As AI helps E&Ps make sense of the torrent of data coming from today’s oil field, there are two areas in which SparkCognition SVP Energy and Sustainability Phillippe Hervé finds extremely significant. One is in drilling optimization, “Which is very, very possible, and we do it with excellent results,” he says. Drilling faster, better, and smarter are the immediate targets toward the overall goal of cutting costs. “Faster” means reducing delays and getting a well into production faster.
Predicting and preventing stuck pipe is toward the top of that list. When drill pipe gets stuck, it can take hours and cost thousands of dollars to remediate, delaying completion and production of the well.
Using AI to predict stuck pipe is really “the visible tip of the iceberg. Because really what you want to do is remove all of the dysfunctions from the drilling operation, such as vibration,” he said, “Getting to your optimum drilling set points.”
According to an SPE paper co-authored by Hervé in 2021, there are two types of stuck pipe. “At a high level, stuck pipe can be categorized into 2 broad categories: differential sticking and mechanical sticking. Differential sticking occurs when the mud column pressure exceeds the formation fluid’s pressure, leading to the pipe becoming embedded in the mud cake, and causing a stuck pipe (DeGeare, 2003). Torque, pick-up, and slack-off can be good indicators to warn against potential differential sticking. Mechanical sticking, on the other hand, can be attributed to pack-off, key-seating, wellbore collapse, and other issues. Standpipe pressure, torque, and hookload are useful as early indicators of mechanical sticking.”
Fishing out the pipe, reconditioning the well afterwards, and making other decisions about what repairs are needed are expensive propositions, Hervé said, adding that a company can lose as many as 10 days of drilling during the process.
The accuracy of AI’s predictive modeling depends on collecting data from all sensors attached to the drilling operation as they look for anomalies in vibration and other things. Making sure the sensors are calibrated is at the center of that effort. “That’s a huge challenge that requires diligence to understand what is happening even when things are not calibrated, which is very, very common in the free world,” he observed. While “There is nothing like perfect data,” advanced AI can indeed overcome an uncalibrated situation. It can also notify the contractor that calibration is needed.
Visual AI Knows If You’ve Been Bad or Good
Many drill sites today are equipped with cameras that record what they see 24/7. These cameras have a number of functions, such as monitoring for theft and providing background data if there is an accident. With visual AI, however, general safety procedures can be monitored, Hervé said.
Currently SparkCognition is monitoring 25,000 cameras in 16 countries. To have humans review all that video would require lots of humans and likely their eyes would glaze over after a few hours.
“How do we look at the hundreds of cameras that exist around the field and on the drilling rig, and analyze this information in real time, and identify violations of things which could be [problematical],” he said.
Predictive and preventing measures are also vital for AI here. If it sees a potentially dangerous situation AI will raise an alarm to “stop a piece of machinery” to prevent an injury, an equipment failure, or something else.
AI can recognize whether workers are wearing proper personal protective equipment (PPE) as required by the Occupational Safety Health Administration (OSHA).
While technically visual AI could recognize faces and allow an operator to take to task anyone not wearing proper gear, Hervé noted that privacy concerns keep this under control. Regulations vary across the countries they work with, and, as Hervé said, “Whatever the legislation requires, we comply with.”
The scenario normally is for the Health, Safety and Environmental (HSE) team to receive AI data saying that a certain number of people on well A were not wearing proper PPE equipment or did not follow some other safety procedure. This would then show the HSE group the need to schedule training for that crew, to remind them to follow safety protocols for the good of themselves and others on the site.
If there is an accident onsite, AI can help interpret what happened and, in that case, show who was where at what time, as well as analyze the issue and initiate training or other measure to prevent a repeat.
“We can only do this because cameras are watching everyone 24/7” and AI is interpreting the data and reporting to management.
Visual AI is used on drones to monitor pipeline leaks and security issues. With proper equipment mounted, they can monitor for visible leaks such as oil spills as well as invisible methane leaks.
“With proper permits, drones can fly up to 400 feet, which makes them invisible to normal people” on the ground, he noted. Drones can be scheduled to launch automatically on a schedule, then return to the charging base, which, Hervé noted, looks like a big barbecue pit.
These systems and more leverage the ability of IIoT and AI to quickly sort and evaluate massive amount of data to let people make decisions regarding safety, efficiency and more.
Efficiency, Efficiency, Efficiency
Leveraging Lean, also known as Six Sigma, is a production process based on the concept of maximizing productivity while minimizing waste. Business Process Engineering Manager for SPM Oil and Gas Adriana Herrera says this is what inspires much of the company’s research and analysis, both internally (themselves) and externally (for clients).
“We’re focusing on our processes, finding out where the wastes are, whether it’s inventory waste, transportation waste, or standard operation management.” Price volatility and rising inflation have made more companies aware of the need to reduce costs by streamlining every process possible and eliminating unnecessary steps. Hererra said SPM Oil & Gas had begun leveraging Lean before the COVID downturn, which put them ahead of the game when clients began asking for help with using IIoT (Industrial Internet of Things) to boost their bottom line.
Michael DiDomenico is Systems Information Manager for SPM Oil & Gas’ Asset Management Program (AMP) system. He explains that AMP is one of the ways the company uses data to create efficiencies and provide actionable information. AMP is designed to facilitate the processes of inspection, inventory, preventive maintenance, and other management issues. SPM Oil & Gas uses this system internally to manage its own assets, as well as giving their clients the tools to manage and track equipment and assets across locations and jobs.
Radio frequency identification (RFID), in which a metal tag with a unique identifier is placed on each piece of inventoried equipment, is a major part of AMP. “What do we do right now that we have to type in or write down? Can we add that to RFID and can we make it overall more efficient?” DiDomenico explained.
Automation will soon be added to the equation. Placing sensors on pressure testing devices would reduce user input errors. With RFID the system already knows each device’s location and history, so tracking its data would add another dimension for the client.
When AMP knows the location of every asset, it can manage the process and prevent errors. “We’re tying RFID and the automated pressure testing together, so now they don’t even have to select a serial number to tell it to automatically pump, it just knows what’s inside the cell and it can pump from there,” he said.
And, “it knows if there are conflicts.” When the system is aware that there are two differently rated pressure vessels involved, it will know the two must be done separately instead of simultaneously.
For SPM Oil & Gas and clients, RFID tracks inspection schedules and asset locations, allowing staff to quickly pinpoint the nearest asset if something is needed on a well pad or elsewhere.
Here is an example of one way RFID inventory tracking pays off. One client told SPM Oil & Gas that a manual yard inventory—just involving product type, not identifying specific pieces separately—at a particular location would typically take three employees three to four days. That project would involve physically listing each piece of equipment on a form.
But with each asset having an RFID tag, using the app and a scanner, the client got specific inventories of the same number of items in two to three hours, DiDomenico said.
Producer interest in IIoT grew greatly during the 2020 pandemic, he noted, as companies were forced to staff down. That meant the remaining personnel had to become much more efficient to manage the workload. “A lot of companies were more open to some of the efficiency items we were providing with RFID, so they could get the same information they got previously, but quicker.”
And since recovery, many workers have not returned, while producers are reopening shut-in wells and ramping up drilling, which increases the workload.
“Being able to know how to use people more efficiently and effectively, not having them waste time inventorying for X-amount of hours, lets us use technology to make that quicker and have them focus on things that provide more value,” Herrera added.
In this sphere, no one can sit still. Today’s cutting edge development will be eclipsed tomorrow, often with companies advancing past their own developments.
By Paul Wiseman