Seismic attributes often provide support for identification of several important parameters that could be applied in geomodelling for constraining reservoir properties such as porosity, permeability and fluid saturation. Additional insight can be gained as a way of detecting depositional features and geometries, and improve qualitative (e.g., data quality, seismic facies mapping) and quantitative analyses (e.g., net sand, porosity prediction). The case study is a deep gas bearing Jurassic fluvial channel sandstone reservoir in an onshore field in Kazakhstan, Central Asia. The data set is scarce, with only a few wells penetrating the reservoir. The available 3D seismic data is weak to a reasonable quality, and an attempt was carried out to use a series of attributes for channel identification and imaging. After testing several attributes, only two of them have shown good results: Generalized Spectral Decomposition (GSD) and Variance . Geobody extraction was also tried, but the seismic quality was not sufficient enough to further pursue such methodology. To overcome this issue, a joint approach for channels detection and extraction was tested by looking at the variance attribute in time slices and applying a workflow for an alternative way. With this technique (Alternative way for extracting geobodies), the facies property was generated in zones where the information was limited. The implementation of this approach, together with core facies interpretation has allowed a way of constraining other properties (e.g. porosity) and undoubtedly increased the robustness of the static model. Such model will support a more accurate estimation of HIIP volumes, future decisions on the most proper development plan and also investment decisions. The alternative way for extracting geobodies was tested in two channels examples and for different zones in Jurassic reservoir with success. The described workflow helped in channel extraction and it can be used in facies modelling when the reservoir data set is scarce. A more robust facies model can be generated by combining the seismic attribute results with well data.

Abstract A large, full-azimuth, shallow-water ocean-bottom node (OBN) survey offshore Abu Dhabi revealed an intriguing azimuth-dependent linear noise pattern. The mud rolls and guided waves appear to curve along surface and near-surface features. Although we have yet to fully understand the mechanism that causes this phenomenon, we need to accurately and effectively eliminate these linear noises to meet the tight processing deadline. In view of the huge amount of data involved, we implemented a data-driven machine learning (ML) technology. Mud rolls and guided waves are dispersive in nature. The standard way to remove them is to compute the dispersion curve (DC) (phase velocity as a function of frequency) from the data, model the corresponding mode, and subtract the model from the data in an adaptive way. In our case, the linear noise patterns are not only related to the azimuth angle of acquisition, but they also vary spatially, which makes removing them even more challenging. Since we cannot manually select the dispersion curves (DC) given the size of the data, we implemented an ML algorithm to classify the noise pattern and design the best noise removal operator for each category. The remaining outliers can then be resolved using customized solutions. In order to transform seismic data analysis into a pattern recognition problem, we need to generate attribute maps. In our case, one of the attribute map is the dispersion curve (DC), which has been calculated from the data. The maps are used as input to the ML stream which identifies classes by unsupervised K-means clustering analysis. Another attribute used is the root-mean- square (RMS) amplitude calculated over a fixed time window for each receiver gather. The corresponding maps are used as the input to the ML flow, which involves two-steps: first reduce the dimensionality of the input data by performing a principal component analysis (PCA), then identify classes through unsupervised K-mean clustering analysis. The generated class determines the common linear noise behavior and its spatial distribution group, thereby effectively driving the linear noise attenuation parameter setup.

Abstract Drilling tools are subjected to mechanical loads while drilling, resulting in loss of performance due to accelerated wear and subsequent tool damage. Two special phenomena are well known to generate high mechanical loads in torsional direction: Stick-Slip and High Frequency Torsional Oscillations (HFTO). Preceeding studies have explained the nature of the interaction of these phenomena, how they can superimpose and how stability maps define safe combinations of operational parameters. This paper will provide a systematic and quantitative study of these effects based on high-frequency measurements recorded during a North Sea deployment. In this paper, for the first time, an in-bit high frequency measuring device is used to validate the theoretical modeling of the stick-slip-HFTO interaction. Field data from the North Sea including surface and MWD measurements are used in a broad study to investigate the influence of operational parameters, bit and formation on the torsional dynamics over the increasing depth of a well. Pure Stick-Slip, pure HFTO as well as the development of HFTO within a slip-phase of stick-slip is observed, described and compared to the theoretical model describing self-excitation. Several hundred instances of stick-slip are used to determine characteristic RPM thresholds for combinations of RPM and WOB. The totality of these thresholds compose a stability map indicating zones of reduced torsional loads. The study shows that stick-slip amplifies torsional loads generated by HFTO up to a certain RPM threshold. Further increasing the RPM leads to reduced loads for certain combinations of bit, formation and operational parameters. In contrast to common belief, higher RPM can lead to lower torsional loads in certain scenarios while also increasing ROP. The stability map summarizes the learnings, indicating zones of safe operation in an accessible manner. The resulting stability map can be used by operators to navigate in stable regimes of operational parameters. This reduces wear, tool damages and enhances drilling performance.

Abstract Asphaltene deposition is a significant flow assurance challenge in Abu Dhabi with over 100 onshore wells impacted. Until recently, there had been no applicable IoT device in the industry for direct measurement of the problem, which prompted the national oil company to sponsor a real-time sensor for asphaltene quantification. A second generation of that device is now available with enhanced capabilities and has been delivered in country, with preparation now ongoing for deployment across multiple onshore wells. Current ways of identifying asphaltene problems via accessibility checks with slickline are reactive, often too late, and thereby increase cost of clean-up and production losses. By quantifying in real-time the asphaltene as it flows through the wellhead, much earlier problem detection is made feasible. Such quantification has been made possible by resonating asphaltene molecules in applied magnetic and GHz fields as those molecules flow through the wellhead. The peak of the resonant signal is directly proportional to the asphaltene in the crude, with decrease in signal indicating potential deposition. Adding cloud-based, machine-learning to the system allows local in-country team to make efficient use of the data. First deployment of the resonant system in Abu Dhabi demonstrated the resolution of asphaltene signature was better than 0.1% when measured at atmospheric pressure and temperature. Subsequent testing in pressure vessels has shown remarkable independence to fluid pressure, even up to 2000 psi. Original system had a sensitivity sufficient to detect only the asphaltene peak. New system reveals multiple smaller peaks in the spectra which can be used for other flow assurance applications - for example, vanadyl porphyrins display a unique characteristic. By mixing solvents and precipitants with crude oil, we have confirmed that the asphaltene peak measures same value whether the asphaltene is in, or out, of solution which is precisely the feature that allows surface data to be representative of the cumulative precipitation downhole. The main advance, however, is that the system data will now be available on multiple wells in the field, which allows Operator to compare and contrast different flow assurance optimizations. This should lead to substantial cost savings in addition to minimizing well downtime and potential loss of production, all in alignment with the Operator’s 2030 "Smart Growth" strategy. The additional spectral information also allows spectrometer use for real-time analysis of water properties (dissolved solids) and rock typing (geochemistry). More generally, the system becomes a real-time platform for advanced chemical analysis at the wellhead. The result is an industrial Internet of Things (IoT) real-time monitoring device, the first of its kind, that not just detects asphaltene deposition but also makes possible the optimization of chemical programs by incorporating surface data into an integrated flow assurance management system.

Abstract Fiber Optic Liquid Leakage Detection (FOLLD) project has permitted of assessing on the DTS (Distributed Temperature Sensing), the DSTS (Distributed Strain and temperature Sensing) and the DAS (Distributed Acoustic Sensing) capabilities, combined with a Standard Telecom Fiber Optic Cable (FOC), for detecting small pressurized (20 barg) and pre-heated (50 °C) liquid (water) leakages on a buried dummy pipeline. DTS/DSTS real time results demonstrated that the soil disruption, the soil and external environmental temperature variation had an influence on leakage detection results robustness. Consequently, only the first three (3) trials out of the ten (10) trials provide some comprehensive results. Real time results shown that two (2) out of three (3) DSTS devices were able to detect a leakage up to twenty (20) km with a detection success rate of 67% in the position P1 or P2 or P3. One (1) out of two (2) DTS device detects up to twenty (20) km with a detection success rate that depends on the fiber optic position (67% for P2 or P3 and 33% for P1). Only one (1) DSTS shows a detection success rate up to fourty (40) km with a detection success rate of 67% in the position P1 or P2 or P3. DTS/DSTS real time results highlight that 100% successful detection rate can be reached up to twenty (20) km with several fiber optic cables in the position: P1/P2 or P1/P3 or P1/P2/P3. For DAS systems, the real time results shown that it was not possible to distinguish a leakage "noise" from other surrounding noise. This was especially due to the water pump noise, which was always occurring simultaneously with the leakage. Only a post processing treatment provide the leakage detection using a DAS technology. Post treatment DTS/DSTS and DAS detection success rate are better than the real time results. This highlight the potential of both technologies. All DTS/DSTS and DAS testing results demonstrated that the detection success rate are Supplier's dependent.

Abstract This paper will describe pre and post workover well diagnostic and the successful water shut-off that has resulted. It also talks about the post workover survey. The well was recently completed to a new targeted reservoir and suffered from high water cut (80%) at the start of production. Proper diagnostics identified the source of water and helped in planning a successful remedial job. After water shut off the water cut dropped to zero. Post workover evaluation confirmed production from the targeted interval. Comprehensive diagnostics are crucial to designing a proper remedial job. A conventional PLT survey will describe the flow characteristics and parameters inside the wellbore only. During the job under consideration, two additional services were deployed; the Spectral Noise Logging and the High Precision Temperature sensing. Their combination with the PLT allows for the assessment of behind casing flow and the identification of active streaks of the formation. The quantification of flow rate behind casing is also achieved by modelling the temperature response from the high precision sensor. The first survey was conducted with the conventional PLT to identify the source of water. Subsequently, a high precision temperature and spectral noise log were recorded under flowing, transient, and shut-in conditions. A crossflow behind casing from the upper unperforated intervals was identified. The main liquid production is coming from the water-bearing formation that is located 15 ft above the targeted reservoir. The cement evaluation surveys that were run previously confirmed the presence of a cement channel above of the perforated interval. Based on these conclusions a remedial job was designed and conducted successfully. Another HPT/SNL survey was conducted post workover. It confirmed the absence of any further flow behind casing and that all liquid production is happening from the targeted reservoir. Surface test data showed a rapid decrease in water production to 1% post workover. This paper highlights the importance of complementing the current practice of running conventional PL surveys with two additional services: the Spectral Noise Logging and Temperature simulation/modelling using a High Precision Temperature measurement.

Abstract With close to 90,000 km 2 of onshore and offshore seismic data being acquired and processed in less than 5 years, ADNOC has introduced new innovative QCs to ensure the huge amount of data is properly vetted at each step of the processing sequence. The pre-stack time and depth processing flows are very demanding and amplitudes must be preserved throughout. We present four key diagnostics for QA/QC methods at critical steps in the processing sequence. These selected QA/QC are meant to achieve a resilient seismic data processing outcome amenable to all types of seismic interpretation, either conventional seismic interpretation or advanced seismic reservoir characterization or seismic quantitative interpretation. Seismic bandwidth is the lynchpin of that QA/QC strategy and it will be kept in the spot light to make sure that no processing artifacts erode the acquired seismic bandwidth and that we ultimately preserve the original six octaves of the recorded seismic data. Additionally utilizing the full seismic waveform, either signal or noise, such as surface waves, ground roll, or mudroll, can help compute long wavelength static solutions. The near surface velocity model is a prerequisite for accurate depth imaging not only for the shallow overburden but also at target reservoir levels. We also explore unsupervised machine learning for the automatic detection of amplitude anomalies in ocean bottom nodes (OBN) and to estimate the velocity of the ground roll or mud roll surface wave noise given the huge amount of data we are dealing with.

Abstract Lateral vibration is the most commonly measured type of vibration in machinery. However, lateral motion in the XY plane does not completely describe all the possible directions of motion of machine components. Torsional vibrations do not get the attention that they should. Since torsional vibration is more difficult to measure than lateral vibration, it is often ignored, due to an "out of sight, out of mind" attitude. In fact, torsional vibration can be quite severe and can produce damaging cyclic stresses that can cause a fatigue failure. Rotating machines are all involved in the delivery or transformation of power, and power delivery requires transmission of torque. The applied torque causes the shaft to twist about its rotational axis. Thus, instead of measuring displacement relative to an equilibrium position, torsion requires measuring the angle of twist of the shaft at different axial positions. All rotating machineries undergo some fraction of degrees of torsional vibration during operations. In many cases, the torsional vibrations are not as easily identified as the translational vibrations, due to lack of simple and direct measurement devices. Typical damages developed under excessive torsional vibrations include shaft cracks, coupling cracks, gear wear, gear tooth failures, key failures, shrink fit slippage, etc. Today torsional vibration detection and monitoring becomes an important step in rotating machinery condition monitoring [ Eshleman 1977 , Muszynska, A. G. 1992 , Naldi, L. B. (June 9–13, 2008) , Naldi, L. G. 2011 ]. Today the machinery diagnostic services team of Baker Hughes, Bently Nevada uses ADRE raw data along with self-developed Matlab executable [ Huageng, L, Howard, B, Menon, A, Péton, N, Roengchai, C. 2013 , Luo, H. C. 2013 ] application for diagnostic purposes based on phase demodulation process to the signals generated by Keyphasor ® or zebra tape of tooth wheels. In this article torsional issues will be illustrated with data and pictures using 6 real cases from the industry.

Abstract Brownfield projects whether topside modifications or underwater modifications are always challenging in Offshore Industry. New facility attachments to the existing jackets in the form of J tubes, Utility Pipes, Risers etc. require clamped connections. It involves underwater removal of marine growth and installing the new clamps on existing jacket members. The allowable span lengths of these attachments pose a challenge to Engineer. Sometimes jacket members are not available to install clamps at the required location. This leads an Engineer to take misaligned and eccentric supports to satisfy the span length requirement. In few occasions this leads to the installation of the new supporting jacket member with clamped connections. This paper presents an enhanced design concept for working out the maximum possible span lengths of these pipes considering associated vortex induced vibrations (VIV). These members are checked for VIV and VIV induced fatigue to determine spacing of the supports on jacket as one of the design criteria. This paper discusses the standard methods of the VIV check and gives a more enhanced methodology by which the spans can be increased without causing VIV excitations.

Abstract This paper presents a novel approach of continuously measuring drilling fluid rheology and density by use of sound signals. A unique apparatus is built with a series of pipe sections designed to exact pre-calculated dimensions to achieve equivalent standard shear rates as stipulated in the American Petroleum Institute (API) Recommended Practice 13D for measuring the rheology of oil-well drilling fluids (from 3 to 600 RPM). Acoustics waves are passed through the fluids of interest and their interaction is recorded and analyzed to deduce the density and rheological properties of the fluids. The concept of resonance as demonstrated by the Barton's pendulums are the basis of the methodology. Sound signals are known to exhibit a damping effect when passing through various media. Pairs of sensors are employed in this set-up and their signal response are first characterized and calibrated with fluids of known properties. Electric current is converted into acoustic signals by piezoelectric sensors mounted of the flowline which are then emitted through the fluids desired to be measured without interrupting the flow. A matching sensor receives these damped signed and reconverts them back to electromotive potentials for recording by a data acquisition unit. The signals are then analyzed by applying statistical techniques to interpret and obtain the fluids physical properties. Owing to the nature of the task, the goal of accurately achieving simultaneous measurement of density and viscosity is attained by applying an ensemble machine learning algorithm, known as Multivariate Random Forest. Pure chemicals and fluids of known properties form the training group on which the predictive model is built for subsequent testing on new mud samples flowing through each section. The pipe sections generate shear rates covering the standard range adopted in oilfield reports. Results from each pair of sensors are analyzed and compared with dial readings from rotational viscometers; these have shown to be within a narrow band of error. As a result of this work, the voltage outputs are sent continuously and in real-time to a processing computer that converts the values to dial readings at standard shear rates, while not disrupting the flow. This can aid in the better monitoring and surveillance of the entire fluid system of the well, which is highly beneficial to well control. The system can also be arranged to acquire gel strengths or how the fluid behaves after a fixed period of rest. Improvements can be made on the current procedures for fluid characterization which have remained relatively static for many years. This work engages the disciplines of rheology, acoustics and machine learning, creating a mechanism for continuous and real-time drilling fluid surveillance critical to the enhancement of safe development of petroleum resources.

Abstract "X-Mas Trees, Wellhead and Down Hole Safety Valves" are deemed as Well HSE critical Equipment in Abu Dhabi Oil fields (alike in many other countries round the world). Ensuring 100%, integrity of this critical equipment has always been a challenge in the industry due to equipment aging (above 25 years in operation) and lack of comprehensive international standard guidelines on this subject. Hence, a comprehensive methodology named as "100% Integrity Assurance approach" was implemented for this equipment in order to ensure sustainability and to prevent HSE incidents throughout the well lifecycle. This methodology in turn contributed significantly towards Life extension of the aging equipment by reducing its failure rates. The 100% Integrity Assurance approach for Well HSE Critical equipment is a broad-spectrum continuous process involving collaboration among multiple disciplines of the organization such as "Well Integrity, Field Development, Field Production and Maintenance teams. The methodology consists of: Introducing a risk based approach to standardize all requirements of Integrity Assurance activities for this equipment (e.g. frequency of activities to be performed, process, responsibilities, Allowable Leak Rates etc.). Establishment of an automated mechanism to track and report each and every pre-defined Integrity Assurance activity, highlighting areas of improvement to meet 100% compliance. A robust road map to overcome all the challenges hindering the 100% compliance target of these activities. This includes introduction of special procedures, right set of equipment and effective tracking system. Introduction of Non-Weather dependent equipment to carry out activities round the year in offshore environment. Implementing this methodology resulted not only in improving performance visibility of Well HSE Critical Equipment to whole organization but it also observed to be contributing significantly to Life extension of the aging equipment by reducing failure rates. Based on last 6-years comprehensive data upto 60% Reduction was observed in the Failure rates of HSE critical equipment/X-Mas Tree Valves. This unique approach involving multiple disciplines of the organization is new and quite fruitful to maintain the integrity of aging Well HSE critical equipment. The methodology is systematic and fulfils exiting gap in the Oil Field Industry, which lacks any comprehensive international guidelines to maintain or extend the life of aging X-Tree and DHSV equipment.

Abstract Numerous enhanced oil recovery (EOR) techniques entail injecting chemical treatments such as surfactants at a predetermined concentration to achieve targeted results. Typically, these formulations are prepared on-site prior to injection into the reservoir using large mixing tanks. The current quality assurance practice for proper concentration of treatment materials uses indirect methods such as interfacial tension (IFT) and/or viscosity measurements. These measurements can be lengthy or only suitable for a specific range of dispersed phase volumes. We examined the prospect of using acoustic measurements as a method to determine surfactant concentration for quality assessment of chemical injections on-site. The advantage of ultrasound is that it can characterize both concentrated systems and low dispersed phase volumes. In this study, we acquired the sound attenuation of an amphoteric surfactant, Cocamidopropyl Hydroxysultaine (CBS), and in-house petroleum sulfonate surfactant nano-capsules (NanoSurfactant), within a frequency range of 3-99.5 MHz. We conducted a set of experiments to establish the relationship between acoustic attenuation and surfactant concentration, and assess the measurements’ dependence on variable factors including applied frequency, suspending fluid, and temperature. The data showed a nearly perfect linear fit in the higher frequency range (65-99.5 MHz) where the coefficient of determination was between 0.979-0.998. Consistent with theoretical predictions, this method showed sensitivity to water salinity where the surfactant, at a constant concentration, displayed higher attenuation values in response to the increase in water salinity, allowing the contribution of surfactants to the acoustic attenuation to be isolated. Our method accurately determined surfactant concentrations using acoustic attenuation regardless of the surfactant used. Adopting such a robust protocol that can determine surfactant concentration and has the potential to improve field quality assurance methods would be beneficial for EOR research and industry applications whether as a standalone system or in conjunction with other commonly available methods.

Abstract A double suction centrifugal fan with rated power 1120 KW driven by variable speed motor using VFD (variable frequency drive) through flexible coupling (shim pack), fan rolling element bearings vibration is monitored by two seismic vibration sensors (one per bearing) connected to the control room. the fan is one of two gas re-circulation fans of water tube boiler generating steam to drive 650 MW power turbine. on 1st of July 2019 after power plant steam turbine major overhaul, fans started with acceptable vibration levels. Within August to September 2019 the fan bearings showed vibration and temperature increase higher than acceptable limits. Customer inspected both fan bearings and replaced the damaged inboard bearing. After one month of normal operation, the same fan inboard bearing failure occurred with the same characteristics based on the previous history, this plan was implemented: Fan bearings’ housings clearances measuring. Old (failed) bearings visual inspection. Fan inboard bearing replacement. Modal impact test to measure the fan natural frequencies. Startup. vibration data collection and analysis. It was observed that, fan inboard bearing radial clearance between bearing and housing was 0.22 mm exceeding the maximum allowable limit 0.01 mm. (clearances adjusted as per manufacturer limits) as the fan inboard bearing is double row type, however the both failed bearings showed the same failure pattern on the closest row to the motor mean while the other row was clear. after fan inboard bearing replacement and machine startup, vibration data collected using online vibration analyzer and voltmeters installed on both fan bearings and motor inboard bearing, fan bearings velocity spectrum showed high vibration component at 141 Hz which is close to the bearing outer race fault frequency, and doesn't coincide with any of the fan natural frequencies. motor inboard bearing horizontal sensor showed severe fluctuations and Ski-slope in the spectrum. visual inspection showed damaged ground brush, and after its replacement, all vibration levels decreased, and the 141 Hz component decreased. based on previous observations it was concluded that, as a result of damaged ground brush, ESD (electrostatic discharge) generated and was responsible for fan inboard bearing housing clearances increase overtime, which led to repeatable bearing failures. this case study is focused on how the root cause analysis performed to avoid the repeatable bearing failure, and the importance of preventive maintenance activities to avoid unplanned failure. also, how the ESD problem was detected through steady state vibration data and applying the appropriate maintenance plan.

Abstract Fusion bonded epoxy coatings (FBEs), due to their zero volatile organic compound (VOC) and less production lines wastes, have increased the customers’ interest in this technology for pipes corrosion protection. However, the high-energy consumption used, to keep the oven at 438°F - 482 °F (225 °C - 250 °C), makes them not very competitive with liquid epoxy coatings (LECs). LECs, cheaper to be applied, using low curing temperatures ranging between 100°F to 212°F (38 °C - 100 °C), faster return-to-service, can perform like FBEs using lower thickness. Here a new LEC able to resist in petroleum, 50 bars CO 2 at 392°F, is reported. The protective behavior of the coating reported, has been investigated using electrochemical techniques too.

Brownfield projects whether topside modifications or underwater modifications are always challenging in Offshore Industry. New facility attachments to the existing jackets in the form of J tubes, Utility Pipes, Risers etc. require clamped connections. It involves underwater removal of marine growth and installing the new clamps on existing jacket members. The allowable span lengths of these attachments pose a challenge to Engineer. Sometimes jacket members are not available to install clamps at the required location. This leads an Engineer to take misaligned and eccentric supports to satisfy the span length requirement. In few occasions this leads to the installation of the new supporting jacket member with clamped connections. This paper presents an enhanced design concept for working out the maximum possible span lengths of these pipes considering associated vortex induced vibrations (VIV). These members are checked for VIV and VIV induced fatigue to determine spacing of the supports on jacket as one of the design criteria. This paper discusses the standard methods of the VIV check and gives a more enhanced methodology by which the spans can be increased without causing VIV excitations.

This paper presents a novel approach of continuously measuring drilling fluid rheology and density by use of sound signals. A unique apparatus is built with a series of pipe sections designed to exact pre-calculated dimensions to achieve equivalent standard shear rates as stipulated in the American Petroleum Institute (API) Recommended Practice 13D for measuring the rheology of oil-well drilling fluids (from 3 to 600 RPM). Acoustics waves are passed through the fluids of interest and their interaction is recorded and analyzed to deduce the density and rheological properties of the fluids. The concept of resonance as demonstrated by the Barton's pendulums are the basis of the methodology. Sound signals are known to exhibit a damping effect when passing through various media. Pairs of sensors are employed in this set-up and their signal response are first characterized and calibrated with fluids of known properties. Electric current is converted into acoustic signals by piezoelectric sensors mounted of the flowline which are then emitted through the fluids desired to be measured without interrupting the flow. A matching sensor receives these damped signed and reconverts them back to electromotive potentials for recording by a data acquisition unit. The signals are then analyzed by applying statistical techniques to interpret and obtain the fluids physical properties. Owing to the nature of the task, the goal of accurately achieving simultaneous measurement of density and viscosity is attained by applying an ensemble machine learning algorithm, known as Multivariate Random Forest. Pure chemicals and fluids of known properties form the training group on which the predictive model is built for subsequent testing on new mud samples flowing through each section. The pipe sections generate shear rates covering the standard range adopted in oilfield reports. Results from each pair of sensors are analyzed and compared with dial readings from rotational viscometers; these have shown to be within a narrow band of error. As a result of this work, the voltage outputs are sent continuously and in real-time to a processing computer that converts the values to dial readings at standard shear rates, while not disrupting the flow. This can aid in the better monitoring and surveillance of the entire fluid system of the well, which is highly beneficial to well control. The system can also be arranged to acquire gel strengths or how the fluid behaves after a fixed period of rest. Improvements can be made on the current procedures for fluid characterization which have remained relatively static for many years. This work engages the disciplines of rheology, acoustics and machine learning, creating a mechanism for continuous and real-time drilling fluid surveillance critical to the enhancement of safe development of petroleum resources.

The objective of this paper is to present a method for monitoring the natural frequency of a fixed offshore platform and to utilize the raw data to proactively monitor and predict structural degradation of the platform using machine learning. This can enable operators to predict critical member failure and guide timing of operational decisions and interventions, thus helping reduce cost of ownership, particularly for ageing assets. A Natural Frequency Response Monitoring (NFRM) system utilizes accelerometers and Fast Fourier Transform (FFT) to measure the sway and torsional frequencies of a fixed offshore platform. A significant structural event is detected by a shift in the natural response of the platform. A machine learning model (or digital twin) can be trained on this accelerometer data to replicate the platform response and determine the fatigue damage accumulation in platform members. The model, which can be re-trained to maintain accuracy, acts as a near real-time fatigue tracker with the NFRM system acting as a verification tool. A proof of concept case study is presented and demonstrates how Finite Element Analysis (FEA) simulations and measured data can be used to develop a machine learning model. The combination of an NFRM system with machine learning has the potential to assist operators in the prediction and identification of structural changes prior to them becoming critical. Knowledge of the actual fatigue damage in key platform members on a near real-time basis allows for proactive planning of inspections or repairs, as well as confidence in the structural performance of the platform, which can justify life extension.

On several occasions during drilling, the drill string or in cases of liner job the liner string comes across an overall regressive environment. By regressive environments, the condition of the well bore returning to its unconditioned and in some cases unstable phase, is implied. The objective of this paper is to see how the regression impacts the circulating pressures in particular and how best to anticipate such conditions to optimize/modify the practices. The circulating pressures in deteriorated/reverted cases start showing spikes in the actual values well beyond the predicted models subsequent to same flow rates at the same depths. Such deviations in the actual vs theoretical values can pose severe complications for the drilling and liner jobs. These resultant complications however can be countered with the help of hydraulic mapping and gel envelope estimation in conjunction with optimized tripping and circulation practices for the respective operations to ameliorate the conditions. This paper explores the impact of regressive hole conditions ranging from constraints, operability, mechanical loads and fluid regimes. It builds upon that impact and delves into how best to utilize the tool of hydraulic mapping for smooth tripping and drilling operations in conjunction with the real time monitoring to define the operational envelopes. Delving into the dynamics of the hole conditions in regards to tripping and drilling operations across open hole, the paper seeks to build upon experience and reach an optimized mantle for tripping, drilling, circulation and conditioning operations in compromised hole conditions without any time delay or complication. During the life span of the well operations, the open hole well bore conditions of the wells become adverse to continuing onwards with the operations without at first conditioning or changing the downhole states altogether. This is the modus operandi for the majority of operations but in instances where the ambient impact is time sensitive or where the operations altogether are too complex and constrained for that to be done, the well conditions worsen, and the complications related to the ongoing operations increase manifold, rendering the operation in extreme cases unfeasible altogether. This likelihood can however be circumvented with the help of preventive intervention tools such as hydraulic mapping. There is no broad stroke solution to operational complications though, techniques and tools vary with each instance and are very case specific in both spectrum of definition and application. The most significant take away though, is how the incorporation of hydraulic mapping ensures the impending operational problems and complications are efficiently and specifically managed without any downtime or operational delay, on the fly. There are cases when the hydraulic and dynamic parameters have been mapped with upper limits built within the model yielding successful execution of operations against odds. In presence of rapid gelation or excessive gel breaking pressures against high differential formations, the practice of rotating prior to circulation or reciprocation prior to circulation are also determined with this tool. Real time operations monitoring also play a pivotal role in benchmarking and tracking the operational parameters of interest that are critical for operations. In order to achieve that however, there needs to be a composite model for the real time broomstick to be in place that reflects the overall picture of the open hole well bore. The hydraulic mapping technique requires minor inputs from the routine operational practices but forms an integral tool that can help execute operations effectively in jeopardized environments, against staggering odds without forfeiting any of the operational parameters or objectives.

Fusion bonded epoxy coatings (FBEs), due to their zero volatile organic compound (VOC) and less production lines wastes, have increased the customers’ interest in this technology for pipes corrosion protection. However, the high-energy consumption used, to keep the oven at 438°F - 482 °F (225 °C - 250 °C), makes them not very competitive with liquid epoxy coatings (LECs). LECs, cheaper to be applied, using low curing temperatures ranging between 100°F to 212°F (38 °C - 100 °C), faster return-to-service, can perform like FBEs using lower thickness. Here a new LEC able to resist in petroleum, 50 bars CO 2 at 392°F, is reported. The protective behavior of the coating reported, has been investigated using electrochemical techniques too.

A large, full-azimuth, shallow-water ocean-bottom node (OBN) survey offshore Abu Dhabi revealed an intriguing azimuth-dependent linear noise pattern. The mud rolls and guided waves appear to curve along surface and near-surface features. Although we have yet to fully understand the mechanism that causes this phenomenon, we need to accurately and effectively eliminate these linear noises to meet the tight processing deadline. In view of the huge amount of data involved, we implemented a data-driven machine learning (ML) technology. Mud rolls and guided waves are dispersive in nature. The standard way to remove them is to compute the dispersion curve (DC) (phase velocity as a function of frequency) from the data, model the corresponding mode, and subtract the model from the data in an adaptive way. In our case, the linear noise patterns are not only related to the azimuth angle of acquisition, but they also vary spatially, which makes removing them even more challenging. Since we cannot manually select the dispersion curves (DC) given the size of the data, we implemented an ML algorithm to classify the noise pattern and design the best noise removal operator for each category. The remaining outliers can then be resolved using customized solutions. In order to transform seismic data analysis into a pattern recognition problem, we need to generate attribute maps. In our case, one of the attribute map is the dispersion curve (DC), which has been calculated from the data. The maps are used as input to the ML stream which identifies classes by unsupervised K-means clustering analysis. Another attribute used is the root-mean- square (RMS) amplitude calculated over a fixed time window for each receiver gather. The corresponding maps are used as the input to the ML flow, which involves two-steps: first reduce the dimensionality of the input data by performing a principal component analysis (PCA), then identify classes through unsupervised K-mean clustering analysis. The generated class determines the common linear noise behavior and its spatial distribution group, thereby effectively driving the linear noise attenuation parameter setup.