Summary: During a drilling operation, a real-time analysis of surface and downhole measurements can give indications of poor hole cleaning. However, it is not always intuitive to understand how and where the cuttings are settling in the borehole because the transportation of cuttings and the formation of cuttings beds are largely influenced by the series of actions performed during the operation. With a transient cuttings-transport model, it is possible to get a continuously updated prognosis of the distribution of cuttings in suspension and in beds along the annulus. This information can be of prime importance for making decisions to deal with and prevent poor hole-cleaning conditions. A transient cuttings-transport model has been obtained by integrating closure laws for cuttings transport into a transient drilling model that accounts for both fluid transport and drillstring mechanics. This paper presents how this model was used to monitor two different drilling operations in the North Sea: one using conventional drilling and one using managed-pressure drilling (MPD). Some unknown parameters within the model (e.g., the size of the cuttings particles) were calibrated to obtain a better match with the top-side measurements (cuttings-flow rate, active pit reduction as a result of cuttings removal). With the calibrated model, the prediction of cuttings-bed locations was confirmed by actual drilling incidents such as packoffs and overpulls while tripping out of hole. On the basis of the calibrated transient cuttings-transport model, it is thereby possible to evaluate the adjustments of the drilling parameters that are necessary to stop and possibly remove the cuttings beds, thus giving the drilling team the opportunity to take remedial and preventive actions on the basis of quantitative evaluations, rather than solely on the intuition and experience of the decision makers.

Drilling fluids perform a number of important functions during a drilling operation, including that of lifting drilled cuttings to the surface and balancing formation pressures. Drilling fluids are usually designed to be structured fluids exhibiting shear thinning and yield stress behavior, and most drilling fluids also exhibit thixotropy. Accurate modeling of drilling fluid rheology is necessary for predicting friction pressure losses in the wellbore while circulating, the pump pressure needed to resume circulation after a static period, and how the fluid rheology evolves with time while in static or near-static conditions. Although modeling the flow of thixotropic fluids in realistic geometries is still a formidable future challenge to be solved, considerable insights can still be gained by studying the viscometric flows of such fluids. We report a detailed rheological characterization of a water-based drilling fluid and an invert emulsion oilbased drilling fluid. The micro structure responsible for thixotropy is different in these fluids which results in different thixotropic responses. Measurements are primarily focused at transient responses to step changes in shear rate, but cover also steady state flow curves and stress overshoots during start-up of flow. We analyze the shear rate step change measurements using a structural kinetics thixotropy model.

Abnormal surface torque and hook load values are symptoms of downhole drilling condition deterioration which can result in unexpected situations. Usually, friction tests are performed at regular intervals and rig personnel uses these measurements to monitor trend variations in order to detect possible risk of poor hole cleaning or increased borehole tortuousity. The quality of the detection can vary greatly in function of the work load and experience of the drilling staff. The availability of real-time measurements through data servers make it possible to automate and systemize the monitoring process and therefore trigger alarms before drilling problems really occurs. This paper presents a computer system used to systematically analyse real-time data in order to monitor downhole conditions. Such a system can utilize much more data than just the above mentioned friction tests, because mechanical, hydraulic and temperature models can calculate predicted hook load and surface torque in any drilling conditions. The numerical models are automatically calibrated (adjustment of drill-pipe linear weight, factors for mechanical and hydraulic friction and heat generation). The evolution of proper calibration factors is used to detect poor downhole conditions. Automatically generated messages are sent to key personnel who can evaluate the potential problems and take necessary actions. To validate this new methodology, the system has been run on recorded data from three wells on an oil field in the North Sea. A data filtering technique has been developed and applied to solve problems with noisy and erratic real-time signals. With correct input parameters, the system has clearly indicated unexpected measurements several hours before a pack-off problem occurred and therefore proven that the methodology could help in detecting the worsening of downhole drilling conditions. Availability of large amount of real-time data at the rig site or in onshore drilling centres does not necessarily facilitate the recognition of drilling problems. However, online interpretation systems, as the one described above, can systematically analyze the logged data to detect as early as possible the deterioration of hole conditions during drilling operations and corrective actions can be taken before any major problem has really occurred.

Accurate characterization of the rheological behavior of non-Newtonian fluids is critical in a wide range of industries as it governs process efficiency, safety, and end-product quality. When the rheological behavior of fluid may vary substantially over a relatively short period of time, it is desirable to measure its viscous properties on a more continuous basis than relying on spot measurements made with a viscometer on a few samples. An attractive solution for inline rheological measurements is to measure pressure gradients while circulating fluid at different bulk velocities in a circular pipe. Yet, extracting the rheological model parameters may be challenging as measurement uncertainty may influence the precision of the model fitting. In this paper, we present a method to calibrate the Herschel-Bulkley rheological model to a series of differential pressure measurements made at variable bulk velocities using a combination of physics-based equations and nonlinear optimization. Experimental validation of the method is conducted on non-Newtonian shear-thinning fluid based on aqueous solutions of polymers and the results are compared to those obtained with a scientific rheometer. It is found that using a physics-based method to estimate the parameters contributes to reducing prediction errors, especially at low flow rates. With the tested polymeric fluid, the proportion difference between the estimated Herschel-Bulkley parameters and those obtained using the scientific rheometer are −24% for the yield stress, 0.26% for the consistency index, and 0.30% for the flow behavior index. Finally, the computation requires limited resources, and the algorithm can be implemented on low-power devices such as an embedded single-board computer or a mobile device.

Summary: Current topside and downhole instrumentation at the wellsite has been developed to manually conduct drilling operations. The emergence of automatic drilling-analysis software shows the limitations of today's measurement capabilities. It is therefore time to analyze the requirements for on-site instrumentation to implement new, efficient drilling-automation technologies. On one hand, drilling is facing more and more difficult conditions with narrow geopressure windows, deepwater, or high-pressure/high-temperature (HP/HT) conditions. On the other hand, unconventional hydrocarbon reserves may require a considerable amount of wells to be profitable. Drilling automation, by means of smart safeguards, automatic safety triggers, managed-pressure drilling (MPD), and ultimately complete or semiautonomous drilling rigs, can provide the solution to safely construct wells in these challenging settings. The common denominator for the vast majority of drilling-automation solutions is the use of physical models of the drilling process in the form of heat-transfer, mechanical, and hydraulic models. By analyzing the requirements of those models for necessary boundary conditions, it is possible to derive which measurements should be made both at surface and downhole to obtain stable and accurate calculations. This analysis also provides a way to estimate the necessary accuracy of the boundary conditions to ensure reaching the target control tolerance. By use of the boundary-condition analysis, it is possible to derive precisely which measurements should be performed and where they should be performed. As a result, a typical organization of sensors that is compatible with the implementation of drilling-automation solutions is derived.

Summary: Abnormal variations of the pit volume are a prime indicator of mud losses and formation-fluid influxes. However, the detection of gain and loss at the pit level can be difficult because of the transient behavior of several components in the hydraulic loop. For instance, while circulating, the increased pressure inside the drillstring and the annulus results in a larger amount of mud taken by the well compared with static conditions because of the compressibility of the drilling fluid. Furthermore, the transport and separation of cuttings also influence volume variations in the pit. Finally, the retention capacity of some of the mud-transport and -treatment equipment (e.g., return flowline, shakers, sand trap, degasser, transfer pit) has a direct impact on the active-pit level. All these effects are transient and can cause substantial variations of the active-pit volume that may interfere with any solutions attempting to automatically detect gains or losses. Most of these effects are well-known and are dealt with in a pragmatic way with fingerprinting between the current pit-volume variation and a reference pattern obtained under similar drilling conditions. Nevertheless, the fingerprinting method has its limitations when the current drilling-parameter sequence does not have an obvious reference pattern. Consequently, automatic gain/loss detection algorithms using pattern matching with previously observed transient periods may have difficulties reducing the number of false alarms to an acceptable level. Measurement of the flow rate out close to the outlet of the well is another way to detect gains and losses. It has the advantage of not being influenced by the side effects of mud-treatment equipment. However, it is an instantaneous value instead of being cumulative as is the pit volume, and therefore requires precise measurements to be reliable. The limitations of the two mostly used flow-rate-out sensors, the flow paddle and the Coriolis flowmeter, are discussed in Cayeux and Daireaux (2013). We present in this paper an analysis of the transient phenomena that limit the existing detection techniques, and explain how those effects can be accounted for with proper modeling. This leads to a significant extension of the domain of applicability of the traditional approaches.