Follow Datanami:
July 26, 2017

Before the Next Disaster Strikes, Get Better at Data Science

Eddie Amos

(James Jones Jr./Shutterstock)

One of the most disruptive events for any industrial organization is asset failure. From halted production and costly physical damage to safety concerns for workers and the environment, the impacts of asset failure remain one of the top risks for industrial companies.

Each year offshore oil and gas operators experience an average of $49 million in financial impact due to unplanned downtime, and this figure can reach upwards of $88 million for the worst asset failures.

Whether from product or equipment loss, cancelled services or a recall, the direct cost of downtime is a primary consideration for industrial companies when thinking about big data strategy. In many cases the indirect costs of damaged reputation and loss of market share are equally, if not more disruptive.

For these increasingly digital-driven organizations, data is one tool that can mitigate the risk of asset failure. With applied data science and advanced software, data can provide critical insight into machine health and help predict mechanical failures before they happen. More than ever, progressive data science is helping industrial operating facilities transition from reactive to proactive maintenance techniques that ensure compliance, improve safety and ultimately save big dollars long term.

The food and beverage industry is one market where data is becoming a critical player. With strict food safety standards in place around the world, data analytics systems such as asset performance management (APM) can help operators remain compliant. Companies are able to assess the risks of various facility and process changes by modeling the data from assets for different scenarios. This is particularly important for temperatures.

In 2014, damaged equipment at a frozen food plant was attributed to a listeria outbreak. The company recalled more than 400 products sold at major retailers and laid off 300 employees as a result.

Studies show that having a data-driven approach to maintenance results in a 36% less unplanned downtime (Azizi Embong/Shutterstock)

If food packaging, storage and transportation assets aren’t operating under precise conditions at the right temperatures, bacteriological contamination will harm customers and the brand. With APM systems in place, asset performance can be managed to reduce contamination risks while maintaining equipment availability.

The benefits of a predictive data analytics strategy in any industry seem obvious, but the journey often seems difficult for many. Fewer than 24 percent of oil and gas operators describe their maintenance approach as a predictive one based on data and analytics. Instead, more than 75 percent take a reactive or time-based approach understanding the high-stakes of failure. Data analytics is daunting for industrial businesses, especially considering that a single blade in a gas turbine can generate 500 gigabytes of data each day and as much content as the print collection of the Library of Congress over the span of 30 days. And that’s just one blade.

A combination of deep domain expertise and technology helps drive more meaningful outcomes. Data scientists are successful in driving meaningful insights when they combine historical information from facility assets with the real-time data produced from the Industrial Internet. To initially overcome data anxiety, many data teams now rely on machine learning in APM technology to code historical asset data and properly categorize manual and automated system inputs for an accurate view of asset performance and health over time. Historical data helps benchmark assets and put real-time present data into context. Without it, operators aren’t able to draw predictions based on performance trends.

APM relies on software platforms, analytics and key insights to turn data into a tool of operational excellence and continued success. Establishing a baseline and continuing to benchmark assets across the organization and industry helps operators identify assets with chronic failures and the causes of failure to drive failure elimination efforts that transition the reactive maintenance approach to a proactive one. By spotting failure trends and characteristics earlier through data, industrial organizations improve overall asset reliability and cut costs.

When industrial operators use a predictive, data-driven approach to maintenance, they experience 36 percent less unplanned downtime than those with a reactive approach. This results in, on average, $17 million dropping to the bottom line annually. Smart sensors and automation technologies promise to make the production processes more efficient and safer than ever before, but only when that data is put to work. Without the proper data science behind a company’s maintenance strategy, failures will always be on the horizon – and operators won’t see them coming.

About the author: Eddie Amos is GM/VP APM Software at GE Digital. He is responsible for GE Digital’s APM software strategy and was previously Vice President at Juniper Networks and a Partner and General Manager at Microsoft. He works with a wide range of oil and gas companies to help them better manage and utilize their big data from equipment sensors to make more strategic decisions and stay ahead of market changes.


Related Items:

Data Opportunities Slipping Through the Cracks

The Data of Things: How Edge Analytics and IoT Go Hand In Hand

IoT Seen Driving AI Adoption