Big data has become the oil of the new digital economy
So it’s not surprising that energy companies, which rely on complex equipment for drilling and oil well maintenance, are in pole position to benefit from automation fuelled by advances in big data and machine learning.
According to a report by Credence Research, the Global Big Data Services Market for the Oil and Gas Industry is expected to reach US$33.5bn by 2023. A typical modern offshore oilfield has more than 10,000 sensors pumping petabytes of data, forming deeply intertwined cyber-physical systems that rely increasingly on algorithms and robotics. As sensor technology becomes cheaper and wireless, there is a premium on the ability to extract and manage large volumes of data in real time. Enter artificial intelligence.
In 2003, Donald Paul, then Chevron’s chief technology officer and C.L. Max Nikias, then USC Viterbi dean, envisioned a paradigm in the world of engineering – CiSOFT – the Center for Interactive Smart Oilfield Technologies, a joint venture between USC and Chevron. Now headed in its 15th year, the partnership redefined the relationship between academic and industrial research.
“As we were expanding sensor technology at Chevron, the Internet of Things (IoT) model was a trend that wasn’t really accounted for,” Paul said. “We needed another structure and another model. One that could produce hybrid engineers who live at the intersection of petroleum engineering and information technology.”
CiSOFT hosts large-scale collaborations between electrical engineering and computer science to tackle complex issues like production efficiency, safety, environmental impact, data integration and automation.Viktor Prasanna, director of Center of Energy Informatics (CEI) and professor of electrical engineering and computer science, is developing a fundamental way of representing networks of very large cyber-physical systems like oil fields and power grids.
“The key driving factor of this research is data science,” said Prasanna. “You can gather these large amounts of data, but can you understand that data?”
Prasanna and his CiSOFT collaborators worked to develop a networked system that went through data pulled from the oil rigs, coupled with weather, historical archives, market conditions, and even tweets.
“If you’re going to institute things like machine learning you need history,” said Paul who explained that the petroleum industry often collects data that can never be collected again. “You drill a well and that location, those measurements that are cut through virgin rock can’t be duplicated.”
Few modern industries can boast historical data that goes back to the turn of the 20th century. Using machine learning techniques, historical and real-time data integration and automation, Prasanna’s team has given petroleum engineers and the machines they operate the ability to respond immediately to problems as they arise. Solutions like this become imperative when minute adjustments can mean preventing field accidents, especially in highly sensitive operations such as underwater or coastal environments.
In the future, Prasanna wants machines to not only collect and process data but also to start making some of the decisions.
“It has a way to adapt automatically and it scales,” Paul said. “It was developed for oil and gas but has rolled over successfully to power grids.”
Smart oilfield research at CiSOFT will spill into the IoT age, when electronic devices will communicate with each other across an internal network for a variety of industries. The need for the integrated data solutions proposed by Prasanna may become invaluable for a variety of applications where real time data processing is needed, from smart cars to open heart surgery.
This is an abridged version of an article from USC Viterbi School of Engineering. To read the full article, log on to www.viterbischool.usc.edu
