The oil and gas industry has picked up on the benefits of digitization and artificial intelligence (AI) in its day-to-day activities, and the health, safety, and environment (HSE) sector is no exception. While AI brings clear benefits, the risks that come with those benefits remain unclear.
While touting the advances of technology in HSE at SPE’s Virtual Annual Technical Conference and Exhibition (ATCE), Olav Skar, director of health, safety, security, and wells at the International Association of Oil and Gas Producers (IOGP), said, “I also see risks, and I remain concerned that we do not truly understand them.”
Skar spoke at ATCE on a panel that included Mohamed Kermoud, Schlumberger’s global vice president for HSE, and Philippe Herve, the vice president of energy solutions at SparkCognition. The panel was moderated by Josh Etkind, Shell’s Gulf of Mexico digital transformation manager.
“A lot of power is in the technology,” Herve said. “The technology is beautiful. How we as humans are going to apply it, we need to think about it. We are thinking about all of the good things that the technology is bringing to humanity. Let’s keep it that way and remove all of the application of artificial intelligence technology that may not be well perceived or beneficial to anybody.”
An early target for digitization in oil and gas HSE has been the most dangerous activity for employees: driving. The IOGP claims that land-transportation-related incidents historically have been the largest cause of fatalities for its member companies. Since 2000, such incidents have accounted for 22% of all work-related fatalities reported by IOGP members.
Schlumberger’s approach to driving safety was outlined in a paper presented at the 2020 SPE International Conference and Exhibition on Health, Safety, Environment, and Sustainability, a synopsis of which appeared in the August 2020 issue ofJPT.
Schlumberger’s approach to improving driver safety includes an advanced driver-assistance system that alerts drivers of maximum speed limits, lane departures, and the proximity of pedestrians and other vehicles. The goal of the system is to effect good driver behavior.
“If you analyze all the data, all the incidents, you find that behavior is always behind it.” Schlumberger’s Kermoud said. “People are trying to save time, to save the day. … The rules are generally perfect, but the behavior is something that we absolutely need to make sure that we impact one way or the other. And using technology will help us.”
In addition to improving behavior, Schlumberger’s approach involves monitoring drivers’ performance with vehicle-mounted sensors. With employee monitoring, however, come concerns about privacy.
Herve cautioned that the Big Brother approach is the wrong approach, adding that workers are likely to try to circumvent systems that feel intrusive. He mentioned an increasing proliferation of wearable devices, such as Apple Watches, that record health statistics.
“More and more, we are going to be able to have sensors that we provide that our employees are wearing,” he said. “But it needs to be very much thought about what information is being gathered by the corporation compared to what information is provided to the employee that he should take his own decision on.”
Despite the privacy concerns, Herve said he sees the benefits of advancing wearable technology in the workplace, but he continues to caution that a line must be drawn. “I think the employee, when he is at the wellsite, should know whether he is having a heat stroke or a potential heart attack or if his oxygen level in the blood is dropping too much,” he said. “And he can have all kinds of sensors to help him do that, and it is very useful in dangerous situations. But that information should not be systematically collected by the corporation.”
Wearable technology also has improved safety at facilities. Location sensors can alert workers when they are entering dangerous zones such as under a crane. And augmented reality and virtual reality can be used to guide maintenance and training; but, hurdles exist there, too.
“A lot of the time, when you have those virtual-reality tools where you can walk around an installation—and we’ve all tried those goggles and those tools—it’s fantastic stuff,” Skar said. “But it doesn’t bring you to a mastery level of competency. It gives you a guided tour of an installation. You can only get real deep competency experience if it’s accompanied by somebody who really knows about the tools or with an active transmission of additional information. And if we can get that through AI, fantastic. It will make us better.”
Analysis of safety data and reports is also on the AI radar. Turning AI loose on this data can generate greater understanding of a company’s safety situation. “When it comes to safety, you can apply today artificial intelligence to improve the safety of your entire corporation, of your employees, and not only the safety but also the impact on the environment, on the health of your employees,” Herve said. “That’s absolutely the first thing you should be looking at is how to you apply AI to the corporation to improve maintenance, to improve our understanding of the safety reports, of the near misses, of the entire corporation to better understand how to improve the safety of our employees.”
“The more data you have, the more reports you have, the more hazards you identify, the less incidents you have,” Kermoud added.
While predictive analytics for tools and machinery, such as pumps or other downhole tools, has been a target of AI for a while now, attaching that same effort to safety has been elusive. Concerning predictive maintenance for equipment, “We are in good shape there,” Kermoud said. “However, we are not applying it at the same level in HSE to predict when could the incident happen. And we have a lot of work to do to be able to get to that level.”
Kermoud said the predictive safety target could be reached by “looking at the engagement of the people, looking at the history of these locations, looking at the training of the different people. And here is where artificial intelligence and big data analytics can help us.”
These efforts, the panel agreed, should not be proprietary. “We are in a very competitive industry and environment, and I think that, if there is one domain where we should not compete, it is HSE,” Kermoud said. “Here, we should work very closely together. If I come up with something that helps save lives, I would be very glad to share it because life is priceless.”
“We are just scratching the surface of this domain,” he added. “I think there is a lot of data that we have to benefit from. We have the tools now. We just have to accelerate, give it a little bit of acceleration.”
