With the long-term price of a barrel of oil projected to hover around $50, all actors in the oil and gas industry have been forced to change or adapt their processes to try to reduce costs and sustain profitability. Accompanying the downturn, the oil and gas industry has entered a new era of digitalization, introducing new buzzwords to our industry such as artificial intelligence, digital twin, machine learning, neural network, industrial Internet of Things, and cloud computing, just to name a few. The industry has already begun its transition, going from more-traditional mechanized systems in the hands of humans to more-computer-controlled drilling systems that will eventually lead to full drilling automation.
If it has been proved in the past that innovative technology is excellent for bringing more-efficient solutions to reduce costs, will digital technology be the panacea for the whole industry? While some segments of the drilling and completion industry could logically and more easily benefit from it, some others require great caution and expertise to leverage digital technology in its many forms and levels of complexity. Being able to digest, filter, and analyze a huge amount of data in real time to provide useful and readable indicators and plots for drilling engineers is fantastic and within reasonable reach, because it obviously facilitates the tasks of drilling engineers dealing with real-time data entry. Machine learning, on the other hand, such as neural networks is not new just for the drilling industry but has re-emerged recently with the growth of big data and requires data-science process skills to implement various algorithms properly. Complex, nonlinear, and multiphysics phenomena that involve many parameters and are difficult to measure directly, such as stuck-pipe events, washout, fatigue, wear, and maintenance of equipment, can be tackled through machine learning with relative success as long as the training set of data is sufficient and representative of all possible situations.
However, machine learning or artificial intelligence should not necessarily be seen as a magic tool applicable to any engineering problem, many of which can still be treated with more-traditional physical models that are more-easily understood by engineers. Even though engineering is still undertaken mostly with traditional and deterministic models, it is highly plausible that this digital transformation will redefine the traditional drilling-engineering job in the near future. An increasing number of traditional tasks will be handled by computers (e.g., data analytics, simulations/modeling), but one must keep in mind that decision making will still require engineers able to understand the fundamental principles of the problem fully. Coupling physical models with machine-learning techniques smartly is probably something to explore. Last, but not least, I think that millennials, digital natives, will have a big role to play in this transformation because they are more prone to adapt naturally to the technology they have always been living with.
This Month's Technical Papers
Optimizing Lateral Well Spacing by Improving Directional-Survey Accuracy
Horizontal Drilling With Dual-Channel Drillpipe
3D Simulator Predicts Realistic Mud Displacement in Deviated and Horizontal Wells
Recommended Additional Reading
SPE 181788 Optimizing the Multistage Fracturing Interval for Horizontal Wells in Bakken and Three Forks Formations by Kegang Ling, University of North Dakota, et al.
SPE 183125 Application of Real-Time Geomechanics on a Horizontal Well by Osman Hamid, Saudi Aramco, et al.
SPE 184875 Evaluating Stresses Along Horizontal Wells in Unconventional Plays by Marisela Sanchez-Nagel, OilField Geomechanics, et al.
