Search

Summary Decision making to optimize the drilling operation is based on a variety of factors, among them real-time interpretation of drilled lithology. Because logging while drilling (LWD) tools are placed some meters above the bit, mechanical drilling parameters are the earliest indicators, although they are difficult to interpret accurately. This paper presents a novel deep learning methodology using mechanical drilling parameters for lithology classification. A cascade of deep neural networks (DNNs) are trained on historical data from wells on a field operated by Equinor. Rather than an end-to-end approach, the drilling parameters are used to estimate LWD sensor readings in an intermediate step using the first DNNs. This allows continuous updates of the models during operation using delayed LWD data. The second DNN takes the virtual LWD estimates as input to predict currently drilled lithology, similar to manual expert interpretation of logs. This configuration takes into account case-dependent [mud, bottomhole assembly (BHA), wellbore geometry] and time-varying (bit wear, wellbore friction) relationships between drilling parameters and LWD readings while assuming a constant rule when using LWD data to classify lithology. Upon completion of training and validation, the system is tested on a separate, unseen wellbore, for which results are presented. Visualizations for true lithology alongside the estimates are given, along with confusion matrices and model accuracy. The system achieves high accuracy on the test set and presents low confusion between classes, meaning that it distinguishes well between the lithologies present in the wellbore. It can be seen that the borders between successive layers of lithology are detected rapidly, which is crucial seen from an optimization standpoint, so the driller may immediately adjust accordingly. It shows promise as an advisory system, capable of accurately classifying currently drilled lithology by continuously adapting to changing downhole conditions. Although we cannot expect perfect estimates of lithology purely based on drilling parameters, we can obtain a preliminary map of the subsurface this way. This novel configuration gives a real-time interpretation of the currently drilled lithology. Thus, the drilling operation can be improved through early information and prompt drilling parameter adjustments to accommodate the current drilling environment.

Graphical abstract, In the streaming learning setting, an agent is presented with a data stream on which to learn from in an online fashion. A common problem is catastrophic forgetting of old knowledge due to updates to the model. Mitigating catastrophic forgetting has received a lot of attention, and a variety of methods exist to solve this problem. In this paper, we present a divided and prioritized experience replay approach for streaming regression, in which relevant observations are retained in the replay, and extra focus is added to poorly estimated observations through prioritization. Using a real-world dataset, the method is compared to the standard sliding window approach. A statistical power analysis is performed, showing how our approach improves performance on rare, important events at a trade-off in performance for more common observations. Close inspections of the dataset are provided, with emphasis on areas where the standard approach fails. A rephrasing of the problem to a binary classification problem is performed to separate common and rare, important events. These results provide an added perspective regarding the improvement made on rare events.•We divide the prediction space in a streaming regression setting•Observations in the experience replay are prioritized for further training by the model’s current error

We present a novel streaming learning approach, utilizing a deep neural network (DNN) to learn from data available during operation to estimate at-bit density using drilling parameters. Since every wellbore is different, the relationship between drilling parameters and at-bit density varies. Equipment used, well trajectory, friction and bit wear are examples of conditions that affect this relationship and makes a pre-trained model unable to represent an accurate input/output mapping applicable to all wells. However, using delayed density log measurements, continuously supervising updates to the model is possible during operation. The algorithm has been tested on drilling data from wells on a field operated by Equinor and compared to a standard deep learning approach, where results show that a streaming learning approach outperforms the traditional method. Statistical analyses have been performed to verify the statistical significance and effect size on the data sets. Data visualizations using a t-distributed Stochastic Neighbor Embedding (t-SNE) indicate that the relationship between drilling parameters and density log indeed vary between wellbores, making generalizability an issue for a traditional supervised learning approach to this problem, and motivating a streaming learning approach. Using the proposed method, more accurate at-bit estimates can be made, providing preliminary indications ahead of the tool placed 20–30 m behind the bit, which, dependent on rate of penetration (ROP), will be available 20–120min later.

As oil exploration and development costs rise, the oil industry increases its efforts to improve oil recovery (IOR) from existing fields. IOR is achieved mainly by drilling more wells, but drilling in partially depleted reservoirs is challenging due to narrow pressure margins. Offshore drilling in harsh environments, such as the North Sea, presents additional challenges, since the heaving motion from a floating rig induces large surge and swab pressures in the well. This brief suggests a remedy for this problem using automatic control of well pressure. Taking advantage of an experimental lab facility recently completed at NTNU, a model of the drilling system is developed using subspace identification methods. The model serves as a basis for state estimation and controller design using model predictive control. Applying the controller to the lab facility, pressure oscillations are suppressed by 70%-90% compared with the open-loop case, depending on the period of the heave motion. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works nb_NO

Background Patients dropping out of mental health treatment is considered a widespread and significant obstacle to providing effective treatment, thus reducing the probability of patients achieving the desired improvement. Here, relative to ordinary treatment, we investigate the effects of providing an educational group programme before mental health treatment on mental health symptomatology and the risk of patients dropping out or prematurely discontinuing treatment. Methods A randomized controlled trial in which adults referred to a community mental health center were randomized to either a Control Group (n = 46) or a pretreatment educational programme followed by treatment as usual (Intervention Group, n = 45). The primary outcome was self-reported mental health symptomatology assessed with BASIS-32. Data were analyzed by multilevel linear regression and Cox’s regression. Results We recruited 93 patients during a 26-month period. Assessments were performed before (0 month, baseline) and after the intervention (1 month, before treatment initiation), and after 4 and 12 months. The net difference in BASIS-32 score between 0 and 1-month was − 0.27 (95% confidence interval CI] -0.45 to − 0.09) in favor of the intervention group. Although both groups had a significant and continuous decline in psychopathology during the treatment (from 1 month and throughout the 4- and 12-month follow-up assessments), the group difference detected before treatment (between 0 and 1 month) persisted throughout the study. Premature treatment discontinuation was partially prevented. The dropout risk was 74% lower in the Intervention Group than in the Control Group (hazard ratio 0.26, 95% CI = 0.07–0.93). Conclusions A brief educational intervention provided before mental health treatment seems to have an immediate and long-lasting effect on psychopathology, supplementary to traditional treatment. Such an intervention might also have a promising effect on reducing treatment dropout. © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/)

Good flow and pressure control is essential for successful Underbalanced Drilling (UBD) operations. This work evaluates the use of Model Predictive Control (MPC) for integrated control of well conditions and the topside separation system during UBD. The downhole well pressure, separator liquid levels, and the separator pressure are controlled by manipulation of the rig pump, the choke, and the separator valves. The control system adheres to downhole and topside constraints. These constraints include pore and collapse pressures, minimum flow rate for hole cleaning, maximum choke pressure, separator pressure, and separator liquid levels. The proposed MPC solution uses simple Hammerstein-Wiener models, where parameters are determined by system identification incorporated into standard drilling procedures. The control system is tested using a high-fidelity multi-phase flow simulator (OLGA) for some common drilling scenarios, including drilling into a producing formation and performing connections. We show that the MPC solution is able to take proactive action to ensure safe and efficient operation without having to enter well control mode or shutting down the separator system. By limiting the amount and variation in influx from the reservoir, we get less Non-Productive Time (NPT), we improve safety, and we may to some extent be able to reduce the footprint of the equipment.

This paper presents a novel application of linear model predictive control (MPC) for pressure and flow control in underbalanced drilling operations. Coordinated control of pump flow and choke pressure is used to control the return flow rate, and the well pressure profile. The control system is verified using a high fidelity drilling simulator for some common drilling operations. The proposed solution shows promising results for operations close to the selected set-points, but the simple models employed have distinct limitations. The control solution is easily extendable to larger control problems, and can be augmented with better models.

A model-based in-/out-flux detection scheme for managed pressure drilling is presented. We apply a globally exponentially stable adaptive observer which estimates the unknown states and parameters of the hydraulic system and in particular quantifies the magnitude of the in-/out-flux and its location in the well. While the observer can be used purely as an in-/out-flux detection system, the paper also presents a simple controller that automatically and effectively stops an in-/out-flux. This novel control scheme departs from common practice for automatic managed pressure drilling, which is to regulate down-hole pressure to some predetermined set point. Experimental results and realistic simulations with a state-of-the-art simulator are provided to show the effectiveness of the method.

Based on experiences with the Controlled Mud Level (CML) system ECD-Management (ECD-M) in the Gulf of Mexico (GoM) in 2014 and 2015, this paper outlines how the reduced riser level was applied to improve kick detection and to reduce drilling risks associated with downhole pressure constraints. CML is a Managed Pressure Drilling (MPD) technology, but it is not only about Constant Bottom Hole Pressure (CBHP). The paper will cover the authors' learning curve from the first to the 4th well. The main focus during the first well was to make everything work and to train the personnel on the rig. The riser level was only reduced by about 150 feet and the ECD-M system was mainly used for improved kick detection by using the riser as a trip tank. Several kick drills were performed with good results showing how the drill crew was able to detect influxes even earlier with the new CML parameters. The use in the second well was extended by mitigating salt creep in the salt section, CBHP during salt exit and reducing hydrostatic head during a cement operation. Further experiences were gained during the last two wells, even though one major new component failed during commissioning. Other potential applications will also be described. The paper will describe the planning of riser levels and controller modes together with the usual drilling parameters mud weight, flow rate, ROP, etc. using hydraulic simulations and pore and fracture gradients taking the different constraints and risk into consideration.

The two-phase dynamics of a gas-kick percolating up a vertical well, drilled using a partly evacuated riser, is described by two ordinary differential equations and algebraic relations. We model the gas as bubbles distributed with a distribution function along the well. The simplicity of the model makes it well suited for estimation purposes and controller design. We present simulation results where an unscented Kalman filter is used to estimate the current location of the gas, distribution of the gas, and rise-velocity. These estimates can be used to predict when the gas will pass the blowout preventer and reach the drilling rig.

This paper addresses control challenges within drilling for the oil and gas industry. Drilling has not been modernized as much as other industries and there is still a great potential for automatic control. Heavy machinery is used to handle pipes and other equipment topside and in and out of the hole. These machines can be controlled remotely with a joystick. Very advanced tools are being used downhole several km away from the rig and these are often controlled remotely. Mud pulse telemetry with low bandwidth is used to communicate with downhole equipment. Drilling involves certain risks and mistakes may have disastrous consequences both for people and economically, e.g. a blow-out. Safety is therefore always the most important issue, and new solutions must be robust and fault tolerant. Extensive testing is required before being used in an actual drilling operation. Managed pressure drilling is a relative new method for drilling challenging wells with narrow margins requiring precise pressure control. Nonlinear model based control solutions and observers have been developed for this technology. This is addressed in this paper together with experimental results. In addition an overview of some other interesting challenges is given.

Summary Automatic control solutions for drilling are expected to become widely used in the future. Both basic and more advanced control tools are well established in other communities such as offshore processing facilities and oil refineries. Drilling systems, however, have traditionally been operated manually. There is a great economic potential for the introduction of automatic control providing reduced drilling time, increased regularity, and improved performance, especially for wells with very narrow pressure margins. One example of automated drilling is automatic control of the downhole pressure by topside choking in managed-pressure-drilling (MPD) operations. Narrow drilling margins, especially in depleted reservoirs, ask for highly accurate pressure control. Statoil applied automatic MPD successfully offshore at the Kvitebjørn field in the North Sea in 2007. This paper presents some MPD results from Kvitebjøm and discusses automatic control requirements for drilling operations. The requirements for MPD operations include a specified accuracy for a set of normal operations, such as rate changes and set point ramping during connections, surge and swab, and some failure operations, such as power loss, gas kicks, and blocked choke. The paper also includes some ideas for the future of intelligent-drilling operations with increasing automation.

We design a leak detection system consisting of an adaptive Luenberger- type observer based on a set of two coupled one dimensional first order nonlinear hyperbolic partial differential equations governing the flow dynamics. It is assumed that measurements are only available at the inlet and outlet of the pipe, and output injection is applied in the form of boundary conditions. Heuristic update laws for adaptation of the friction coefficient and leak parameters are given, and simulations demonstrate their ability to detect, quantify and locate leaks. Particular attention is given to time-varying boundary conditions, such as during pipeline shut-down. A scenario consisting of leak detection followed by pipeline shut-down during which the leak is accurately quantified and located is successfully simulated for both liquid and gas systems.

This paper presents the development, qualification, and first use of ECD-Management (ECD-M); a new managed pressure drilling system developed to solve major challenges when drilling in deep waters. ECD-M combines EC-Drill from Enhanced Drilling (EDR), a Controlled Mud Level technology where a subsea pump module is installed mid riser, with the Delta Seal Riser Module (DSRM) from Managed Pressure Operations (MPO) having a Quick Closing Annular (QCA) installed just above the subsea pump. This enables extending the operating envelope of the technology to a point where loss of circulating capability could otherwise cause an underbalance in the well. The first ECD-M well was partially drilled with a reduced riser level, but without the Delta Seal Riser Module. The paper describes the rig modifications, interfaces between vendors, rig contractor and service companies, training, process with BSEE, qualification, success criteria and finally the results and experiences from first use.

To increase the oil production for the Tordis subsea oilfield located at the Norwegian Continental Shelf, a subsea separation and boosting station will be installed. Most of the water will be injected into a subsea reservoir instead of being transported up to the platform. Several challenges conserning process control need to be addressed before the implementation process, and dynamic simulations have therefore been performed in order to develop and test different control strategies to deal with these challenges. The results from some of these simulations will be presented in this paper.

The challenge of handling intermittent flow with liquid slugs followed by gas pockets in multiphase flow lines becomes more important when the number of satellite fields increases. Oil, gas and water are transported from the wells in some km long flow lines along the seabed and up through a riser to the oil rig. Slugging may cause several problems for topside processing. This paper concerns suppression of slug flow by active use of the topside choke. Process measurements such as pressure and density are used in a PID controller. Slugging has been reduced significantly with such a system in operation offshore since April 2001. This paper contains results within simplified modelling of flow dynamics. New experimental results verify the dynamic model. A control scheme in operation offshore has been tested and a new control scheme independent of subsea measurements has been developed and tested in experiments.

This paper presents ongoing work as well as plans and ideas for future development of a toolbox of managed pressure drilling systems that solves major challenges when drilling in both mid- and deep waters. The toolbox contains three applications based on field proven Controlled Mud Level (CML) technology. Additional benefits can be achieved by combining controlled mud level technology with other technology elements such as a fast closing annular or a sealing element in the riser, usually referred to as a rotating control device (RCD).The first application is called CML. A subsea pump module (SPM) is used to pump the mud returns to surface in a separate mud return line (MRL). The SPM is used to regulate and manage the bottomhole pressure by adjusting the mud level in the riser. High bottomhole pressure or Equivalent Circulating Density (ECD) due to friction and other effects can be avoided by reducing the mud level in the riser accordingly. One of the objectives with this technology is to avoid or reduce losses both during drilling and other operations such as cementing.In the second application called ECD-Management, CML is combined with a fast closing annular installed in the riser just above the SPM. This enables extending the operating envelope of the technology to a point where loss of circulating capability could potentially cause an underbalance scenario in the well. The hydrostatic pressure from the drill string or MRL can be trapped within a few seconds by closing the fast closing annular element, e.g. on unplanned rig pump stops or formation pack off situations. A second annular will also be installed to be used during connections in order to save time filling up the riser to compensate for the ECD effect (friction losses in the annulus).The third application in the toolbox is called ECD-Control. Here a subsea RCD or a sealing element is installed in the riser and the SPM can then be used to manipulate the pressure below the RCD to compensate for ECD variations. The riser is always topped up with mud and hydrostatic overbalance is maintained.The paper will present simulation results comparing the different technologies and address benefits and challenges with the different methods.

In this paper two non-linear control laws for ships are derived by using a non-linear ship model which includes the hydrodynamic effects due to time-varying speed and wave frequency. The non-linear ship model does not have the structural properties of symmetric inertia and positive damping at high speed, which are common assumptions, when designing non-linear ship tracking control systems. Backstepping control designs are used to circumvent the problem of a non-symmetrical inertia matrix. The first case is ship tracking with full actuation in surge, sway and yaw while the second case discusses an emergency situation where a ship is supposed to track a time-varying trajectory by using the bow thruster and one of the main propellers as actuation devices only. The application of speed- and wave-dependent models for ship control is a first step towards the design of a new generation of non-linear ship control systems that can operate under time-varying speed and wave frequency conditions. Previous work has often relied on the assumption of constant speed, and practical solutions have been based on switching between different control schemes at different speeds. Slowly varying environmental disturbances, such as ocean current, second-order wave-induced disturbances and wind forces, are compensated for by using adaptive backstepping. Global uniform asymptotic stability (GUAS) is proven for the fully actuated case by using a recent theorem for GUAS when backstepping with integral action, whereas asymptotic tracking of position and velocity are shown for the degraded control case. Computer simulations illustrate the performance and the robustness of the control laws for an offshore supply vessel. © 1998 John Wiley & Sons, Ltd.

In this paper a nonlinear control law for an underactuated ship is derived by using a nonlinear ship model which includes the hydrodynamic effects due to time-varying speed and wave frequency. The nonlinear ship model does not have the structural properties of symmetric inertia and positive damping at high speed, which have been a common assumption when designing nonlinear ship tracking control systems. Backstepping control designs are used to circumvent the problem of a non-symmetrical inertia matrix. In this paper we have considered a ship with one propeller and one bow thruster. This is actually an underactuated system since only two actuators are used to control three degrees of freedom. The application of speed and wave dependent models for ship control is a first step towards the design of a new generation of nonlinear ship control systems that can operate under time-varying speed and wave frequency conditions. Previous work has often relied on the assumption of constant speed, and practical solutions have been based on switching between different control schemes at different speeds. Compensation of slowly-varying environmental disturbances, such as ocean currents, 2nd-order wave drift and wind forces are done by using adaptive backstepping. For tracking applications, global convergence is proven by using Barbalat's lemma. Computer simulations illustrate the performance of the control law.

As oil exploration and development costs rise, the oil industry increases its efforts to improve oil recovery (IOR) from existing fields. IOR is achieved mainly by drilling more wells, but drilling in partially depleted reservoirs is challenging due to narrow pressure margins. Offshore drilling in harsh environments, such as the North Sea, presents additional challenges, since the heaving motion from a floating rig induces large surge and swab pressures in the well. The paper suggests a remedy for this problem using automatic control of well pressure. Taking advantage of an experimental lab facility recently completed at NTNU, a model of the drilling system is developed using subspace identification methods. The model serves as a basis for state estimation and controller design using model predictive control. Applying the controller to the lab facility, pressure oscillations are suppressed by 70-90% compared to the open-loop case, depending on the period of the heave motion. © 2016. This is the authors’ accepted and refereed manuscript to the article. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/

We apply a PDE observer to a hydraulic system to detect and quantify an in- or outflux. We show that the PDE model of the hydraulic system can be expressed as a 2×2 linear hyperbolic system of PDEs with spatially varying coefficients coupled with an ODE at the inlet boundary that models the in- or outflux. Using the method of backstepping, we design an observer which is exponentially stable at the origin in the ℒ2-norm while relying on measurements taken at the outlet boundary, only. Simulation results verify the validity of the observer.

As production on the Norwegian shelf enters tail production, drilling wells with vanishing pressure windows become more attractive. This motivates use of automatic control systems for improved control of downhole pressure using Managed Pressure Drilling (MPD) techniques. PID SISO control solutions for MPD are by now relatively standard, and well understood. This article explores the potential benefits of using linear Model Predictive Control (MPC) for MPD. It is shown that in combination with wired drill pipe, the downhole pressure can be controlled at multiple locations in the open wellbore, by using both pumps and choke in applied backpressure MPD. Also, downhole pressure constraints (pore and fracture pressures) fit naturally in MPC. Illustrative simulations are presented from using a high fidelity well simulator called WeMod, and Statoil's MPC software SEPTIC. ©2013 Norwegian Society of Automatic Control. All articles in MIC are published with the Creative Commons Attribution 3.0 Unported (CC BY 3.0) license. See: http://creativecommons.org/licenses/by/3.0/.

A model-based in-/out-flux detection scheme for managed pressure drilling is presented. We design a globally exponentially stable adaptive observer which estimates the unknown states and parameters of the hydraulic system and in particular quantifies the magnitude of the in-/out-flux and its location in the well. While the observer can be used purely as an in-/out-flux detection system, the paper also presents a simple controller that automatically and effectively stops an in-/out-flux. This novel control scheme departs from common practice for automatic managed pressure drilling, which is to regulate down-hole pressure to some predetermined set point. Simulations are provided to show the effectiveness of the method.

An adaptive control scheme for the attitude control of a rigid spacecraft is derived using a linear parameterization of the equation of motion. The tracking error is described with the Euler parameter vector. Global convergence of the tracking error to zero is shown using passivity theory. This allows for the use of time-varying positive-definite feedback gain matrices, and the results can easily be extended to other passive parameter update laws. >

Control of the annulus pressure during drilling of oil and gas wells can be a challenging task, due to the complex dynamics of the multiphase flow. Since well kicks and fluid losses are potentially dangerous, it is important to improve the performance and the capabilities of estimation and control and to initiate an automatic well control system. Novel adaptive observers for estimation the flow rates through the well are developed for kick and loss detection, and asymptotic convergence properties are established. A switched control algorithm for feedback control of the choke and the back-pressure pump is proposed for pressure control and kick attenuation and asymptotic convergence is achieved. A new observer is developed to estimate the reservoir pore pressure for improved kick management. Simulation results obtained with a high fidelity drilling simulator are presented to demonstrate the effectiveness of the proposed observer and control scheme. The results show that the proposed observers effectively detect kicks in the early phase and that the automatic control scheme improves the kick handling.

When drilling into a petroleum reservoir it is critical to maintain a required downhole pressure to avoid well bore fracture or influx of petroleum fluids. Drilling into mature and often depleted petroleum fields may be difficult due to tight pressure margins.

In this paper, we present an algorithm for optimizing reservoir production using smart well technology. The term smart well is used to indicate an unconventional well equipped with down hole inflow control valves (ICVs) and instrumentation. This additional instrumentation extends the degree of freedom in the field production planning, since production can be efficiently distributed on the different well segments available. By proper utilization of the ICVs through optimal production planning, an increased oil recovery for the reservoir can be expected.We propose a method for optimal closed-loop production known from control theory as model predictive control (MPC). A commercial reservoir simulator, ECLIPSE, is used for modeling and predictions. MPC is chosen for its ability to provide an optimal solution for the constrained multivariable control problem. To compute the optimal ICV settings, we propose using a nonlinear MPC (NMPC) application, which can handle the severe nonlinearities found in reservoir models. The NMPC uses a single shooting multi-step quasi-Newton (SSMQN) method to solve the optimization problem. As the term multistep suggests, this is an iterative method which solves a sequence of quadratic problems (QPs) in each time step.We apply our method to a benchmark reservoir model with multiple geostatistical realizations. This model has already proven potential for increased oil recovery by using optimization techniques. We show an even additional increase over the former approach in production totals, using the SSMQN method, with as much as 68% increase in one case, and 30% on average compared to a reference case.

A control law for tracking of underactuated surface vessels is presented. Optimal trajectories for surface vessels are given as combinations of arcs of circles and line segments. It is shown how optimal trajectories can be made smooth by a technique called splines, and how these trajectories can be stabilized by backstepping. The control law depends on the forward velocity of the surface vessel always being positive, which is not a major limitation for tracking purposes. It is also shown how this work can be part of a hierarchical hybrid system for control of surface vessels.

A continuous feedback controller with a periodic time varying term for the attitude control of an underactuated rigid spacecraft is derived. The attitude is described with the Euler parameter vector. Due to the nondifferentiable nature of exponential stabilizers, sampling of some measurements is used. Global exponential convergence is indicated with analysis and simulations.