Predictive analytics used to cut costs in oil sector

In January, UK-based market research firm GlobalData estimated that capital expenditure in the oil and gas industry would increase by 13% during 2012 to reach $1.03 trillion globally (£650 billion at current exchange rates).

As well as being a sign of the growing confidence about business conditions within the oil and gas sector, that level of spending demonstrates the scale of opportunity for using predictive analytics software to help extract maximum value from the high capital costs that mark out the industry.

Warren Wilson, the US-based leader of analyst firm Ovum’s energy and sustainability technology team, said avoiding unplanned downtime is the main driver for the adoption of predictive analytics by oil and gas companies for asset maintenance uses. “It’s an extremely capital-intensive industry,” Wilson said. “The kinds of problems that we’re talking about have huge financial effects.”

Many oil and gas equipment manufacturers and specialist services firms, including suppliers such as SKF, Rovsing Dynamics and General Electric, bundle analytics software as part of their product offerings. From their own laboratory data, the suppliers can build models to help predict component failure, explained Paul Wheelhouse, a lecturer at Manchester University in England and director of Red Wheel Solutions, an asset management and maintenance consultancy. Input data from sensors measuring physical characteristics, such as vibration, temperature or the condition of lubricants, is compared with the model to assess the likelihood of failure of the equipment being monitored.

“You need to look at the failure modes that you want to cover [and] put in place the parameters to see if that failure mode is taking place,” Wheelhouse said.

With the help of the predictive analytics tools, equipment can be replaced based on its actual condition and not according to a timetable set out by the manufacturer, potentially saving on replacement costs and eliminating unplanned downtime or catastrophic failure -- for example, when the software flags components that need replacing ahead of schedule.

Predictive analytics software takes Nexen beyond time

Canada-based oil and gas company Nexen develops energy resources in offshore oil fields in the UK North Sea, the Atlantic Ocean off the coast of western Africa and the Gulf of Mexico.  It also drills for shale gas and processes “oil sands” in western Canada.

From a base in Aberdeen, Scotland, Marjorie Chamberlain, a reliability specialist at Nexen, uses analytics software to help predict when equipment needs replacing or cleaning, as an alternative to relying on a time-based maintenance schedule.

“With time-based maintenance, you might change a seal on a pump every two or three years so that it would not fail,” Chamberlain said. “Instead, you can employ predictive techniques [for] condition monitoring to see when that starts to degrade, and then you can change it out so that you’re not just pulling it apart for the sake of it.”

Unnecessary maintenance not only adds extra cost to the business, it also can lead to equipment problems from introducing contaminants to the system or from parts being re-assembled incorrectly, she said. 

The predictive analytics software that Nexen uses was developed by SKF, a provider of bearings and seals to OEMs that supply the oil and gas industry. SKF offers the @ptitude product as part of a service package and will either host the software and do analysis work for customers or license it for on-premises installation.

While such analytics tools can help improve the productivity of individual assets, such as an oil rig or refinery, users implementing the technology often fail to take a companywide view, said Duncan Slater, a manager at Accenture’s information management services unit in London.

“However, what [Accenture is] starting to see is more of an enterprise way of thinking about this, an intention to exploit the data which large organisations have at their disposal,” Slater said. “There is an intention to make this more systematic, to start to deploy analytics in a more coherent, cohesive way across all of the assets.”

When it’s done properly, he added, “predictive asset maintenance is a shining example of how looking more closely at the data you have got at your disposal can bring significant value at an enterprise level.”

Case study: Predictive analytics helps to minimize environmental damage

Statoil, a Norway-based oil and gas company, is also moving to deploy streaming and predictive analytics software. But it has a more environmentally minded application in the works for the technology.

Statoil has begun a three-year project with IBM, oil and gas engineering services provider Kongsberg Group and risk management company Det Norske Veritas to use analytics tools to minimise the environmental impact of drilling and exploration.

The project will result in a real-time monitoring system designed to capture, process and analyse large amounts of physical, biological and chemical data generated by sensors and cameras installed around one of Statoil’s offshore drilling facilities, with a goal of making it easier for the company to predict, detect and respond to operational issues that could cause environmental problems.

Vidar Hepsø, principal researcher and project manager for environmental monitoring at Statoil, said officials think the monitoring and analytics programme could contribute to the company’s winning consent from regulatory authorities for proposed new drilling operations. “We firmly believe that this will increasingly become a way of doing business in the oil and gas industry,” he said. “Those oil companies that are the first to take these kinds of measure into operations will be those that get the best acreage, for instance, in the Arctic waters.”