Search

In the current challenging downturn market scenario, many companies have turned their attention towards enhancing the existing training and knowledge transfer programs for their engineers to reduce costs and improve efficiencies. Keeping in mind the increased spending cuts that the industry is facing, the SPE Gulf Coast Section Continuing Education Committee collaborated with leading technical experts in the drilling domain, and came up with a novel development model to transfer knowledge at reduced costs, while simultaneously raising funds for student scholarships. In order to provide exposure to advance technologies that focus on improving the drilling performance, a ‘Drilling Workshop' was planned for both the young professionals and the experienced engineers. With the objective of maximizing knowledge transfer and learning, the workshop was scheduled as a one-day event having two separate 3-hour sessions, and focused on current interest areas associated with emerging technologies in drilling. The registration fees for the workshop per attendee ranged from $150 to $250 depending on other organizational costs. Overall overheads were low as compared to some of the existing industry training sessions as no expenses were incurred on instructors' fees or venue costs. As an additional incentive for the companies to allow their employees to attend the workshop, free copies of a textbook related to this topic and valued equal to the registration fees, were provided to all the attendees for further enhancing their engineering fundamentals. This paper will present detailed features of this knowledge transfer model and will highlight ways in which this model can be effectively incorporated and used by other SPE sections to benefit young engineers and experienced professionals in this current downturn market. Over the course of the last 24 months, 10 such workshops were conducted in various SPE sections in the United States to successfully raise about $60,000 in student scholarships for SPE. In addition, a team of five students was financially supported to attend the SPE Annual Technical Conference & Exhibition in Amsterdam in 2014 using the funds raised from the workshop. This training and development model has the potential of being a great means of knowledge dissemination across the board in the oil and gas industry. These workshops successfully provided a cost-effective means to revisit the technical fundamentals, along with an avenue for continuous learning for engineers, keeping them updated about the latest available technologies in the industry. This paper will be very useful for all the SPE sections globally, helping them adopt this model for knowledge transfer in challenging times.

As the drilling industry enters the era of big data, it has become necessary to find ways to organize and understand the vast amounts of real-time high frequency data recorded. Shor et al. (2014) show that real-time accelerations recorded down hole actually differ from those predicted by various models, either in the high frequency range or at peak values. These differences in vibration levels can be attributed to noise from various sources, such as the motor or surface equipment, design of measurement devices, uneven shape of the borehole, and/or dynamic effects, such as dampening/cushioning of fluids, bit rock interactions, drillstring-borehole wall interactions, etc. These differences can cause real-time drill log analysts to miss instances of actual failure or predict false failures. The industry currently focuses on the peak and average values of the accelerations recorded down hole, and there is no established method to examine the trend of these accelerations. The methodology proposed in this paper first focuses on obtaining the actual unmodulated acceleration values from the recorded downhole gauge data and then processes these values to understand the trend of acceleration as the drilling proceeds. This demodulation (or deconvolution) of the recorded data can reduce the likliehood of false predictions and, at the same time, increase the credibility of the mode of acceleration predicted (and hence more accurately predict failure), which currently relies only on the experience of the engineer. Thus, this technique can make real-time data monitoring more reliable and simple. Further, if combined with a gamma ray log to know the lithology of the formation being drilled, this data monitoring technique can reveal a significant amount of information. In addition, this paper suggests calculating instantaneous jerk intensity and decomposing it to its monotonic intrinsic mode function. This reveals that these monotonic functions follow a certain trend, which might be useful in future endeavors to understand the underlying physics of drillstring failure.

With increased well complexity, difficulty matching the actual well profile with the planned well profile increases. This study discusses building a workflow for automating the bottomhole assembly (BHA) to take corrective actions upon detection of a deviation from the planned well path in a 3D coordinate space, and constructing and programming a prototype demonstrating the proposed theory. The corrective paths would be conventional constant curvature curve and unconventional curves, such as catenary and spline, which are proven to be smooth, thereby reducing torque and drag in the well. The main constraint governing the unknown parameters of the corrective path is the minimum well profile energy criterion, which is excellent in terms of producing smooth curves because it minimizes the curvature and torsion of the curve together. Balanced tangential and minimum curvature methods are used for building the corrective trajectory. Previous work on downhole drilling automation showed that the minimum energy method, compared to proportional, integral, and derivative (PID) control and fuzzy control, where the model was built for 2D and 3D coordinate space by considering conventional constant curvature curves for correction, yields much smoother wellbore trajectories. The work was continued for unconventional well profiles in 2D wells, where spline, catenary, and clothoid curves were used to return the deviated well to the original planned path. This work focuses on building the model to determine the corrective paths based on unconventional curves for wells in a 3D coordinate space. The model first determines the point on the planned path where the bit needs to join and then estimates the best possible path to do so. The whole corrective workflow is demonstrated with a small scale working prototype. Finally, the results of various optimization models obtained by combining different unconventional curves and trajectory calculation methods are compared.

Predictive data-driven analytics has driven massive interest, primarily because of its successful implementation in several industries. Data-driven modeling and its application to predict downhole environment will prove to be the future of drilling operations, as they hold the potential to optimize highly complex drilling operations. In general, rate of penetration (ROP) optimization involves adjustment of the weight on bit (WOB) and rotary speed (RPM) for efficient drilling. ROP follows a complex relationship with several other parameters such as formation properties, mud properties, mud hydraulics, borehole deviation as well as the size/type of bit. Traditional regression analysis models have limitations and have limited accuracy while attempting to describe the dependence of one observed quantity on another observed quantity. On the other hand, the artificial intelligence methods face drawbacks trying to understand the physics behind the operations. To ensure the physical and technical feasibility of the prediction, coupling conditions between the two have been developed for the ROP optimization. In this study, a new model based on statistical regression and artificial neural networks (ANNs) was designed to predict ROP using field data gathered from the North Sea Horizontal Wells. Exploratory analysis was performed to find correlations between variables, followed by measurement of predictor importance to infer relative contributions and weights of inputs. Cross-validation was employed to prevent overtraining of models. Several models such as Step-Wise regression, neural networks (NN, KNN), support vector regression (SVR), classification regression trees (CART), were applied for prediction. Ensemble methods such as Random Forests (RF) and Boosting helped increase accuracy and reduce errors. The algorithm designed was further tested on other wells and was shown to predict with significant accuracy. This proves that several parameters need to be comprehensively considered while optimizing the ROP. Since the algorithm presented here doesn't depend on a single model but rather several predictor models, it can be effectively employed independent of location or formation.