Refining oil, in the cloud

From robots to cobots, Artificial Intelligence to Augmented Reality, analytics to machine learning – the fourth industrial revolution is upon us.

Industry 4.0 is already creating incredible opportunities for manufacturers, employees and their customers. New technologies are enabling industry leaders to maximise resources, empower employees and even predict breakdowns before they happen.

But the true potential of Industry 4.0 will only be realised through smart manipulation of data. Manufacturers have recognised the potential of sensors and connected devices for some time, and as a result, they’re enjoying access to more relevant, real-time information than ever before. But such vast quantities of data can be tough to store, visualise and understand, let alone guide business decisions. Luckily, technologies like machine learning make it possible to turn this raw data into actionable insights.

At MOL’s Danube Refinery in Hungary, Microsoft technology has transformed the way data is being collected and analysed by one of the largest petrochemical refineries in Eastern Europe. The Danube Refinery has 54 major processing plants, which use complex processes to turn raw materials into chemicals such as ethylene, propylene and benzene that are in turn used to make resins, fibres and plastics.

One such processing plant is called a delayed coker. This is where oil is heated to a high temperature to produce gas oil and petroleum coke – a by-product of the oil refining process and a valuable carbon material that might otherwise go to waste. However, the process also creates a risk of steam eruptions. When an eruption occurs, less valuable petroleum coke is carried further into the refining process and the whole system must be shut down until the equipment can be cleaned. Steam eruptions are bad news for people running the processing plant.

“We wanted to extract the most value from the materials produced by the delayed coker,” says Tibor Komróczki, head of the process information and automation team at MOL, “while also minimizing the occurrence and impact of steam eruptions.”

The answer was hidden in data. Each plant at the Danube Refinery is equipped with thousands of sensors for monitoring the performance of the equipment and quality of the product produced. The refinery records nearly 100,000 data points every single minute – not bad for a refinery that first started production over fifty years ago. But the MOL team needed a way to turn that sensor data into useful, actionable insights.

Working together with Microsoft Hungary, MOL put several years’ worth of historical plant data through the Azure Machine Learning algorithms, part of Microsoft’s cloud services. Although MOL’s team of experts includes electrical and chemical engineers, as well as oil industry professionals, this level of data analysis was something MOL simply wasn’t able to do alone.

“The main advantage of the Microsoft Cloud is the user-friendly interface that makes Machine Learning widely accessible,” says Dániel Percze, head of BI and Big Data at the IT department of MOL Group. “Using Azure Machine Learning, we can leverage advanced analytics with relatively low risk and investment. Business users expect clear and easy-to-use tools, and this is exactly what Azure Machine Learning offers.”

Microsoft Azure Machine Learning has already confirmed several hypotheses on how to improve the coking process and reduce the risk of steam eruptions. In fact, the initial analysis was so useful that MOL plans to build a more comprehensive data analysis platform that can be used by even more employees. Microsoft Azure Machine Learning can be easily integrated with the refinery’s existing operational intelligence system, OSIsoft’s PI System, meaning more operational data can be analysed quickly without needing to invest in a whole new system.

“Our plans for the next phase include training more people on Azure Machine Learning, so that they can gain experience with model design and data analysis,” Percze explains. “We would like to provide business users with advanced analytics tools that can help them drill down into refinery data, evaluate the results, and use them to optimize more production processes.”

MOL’s Danube Refinery is just one example of an industrial business partnering with Microsoft to embrace Industry 4.0. You can discover more stories from European partners here.

Tags: , ,

Related Posts