The oil and gas industry has a reputation of being relatively slow to adopt new technologies. Within the past few years, though, it has moved beyond reliance on paper and pencil, physical labor, and machines.

Volatility in world commodities markets has pushed companies to optimize efficiency and reduce costs. And because of the Fourth Industrial Revolution — the wave of digital innovation that has spurred the development of artificial intelligence (AI), augmented reality (AR), the internet of things (IoT), and the like — industry players are finding attractive solutions.

This article examines the ways in which oil and gas companies are looking at AI as a tool to optimize performance.

AI has attracted strong interest from major players in the industry — well-capitalized international giants such as ExxonMobil and Royal Dutch Shell and leading service providers like Halliburton. But it also creates new opportunities for operators of all types and sizes, including tech startups like Tachyus and established service companies such as National Oilwell Varco.

It also affects every sector of the industry — upstream, midstream, and downstream — and promotes the development of solutions that can integrate every link in the value chain. We’ll present examples to show how AI can improve performance at every level, from drilling sites on down to retail sales.


AI’s main advantage is its ability to process large sets of data quickly to find relevant patterns; for example, CNCs and PLCs operating on a manufacturing assembly line. The tech has obvious utility for upstream operators and their contractors, who accumulate large amounts of data before they even begin production.

During the exploration phase, they collect seismic and other geophysical data from prospective sites in order to determine where to drill. During the drilling phase, they collect logging data to assess the size and potential of their targets.

Until recently, most of these data had to be interpreted after the fact. They were rarely available in real time and could not be uploaded easily into devices with analytics functions. Now, wireline equipment can integrate IoT sensors that utilize logging data in real time, automatically sending information to AI systems that can use it for a variety of purposes.

Some systems apply these data to generate accurate models of the fields in play — and they do so far more rapidly than human experts and standard computers.

Houston-based Tachyus, for instance, develops solutions that use AI and machine learning capabilities to develop accurate predictive models of subsurface hydrocarbon reservoirs and draw up production plans in line with the operator’s business goals. Similarly, Austin-based Novi Labs designs AI-driven predictive analytics software that devises optimal drilling strategies and schedules for unconventional oil and gas fields.

Both of these companies are startups, but the majors are also keen to explore the potential of AI. Baker Hughes, one of the world’s biggest oilfield service providers, teamed up with NVIDIA in early 2018 to develop neural networks that process data from smart sensors that monitor flow rates, pump pressure, temperatures, and other operational metrics.

These networks use deep-learning and machine-learning algorithms to model reservoirs, optimize drilling and production plans, predict equipment failures, and ensure the availability of transport capacity.

ExxonMobil is taking AI in another direction. Several years ago, it began collaborating with Massachusetts Institute of Technology (MIT) to automate certain underwater exploration tasks. The partners have developed an AI system that uses machine-learning algorithms to interpret seismic data collected by unmanned remote robots from offshore hydrocarbon seeps. This solution helps ExxonMobil assess the potential of its assets more quickly and cheaply — and more safely, since it can do so without trained diving crews or submarine drivers.


Like their upstream counterparts, midstream operators generate large amounts of data. They outfit pipeline networks with sensors to monitor pressure levels, temperatures, flow rates, structural integrity, and ambient conditions. They install similar devices on rail tank cars and tanker trucks, and they require drivers and other operators to submit forms detailing their activities.

These processes have become more automated over time, and the logical next step is to integrate them into AI systems.

Oil and gas companies can use IoT devices — smart monitors built into the components and control systems of physical infrastructure such as pipelines and pumping stations as well as drones and external monitors equipped with cameras and other sensors — to collect information and upload it to neural networks. Those networks can then use these data to make decisions in real time, using machine learning and deep learning to operate controls in ways that improve efficiency, generate predictive and preventative maintenance schedules, identify safety risks, and develop real-time solutions to emergency situations.

Additionally, they integrate these data with information from other links in the value chain. They can use reports on operating conditions at fields to ensure that pipelines accommodate changes in flow rates, and they can take predictions of consumer demand for fuels into account when setting schedules for road and rail transport.

These changes are already in the pipeline, as it were. U.K.-based Modcon Systems’ ANACON-AI package uses a combination of big-data analytics, process knowledge and remote analysis technologies to monitor oil and gas transportation networks, compare current conditions to past reports, and make real-time decisions that optimize operational efficiency.

Norway’s DNV GL Software is developing solutions that generate predictive maintenance schedules on the basis of machine learning and datasets collected from IoT devices, including smart pigs that monitor internal pipeline conditions and external sensors that check ambient factors that contribute to corrosion. In 2017, Colombia’s national oil pipeline operator tested the use of a fuzzy-logic AI system to assess the risks facing different parts of its network.


Since downstream operators also produce large amounts of data, they can also benefit from new technologies. AI systems let them monitor operating conditions at refineries, storage depots, wholesale facilities, and retail outlets — and integrate the information with big data to respond to changes in demand and pricing to generate predictive and preventative maintenance schedules, and to respond to emergencies.

Again, IoT devices — smart sensors built into physical plants and vehicles and external monitors such as drones — can upload this information to networks that generate analytics and make real-time decisions in response to equipment breakdowns, changes in consumer demand, or shifts in the quality of feedstocks and refined products. These networks, in turn, can draw on big data to take account of conditions elsewhere in the sector and to compensate for disruptions such as upstream drilling delays, midstream outages, or inadequate storage capacity.

Spain’s Repsol stepped up its use of these innovations in mid-2018, when it struck a deal with Google Cloud on the use of AI and big-data services at its refinery in Tarragona. The new system built on the plant’s existing digital infrastructure and made it much more powerful, bringing the number of variables monitored and controlled automatically up from about 30 to approximately 400.

Meanwhile, Shell began using AI as a tool to improve customer satisfaction in 2015, when it introduced the Virtual Assistant service for lubricant customers and distributors. Virtual Assistant helps buyers select, find, and make proper use of Shell-branded lubricants.


None of these technologies will change the fundamental nature of the oil and gas industry. Upstream operators will continue to explore and develop hydrocarbon reservoirs, midstream operators will still move commodities from one place to another, and downstream operators will go on with the processing and distribution of fuels.

Nevertheless, AI will reshape the industry by changing operational patterns — that is, by using new tools to make the business more efficient, less costly, and quicker to react to new developments and emergencies.

Additionally, it will afford operators a more comprehensive view of conditions across sectors, allowing midstream operators, for example, to use big data, remote analysis, and other tools so that they never miss a step because of problems upstream or downstream. National Oilwell Varco has described its NOVOS process automation platform as a means of creating an “interconnected ecosystem” in which drill bits, pipes, and other equipment can work together to respond automatically to conditions within wellshafts.

As a result, it will make oil and gas companies, along with the service providers that support them, more nimble and responsive — this is a desirable outcome, given the volatility of fuel prices in recent years.