Detecting persons overboard

How an intelligent detection system on a jack-up rig provides automated person overboard detection and alerts.

Workers on offshore rigs have to contend with severe weather and rough seas, and ensuring their safety in these conditions is a key concern for oil and gas operators. Although accidents such as falling overboard are infrequent, they remain a potential risk. To prevent such incidents, one of the world’s largest offshore rigs in the North Sea, the Valaris Stavanger, has adopted an AI-powered system from solutions company Zelim that automatically detects and tracks persons overboard at sea.

An intelligent AI solution

Called Zoe, Zelim’s solution is a person in water (PIW) technology that helps search and rescue operations in the ocean. It uses a suite of cameras and sensors to monitor the area around a vessel, raising alerts if someone falls overboard.

Zoe was deployed on the jack-up rig Valaris Stavanger in February this year, and installed a tailored solution that included seven infrared and optical cameras placed around the vessel, allowing the entire rig to be monitored. Following an initial trial period, Valaris will evaluate the suitability of the system for use on other assets in their fleet.

“The harsh weather in the North Sea can change quickly. Traditionally if someone falls into the water, the alarm is raised, and a person acts as a spotter to maintain visual contact. We have been totally reliant on line of sight,” Glen Spearman, offshore installation manager onboard Valaris Stavanger, explains. “However, now Zoe gives us a 360-degree view from the rig at any one time. It will reduce our response time and improve the chances of a successful rescue in the event of an accident.”

How it works

When someone falls overboard, Zoe transmits an automatic alert to the radio operator, who can view a 10-second clip taken when the alert was triggered while simultaneously viewing live footage. Zoe continues to track the person in the water, thereby increasing the chance of  a successful rescue.

Mayday call geolocation

The system then logs the position of the vessel and the person overboard at the point of detection for mayday call geolocation, before providing the radio operator with a mayday script and actions checklist, ensuring fast and accurate reporting.

“Zoe has been designed and developed in collaboration with the US coast guard,” says Doug Lothian, chief technology officer at Zelim. “The system has been through extensive testing in adverse conditions and given a 96.8% successful detection rate of up to 337m, above standard requirements. We worked closely with Valaris on this project in their working environment. This meant we were quickly able to understand their specific customisation requirements.”

Data collection

Data from Zoe is processed onboard the vessel for security and full functionality with no internet connection. The system’s AI models have been built on a dataset of over 4.9 million labelled maritime rescue images across a wide range of ocean conditions. According to Zelim, the technology could be used in many other settings, including automatic security threat detection and alerting as well as red zone monitoring.

Alongside Zoe, Zelim is developing hardware to aid search and rescue operations, including a rescue ‘conveyor’ called Swift and, additionally, an unmanned rescue vessel called Guardian.