The arrival of the Industrial Internet of Things (IIoT) has led to the collection and transmission of vast amounts of new and complex data, in the energy sector just as elsewhere. However, this huge inflow of data, gathered via connected devices such as sensors placed along pipelines in oil and gas fields and power plants, needs to be processed, analysed and interpreted if it is to be useful.
A handful of oil majors have turned to cloud computing at their in-house data centres to take on the challenge. BP’s state-of-the-art Centre for High-Performance Computing (CHPC) in Houston, for example, is capable of over 11.2 petaflops of processing speed, while the Eni Green Data Centre near Milan now has a computational peak capacity of 22.4 petaflops.
These tend to be exceptions rather than a trend in the energy sector. Somewhat surprisingly, companies have been slow to warm to technologies now commonly found in factories, telecommunications and transportation. As noted in Global Human Capital Trends in 2019, artificial intelligence (AI), robotics and automation are only now beginning to be adopted by the oil majors, power companies and grid operators, and only in limited roles.
Ian Phillips of the UK Oil & Gas Innovation Centre describes the current situation: “AI is still in the very early stages of application in oil and gas – generally doing data processing such as analysing trends in a pressure measurement to identify an issue.”
Meanwhile, AI-enabled robots are mostly in the research and testing stage. The £36m Offshore Robotics for Certification of Assets (Orca) Programme is developing autonomous and semi-autonomous AI-enabled robots, capable of inspecting, repairing, maintaining and certifying offshore energy installations.
Currently being tested by Total in the North Sea, these sophisticated robots are equipped with sensors to understand their environment, while algorithms allow the robot to plan, act and interact with humans and the environment.
Artificial intelligence, smart software, cloud- and super-computing are bringing once unimaginable speed to processing, streaming, analysing and interpretation of the flood of data the energy sector generates.
The prospect of greater speed and lower processing costs makes the application of AI very attractive for oil companies, which need a fast and cheap way to determine if there is any oil at the bottom of a drilled well. A well’s geothermal temperature profile, hole size and shape, rock density, porosity and formation, and fluid resistance are among the data that can be analysed by bespoke AI algorithms faster and more accurately than by humans.
Power grid operators also collect a variety of data that requires constant monitoring, analysis and interpretation to ensure the electricity supply meets demand, as well as ensuring the smooth management of the varying contributions from renewables, nuclear and fossil fuel energy in the system.
For their part, wind farm operators collect real-time weather conditions and wind speeds, actual output of the turbines and the amount exported to the grid.
Dedicated AI takes the pain out of data analysis and, according to Harry Block, chief financial officer of Australia’s VROC AI, “significant declines in the cost of data processing and processing power are having a key impact on the use of AI technologies”.
However, being characterised as a late adopter of these tech innovations, which after all lie outside the core competencies of the energy sector, might not be as damaging as appears at first sight. Early adopters of new tech are at risk of being exploited. Time is essential to develop the requisite bespoke algorithms that are needed to programme and construct robots for particular tasks.
Nevertheless, the very size of the oil and gas industry, and the power and renewable energy sectors, does represent a huge market opportunity for AI scientists, engineers and software creators. A snowballing of technological developments in the years to come is almost a cert.
Indeed, Shell’s Daniel Jeavons, from the Advanced Analytics Centre of Excellence, is confident that “throughout our value chain there are big opportunities to exploit AI and, in particular, machine learning”.
Once AI-enabled robots and machine learning are more widely adopted, these innovations have the potential to revolutionise the operations and cost structure of energy companies as well as reducing risk, improving health and safety and changing the skill set of the sector’s human resource requirements and perhaps even the sector’s gender profile.
The acceleration of the introduction of AI and robotics in the energy sector has involved external specialist contractors being brought in to work on projects ranging from oil and gas extraction and power generation to renewable energy and storage and coal and uranium mining.
The power sector has long held a well-developed culture of sharing anonymised data sets, which enables innovators and contractors to develop new AI robotics solutions for managing the power supply. Weather forecasts are an essential ingredient of forecasting customers’ demand for heat and power, and underpin management of both power assets and the grid network. For example, the UK’s power and pipeline grid operator, National Grid, is currently working with Google subsidiary DeepMind to introduce AI technology to the country’s grid systems.
This project involves processing of massive amounts of weather forecast data in order to develop more accurate and predictive models which are better able to anticipate surges in demand for power and calibrate supply and demand more efficiently. The use of AI in this way could reduce Britain’s energy usage by 10 per cent.
To save energy further, AI-enabled smart meters, manufactured by Swiss-based Landis+Gyr, and smart thermometers such as NESTE, are helping households and businesses in many countries to reduce their energy consumption.
Meanwhile, power traders in Europe are using blockchain, provided by Sussex-based Origami Energy, to buy and sell energy on behalf of utilities and employ machine learning to predict power availability and market prices in near real-time. Such use is to improve their bidding on the European Wholesale Power Frequency Response markets.
Utilities around the world are employing leading Texan AI services contractor Spark Cognition to pinpoint the new ‘normal’ in wind turbine performance.
Traditionally, turbine maintenance managers have depended upon condition-based monitoring (CBM) to alert them to wear and tear of wind turbines and to warn of necessary repair requirements. But, AI provides a more reliable way to predict equipment failures.
By providing increased understanding of deviations from normal behaviour, AI-powered predictive maintenance has detected some failures months in advance with up to 90 per cent accuracy. Likewise, Spark Cognition is using AI to detect manufacturing faults and to predict equipment failures in energy storage and turbines used for hydroelectricity generation.
Oil and gas majors including Chevron, BP, Total and Shell have similarly outsourced their AI applications to external contractors. For example, Chevron has a seven-year partnership with Microsoft to speed up its data analytics and employ the IIoT to monitor and optimise field performance and raise efficiency.
BP is working with Silicon Valley AI start-up Beyond Limits in order to gain ‘first mover’ advantage by accessing AI software used in deep-space exploration missions. BP has also invested in DeepMind to use AI to screen seismic images and geological models to increase the success of its drilling operation. Indeed, BP is banking on AI bringing in a step-change to the way it locates and develops reservoirs, produces a refined crude oil and subsequently markets its portfolio of refined products.
Likewise, Total and Tata Consultancy are jointly creating a digital innovation centre in Pune in Maharashtra state, India, to explore how AI, real-time data analytics, Internet of Things (IoT), automation and agile method can, when combined, increase the efficiency of operations.
Last, but not least, and in contrast to its peers, Shell is applying in-house expertise in AI and machine learning to its oil and gas exploration, production and refining operations. AI is advising Shell geologists to chart the course of drilling efforts and enable the Geo-steerer – the human operator of the drilling machine – to steer the drill through the oil and gas field plays thereby boosting productivity and cutting costs.
The growth in gas production in countries as far apart as North America, Russia, Australia, the Middle East and Malaysia has not only increased supply but also stimulated regional and international trade in liquid natural gas (LNG), carried in specialised LNG tankers.
Rudolf Huber, president of LNG Austria, says: “I could see whole fleets of autonomous robotic tanker barges carrying LNG. Such vessels roaming the seas would meet where AI tells them the next demand burst will be. Thus giving a new meaning to the term ‘line pack’.” Line pack is a procedure for allowing more gas to enter a pipeline than is being withdrawn, thus increasing the pressure, packing more gas into the system, and effectively creating storage.
Automation is eroding low-skilled jobs on oil and gas rigs, in power plants and mines, and will inevitably lead to future job losses. For example, a typical 1MW power plant employs around 500 members of staff, but a power plant opening in Detroit in 2022 is likely to be run by just 31 people.
The coming of widespread adoption of AI and automation throughout the energy sector will, according to a Black & Veatch study, lead to the loss of many on-site in-house operational staff as operations are supervised from remote centres.
Likewise, Deloitte’s Tech Trends 2019 attributes the coming loss of transactional and routine process tasks to the impact of AI-enabled robotics with industry-wide consequences, as Julia Harrison, lead partner in resources and industrials human capital, says: “The industry will need to redesign the world of work.”
On the plus side, the application of AI, robotics and machine learning will free employees to focus more on human aspects of job tasks, requiring greater use of emotional intelligence, on people leadership and verification of insights and solutions generated by automated equipment and sensors.
For example, decision-making and maintenance tasks will, in the future, be concentrated in regional control centres. Operators, wearing Google glasses or using tablets, can operate or maintain equipment, a plant or even a whole system.
Shell is already testing Silicon Valley-based AI predictive analytics and IIoT applications to predict when maintenance is needed on compressors and valves in Australia.
Consequently, AI robotics is changing the role of energy workers from one focused on finding problems to one of solving them; moving away from low- to high-skill work as well as a suite of new tech jobs.
As Phillips notes, “a study about to be published by Robert Gordon University and OPITO (skills body for the energy industry) suggests a massive change in skills requirements for the industry over the next 10 years – with 30 to 40 per cent of jobs in 10 years’ time simply not existing right now!”.
Because these recent tech innovations are essentially foreign to energy companies they are, on the whole, being forced to recruit a wholly different and unfamiliar set of skilled recruits and poach from Silicon Valley organisations such Apple, Google and Microsoft.
However, such skilled people are in short supply. Element AI Inc, an independent lab based in Montreal, estimates that there are currently just 10,000 qualified AI expert researchers, product managers and software developers in the world. Therefore, human resources departments of energy companies will have to change in order to acquire the expertise needed to recruit and train the workforce of the future.
For example, predictive maintenance is typically dependent on spreadsheets and manual models using brute-force analysis by people with high skill sets in mathematics and statistics. AI allows for real-time dynamic modelling, which adapts as it learns new things and requires tech skills such as algorithm design.
In sum, the energy industries of today will be transformed by AI, robots and supercomputing. From a late start and slow beginning, the energy sector is beginning to recognise the advantages that AI, machine learning and robotics can contribute to most sectors in the energy field.
Still on the horizon lie developments in quantum computing, which will enable the robots of the future to be enhanced to become autonomous decision makers, capable of ensuring the smooth running of operations and maintenance.