When engineers at Shell’s Ormen Lange gas processing plant in Norway wanted to inspect the 70 metre tall flare stack, the plant had to shut down for up to two weeks. Tasked with carrying out fault inspections, a team had to abseil down the stack section by section in a lengthy and potentially dangerous process.
Today, this can be done in a matter of hours and the plant remains fully operational. Aerial drones fitted with cameras and other sensors are can effortlessly access those areas that previously required complicated rigging or scaffolding in order to accommodate the human inspectors.
Like what you’re reading? Get the free UnHerd daily email
Already registered? Sign in
Falls from a height were responsible for 28% of fatal injuries to workers between 2012-2017, according to the UK Health and Safety Executive – making them the leading cause of death at work. These deadly falls aren’t specific to industrial inspection, but it’s a stark reminder of the dangers involved in sending workers to climb tall buildings or equipment, and an indication of the lives that could be saved by reducing such work to a minimum.
That’s why for companies involved in the field, incorporating aerial drones and other kinds of unmanned vehicles into security, surveying and maintenance practices has the potential to both lower costs and increase safety at the same time. That’s one of the reasons why in 2017, PwC estimated that the global market for drone-powered solutions for the power and utilities industry was worth around $9.46 billion per year, with that figure set to grow as the technology is more widely adopted into standard practices.
One company which has already carved out a niche in this market is Cyberhawk, which specialises in aerial surveys of power infrastructure systems.
Founded in 2008, Cyberhawk provides data collection and photographic analysis of power stations, wind turbines, and oil and gas rigs. To conduct these surveys, an experienced pilot (counting a minimum of either 18 months’ experience or 500 hours of flight time) partners with an inspector to carry out an onsite examination.
For the power generation and resource extraction industry, surveying has historically been a costly and dangerous job. But as more parts of the process can be outsourced to machines, this is rapidly changing, as Patrick Saracco, VP for Technology Solutions at Cyberhawk, explains to me.
Before the current era, Saracco says, “Either you had people on rope access or scaffolding, or they were unable to do it at all. So now the UAV [unmanned aerial vehicles] is taking people away from risk, which is one of the biggest benefits, and also allowing the client to obtain much better and more detailed information about their specific assets.”
Use cases like these are a counter to the somewhat dystopic image of drones that has made its way into popular culture, in which the hovering robots have come to be seen as a symbol of surveillance culture or an intrusive public nuisance.
A recently published report produced by the Foundation for Responsible Robotics, titled Drones In Service Of Society, argues that there are numerous applications for drones in the fields of journalism, environmental research, political activism and humanitarian aid, all of which can bring large benefits to society provided that ethical and technical regulation is implemented correctly. “I have long campaigned against the excessive use of drones in armed conflict and policing as well as their erosion of our privacy,” said Professor Noel Sharkey, co-director of the Foundation, on the report’s launch. “However, we mustn’t throw the baby out with the bathwater.”
Most current uses of drones pre-suppose a human operator in direct control of the vehicle, whether in line-of-sight or elsewhere at a remote control station. Over time however, as the field evolves, a gradual shift is underway towards the deployment of more autonomous drones, which will be able to carry out detailed safety observations with a minimum of human involvement – a process that has just as much to do with breakthroughs in software as hardware.
As one example, in Boston, Massachusetts, autonomous inspection company Avitas Systems has been developing robotic systems capable of detecting faults with a minimum of human intervention.
According to the MIT Technology Review, Avitas Systems makes use of drones, wheeled robots, and autonomous underwater vehicles to collect images required for inspection from oil refineries, gas pipelines, coolant towers, and other equipment. The data collected is then fed into a neural network which has been trained with a large dataset of images of faulty equipment and, in theory, can be flagged as potentially dangerous if any of the same features are detected.
And while the machine learning technology supplied by Nvidia is an ‘off-the-shelf’ solution, even more can be achieved with custom development. In one example, a research team from MIT recently succeeded in producing an ultra-low power chip with a size of less than half a square centimetre, which can process camera images and inertial measurements in order to pilot drones about the size of a bee.
Though this particular technology isn’t yet ready for commercial deployment, it points toward a future where small, cheap surveying robots will be a standard part of industrial plants of all kinds, able to perform regular fault detection routines without human supervision.
While there are concerns about the jobs that such autonomous systems will replace, large-scale electronic data capture can provide a level of reliability beyond the ability of even the most experienced human surveyor – and without the drawbacks of fatigue or lapses in attention; a huge bonus for the maintenance of critical infrastructure and much safer for workers.
Although some market analysts have drawn attention to the fact that adoption has been somewhat slow so far, the potential gains for the power and utilities sector are unambiguous. We’re still a good few years, perhaps even a decade or two, from the era of ubiquitous drone use in the power sector, but at this stage the question is not one of “if”, but “when”.