New IBM Software Aims to Predict Mechanical Failures
April 1, 2013 Alex Woodie
Companies in the manufacturing, distribution, energy and utility, and oil and gas industries may be interested in a new analytics package recently introduced by IBM. The software and services offering, called Predictive Asset Optimization, aims to forecast when assets in the field–such as trucks, water pipes, or robotic machines–are likely to fail, so that preventive maintenance or preemptive engineering changes can be made, saving the customer millions of dollars.
Well-run companies have always paid attention to maintenance. But the fact is, in this physical world, stuff happens. Cogs in the wheels of industry get gummed up. Engines will get old and break down. Batches of product will occasionally be unmarketable due to glitches in production. Companies with millions of dollars invested in physical machinery realize that repairs, replacements, and rejects are simply the costs of doing business.
For more than 100 years, IBM has sold tabulating machines that help companies keep track of their maintenance schedules. The programs weren’t any smarter than the technicians, who had their fingers on the pulse of their equipment, and who knew that his big rigs could go so many miles between oil changes or that the bearings on a printing press had to be repacked every so often.
Now, IBM is saying that it has developed software that may just be smarter than those wily old oil- and ink-stained mechanics.
Predictive Asset Optimization (PAO) combines IBM technical services and software, including its Hadoop-based IBM Big Data platform, as well as components cherry picked from its DB2, Cognos, SPSS, InfoSphere, and Tivoli product families. This data warehouse is fed with a variety of static and streaming information, including maintenance logs, asset usage data, data from equipment sensors, telematics, and environmental and facilities monitoring systems.
IBM’s Hadoop warehouse (based on its BigInsights distribution) takes all this data in, integrates it, runs it through predictive statistical models, and outputs its predictions about likely mechanical failures, so that preventive maintenance can be undertaken to avoid downtime and the need for emergency maintenance. According to IBM, emergency maintenance costs three to ten times as much as preventive maintenance.
The software can also be used to improve manufacturing efficiencies, and provide early detection of product defects that increase manufacturers’ warranty costs. And once customers have a predictive data warehouse in place, they can begin getting answers to all sorts of questions they might have, such as “How can I detect warranty issues sooner?” and “How can I optimize my maintenance plan?”
“The world is entering a new era of smart, where decisions will be based on facts, data, and increasingly on the ability to apply analytics to massive data sets and extract very precise business insights,” Fred Balboni, senior partner in IBM Global Business Services’ big data analytics practice, said in a press release.
Predictive Asset Optimization is one of the solutions IBM is offering through the Advanced Analytics Center that it formed in November in Columbus, Ohio. It’s also one of a new breed of “Signature Solutions” from IBM. Other Signature Solutions include Next Best Action, Performance Insight Solution, and Anti-Fraud, Waste, and Abuse Solution.
BMW is an early beneficiary of PAO. By using PAO to analyze historical repair data and data collected from individual vehicles, BMW was able to reduce warranty cases from 1.1 per vehicle to 0.85, which corresponds with €30 million in annual savings. BMW also used SPSS in its light-alloy foundry to reduce the scrap rate by 80 percent, IBM says.
ConocoPhillips also uses IBM PAO to improve the efficiency of its oil and gas drilling operations in the Arctic Ocean. The company feeds PAO with data about the location of icebergs and prevailing weather patterns. PAO crunches the data and gives ConocoPhillips accurate predictions on iceberg locations, thereby saving the company on costs associated with moving drilling rigs.
The ability to take advantage of analytics is a strong predictor of business success. According to an IBM presentation, an organization that uses analytics is 3.6 times more likely to outperform its competitors than if they didn’t use analytics.
This new offering is aimed at customers in the automotive, electronics, aerospace, defense, manufacturing, mining, transportation, telecommunications, and energy and utilities industries. There isn’t an IBM i component to the solution, except for the fact that many IBM i customers work in real-world industries mentioned above.