How Analytics Drives Prescriptive Maintenance

Nov. 11, 2019
Drawing on technology from its recent acquisition of MaxGrip, AVEVA is delivering its prescriptive maintenance capabilities to help customers save major costs through unplanned downtime avoidance.

At the AVEVA World Conference in Orlando this week, AVEVA has booths devoted to predictive and prescriptive maintenance capabilities, artificial intelligence (AI), and the application of asset performance management (APM). While much can be said about each of these technologies on their own, it’s important to understand that they’re also inextricably connected—as manufacturers increasingly use AI-driven predictive and prescriptive analytics to better understand and improve the performance of their assets.

AVEVA’s predictive maintenance is a mature technology, having been in use now for about 14 years, says Jim Chappell, vice president of information solutions for AVEVA. The prescriptive analytics technologies AVEVA is showcasing at the event this year, however, are relatively new additions, coming largely from the company’s acquisition of MaxGrip last spring. Although AVEVA had some prescriptive-related offerings, MaxGrip “bulleted us forward,” Chappell says, with its 20 or so years of experience in fault diagnostics. “We’re merging all that together with our predictive analytics portfolio.”

The bulk of AVEVA’s work with predictive maintenance began with customers in the power industry, who were well positioned with their historian data to take advantage of analytics, Chappell says. This type of use has spread in the past five or six years to the oil and gas industry, and more recently to food and beverage, water/wastewater, mining, and others.

“We’re in all these industries with predictive analytics now,” Chappell points out. “We’re infusing AI across all of our business units, from both SCADA and MES perspectives.”

To highlight how these technologies are being applied, Chappell shares an example of a customer’s steam turbine that was showing unusual vibration on one of its blades. Using AVEVA analytics to drill down into the turbine’s data, the customer discovered that the major contributor to this anomaly was a hairline crack in a blade that was almost undetectable at first glance with the human eye. “It could’ve caused major damage,” Chappell says. “They calculated the avoidance of potential damage—had the cracked blade gone unnoticed—to more than $34 million.”

For the food and beverage industry, which is relatively new to using analytics, manufacturers have a general desire to improve their maintenance and better react to operations, Chappell says, but they don’t necessarily know what problems they need to be looking for.

As an example of how the food and beverage industry is beginning to use analytics, Chappell shares the story of an AVEVA customer whose oven oxidizer was experiencing larger-than-expected pressure differentials. Leveraging insights provided by AVEVA analytics, the manufacturer discovered that the moisture bleed tubes were freezing when the outside temperature dropped below zero. This caused the lines to back up, resulting in a pressure delta across the catalyst. 

“This was not something they were looking for,” Chappell comments; but once they found it, they were able to go back and fix the issue at other plants too.

Explaining the key distinctions between predictive and prescriptive maintenance, Chappell says predictive provides early warning through anomaly detection, using machine learning to create a digital twin of the asset or process to look for anomalies. It works hand-in-hand with performance analytics, which leverages first principles to simulate the behavior of assets or processes, first by autotuning the simulation and then by using the simulation to look for deviations from the baseline.

Prescriptive maintenance, Chappell says, goes a step beyond anomaly detection to look at which specific tags are contributors to the detected anomalies. With prescriptive, you not only know you have an issue that needs fixing, but you also get a better understanding of what the likely root cause is and how to fix it, he says.

“Predictive tells you that you have a problem, it’s getting worse, and it’s time to fix it. You can schedule maintenance with your next planned outage or at least take controlled action,” Chappell says. “With prescriptive, it tells you what you need to do to fix it.”

Companies in this Article