Securing the Energy Ecosystem through AI-driven Applications

Presented by:

Editor's note: This article appears in the E&P newsletter. Subscribe here.

Recent cybersecurity breaches have unmasked a vulnerability in oil and gas infrastructure that has operators looking to sophisticated artificial intelligence (AI) programs to stay secure and operating. Through predictive analytics, operators are given greater visibility into their operations to develop actionable risk mitigation strategies that prevent unexpected downtime and the associated expense, according to SparkCognition Founder and CEO Amir Husain.

“There is no shortage of ways in which the energy industry can leverage their data and AI technology to improve operations,” Husain said in a recent interview with Hart Energy. “By employing a predictive solution on well data, customers can foresee events with enough time to take preventative action, saving millions of dollars in drilling costs and downtime, and alleviating, if not eliminating, overall nonproductive time.”

Founded in 2013, the Texas-based startup has raised more than $163 million in private funding to build AI technology solutions for industrial applications. 

Targeting the infrastructure space, Husain founded SparkCognition when he saw a burning need for a company that would wrap physical infrastructure with intelligent, AI-powered software fabric “to increase efficiency, protect OT [operational technology] investments and make infrastructure smarter.”

“Infrastructure is a 100T+ problem with four thematic challenges: environmental, increasingly sophisticated cyber threats, widening skills gaps in the workforce and the need to deliver constant growth to satisfy financial markets,” he added.

The company developed SparkPredict to help revitalize global infrastructure. This turnkey solution analyzes sensor data and uses machine learning to return actionable insights, flagging suboptimal operations and identifying impending failures before they occur.

Onshore case study

A major utility company turned to the tool to protect a $2 billion turbine investment from critical unknown defects. SparkCognition analyzed the turbine’s streaming data and was able to identify a unique failure event with one-month advance notice, avoiding roughly $500,000 in repairs, according to the case study.

Within four months of deployment, SparkPredict identified a persistent anomalous behavior in the area of the initial stages of the compressor end of the turbine. Further investigation revealed several factors that pointed to a compressor blade rubbing against the inlet guide vane wheel.

“It’s important to stress that this failure had never been seen before and was not detected by existing monitoring systems,” the report stated. “Because SparkPredict found an unknown relationship from the utility’s own data through unsupervised learning and could clearly indicate the problem area, the utility’s operations teams were able to see exactly where the problem was.”

Offshore case study

In the offshore space, SparkPredict helped an oil major resolve recurring failures in production caused by bottlenecks in its fluid separation process. The failures resulted in more than 10% in downtime and millions of dollars of lost production, according to the case study.

The operator shared two years of blind sensor data from their gas system with SparkCognition to build the initial model. Using a machine learning model, SparkPredict identified 75% of production-impacting events on an average of eight days in advance.

“This initial success was verified on additional assets, and since then, deployment has been occurring across their fleet to predict impending failures and optimize maintenance activities,” Husain said. “Ultimately, by using the SparkPredict product, the customer has improved production by 1% to 4%, or up to $30 million annually, per platform.”

Prior to adopting the software, the study found that the operator would observe five to 10 unique failure events, which resulted in an estimated 10% to 15% downtime and up to $8 million in lost production per event.

According to the study, 80% of these failure events are attributed to three subcomponents—one glycol system and two export compressors—each of which is instrumented differently and operating at different stages in their life cycle.

Cybersecurity

Additionally, intelligent software is the base of its AI-built cybersecurity tool, DeepArmor. Leveraging AI techniques, the application provides pre-execution attack prevention against zero day threats and conducts cybersecurity threat analysis and detection. 

DeepArmor exclusively uses AI to prevent file-­based, file­less and in­-memory attacks.

“DeepArmor pioneered the use of AI models in this space, having been first to use AI to detect phishing attacks using office documents, weaponized PDFs and direct­ to memory powershell attacks,” according to the company’s website.

In June 2020, the company launched an updated version of the application for OT called DeepArmor Industrial. It leverages cognitive models, developed by SparkCognition, to analyze file types, asses the threat level and ultimately take action.

“This product is specifically designed to protect the energy industry’s OT assets, which are notoriously difficult to secure due to legacy operating systems or hardware, often operating in limited connectivity environments,” Husain said. “With the rapid evolution of the cybersecurity landscape, true protection often requires updates and patches to stay ahead of the most advanced threats. But in limited or no connectivity environments, this presents a massive challenge for OT security.”

Predictive models

Using machine learning to generate evolutionary algorithms, known as neuroevolution, SparkCognition’s Darwin platform allows non-technical users to build predictive models. It automates the process of building, testing, deploying and maintaining models for a dataset, providing an intuitive environment that takes users quickly from data to meaningful results.

“The Darwin product is a development platform that accelerates the process of building highly accurate, generalized models,” Husain said. “It allows engineers, analysts and data scientists to rapidly iterate through thousands of different models and architectures and generate optimal solutions that scale across large operations. These models are essential for obtaining the actionable insights demanded by the oil and gas industry, such as predicting workover and production in wells.”

Utilizing AI for ESG goals

As ESG gains popularity, Husain believes AI systems will be a useful tool to achieve those low-carbon goals.

“AI brings risk mitigation and optimality to operations,” he said. “It lowers cost, reduces risk and increases output all at the same time and has the potential to improve the economics of ESG projects.”

Husain noted, however, that the level of success will depend on the type of AI systems that are implemented by operators, how effectively they are integrated with the infrastructure and how well AI software companies support ESG-specific goals and strategies.

“Virtually all companies want to become more energy efficient in their operations, and they can certainly utilize AI to achieve that goal like leveraging predictive models that suggest operating parameters ideal for low-power draws and predictive maintenance to identify and mitigate efficiency shortfalls,” he added.

Husain said new AI systems can expand on that idea enormously, “both in their own right and by integrating with any currently deployed AI where that makes good sense.”

“For instance, one possibility would be to leverage AI-generated knowledge graphs that reveal subtle relationships at different operational levels—from individual assets, all the way up to the entire business model," he said. "On that basis, you could then implement ESG strategies in a larger and more comprehensive way than ever before, with full explainability."

Moving forward, the company intends to continue to investing in green energy and adding capabilities to become more sustainable. 

“We know that the future of the energy industry lies in renewable power generation,” Husain concluded. “We’re only scratching the surface of the profound possibilities that will emerge as AI becomes more and more tightly interwoven into the fabric of business.”

RELATED CONTENT:

July 13, 2021: Exploration: Leveraging Digitalization to Maximize Results in the Field

July 9, 2021: Opinion: The Future is Exponential, The Future is Digital

July 6, 2021: VIDEO: The Cyberphysical Realm: What’s Next for the Upstream Sector?

June 14, 2021: Artificial Intelligence Platform Accelerates Reservoir Opportunity Identification

March 1, 2021: C3 AI, Baker Hughes to Collaborate with Petronas on Digital Transformation Program

Sept. 2, 2020: Sixgill, CleanConnect.ai Partnership to Unlock Oil Industry’s AI-powered Future

July 24, 2020: E&P Solutions: Reduce Costly Regrets through AI-powered Digital Twins