E&P Plus Digital Solutions: AI-powered Gas Leak Detection Technology Improves Workplace Safety

[Editor's note: This article originally appeared in the January issue of E&P Plus. Subscribe to the digital publication here.]

U.S. oil and gas companies are seeking new technologies to address an old problem: gas leaks. Many organizations have long relied on fixed hardware or manual assessments to monitor leaks. Naturally, the occurrence of a major leak requires immediate response, but detecting the source of a leak from the beginning can ensure a quick and safe correction or evacuation.

Blackline Vision, the data science team of Blackline Safety, is developing its AI Gas Leak Detection module, which resolves this issue by leveraging artificial intelligence (AI) to automate the process while providing situational awareness and connectivity. The module provides advance alerts of gas leaks by identifying patterns in low-level gas readings streamed to the Blackline Safety Cloud from G7 wearable gas monitors.

“The module’s algorithm automatically gathers a lot of data from all of the different devices and indicates where a leak or process fault is happening,” said Lohrasp Seify, lead of Blackline Vision. “This advance notice of a gas leak helps minimize unexpected future downtime while keeping workers safe.”

Traditionally, health and safety professionals would conduct assessments to identify the riskiest places of gas exposure or process faults and then determine where to put fixed systems. A small leak, he noted, would take weeks or months to hit a point where a significant amount of gas has escaped and thereafter cause instrumentation to go off.

“With our system, you no longer have to do guesswork,” Seify said. “You don’t have to think about where the possible leaks are going to be and then make an investment. All you have to do is equip your people with the devices, and before leaks become real problems, we automatically detect and report them in their initial stages.”

Blackline’s module provides alerts in the early stages of a leak and does not require any additional infrastructure costs or equipment, eliminating the risk of human error and maximizing an organization’s uptime while keeping workers safe.

“We took what was already equipped on workers and increased the value by an order of magnitude,” he said. “There is no change management required, so our clients quickly go from wanting to use our technology to seeing the value. And in our industry that’s kind of unheard of because usually you have to keep setting up infrastructure.”

Each worker is equipped with a personal connected gas detector, which enables Blackline to monitor an entire site and collect residual, low or high gas readings in real time. The georeference data are accumulated and sent to a central server, Seify said.

“To date, we’ve collected 130 billion datapoints, so it’s not easy to just look at it and determine if there is a leak,” he said. “You have to automatically study a bunch of different dimensions and our AI can do that.”

In addition, the G7 device indicates if the worker is using the technology efficiently like administering regular calibrations. It also features SMS messaging and voice-calling capabilities, so in the event of a safety alert, the user can communicate directly with the Blackline Vision team.

The module is still in the testing phase and will remain there until early 2021. But the company intends to use this period to gather feedback on how to improve the model and how to increase the learning capacity of the AI.

“We would like to understand what the uptick is going to be like in terms of what value our technology is going to add, and I think that’s going to crystallize over the next six months and reveal where it will be most useful,” he said. “After this loop of self-learning is established and the AI successfully learns from the feedback that our end users give it, we’re going to make it available to all of the Blackline Safety customers, and they will be able to leverage this module to be safer and avoid leaks and process faults proactively.”