The South African energy and chemicals company has implemented a data analytics infrastructure that will help it win its digital transformation ‘micro battles’. Jessica Twentyman reports.
When it comes to digital transformation, executives at South African integrated energy and chemicals company Sasol pick their battles wisely. In fact, they like to talk in terms of “micro battles” – strategic technology projects, designed to tackle specific business challenges, undertaken as part of a wider campaign of digitisation.
Today, Sasol has several such micro battles underway – but more of those later. First, the company needed to put in place a data analytics infrastructure capable of bringing together a wide range of information, from sensor data collected from plant floor machinery to business information held on enterprise resource planning (ERP) systems, for example.
That in itself was an enormous task, which Internet of Business discussed this week with Ana Moreira, senior manager for data services management at Sasol. Moreira was speaking at data warehousing company Teradata’s annual European conference in London.
A data lake and warehouse, located in the cloud
As Moreira explained, at the heart of Sasol’s new, integrated analytics system is a data lake, based on Cloudera’s distribution of Hadoop, and a data warehouse from Teradata. Together, they handle data from over 150 critical systems. Fourteen of them are individual ERP instances from SAP, but the vast majority are plant systems: data historians that collect sensor data from operational technology; planning and scheduling systems; maintenance applications, and so on.
“We had to evolve an analytical ecosystem to cope with all this structured, semi-structured, and unstructured data, so that we could be ready for the digital onslaught that’s coming our way,” says Moreira.
The technology that the company had prior to this implementation was never going to cope in a new age of industrial IoT (IIoT), as it was predominantly based on spreadsheets and many, disparate SQL Server databases.
“Sasol needed an architecture that would sustain us going forward,” Moreira explains – plus it needed to be highly scaleable, in order to accommodate new needs as they emerge. For that reason, the team decided to implement both Teradata and Cloudera on the Microsoft Azure cloud. The decision also gave some valuable ‘wriggle room’ if these technologies proved unable to cope with Sasol’s demands.
Says Moreira: “We wanted the infrastructure to be elastic, but we also wanted to be able to shut it down if it didn’t work, without having made a mess of our capital investment.”
Let battle commence
The analytics environment has now been up and running for about a year, enabling Sasol to forge ahead with those micro battles. The first focuses on asset health, with a view to moving steadily in the direction of predictive maintenance.
Right now, Sasol has to shut its plants down annually, typically for the month of September, in order to check all equipment and replace older machines, or fix those with issues. That’s a costly decision to take. “So being able to predict what plant machinery really does need replacing is massive for the company,” says Moreira. “We’re seeing other oil and gas companies doing this, and understand that this is the way that the industry is moving.”
The second micro battle focuses on the use of drones to inspect plant-based boilers, to check for corrosion. A camera on each drone captures images, and then image recognition and machine learning is deployed to identify potential problems. Soon, this may be extended to include gas pipeline inspections, with drones flying the vast distances involved, carrying sensors that detect gas leaks.
These battles would be hard – if not impossible – to win without a rock-solid analytics platform in place, according to Moreira. Information management is, she says, “an important pillar” of digitisation, adding, “Our business, oil and gas, is being disrupted and we had to make a step-change in our approach to analytics that would enable us to stay ahead of the game.”
Internet of Business says
For industrial and energy companies in particular, predictive maintenance via the IoT promises a more efficient, sustainable, and profitable future. The smart mix of enterprise asset management (EAM) systems, the IoT, sensors, data analytics, and – increasingly – associated technologies, such as digital twins and even virtual/augmented reality headsets and smart glasses, can give insights into how industrial assets are performing, and are likely to perform in future.
For example, if every asset is logged in EAM and exists as a digital twin, then each IoT-linked machine can tell operators how it is performing, which component needs fixing, where it is located, and how it can be removed or replaced. For complex machinery, engineers can see schematics overlaid on the real world via smart glasses, and be trained how to fix devices or locate components.
Over time, more and more gathered data will mean that predictive maintenance can kick in, which not only helps the plant or installation perform better, but also helps to optimise the supply chain that keeps it running. In this way, machines can tell engineers when they are going to fail, and which part needs to be replaced. Via smart supply chains, those components can be ready for installation before the machine is forced offline by a critical failure.
Perhaps the world’s leading example of this type of system can be found at CERN in Geneva, where every component in the world’s biggest machine, the Large Hadron Collider, is logged in an EAM system as a digital twin, as is everything else on the campus. CERN’s analytics and smart supply chain means that specialist parts can be ordered and manufactured and be ready for installation before the LHC fails.
Big data meets big science – and keeps it running.
For more on cybersecurity in manufacturing, and much more besides, attend our Internet of Manufacturing event in London on 5-6 June.