AI delivers amazing results in diagnosing sick humans, imagine what it can do for diagnosing sick oil & gas wells.
I attended a presentation on well intervention during the completion phase, and it struck me how similar well intervention is to healthcare, and how valuable AI can be in the diagnosis and treatment of wells. Whether you’re dealing with a sick well—casing breach, stuck tools, production drop-off, etc.—or a sick human, you need to diagnose and address the problem quickly. In the drilling and especially in the completion phase, well intervention is like trauma surgery, you’re on the clock, a very expensive clock. You’ve got to diagnose and resolve the issue now!
Of course, these issues invariably arise in the middle of the night, when the engineer doesn’t have the collective wisdom of company leadership to help. This is where AI can not only aggregate the collective wisdom of the company, and the industry, but also the specifics of the basin, and the actual well. But unlike medicine, oil companies don’t yet appear to be exploiting AI in the realm of well intervention. They should, it can deliver amazing results.
Just as doctors rely on symptoms and diagnostic tests to understand what's happening inside the human body, well intervention specialists use similar methodologies for oil wells. In both cases, perfect visibility is a luxury rarely afforded. For wells, this might mean dealing with high-pressure zones, unexpected geological formations, or equipment failures deep underground. In humans, the complexity comes from the body's vast network of organs, tissues, and the myriad ways they can malfunction.
Like the human body, you need to take a holistic approach to well intervention. You need to consider the impact on service providers, unique basin and well issues, the specifics of the well, and availability of diagnostic and well intervention teams in the area. For example, the ideal diagnostic equipment might be in Saudi Arabia and it’ll take a week to clear customs and get to Midland, that’s not an option. What’s available within my window of time and how do the well-specific measurements, such as downhole pressure, stage, stuck items, etc. impact the decision? What happens if we bring in coil tubing for a drillout and it displaces and delays the ongoing frac work? It’s a multi-variate problem, something AI is particularly good at.
In the healthcare world, AI is being used for these same multi-variate problems, such as evaluating health variables, gender, lifestyle, age, pre-existing conditions, drug allergies, baseline diagnostic numbers versus current numbers and more. Feeding the AI models with health histories, diagnostic data and statistics, and the like enables them to narrow the issues and use testing to focus on the true diagnosis and resolution. This is exactly what completion, drilling, and production engineers need to diagnose and treat their sick wells.
In healthcare, doctors consult medical literature, past cases, and peer discussions to treat patients. Similarly, in well intervention, engineers leverage past experiences, case studies, and expert advice. However, the frequency of such interventions isn't daily, making each event a learning opportunity but also a challenge due to the irregularity of practice. The way engineers learn about well intervention starts with leveraging the company experts. Then these engineers share their well intervention war stories over drinks. Learning from the mistakes of others is always cheaper and less painful to your career.
Here's where AI, particularly through the application of LLMs, can make a significant impact:
Imagine this. You’re a completion engineer. You get the call at 1:00 AM that your well has a problem. You open the AI Agent. It has already loaded all the data about the well. It knows everything. In a chat, it asks for any details you’ve heard from the wellsite and any mitigating factors like how services providers might work around it to minimize costly downtime. It processes this in less than a second and suggests the most probable issues with the well, along with the statistical possibility of each and the diagnostic tests, and associated costs, to help you narrow the diagnosis. You and the AI agree on the proposed steps. It then reaches out to available service providers in the area to execute the plan, adjusting as needed based on availability and response time. In continues like this through the troubleshooting process. Once the issue is resolved, the AI interviews you for additional insights and then writes up an incident report. Problem solved.
Implementing AI in well intervention involves:
The potential impact includes not just cost savings through more efficient interventions but also enhancing safety, reducing downtime, getting wells on production faster, and extending the productive life of wells. In healthcare, similar applications have led to better diagnoses, personalized medicine, and improved patient outcomes.
The analogy between well intervention and human healthcare is striking, especially in how both can benefit from AI innovations. By learning from one another, these fields can advance their practices, making interventions smarter, safer, and more effective. As we continue to develop AI technologies, the synergy between these disciplines could lead to groundbreaking advancements, ensuring that both wells and humans can be 'treated' with unprecedented precision and foresight. If you are interested in leveraging AI for well intervention, let us know, we can schedule a workshop to figure out how we can build this solution with you.
Subscribe to our newsletter to receive the latest updates and promotions from MPH straight to your inbox.
