Those companies that are willing to make the necessary changes in their operations will be able to use Artificial Intelligence to improve operation times and reduce their costs.
By: Christophe Dupuis
Division Manager Ecuador, Colombia, and Peru, SLB
The revolution of artificial intelligence (AI) has arrived. Achieving significant advances in consumer markets, from ChatGPT to the digital assistants deployed by all major tech companies.
However, in our professional lives, progress is slower. Despite the speed of this technology, all companies have a digital strategy that includes AI and an increase in digital budgets.
During the initial phase in recent years, companies equipped themselves with AI specialists and toolsets to try to solve specific problems through a series of proof of value (POV) projects. In some cases, these were successful, but the next big test for these strategies was scalability. The promises of quick and significant gains turned into delays, disillusioned leaders, and frustrated teams.
Amid this situation, a minority of successful players emerged from the group. They now know the truth: to harness the promise of AI, companies must be ready for a deep change. There are no shortcuts. First, they must decide whether they will be shapers, takers, or makers of technology, and then prioritize focus areas that can capture quick value and avoid falling behind.
Secondly, companies need to build capabilities and adopt a change management mindset that allows them to transform some of their processes and structures to operationalize AI. For example, with a real MLOps approach (Machine Learning Operations), which marks the transition between experimentation and sustainable industrialization. For this, they must evaluate and acquire a platform that supports their ambitions. If a company skips any of these elements, it will lose years of progress; while its competition accelerates, creating a gap between the two that will be almost impossible to close.
The time for experimenting with POV must stop. Those who are visionary enough to understand the current urgency of business changes are the only ones who will be boarding the AI train, a rather fast one.
The reward for doing this well is worth the effort, and the results are no longer science fiction. Today, underground, we have managed to improve the accuracy of the ownership model by 60%, compared to geostatistics. We also managed to reduce the response time by 75%, turning a week’s work into one day’s work. This significantly reduces the risk in well placement and seriously improves any economic decision based on reservoir simulation. That’s why it changes the reality for the asset team and the company executives.
For a few fortunate companies onshore and offshore, drilling planned trajectories without human intervention is no longer old news. Operations have paid for themselves by increasing consistency among teams while improving penetration rates, connection times, and reducing costs of well construction events during pipe changes and casing, not to mention improvements in health and safety (HSE) statistics. One of the most complex operations, geonavigation, can now be performed autonomously, something many experts considered impossible just a year ago.
In addition to drilling operations, planning has been digitized to increase the speed and quality of displacement analysis, design, and interactions of the planning team. With new cloud-based platforms, wells are re-evaluated on the fly, allowing the team to review significant changes highlighted by an intelligent platform. With each new activity, the platform learns and improves itself, and the rest of the operation, of course, is always under user control. The impact, again, translates into high consistency among engineers, making each well better than the last, and giving confidence to the company heads due to improvements in time and budget.
All these applications are powerful, but they may not match the return of a transformation of production operations, where innovators are already reaping daily gains and will do so for years. A handful of pioneering companies have managed to change the game with smart wells, improving daily production and safe restarts after force majeure events, and working to extend the lifespan of the lifting system and the well. Managing pump parameters and chemical injection every second based on the actual flow pattern in a closed loop is the best way to keep the well healthy, producing, and free from repairs.
Prevention is undoubtedly the best strategy, similar to the exercise of brushing your teeth, rather than going to the dentist only when the pain is unbearable, to receive an equally undesirable bill. This should be the decision of every manager who wants to increase production while reducing operational costs and exposure in HSE.
Beyond wells, autonomous management of drilling platforms is already a reality. For example, balancing injection and production patterns with AI to improve water flood recovery, something that typically requires complex analyses and interpretations across departments, numerous meetings, and often a bit of improvisation. Upgrading the drilling platform also provides the opportunity to eliminate flaring, addressing a significant source of emissions that the industry needs to tackle as a top priority. AI also helps manage safety at many drilling sites today, with video analysis successfully deployed in the region.
We are now on the path to reaching the final level: the autonomous management of a field. In a rapid cycle, this means implementing a digital twin of the production system underground and on the surface, to deliver superior production at the lowest lifting cost. Another cycle will recommend field development decisions in a more systematic, more frequent, and more economical manner than a traditional manually updated workflow, freeing our time for tasks that no AI can achieve.
These few examples offer tangible benefits of how AI is revolutionizing the oil and gas industry. None of them have existed for long, highlighting the speed of development of these technologies. They were selected from a library of over 100 deployable applications built on the best available platforms. The first shapers built them in close collaboration with specialized teams worldwide, combining expertise in data science and decades of domain knowledge.
If your company seeks to be a shaper, this collaboration is available at any location to help you connect your corporate data and intelligence with the latest evolutions in machine learning and GenAI. There is no need to reinvent the wheel; you can focus the budget on perfecting it.
If your company’s strategy is more of a taker, the good news is that quick benefits are possible without risk and without the need for the fundamental changes mentioned earlier. It only takes deploying some of the many proven solutions. For example, retraining and validating an existing machine learning algorithm for predicting failures of electric submersible pumps (ESP) could be the solution. Or the conversational insights from your corporate document base are an easy entry point to GenAI. The most important thing is to start; you can always become a developer later.
In conclusion, the fundamental impact of AI in an industry like oil and gas is that the competitive landscape will be reorganized before 2030. Time is running out, so if you are still at a station, why not take the next step? The only regret will be not having done it sooner.
Colombian Association of Oil and Gas