Analyze Big Data with Prescriptive Maintenance
Grant Gerke | February 9, 2017
By Grant Gerke, Contributing Editor
As manufacturers modernize plants and retrofit equipment with additional sensors, reliability and maintenance managers are working to develop Industrial Internet of Things (IIoT) strategies that effectively manage new data streams. In the past, manufacturers would employ operational consultants or specialists to analyze the mountains of data these sensors generate. Today, limited resources are driving reliability professionals to explore prescriptive maintenance.
Prescriptive maintenance is a component of the IIoT. This discipline uses machine learning and automated data review to prevent equipment or device failure. Some industry experts call it preventive maintenance with built-in intelligence.
It’s the next bridge for reliability teams to cross as referenced in the January 2017 edition of Maintenance Technology’s “On the Floor.” The focus was on regrets and hopes. One industry consultant stated, that among his clients, “the biggest regret seems to be PM/PdM compliance and not doing what they planned to do to prevent breakdowns.”
The consultant added, “that one client increased training and invested in maintenance employees but still hasn’t realized the returns on that investment.”
One reason for the lack of follow-through could be the ability to promptly act on plant-floor data, also known as perishable data in the field or factory floor.
In a recent article on 2017 IIoT trends, Abedayo Onigbanjo, director of marketing at Zebra Technologies, Lincolnshire, IL (zebra.com), stated “businesses must make sense of data before it expires. Enterprises are losing valuable insights with many disjointed sources generating and collecting data on their own, contributing to only bits and pieces of the big picture, instead of rendering a broad view.”
Legacy cultures and platforms are the main culprits. Take the process industries, for example. Many operations, including large chemical plants and oil fields, are relying on 4- to 20-mA fieldbus networking solutions. Identifying device defects is difficult in these facilities.
Procentec, Wateringen, The Netherlands (procentec.com), provides asset-management solutions that accelerate the identification of malfunctioning devices. In 2015, the company updated firmware for its Foundation Fieldbus Diagnostic module. The benefit was a “live list” of all operating devices in one overview, and device-type viewing in the oscilloscope images.
Documenting downtime is sometimes painful, but essential. “For a steel producer in Europe, the total estimated revenue for downtime was in the neighborhood of [$1,600] 1,500€ a minute,” stated Matthew Dulcey, global sales manager for Procentec at a 2016 PROFI networking conference.
During the presentation, Dulcey also provided examples of how 1% of downtime for a plant running 24 hours/day equals about 78 hours of unscheduled downtime. According to Dulcey, a steel-plant maintenance team demonstrated that one preventive downtime event could pay for new diagnostic tools.
Other non-networking solutions for manufacturers include Panoramic Power’s (New York City, panpwr.com) Device Analyzer and its machine learning platform. This platform—PowerRadar version 2.0—learns usage patterns for devices in production lines and allows users to view an operational device state in real-time. The system collects device-level energy data, automatically learns device patterns and, after training, automatically identifies the different operational states of each instrument.
In the big picture for manufacturers, this is just the beginning for big data and IIoT. The road seems to lead to easy-to-use analytical tools and operational automation. MT
Grant Gerke is a business writer and content marketer in the manufacturing, power, and renewable-energy space. He has 15 years of experience covering industrial and field-automation areas.