Pd M Strategies

Right-size your predictive maintenance strategies

Oct. 14, 2022
Five case studies show that anyone can benefit from this proven strategy, from targeted tools to complete solutions.

The value of predictive maintenance is firmly established and methods to achieve it keep getting better. Before it had a name, it was a technique performed intuitively by operators and maintainers using their senses: Is there an unusual sound or smell, excessive heat, or visible vibration coming from a machine in operation? If so, inspection and maintenance are probably in order.

Today, there are countless technologies supporting this best practice that are faster, easier, and reveal insights much earlier. They range from purpose-built portable condition monitoring devices to real-time streaming of millions of condition data points for predictive and prescriptive analytics.

Whether targeting troublesome machine types such as rotating equipment, crucial practices such as leak detection or electrical inspections, selected critical machines or processes, or plant-wide maintenance optimization, companies of any size or budget can apply predictive maintenance (PdM) to reduce unplanned downtime.

The need is clear because the impacts of poor asset management are so widespread. “The key aspect is preventing unplanned downtime because unplanned downtime results in lost revenue and shorter asset life (see Figure 1). It also has other impacts like WIP losses, safety/environmental, and quality/yield/rework, but if you want to financially justify smart asset management, use revenue and asset longevity,” suggests Ralph Rio, VP of enterprise solutions at ARC Advisory Group. “That will get executive attention because it’s in their KPIs.”

The five case studies summarized below describe how companies are becoming more predictive and proactive in their maintenance and reliability to avoid dealing with the consequences of asset degradation and failure. Each is taking a path that suits them, but all are benefiting from a common goal: increasing asset uptime and all the benefits that come with it. But first, let’s look at some impressive enabling technologies.

Maintenance never looked so good

Portable PdM tools are getting smaller, lighter, less expensive, and more intelligent—and their scope of functionality keeps advancing. Early devices mostly alerted to a single asset or component condition, such as excessive vibration, elevated temperature, or degraded oil quality. Now, many include capabilities such as:

  • assessing multiple condition variables
  • conducting on-device analytics
  • producing intuitive visual depictions of conditions and diagnostics
  • displaying colorful pattern trends and alerts
  • acquiring photographs, video, or location data
  • enabling direct access to supporting records or subject matter experts
  • transmitting the data for centralized analytics
  • triggering PdM work requests or work orders in the asset management system (EAM/CMMS).

Online, real-time condition monitoring and predictive analytics as well as PdM program support are enabled by Industry 4.0 and industrial internet of things (IoT) developments, such as:

  • wired or wireless machine condition monitoring sensors that automatically track relevant machine condition variables
  • holistic condition and process monitoring for predictive analytics throughout the asset lifecycle through asset performance management (APM) solutions
  • artificial intelligence and machine learning (AI/ML) that improve issue detection, predictive analytics, and prognostics over time, and develop prescriptive recommendations on how to resolve identified issues
  • cloud-based platforms that simplify PdM deployment, centralize the data, and provide cost efficiencies
  • collaborative data hubs that enable knowledge sharing among trusted organizations
  • strategic intelligence from OEMs and third-party service providers who can harness and package their cross-industry asset and reliability data
  • drones or robots able to gather condition data when equipped with payloads such as sensors or a portable condition monitoring tool
  • augmented, virtual, or mixed reality (AR/VR/MR) solutions that provide PdM knowledge support for asset-facing individuals
  • virtual representations (digital twins) of machines, parts, or processes to simulate how changes impact asset health and reliability.

These capabilities and more are expanding the potential for PdM to deliver significant performance and bottom-line benefits. Let’s now look at some real-world applications.

WV aluminum producer’s predictive lubrication

Global aluminum products manufacturer Constellium is moving away from reactive maintenance. Its Ravenswood, WV, site chose to replace its inefficient time-based lubrication PMs with a PdM program to solve bearing lubrication issues that can lead to catastrophic failures. The results of this and complementary initiatives were presented at the 2022 Leading Reliability Conference by Roger Carpenter, a casting reliability millwright at Constellium.

With management support, a condition monitoring route was created for 98 furnace bearings in the site’s five DC furnace complexes. The goal was to significantly reduce unplanned downtime from critical failures on combustion fan bearings and motor failures due to incompatible and improper greasing. One critical combustion fan bearing failure can cost 12-16 hours of lost production and involve the use of a 70-ton crane for its replacement, explained Carpenter.

The route is conducted during normal operating conditions using handheld ultrasound-based solutions from UE Systems. Ultrasound technology detects or “hears” telltale signals that are inaudible to the human ear and emerge earlier than vibration analysis can detect a problem (see Figure 2).

First, the Grease Caddy 201 was used to simultaneously lubricate and monitor ultrasound levels to avoid over- or under-lubrication of the bearings. Later the Ultraprobe 9000 and Ultraprobe 15000, with included software, were added to listen to the bearings, establish trends, and generate alarm reports when friction increases by 8 dB above baseline (greasing is needed) or 16 dB above baseline (the bearing is approaching failure mode and needs greasing and internal inspection).

This and other reliability initiatives not only succeeded in reducing costly failures and unplanned downtime at the plant, but are also increasing safety and efficiency, lowering contract maintenance costs, and extending asset life.

IL electricity provider’s risk-based maintenance

As part of its mission to “power quality of life,” Ameren Illinois needed to better predict asset health and be more proactive in its maintenance to prevent vital substation equipment from failing. The utility serves 1.2 million electric customers across 80% of the state and operates 1,200 substations, including some with transformers that are more than 40 years old. Its millions of existing asset health data points were distributed across organizational silos, constraining equipment analysis, asset life predictions, and performance forecasting efforts.

To help aggregate data in near real time and move from time-based maintenance to risk-based maintenance, it chose to deploy an asset performance management (APM) system. Lumada APM from Hitachi Energy is predictive in detecting the underlying conditions that can lead to failure; prognostic in helping to anticipate when, what if, and how an asset may fail; and prescriptive in providing recommendations and priorities for resolving, analyzing, and mitigating failures.

Within 90 days, more than 2,000 transformers were entered into the software, prioritized based on risk, and yielding actionable insights—including identifying a failing transformer that was promptly scheduled for maintenance, preventing a customer outage event, emergency repairs, and lost revenue.

Substantial time savings are also being gained from having fleet-wide visibility, embedded asset performance models, and configurable predictive analytics. “The asset models built into APM software enable us to reduce ineffective time-based practices and maximize resources, eliminating a huge backlog of data input for our maintenance engineers,” observed Donald Borries, supervising engineer of substation maintenance at Ameren Illinois.

KY aluminum factory’s early fault detection

The Logan Aluminum rolling mill near Bowling Green, Kentucky, is the largest supplier of aluminum can sheet in North America. It produces about two billion pounds of aluminum a year at a very high rate of speed, so reliability is in constant focus. Fifteen years ago, the facility began vibration analysis and predictive maintenance. In the last two years, an APM journey began to integrate the facility’s asset condition and process sensor data to enable earlier fault detection and AI, with the goal of a five percent reduction in unplanned downtime.

The site’s reversing mill, finishing mill, a slitter, and coating line were the first to be added to the APM 360 solution from SymphonyAI Industrial. “From the thousands of measurement data points going into the system, machine learning detects anomalies early and APM provides the causes and recommendations,” said Vijay Kamineni, chief innovation officer at Logan Aluminum in his presentation at the 2022 ARC Industry Forum.

Within four months, with one use case, the system paid for itself multiple times over. It alerted to an issue in the reversing hot mill with an axial bearing of a pinion gearbox for a 12,000 hp motor. The subsequent inspection revealed spalling—an unseen internal anomaly predicted early with AI—that was able to be corrected before failure.

That one incident could pay for everything Logan Aluminum ever invested in vibration analysis, predictive maintenance, and APM, suggested Kamineni. The motor is so big that it takes two trucks to send it out for rewinding, so proactive maintenance and failure prevention is critically important, he added.

The positive outcomes from the first four machines using APM 360 cinched management support, and over the next four years, the plant plans to add the next 20 machines to the solution.

ID fertilizer producer’s digital predictive inspections

Itafos Conda a fertilizer manufacturer in Soda Springs, Idaho, underwent an asset reliability program review after Itafos acquired Conda’s assets in 2018. It quickly became apparent that the plant’s manual, siloed equipment data gathering processes did not adequately foster effective asset management, insurance or safety compliance, or long-term sustainability. A digital asset management and reliability program was needed to better monitor for and prevent functional failures and care for the facility’s more than 9,000 assets.

A project was scoped to develop asset care plans that would make necessary information readily accessible, be quick to deploy, and provide a consistent framework across all operating units in the plant, said Nicholas Hofeldt, senior reliability engineer at Itafos Conda’s Phosphate Operation. Highly critical and safety critical assets would be prioritized.

The asset care plans were developed using AssetWise Asset Reliability, an APM solution from Bentley Systems, based on both historical learnings and current best practices. The tablet-based plans guide operators and maintainers through their inspection routes, with the findings entered in the electronic checksheet and stored in a single digital environment accessible to all stakeholders (see Figure 3). Asset abnormalities, identified based on alarm states, automatically trigger email notifications and integration with the facility’s work management system for planning and scheduling.

With this solution, data collection time dropped from days to minutes and maintenance is properly planned, scheduled, and completed before failure. Further, by continually expanding understanding of the assets’ current operating context, reasonably likely functional failures, and their potential effects, the software also facilitates continuous improvement.

The project, which was chosen as a finalist in Bentley’s 2021 Going Digital Awards in Infrastructure, has succeeded in enabling Itafos Conda to be a more sustainable, safe, and reliable site. “Over 95% of our highest criticality assets have implemented asset care plans in place, and we have achieved a site checksheet completion compliance of over 98% year to date,” said Hofeldt.

Ecuador oil company’s prescriptive intelligence

A crude oil transport company, Oleoducto de Crudos Pesados (OCP) Ecuador, fast-tracked an upgrade of calendar-based maintenance activities to an AI-powered solution to better predict and prevent equipment failures, reduce costs, and increase regulatory compliance. It was implemented within six weeks across 31 crucial remote assets installed along the 485-kilometer pipeline the company operates, which runs from the Amazon Rainforest to the Ecuador coastline (see Figure 4).

To strengthen its condition monitoring outcomes, OCP Ecuador selected a prescriptive maintenance solution with AI and machine learning capabilities known as Aspen Mtell from Aspen Technology (AspenTech). By analyzing both equipment and process data, it can detect even the slightest changes in asset behavior and accurately determine the probability of failure.

Several advantages were realized from detecting critical issues that previously went unnoticed, even with only three to four months of data. For example, engine overhauls that used to occur at 16,000 hours are now conducted at 19,200 hours—a 20 percent improvement in uptime. This consequently reduced maintenance costs by 25 percent, or almost $500,000 per year.

Similarly, extending asset life lowers spare parts and components inventory requirements. “Based on equipment availability insights derived from Mtell, we further reduced costs by updating our spare parts guidance for camshafts from 20,000 hours of life to a new benchmark of 50,000 hours,” said David Mafla, maintenance analysis and monitoring supervisor at OCP Ecuador.

Additional benefits gained were the prescriptive maintenance solution’s ability to deploy quickly at scale, to detect impending asset failures up to 20 days in advance, and to return three times the initial investment within five months. OCP Ecuador plans to scale the solution to additional assets in the coming years.

Understanding conditions

When predictive maintenance is performed haphazardly—or not leveraged at all—companies suffer operationally and financially, and the impacts can be catastrophic. The above examples provide a small taste of the many methods available to reduce the risk of unplanned downtime and asset failure through PdM. It all comes down to heightening understanding of critical asset conditions and becoming more proactive in reliability and maintenance practices.

About the Author

Sheila Kennedy | CMRP

Sheila Kennedy, CMRP, is a professional freelance writer specializing in industrial and technical topics. She established Additive Communications in 2003 to serve software, technology, and service providers in industries such as manufacturing and utilities, and became a contributing editor and Technology Toolbox columnist for Plant Services in 2004. Prior to Additive Communications, she had 11 years of experience implementing industrial information systems. Kennedy earned her B.S. at Purdue University and her MBA at the University of Phoenix. She can be reached at [email protected] or www.linkedin.com/in/kennedysheila.