|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|
This paper covers the development of a key component of an internal system to report invisible lost time (ILT) metrics across drilling operations. Specifically this paper covers the development of a generalizable rig state engine based on the application of supervised machine learning. The same steps used in the creation of the production rig state engine are appled here to a smaller data set to demonstrate both the tractability of the problem and the methods used to create the rig state engine in the production system.
The project objective was to provide efficiency and engineering metrics in a central repository covering operated regions. The system is designed to require minimal user configuration and management and provides both historic and near real time analysis to deliver a rich resource for offset comparison and benchmarking.
Identifying rig-state is at the heart of every performance and engineering analysis system. This can be thought of as a machine learning classification problem. A large supervised learning set was constructed and used to train classification models which were compared for accuracy. A key success metric was the ability to generalise the selected model across different operations. Output from the rig-state classifier was then used to derive KPI data which was presented through a web based front end. A pilot system was then developed using agile principles allowing for rapid user engagement. Testing demonstrated that the system can support all real time operations within the company simultaneously and rapidly process historic well data for offset benchmarking. The cloud-based architecture allows rapid deployment of the system to new groups significantly reducing deployment costs. The system provides a foundation for onward data science and more advanced functionality.
Minimal configuration, cloud storage and processing, combining contextual data with real-time rig data, near-real-time and historic analysis capabilities, rapid deployment, low cost, high accuracy and consistent metrics are all key and proven value drivers for the system. The output data is aso a valuable resource for additional machine learning and data science projects.
When drilling complex wells, such as those with long lateral sections, the friction forces become significantly high, which can impede advancement of the drill string and reduce drilling performance. In these situations, Axial Oscillation Tools (AOT) could be used to introduce an axial vibration to the drill string. By locally reducing the friction forces, better transmission of weight to the drill bit is possible and an increase in the rate of penetration occurs. However, to optimize the use of these tools, predictive modeling is necessary to assess their effect on drilling characteristics.
A new modeling approach is proposed to accurately model the effect of the AOT on drilling operations without the need to carry out resource-intensive and time-consuming dynamic computations. To estimate the influence length (
The model was applied to a real case study, and an agreement between the modeling results and field measurements regarding the influence of the AOT was obtained. Moreover, the effect of the excitation force and rate of penetration on the drill string tension profile was investigated. This work should enable drilling engineers to better optimize the position of AOT along the drill string and to maximize its efficiency.
Gilles, Pelfrene (Varel International Energy Services) | Olivier, Stab (Mines ParisTech, PSL University, Centre de géosciences) | Danny, Tilleman (Varel International Energy Services) | Thomas, Gallifet (Varel International Energy Services) | Bruno, Cuillier (Varel International Energy Services) | Julien, Carlos (Varel International Energy Services)
The bit-rock interaction has long been studied to assess PDC drill bit performance, which is driven both by cutting and non-cutting parts of the drill bit. While the cutter-rock interaction has been studied by many authors in the literature, only a few studies have focused on the interaction between the rock and non-cutting parts of the drill bit.
In this paper, we introduce a new method designed to model the interaction between the whole drill bit and the rock formation within a full three-dimensional framework. This approach is based on a generic computational geometry algorithm which simulates the drilling process considering both the drill bit and the hole being drilled as a set of 3D meshed surfaces. The volume of rock removed by the PDC cutters as well as the area and the volume of contact between the rock and the non-cutting parts of the drill bit can be computed with a high accuracy based on the 3D CAD model of the drill bit.
The in-house drill bit simulator implementing this algorithm primarily allows the engineer to estimate how bit-rock interactions distribute between cutting and non-cutting parts of the drill bit and to balance the bit design in the 3D space accordingly over a given range of drilling parameters. This approach has been brought to the field in order to address cutter breakage based on rubbing contacts optimization. Field results associated to some case studies in US shale plays and Canada are described and clearly show that contact points predictions closely match field observations. Moreover, design modifications applied following this process have led to an overall increase in bit performance and bit durability while preventing core-out issues.
The bit design methodology presented in this paper is dedicated to design drill bits whose interaction with the rock formation is predicted with a higher accuracy by accounting for the exact 3D shape of the drill bit.
Gul, Sercan (The University of Texas at Austin) | Johnson, Mitchell David (The University of Texas at Austin) | Karimi Vajargah, Ali (The University of Texas at Austin) | Ma, Zheren (The University of Texas at Austin) | Hoxha, Besmir Buranaj (The University of Texas at Austin) | van Oort, Eric (The University of Texas at Austin)
Managing drilling margins in challenging wells requires precise prediction of frictional pressure losses and equivalent circulating density (ECD). Current hydraulic models in the industry fail to accurately predict the frictional pressure losses of certain mud formulations in turbulent flow due to the complex behavior of long-chain polymer additives. These additives facilitate friction reduction in certain flow regimes. This reduction depends on several parameters, such as molecular weight and chemical composition of the polymers, making it difficult to quantify using existing models. In this paper, a data-driven approach is proposed to precisely predict frictional pressure losses for polymer-based fluids.
A flow-loop was constructed to measure frictional pressure-losses of several polymer-based non-Newtonian fluids under laminar, transitional, and turbulent flow regimes. Pressure loss data was obtained for fluids with different polymer concentrations at various temperatures using differential pressure measurement. A database of the experimental data was compiled and used to build a predictive model for frictional pressure prediction using advanced machine learning techniques. The proposed approach has general validity and can be extended to any type of well construction fluid (used in drilling, completion, stimulation, or workover).
Results for frictional pressure loss predictions from the proposed data-driven approach were compared with both the experimental data and widely used industry models. An excellent agreement was observed between the proposed approach and the experimental results, demonstrating the applicability of this approach for hydraulic modeling of polymer-based fluids. The improvements are particularly noticeable at higher polymer concentrations in the turbulent flow regime, where the average percentage discrepancy between the existing models and the experimental data can be as high as 45%.
The proposed approach in this study is particularly valuable for wells with a narrow drilling margin and concerns about the ability to manage ECDs (such as slim hole wells, deepwater wells or extended reach wells). It can assist with better planning and avoiding non-productive time and drilling problems such as lost circulation, stuck pipe, wellbore instability, and well control events. Its adaptability to a wide range of fluids using an expanded database makes it particularly attractive as a practical solution to this challenging problem.
Solid particles in suspension in a fluid, like barite, lost circulation material (LCM), cuttings or cavings, influence the pressure losses that are experienced when pumping or moving a drill-string in a borehole. As the volume fractions of those different solid particles varies along the hydraulic circuit, it is desirable to estimate the impact of local solid concentrations on pressure drops.
The influence of solid particles on the rheological behavior of fluid has mostly been studied for Newtonian fluids, but very little experimental work has been published for non-Newtonian fluids like drilling muds. For that reason, a series of measurements, made with a scientific rheometer has been conducted on a typical KCl/Polymer water-based mud. The experimental investigation covers the effect of particle concentration on the rheological behavior of the mix, in conjunction with the particle size.
The change of rheological behavior is slow at low solid concentrations but increases exponentially with larger proportions of solid in suspension. Furthermore, the increase of effective viscosity is larger with fine particles than with coarser ones. Empirical formulas are proposed to describe how the original Herschel-Bulkley rheological behavior of a base fluid can be modified to incorporate the effects of the variation of solid concentrations in the fluid mix.
All these results are based on measurements made with a scientific rheometer. As computerized and high precision rheometers are usually not available at the rig site, we describe a methodology to utilize standard model 35 rheometer measurements to estimate the pressure loss gradient as a function of the volumetric solid content.
The use of advanced solid-state gyroscopic sensors has now become both a viable and practical option for high accuracy wellbore placement, with the potential to out-perform traditional mechanical gyroscopic systems. This paper describes how the contributions of the new gyroscope technology are causing service providers to reconsider current survey practices, and to examine how the new gyroscopic survey tools can be best used for wellbore surveying and real-time wellbore placement.
The simultaneous application of multiple survey tools, largely made possible as a result of the unique attributes of solid-state gyroscopic sensors (including small size and significant power reduction), has clear benefits in terms of enhanced well placement, reliability and the detection of gross errors in the survey process. Further benefits accrue through the combination of different, but complimentary survey methods. This paper focuses mainly on the benefits of combining gyroscopic and magnetic measurements to reduce or remove the known errors related to the Earth's magnetic field to which magnetic survey systems are susceptible; errors in total magnetic field, declination and dip angle.
In this context, the use of statistical estimation techniques based on performance models of the survey systems used is described. For post-drilling surveys (using drop survey tools or wireline-conveyed tools for example), post-run analysis of the data using least-squares estimation techniques is appropriate. Alternative methods capable of achieving real-time data correction during drilling are also described and results are presented to demonstrate the potential for enhanced magnetic survey performance.
The principles described may be used when running basic magnetic measurement while drilling (MWD) systems, and for systems that employ field correction methods, such as the various in-field referencing (IFR) techniques, that are frequently used. The proposed methodology is of particular benefit in the former case, allowing enhanced magnetic surveying to be achieved without the need for expensive and complex magnetic field correction procedures. The potential also exists either to identify or to correct possible errors in the IFR data when such methods are used. This information may be of great value for the safe drilling of additional wells in the same region.
Driller's instructions are written with the expectation that the specific procedures they contain will be followed by the driller during execution. Compliance with these instructions, such as connection practices, is rarely monitored as this would require either the company man to be next to the driller, or an interpretation engineer to examine multiple measurement curves remotely. This paper will present an automated system capable of generating narratives of the driller's action from real-time drilling data. These drilling narratives are intended to provide a consistent interpretation of the driller's actions in context, with a goal of improving the effectiveness of the procedures.
To avoid the need for intensive interpretation of multiple measurement curves, several algorithms are used to segment the drilling data into partitions that encompass the sequences of operations during drilling or tripping. In each of these partitions, surface measurements are used to infer the driller's actions. Different combinations of the actuation of the block, mud pumps, top drive rotation, and their corresponding responses, are assigned pre-defined labels such as taking a survey, working the pipe, or conducting a friction test. These labels are later combined sequentially to form the drilling narrative.
Automating the generation of the narratives allows the drilling engineer to quickly search, extract and load events of interest, making it easier to measure the effectiveness of the planned drilling procedures outlined in the drilling program. In addition, it will also allow driller's action to be translated from a set of real-time data curves to a structured narrative that could be understood and read as a quick summary of drilling events in real time.
This automated, scalable method should be beneficial to drilling engineers, who currently spend many hours on analyzing drilling operations data, allowing them to come up with better plans and procedures for the next well.
Drilling interbedded formations can induce torsional vibrations that result in inefficient drilling and damage to drillstring components. A common bit choice for these applications is a standard polycrystalline diamond compact (PDC) drill bit; however, PDC bits due to its shearing action often exhibit some level of torsional dysfunction.
Historically, the most effective method to mitigate torsional vibrations in PDC bits is to use fixed depth-of-cut (DOC) control technology that restricts the PDC bit formation engagement at a pre-determined ratio of rate of penetration (ROP) and drillstring RPM. The challenge with using fixed DOC control is finding a compromise between limiting vibrations through targeted sections without limiting ROP in others. To address this, a self-adaptive DOC technology was developed. The self-adaptive DOC technology automatically adjusts the DOC engagement threshold as drilling conditions change, eliminating manual parameter adjustment required at surface to manage torsional dysfunctions.
This paper will cover self-adaptive bit runs from deepwater Gulf of Mexico wells. In a recent run, a 12¼-in. bit drilled past 30,000ft measured depth (MD) in an abrasive and interbedded section. The self-adaptive bit delivered a 48-percent improvement in ROP over the best offset, saving 23 drilling hours while exhibiting 97-percent smooth drilling concerning stick-slip and 100-percent smooth drilling to axial and lateral vibrations. Another application yielded excellent results in a section featuring bottom-hole coring work. In three separate runs, the self-adaptive bit drilled a sand/shale formation with 98-percent smooth drilling concerning lateral vibrations, axial vibrations, and whirl. It also exhibited 97-percent smooth drilling concerning stick-slip. The self-adjusting technology helped to return to drilling despite the coring disrupting the bottom-hole pattern.
Real-time drilling dynamics data measured downhole is used for demonstrating the effectiveness of self-adaptive DOC control technology for sustained drilling performance improvement in deepwater wells.
Uncertainties in the drilling process result in safety factors or safety margins sufficient to minimize risks in the drilling process. These safety margins represent inefficiencies in the system. This paper will discuss a method for reducing uncertainty as it relates to well bore pressures and hole cleaning to eliminate or reduce these inefficiencies, quantify the rates of penetration that can be achieved, and illustrate the expected wellbore pressures generated by these rates of penetration.
When data is collected manually, the nuances of fluid changes are lost between property measurements. This paper will illustrate the difference between calculating equivalent circulating densities (ECD) with manually collected mud report data and fluid properties collected in real time and the impact that this can have on optimizing the rate at which the operator can drill and trip pipe.
A patent-based methodology will be presented, in which real-time drilling and fluids data are captured and utilized to model ECD pressure data related to the bore hole. The actual and modeled data are statistically analysed to infer information about how rapid a rate of penetration (ROP) may safely be employed to optimize drilling results.
Data will be presented demonstrating the impact that small improvements in fluid parameters and drilling operations can have over the course of drilling a well.
The role that a real-time hydraulics software model plays in providing predictive analytics for ROP optimization will also be discussed. Predictive analytics enable operators to look several stands ahead of the bit to determine if the ROP drilled will cause issues in the future. This enables the identification of the maximum ROP that can be drilled versus optimizing instantaneous ROP. This enables operators to optimize casing-to-casing time.
To improve magnetic disturbance rejection and robustness of wellbore survey measurements, an adaptive neuro network-based fuzzy inference system (ANFIS) filter for wellbore position calculation is presented. This technique significantly improves magnetic disturbance rejection and reduces sensor error influence for borehole survey measurements. The new approach for the ANFIS filter is based on two redundant sets of IMUs which are located in different positions in the BHA at a known, constant distance. The distance between these two sets of IMUs will physically fade the effect of the magnetic disturbances. Each IMU set outputs position estimation based on the splines method which is then input into an ANFIS filter. The inputs of the splines calculation are azimuth, inclination angles and measurement depth, and the outputs are moving distance in three directions (Northing, Easting and True Vertical Depth). However, the accuracy of the splines method highly depends on the accuracy of the inputs, which are difficult to obtain during the measurement while drilling process even under pure clean environments (without any magnetic disturbances). Furthermore, the distorted azimuth caused by magnetic interference affects the borehole position accuracy. In order to deal with those problems, the designed ANFIS filter has a two-level structure. First a local level position estimation (splines method or well trained local ANFIS based on the sensor accuracy) for two sensor sets is used. If the sensor measurement accuracy is low, this local ANFIS will correct the position estimation. Then the outputs of the local modules were input into ANFIS for second level filtering (global filter) to remove the error which caused by unknown magnetic disturbances. According to the judgement of the ANFIS, the IMU set with the smaller magnetic disturbance is given greater weight to reduce the interference effect on the borehole position estimation. This two-level filter is compared to the traditional splines method under different tests situations. First, we evaluate this method by comparing with GPS positioning, from this test we know that the ANFIS filter shows a good performance when the magnitude of magnetic disturbance is within the training magnitude range. Even when the magnitude of magnetic disturbance is above the training range, the ANFIS filter shows a higher robustness than the traditional splines method. Also, this method was applied to borehole data with two IMU containing accelerometers and one magnetometer measurements. In order to apply our method, we duplicated one more magnetometer measurement data under magnetic interference for assessment. The results proved its magnetic disturbance robustness in borehole position estimation. Finally, we demonstrate the full potential using a laboratory experimental setup.