Real-World Use Cases of AI and ML in the Oil and Gas Industry

Take a look at 10 real-world use cases that demonstrate how AI and ML are already being used in the oil and gas industry, and how AI innovation can help renew the industry.

Real-World Use Cases of AI and ML in the Oil and Gas Industry
Real-World Use Cases of AI and ML in the Oil and Gas Industry

Artificial intelligence (AI) is pretty much the first in line as the key technology with all the potential to unlock a productivity revolution in the oil & gas industry. The applications of AI and machine learning (ML) in the oil & gas industry are many, and some of them can make the difference in a sector that is seeking to renew itself.

Take a look at 10 real-world use cases that demonstrate how AI and ML are already being used in the oil and gas industry, and how AI innovation can help renew the industry.

Content Summary

Predictive Assets Maintenance
Improving the Health And Efficiency of BOPs
Using Intelligent Drones for Aerial Inspections
Better Forecasts and Optimized Financial Planning
Easing the Life of Oil Tankers
Optimized Reservoir Management
Transferring Critical Human Knowledge
Digital Twins and Prediction Models
Detecting Natural Oil Seeps with SmartUnderwater Robots
Virtual Assistants and Chatbots to Improve Customer Experience
Conclusion

Despite the widespread diffusion of renewable energy technologies, oil and gas are among the most highly valued commodities in the energy sector. However, in the age of global warming, many are worried about the tremendous environmental impact of less-green energy sources. Newer technologies are highly sought by oil & gas companies, who are actively trying to enhance and optimize the consumption and production of these resources to always stay ahead of the curve. Nouman Ahmad, CEO of Validere explained, “oil and gas development increasingly resembles a manufacturing approach, where specific margin enhancement initiatives are the key differentiators between peers.”

Artificial intelligence (AI) is pretty much the first in line as the key technology with all the potential to unlock a productivity revolution in the oil & gas industry. The AI market in this industry has been predicted to grow to $2.85 billion by 2022 at a compound annual growth rate of 12.66%. North America represents the leading and most advanced market – most of the top AI players in the oil & gas industry are based in this region, with names such as Google, IBM, Microsoft, and Oracle.

The applications of AI and machine learning (ML) in the oil & gas industry are many, and some of them can make the difference in a sector that is seeking to renew itself. Just by leveraging the potential of predictive analytics, big data, and ML in upstream oil & gas activities alone, costs could be cut by $50 billion. We can only imagine what the future may hold for this sector after this fruitful marriage between titanic machines and smart computer intelligence is celebrated. Let’s have a look at some of the most interesting current use cases and applications of AI and ML in the oil & gas industry, and how they’ve already started to make a change.

Predictive Assets Maintenance

Unplanned downtime represents one of the primary causes of loss for oil & gas companies. Just 3.65 days of unplanned downtime a year can cost $5.037 million, and as the average offshore company experiences roughly 27 days of unplanned downtime a year, the losses quickly mount up to a whopping $38 million. To mitigate the risk of unexpected equipment failures, predictive maintenance solutions represent the current answer across all three segments of the oil and gas industry: upstream, midstream, and downstream.

Expected to grow into the largest segment in AI in the oil & gas market, predictive maintenance solutions are currently the principal application of this technology. They help operators improve operational safety and turn higher profit margins at the same time. AI models that can predict equipment failure are available across all the streams of the oil & gas industry as they reduce the risk of costly accidents, minimize operating expenses by reducing downtime, and improve compliance to safety standards. All the 2.5 million miles of pipelines that distribute oil and gas across the United States come packed with the industrial internet of things (IIoT) sensors that measure a broad range of parameters, such as pressure, flow, soil movement, and corrosion. All these sensors generate terabytes of streaming data every day, which can be proactively collected and analyzed by AI to generate immediate responses.

For example, FogHorn’s predictive maintenance solution is currently used to monitor the operational data gathered from the electric submersible pump, a gigantic piece of equipment used to extract oil from the bottom of the well and pump it to the surface. Advanced analytics is applied to detect potential failures and automatically stop the pump before damage occurs, but only relevant, preprocessed data is sent to the cloud to minimize cloud transferring costs. Pipeline integrity is monitored in real-time, and valves can be immediately shut down when sudden variations of pressure expose the pipeline to serious damage. Even better, the AI is so intelligent that it can predict and prevent cavitation – a phenomenon occurring in centrifugal pumps when a sudden reduction in fluid pressure results in the production of bubbles that can later collapse and cause dangerous shock waves. The system can automatically move the flow of the fluid to a different pump and prevent bursting bubbles from damaging the pipeline.

Other solutions such as the one developed by LiquidFrameworks can be used to predict the next probable failure event for fracking equipment, generating preemptive work orders when an asset approaches a high probability to fail threshold. The equipment can then be removed from service and overhauled or retired before it causes an incident. As Travis Parigi, CEO of LiquidFrameworks explained, “Predicting the next probable failure event for fracking equipment drives significant value for our customers and more importantly radically improves safety at the job site.”

Predicting the next probable failure event for fracking equipment drives significant value for our customers and more importantly radically improves safety at the job site. ―Travis Parigi, CEO of LiquidFrameworks

Improving the Health And Efficiency of BOPs

Maintaining a well’s seal on the seafloor is complicated and risky. Blowout preventers (BOPs) are a fundamental piece of equipment that is responsible for keeping the seal in place in the most difficult conditions of pressure, or against uncontrolled flow/formation kicks that may occur during drilling. A BOP’s performance must be optimal at all times to ensure its ability to seal and control wells and prevent tragic accidents. Failure of BOP systems are catastrophic, but because of their remote locations, monitoring capabilities are very limited, and troubleshooting is extremely hard when manual reporting seems to be the only viable option. In the aftermath of the Deepwater Horizon catastrophe of April 2010, the largest accidental offshore oil spill so far, it was immediately clear how scarce readily available information was about the BOP.

Digital companies such as Deepwater Subsea are currently harnessing the power of AI and ML to understand the condition of BOPs in real-time and reduce rig non-productive time (NPT). Through pattern recognition, the AI may leverage all the available rig’s data coming from faults, alarms, and subsea control systems. BOP pressure testing is the core of this technology which uses the data analytics platform TrendMiner. It is used on rigs across the Gulf for customers such as Chevron, Pacific Drilling, and TransOcean. TrendMiner collects all trends and compares them to optimal performance profiles. Whenever a deviation from this “golden fingerprint” occurs, such as when degradation of the equipment starts to happen, the maintenance team is notified in real-time. Historical trends may also be utilized to determine the ideal average test time at each pressure level, understand the factors causing sudden pressure changes, and optimize all BOP pressure tests.

Using Intelligent Drones for Aerial Inspections

Oil & gas operators must run regular inspections of their assets, such as pipelines and wells, to ensure their proper management and maintenance. Even if ML can be used to analyze all the data generated by these assets, sensors alone are not enough to extract all the necessary data. Traditional methods of inspection, however, are costly, time-intensive, or dangerous when roustabouts on trucks are sent to inspect equipment manually. Even air-based missions are a suboptimal solution: The cost to deploy a manned aircraft that can perform a full visual inspection is prohibitive, and helicopters must deal with rugged terrain, bad weather conditions, and noncompliant aircraft out in the oilfield.

AI comes to the rescue even in this instance and helps reduce all the dangers, time, and effort needed to perform these inspections through cheap, safe, and intelligent drones. This solution has been provided by PrecisionHawk, a company that combines drone technology, aerial mapping, and modeling software, along with AI and machine vision to replace dangerous manned aircraft inspections with drone deployments. AI is also used to incorporate, process, and report all data collected during the inspections into a streamlined flow of information. Hours of tedious images are examined accurately through ML, avoiding unnecessary data overload. High-resolution remote sensing is also much more efficient at detecting and pinpointing the type and extent of damage than visual inspections.

Better Forecasts and Optimized Financial Planning

A better forecast (whether it’s demand, sales, or financial planning) allows companies to better plan their production and distribution of products. Having the right product in the right place at the right time is key to increase their fulfillment levels and lower working capital at the same time. Some companies such as the Silicon Valley-based KAPUA Inc. have developed ML-based solutions to provide companies with forecasting platforms that can operate at unmatched speed, achieving unprecedented accuracy.

AI can predict demand for a specific product down to the SKU level for a region, helping companies make sure they have the right supply to serve customer orders on time, and know on which strategic regions they want to focus. Optimized forecasting is helpful at the micro-level as well, rather than just the macro one. Companies running gas stations may, for example, reap the benefits of predicting their demand for consumer goods (snacks, drinks, other products), reducing their inventory capital and safety stocks to release larger working capital.

Easing the Life of Oil Tankers

Oil tankers are a fundamental asset for oil & gas operators. The Atlantic hurricane season represents a serious threat to all operations in the Gulf of Mexico, one of the most active regions in the world for oil & gas operations. Tropical storms may cause oil tankers to sink, negatively impacting the environment as well as oil supplies around the globe. AI such as Dataminr’s, has been used to mine data coming from social media to provide vital alerts that could mitigate risks and save ships from natural disasters.

One particular instance involved an oil tanker delivering crude in Corpus Christi, TX that had to return to sea following a hurricane threat notification. A Facebook post showed a photo of the Searanger crude oil vessel after offloading at the Jean-Gaulin refinery in Quebec. Despite Tropical Storm Gordon threatening the continental U.S., as well as Hurricane Florence reaching peak intensity in the middle of the Atlantic, thanks to the real-time Dataminr alerts, this particular crude vessel was able to offload as planned and presumably make its way back to port in Corpus Christi, TX. This threat was never reported on in the news but had tremendous relevance and impact to the region and industry. The AI was able to provide this alert when it mattered, offering vital information that was not on the radar of traditional sources to make quick, reactive crisis decisions.

Other forms of AI include the so-called cognitive AI currently employed by Beyond Limits, a company whose software solutions have already been used to support NASA and space programs. Cognitive AI can track tankers from the moment they leave port to the time when they unload their cargo. Knowing all about their shipments, the amount of petroleum transported, routes, and destinations can help traders make smarter, more informed decisions that can remove friction from port scheduling operations.

Optimized Reservoir Management

Accurately identifying targets for new wells is one of the principal challenges faced by oil & gas operators. When most exploration operations take place offshore in deep water, drilling even a single exploratory well may cost up to $150 million. Near-perfect characterization of geological structures deep under the surface and fluid dynamics is necessary for efficient field development. However, even when this data has been collected preemptively over several decades, it is often siloed, unstructured, and highly disorganized. It comes as no surprise that success rates rarely get beyond a mere 20%.

AI has been deployed by companies such as Lucidworks and Beyond Limits to solve these issues. It helps spot better wells with reduced amounts of water and gas and increases oil production by making the discovery process more accurate and quick. ML can be applied to drilling to draw significant information from strata permeability, thermal gradients, pressure differentials, seismic vibrations, and much more. Once “digested” this immense amount of data can be used by geoscientists to make more informed decisions, discover new energy sources, and reduce the environmental impact of new wells. AI-powered reservoir management can leverage seismic data and historical geographic information to improve the chances of pinpoint drilling opportunities with maximum precision.

Transferring Critical Human Knowledge

The health of productive wells depends on the ability of an organization’s senior workforce to quickly detect ongoing problems, predict potential issues, and ultimately maintain them in pristine conditions. The ability to diagnose problems, make adjustments, and determine the proper operation and maintenance procedures often rely heavily on the experience of a few master experts who learned their trade over many years. When these people retire, their knowledge and knowhow are lost, and new employees must rely on procedural information that is often buried under hundreds of thousands of pages of documentation. There’s no need to explain how this situation is not scalable when 50% of oil & gas professionals are set to retire over the next five to 10 years.

IBM used its Watson AI technology to create a digital assistant that could help all less-experienced employees during their day-to-day work. The AI was fed with over 600,000 pages of documentation, reports, and even correspondence about drilling operations and is now able to provide expert-level answers to all the most common types of questions asked by technicians. After the implementation of Watson, engineers spent 75% less time on researching possible solutions or hazards, significantly improving their productivity. On top of that, the AI is still able to learn new information that can be fed in by new employees as they become more experienced, creating a solid bridge to connect past and future generations.

Digital Twins and Prediction Models

Many offshore structures have already exceeded their original design lifetime. Their reduced productivity is not the only problem since they also represent a liability in terms of human safety and environmental impact. The choice is between decommissioning them and losing the oil and gas they currently produce, or facing substantial investments to upgrade or reinforce them. Over the years, the offshore oil & gas industry has relied on digital copies of a structure – also known as digital twins – to monitor the health of their physical assets such as pipelines, rigs, valves, and other equipment. By using LiDAR to generate 3D point clouds and analytics for plant construction, experts can predict the behavior of a structure and assess its maintenance needs, thus prolonging its lifespan by a significant margin.

However, these models are somewhat “static” and do not take into account all the changes of the actual, real-life physical conditions of an asset which may influence its performance over time. New prediction models have been developed to a couple of information about the actual environmental loads coming from IIoT sensors with the virtual replica of the asset. Drawing on machine learning, data mining, and pattern recognition, new technologies such as the one developed by Ramboll identify the real conditions through structural monitoring and update the digital twin accordingly. This way, the entire structure can be tested, assessed, and optimized as needed in the virtual world, reducing the costs for full-scale testing in the field under real operational conditions.

Detecting Natural Oil Seeps with SmartUnderwater Robots

Crude oil and natural gas naturally leak out of the ocean floor through cracks and fractures known as “seeps.” Like a freshwater spring, hydrocarbons then reach the surface, forming an oil seep. With nearly 160,000 tons of petroleum entering waters in North America this way every year, natural oil seeps constitute at least 60% of the oil found underneath the earth’s surface in this region. Oil from underwater seeps can damage the environment, both because they can release methane and other light hydrocarbons in the atmosphere and because the slicks can form tarballs and mats that may eventually come ashore. Tapping these seeps, on the other hand, can both protect the ecosystem and provide oil & gas operators with a profitable source of energy.

ExxonMobil is currently working together with MIT engineers that developed the AI software used for NASA’s Mars Curiosity Rover to design AI-powered underwater robots suited for ocean exploration. These smart, self-learning machines use AI to monitor the depths of the oceans as well as a map and analyze them. Although very little about these submersible robots has been disclosed so far, it is known that they can detect natural seeps, navigate the oceanic regions, and identify potential energy sources.

Virtual Assistants and Chatbots to Improve Customer Experience

In May 2018, Royal Dutch Shell announced its launch of LubeChat, a smart AI-powered virtual assistant (VA) for B2B lubricant customers and distributors. Although it seems like old news, it must be clarified that the players of the oil & gas industry rarely (if ever) invest in this kind of customer care – customers are usually left wandering among vast databases of products to choose their ideal lubricants. The idea of using an AI that employs natural language to help clients discover products are definitely not new to other sectors, but the oil & gas field was somehow lagging.

Shell’s VA is currently available in the United States, China, and India, but the company plans to launch it in the United Kingdom as well. It provides a broad range of information about products, their availability, specific technical and safety properties, and the different commercial offers.

Conclusion

There are a lot of other interesting applications of AI in the oil & gas sector that still need to be discovered and developed. As AJ Abdallat, CEO of Beyond Limits explained, “The modernization of the oil and gas industry is a global landscape rich in opportunities for protection of the environment, more efficient discovery of energy sources, workplace safety, plus diagnostics for more informed decision-making.”

As much as it is benefiting from the implementation of AI and ML-based solutions, this industry can drive forward the advancement of this technology in this and other fields as well. Most of the leaders in the oil & gas sector have the expertise and budgets required to invest a significant amount of financial and human resources that could “fuel” the ongoing AI revolution.

The modernization of the oil and gas industry is a global landscape rich in opportunities for protection of the environment, more efficient discovery of energy sources, workplace safety, plus diagnostics for more informed decision-making.

Source: Techopedia

Thomas Apel Published by Thomas Apel

, a dynamic and self-motivated information technology architect, with a thorough knowledge of all facets pertaining to system and network infrastructure design, implementation and administration. I enjoy the technical writing process and answering readers' comments included.