a result, production goals are not as expected. Due to limited core data and imaging logging data, conventional methods cannot characterize fracture information in this oilfield accurately and comprehensively. In order to solve the problems mentioned above, this paper makes full use of logging, geology, seismic and other data to carry out fine characterization of fracture with multi-information fusion fracture modeling based on neural network. Firstly, neural network technology is used to predict the fracture density curves of wells with conventional logging data. Secondly, combined the nonlinear fusion of multiple pre-stack seismic attributes including Tectonic Stress Field, P-Wave AVO and Seismic Discontinuity Detection based on neural network, the fracture density probability model is created to predict three-dimensional distribution of fractures. Thirdly, the fracture Intensity model is set up under the constraint of strict variation function analysis and fracture density probability model. Finally, fracture parameters are obtained according to core observation and imaging logging data, and the fracture network is established by geostatistics modeling method and coarsened equivalent into fracture property model. Blind well validation and dynamic validation are used to verify the validity of fracture model. Under the guidance of research results in this paper, the oilfield development plan is optimized and adjusted. It has also been proved that fracture characteristic of new wells is consistent with pre-drilling prediction result. Most areas with relatively high fracture development degree are corresponding to well groups with faster water cut rising rate. Against the backdrop of global oil economic downturn, fracture modeling technology studied in this paper provides strong technical guidance for oilfield development plan, therefore reducing oilfield development risks and maximizing economic benefits.

Real-time location information is essential in the hazardous process and construction areas for safety and emergency management, security, search and rescue, and even productivity tracking. It's also crucial during pandemics such as the COVID-19 pandemic for contact tracing to isolate those who came to the proximity of infected individuals. While global positioning systems (GPS), can address the demand for location awareness in outdoor environments, another accurate location estimation technology for indoor environments where GPS doesn't perform well is required. This paper presents the development and deployment of an end-to-end cost-effective real-time personnel location system suitable for both indoor and outdoor hazardous and safe areas. It leverages on facility wireless communication systems, wearable technologies such as smart helmets and wearable tags, and machine learning. Personnel carries the client device which collects location-related information and sends it to the localization algorithm in the cloud. When the personnel moves, the tracking dashboard shows client location in real-time. The proposed localization algorithm relies on wireless signal fingerprinting and machine learning algorithms to estimate the location. The machine learning algorithm is a mix of clustering and classification that was designed to scale well with bigger target areas and is suitable for cloud deployment. The system was tested in both office and industrial process environments using consumer-grade handphones and intrinsically safe wearable devices. It achieved an average distance error of less than 2 meters in 3D space.

In PTTEP's offshore fields, more than 100 units of hybrid Solar-TEG power with individual VRLA battery banks are installed on wellhead platforms for powering the process. With large number of various equipment in remote locations, traditional maintenance approach highly consuming resources is, thus, not cost effective especially in oil price crisis. In 2019, "Hybrid Power Solar-TEG Predictive Maintenance" project is established to develop a predictive model and transform maintenance process to total predictive maintenance. The project was begun with three platforms as a pilot project. The operating model was built by machine learning using various historical data recorded in PI system, records of maintenance data and other relevant information such as manufacturer manual, international standard and related white papers. The modelled algorithm was embedded in an application which was developed by Python to predict the ageing and performance of battery banks on pilot wellhead platforms. In 2020, the project continues to build the model of Thermo-Electric Generator (TEG) and extend the coverage location for additional thirty-seven (37) platforms. Lower Depth of Discharge (DoD), higher ambient temperature and lower charging performance are signs of battery’s deterioration while lower supplied current from power source is sign of their underperformance. All parameters were ingested to conduct pattern recognition to make algorithm be able to predict the remaining life of the key equipment. The Eyeball method is conducted to train algorithm the various charging patterns by the developers with aim to evaluate the DoD of battery bank. Apart from battery life prediction, DoD is employed to determine the energy left in battery from night operation to indicate the remaining run time duration. By leveraging machine learning, all failure patterns are recognized. The application is operating real?time and provide early alarm to all person-in-charge when failure potential is realized. The results are visualized on PowerBI to provide the latest status of power units of each platforms. From above, the maintenance approach is thus completely converted from Run-to-Failure to Predictive Maintenance. The long lead spare parts e.g. battery cells could be procured in advance. Spare inventory can be optimized per actual demand. In addition, the offshore supervisor could accurately identify the defective battery banks and proactively recover them in time to minimize unplanned shutdown. The modelled algorithm was in-house developed based on technical information and maintenance records. Although the system goes live, the preventive maintenance according to IEEE1188 is still retained for further collecting more field data to improve accuracy of the model. In addition, the model’s analyzed information, such as battery run time and DoD, has revealed the hidden actual design margin of power system. The platform CAPEX can be thus deducted by removing such excess margin.

Tight gas sandstone reservoirs vary widely in terms of rock type, depositional environment, mineralogy and petrophysical properties. For this reason, estimating their permeability is a challenge when core is not available because it is a property that cannot be measured directly from wire-line logs. The aim of this work is to create an automatic tool for rock microstructure classification as a first step for future permeability prediction. Permeability can be estimated from porosity measured using wire-line data such as derived from density-neutron tools. However, without additional information this is highly inaccurate because porosity-permeability relationships are controlled by the microstructure of samples and permeability can vary by over five orders of magnitude. Experts can broadly estimate porosity-permeability relationships by analysing the microstructure of rocks using Scanning Electron Microscopy (SEM) or optical microscopy. Such estimates are, however, subjective and require many years of experience. A Machine Learning model for the automation of rock microstructure determination on tight gas sandstones has been built using Convolutional Neural Networks (CNNs) and trained on backscattered images from cuttings. Current results were obtained by training the model on around 24,000 Back Scattering Electron Microscopy (BSEM) images from 25 different rock samples. The obtained model performance for the current dataset are 97% of average of both validation and test categorical accuracy. Also, loss of 0.09 and 0.089 were obtained for validation and test correspondingly. Such high accuracy and low loss indicate an overall great model performance. Other metrics and debugging techniques such Gradient-weighted Class Activation Mapping (Grad-CAM), Receiver Operator Characteristics (ROC) and Area Under the Curve (AUC) were considered for the model performance evaluation obtaining positive results. Nevertheless, this can be improved by obtaining images from new already available samples and make the model generalizes better. Current results indicate that CNNs are a powerful tool and their application over thin section images is an answer for image analysis and classification problems. The use of this classifier removes the subjectivity of estimating porosity-permeability relationships from microstructure and can be used by non-experts. The current results also open the possibility of a data driven permeability prediction based on rock microstructure and porosity from well logs.

The execution phase of the wells technical assurance process is a critical procedure where the drilling operation commences and the well planning program is implemented. During drilling operations, the real-time drilling data are streamed to a real-time centre where it is constantly monitored by a dedicated team of monitoring specialists. If any potential issues or possible opportunities arise, the team will communicate with the operation team on rig for an intervention. This workflow is further enhanced by digital initiatives via big data analytics implementation in PETRONAS. The Digital Standing Instruction to Driller (Digital SID) is a drilling operational procedures documentation tool meant to improve the current process by digitalizing information exchange between office and rig site. Boasting multi-operation usage, it is made fit to context and despite its automated generation, this tool allows flexibility for the operation team to customize the content and more importantly, monitor the execution in real-time. Another tool used in the real-time monitoring platform is the dynamic monitoring drilling system where it allows real-time drilling data to be more intuitive and gives the benefit of foresight. The dynamic nature of the system means that it will update existing roadmaps with extensive real-time data as they come in, hence improving its accuracy as we drill further. Furthermore, an automated drilling key performance indicator (KPI) and performance benchmarking system measures drilling performance to uncover areas of improvement. This will serve as the benchmark for further optimization. On top of that, an artificial intelligence (AI) driven Wells Augmented Stuck Pipe Indicator (WASP) is deployed in the real-time monitoring platform to improve the capability of monitoring specialists to identify stuck pipe symptoms way earlier before the occurrence of the incident. This proactive approach is an improvement to the current process workflow which is less timely and possibly missing the intervention opportunity. These four tools are integrated seamlessly with the real-time monitoring platform hence improving the project management efficiency during the execution phase. The tools are envisioned to offer an agile and efficient process workflow by integrating and tapering down multiple applications in different environments into a single web-based platform which enables better collaboration and faster decision making.

The technique of interwell tracer testing is considered as one of the most effective method to identify the thief zone (TZ) in reservoirs. However, in heavy oil reservoirs, tracer breakthrough curves are mostly parabolic and unimodal, thus resulting in slight differences between curves. It is inefficient and inaccurate to identify different types of curves with traditional methods applied to characterize the levels of TZs. In this paper, convolutional neural network (CNN) is applied to construct a classification model for the automatic identification of the levers of TZs. According to the TZs criteria specified on the field, the analytical tracer transport model was applied to generate 3000 curves as the sample, which can meet the requirements of model training accuracy. In the meantime, One-hot encoding, Xavier initialization, Adam optimizer, and mini-batch normalization were used to construct the model, and the key parameters are optimized to improve the performance of the model. The results show that the appropriate activation function is ReLU and the optimal dropout rate is 0.5. Moreover, the construction of CNN with discrete data points (DDP-CNN) as input contributed to a further improvement of classification accuracy of tracer curves. The accuracy of DDP-CNN in training set is 0.96, which is 14% and 23% higher than random forest (RF) and k-means, respectively. In practical applications, DDP-CNN proves capable to correctly classify 88 of the 100 curves.

Sonic data are commonly acquired in exploration, appraisal, and development wells using wireline, logging-while-drilling, or through-the-bit conveyance for applications within petrophysics, geophysics, geomechanics, and geology disciplines. The measurement data require processing to obtain elastic wave slownesses (inverse of velocity) and associated attributes before the results can be used in petrotechnical workflows. The objective of the digital transformation is to streamline and automate the processing workflow to reduce user intervention and turnaround time while increasing the accuracy of results and possibly extracting more answers by fully utilizing all waveform attributes, which consequently benefits downstream applications. There are four workflows that are the foundation of the transformation. They support the overall goals of reducing user interactions and providing robust results in a timely manner for continuous slowness logs. First, data-driven inversions done during acquisition with automatic quality control and interpretation flags immediately provide assurance about the data quality and identify formation intervals that require further evaluation. Second, automatic dipole-flexural shear extraction is done using physics-based machine learning (ML) where purely data-driven models are inadequate due to borehole or geological conditions. The physics-based ML utilizes cloud-based computing that is needed for large volume synthetic data generation and neural network training. Third, a multiresolution analysis of the monopole waves for the compressional slowness uses automatic peak detection on multiple receiver levels removing any subjective manual labeling after the semblance processing. Finally, the multimode (flexural and Stoneley) inversion determines anisotropic constants and accounts for mud-speed variations in the borehole, including detailed uncertainties. The new methods address underlying concerns most users and waveform processing experts already observe in their sonic deliverables. Enabling wellsite algorithms to be more automatic and data driven improves the robustness of the field deliverables and provides insight into the quality of the data. For the shortcomings with regards to borehole or geological conditions such as laminations, sharp lithological transitions, or the presence of anisotropy, the physics-based ML is shown to honor the physics of the dipole flexural mode, while the multiresolution for the monopole provides physics-based reasoning for discrepancies between the geological layering and receiver aperture. By incorporating the range of results derived from the inversions with advanced interpretations such as transversely isotropic constants, these uncertainties can be further used in stochastic models in downstream workflows. All these methods are fully automated and can be done in a short timeframe to be used without doubt in operations.

The restriction or inability of the drill string to reciprocate or rotate while in the borehole is commonly known as a stuck pipe. This event is typically accompanied by constraints in drilling fluid flow, except for differential sticking. The stuck pipe can manifest based on three different mechanisms, i.e. pack-off, differential sticking, and wellbore geometry. Despite its infrequent occurrence, non-productive time (NPT) events have a massive cost impact. Nevertheless, stuck pipe incidents can be evaded with proper identification of its unique symptoms which allows an early intervention and remediation action. Over the decades, multiple analytical studies have been attempted to predict stuck pipe occurrences. The latest venture into this drilling operational challenge now utilizes Machine Learning (ML) algorithms in forecasting stuck pipe risk. An ML solution namely, Wells Augmented Stuck Pipe Indicator (WASP), is developed to tackle this specific challenge. The solution leverages on real-time drilling database and supplementary engineering design information to estimate proxy drilling parameters which provide active and impartial pattern recognition of prospective stuck pipe events. The solution is built to assist Wells Real Time Centre (WRTC) personnel in proactively providing a holistic perspective in anticipating potential anomalies and recommending remedial countermeasures before incidents happen. Several case studies are outlined to exhibit the impact of WASP in real-time drilling operation monitoring and intervention where WASP is capable to identify stuck pipe symptoms a few hours earlier and provide warnings for stuck pipe avoidance. The presented case studies were run on various live wells where restrictions are predicted stands ahead of the incidents. Warnings and alarms were generated, allowing further analysis by the personnel to verify and assess the situation before delivering a precautionary procedure to the rig site. The implementation of the WASP will reduce analysis time and provide timely prescriptive action in the proactive real-time drilling operation monitoring and intervention hub, subsequently creating value through cost containment and operational efficiency.

Traditional two-phase relative permeability upscaling requires the fine-scale two-phase flow simulation over the target regions/blocks. It can be very computationally expensive especially for cases with multiple (hundreds of) geological realizations (as commonly used in subsurface uncertainty quantification or optimization). In this paper, we develop a machine learning assisted relative permeability upscaling procedure, in which the full numerical upscaling is performed for only a portion of the coarse blocks, while the upscaled functions for the rest of the coarse blocks are calculated by the machine learning algorithm. The upscaling procedure was tested for generic (left to right) flow problems using 2D models for scenarios involving multiple realizations. Numerical results have shown that the coarse-scale simulation results using the newly developed machine learning assisted upscaling procedure are of similar accuracy to the coarse results using full numerical upscaling. Because the fine-scale numerical simulation is only performed for a small fraction of the model, significant speedup is achieved.