From WWII to IIoT: The evolution of data-driven maintenance

Condition monitoring (CM) and predictive maintenance (PdM) are key subtopics of the broader leading-edge manufacturing topics of the Industrial Internet of Things (IIoT), smart manufacturing and artificial intelligence (AI). Yet the foundation for data-driven CM and PdM includes practices developed more than 80 years ago. Over time, equipment builders’ and operators’ goals have remained similar: improve asset utilization and availability, reduce unplanned downtime, improve quality and maximize overall equipment effectiveness (OEE). Technological innovations including advanced sensors, industrial networking, high-powered controls/processors and cloud computing are making it easier, faster and less expensive to achieve these goals:

• Data acquisition: Advanced sensors and networking enable gathering equipment data and communicating it to where it’s needed.
• Data processing: Edge gateways, more powerful processors and new software tools, including AI, have made it simpler to manage this large stream of data.
• Deployment cost/accessibility: Innovations have made it more cost effective and easier for machine builders to implement proactive maintenance approaches including condition monitoring and predictive maintenance and to try an even more advanced approach—prescriptive maintenance.

Data-driven maintenance’s historical foundation includes its key role in World War II’s Battle of the Atlantic. In 1942, the British faced a crisis: the U-boats had dramatically increased sinkings of Allied ships, and there were not enough resources to counter the threat, according to Admiralty's "The Anti-Submarine Report" (CB 4050/Series), a highly classified periodical published by the British Admiralty's Torpedo, Anti-Submarine and Mine Warfare Division during World War II. To address this challenge, the British gathered scientists from a wide range of fields into an operational research section (ORS) to come up with methods to combat the U-boats.

ORS data analysis revealed that the long-range bomber could be one of the most effective anti-U-boat tools; yet there were not enough planes to make a difference. The ORS therefore focused on what they could accomplish with limited resources, using a data-driven approach. They came up with several tactical solutions appropriate for aircraft operations vs. ships, including shallower depth-charge settings, white camouflage and tighter bomb dispersion patterns.

But they also investigated aircraft OEE: how to maximize flying time and the number of planes available for missions. Aircraft maintenance and part failure analysis led ORS to a counterintuitive conclusion: preventive maintenance (PM) was reducing, not increasing, aircraft availability and uptime.

There were several root causes:

• increased equipment failures shortly after scheduled PM
• long queues before starting scheduled PM
• bottlenecks in the maintenance process due to scarce resources in some departments and underutilized resources in others
• long maintenance times due to overly detailed procedures and maintenance of parts with no/low probability and/or no/low impact of failure
• wait times for correct parts, incorrect parts on hand.

The ORS implemented several policies and processes which today are being “rediscovered” in factory and warehouse maintenance, including:

• addressing bottlenecks and optimizing personnel/resource scheduling using actual production data
• analyzing failure data and using it to implement CM and/or PdM
• implementing longer, rather than shorter, maintenance intervals when justified by data
• considering a run-to-fail approach for non-critical components with very low failure rates
• optimizing spare parts inventory based on actual failures.

These findings were not just wartime operational improvements; they form the foundation of modern data-driven maintenance. Today’s machine builders face similar constraints: limited resources, variable asset performance and pressure to maximize uptime and improve OEE. The difference is that what ORS did manually can now be automated at scale using sensors, networking and analytics.

Unlike the Royal Air Force’s ORS, machine builders can automate data collection, transmission and analysis in real time. What once required manual data gathering and statistical analysis can now be embedded directly into the machine via smart sensors, industrial networks and edge computing platforms.

Get your subscription to Control Design’s daily newsletter.

Despite these advances, machine builders often hesitate to integrate CM/PdM into their designs due to:

• cost of implementation
• complexity of integrating into existing controls architectures and risk of negatively affecting the system
• lack of clear payback and ROI
• end-user concerns about data security, risk of machine hacks
• “if it ain’t broke, don’t fix it” mentality.

These concerns are valid but can be overcome by close cooperation with technology providers and by considering the opportunities for machine builders. Implementing CM/PdM solutions can help them in several ways:

1. Design and commissioning

• Fine tune equipment before shipment
• Validate the performance before and after the machine arrives at customer site, and provide documentation to the machine builder and the end user
• Provide benchmarks for future customer site tune ups and/or troubleshooting.

2. Service and support

• Enable remote troubleshooting and pre-support visit analysis
• Prepare technicians with appropriate tools and spare parts before customer site visits
• Reduce service time and costs
• Use data to check possible warranty claims.

3. Continuous improvement and optimization

• Adjust PM procedures based on actual component performance data
• Identify failure patterns and root causes
• Optimize spare parts inventory based on actual data
• Improve machine performance, especially in the availability component of OEE.

4. New revenue and business models

• Offer CM/PdM as a paid-for add-on option
• Offer more comprehensive fee-based services: support contracts, remote diagnostics, troubleshooting and cloud-based support
• Build the foundation for prescriptive maintenance: include tools and documentation into the solution to guide users through the failure resolution for the failure anticipated by CM/PdM
• Participate in customers’ IIoT and MC/PdM initiatives.

Data-driven maintenance is not a new concept; it is a proven approach refined over decades. What is new is the accessibility of the technologies required to implement it. Machine builders who embed sensing, connectivity and analytics into their equipment can move beyond reactive and scheduled maintenance toward truly proactive strategies.

By combining time-tested principles with modern IIoT technologies, builders can deliver more reliable machines, create new service opportunities and provide measurable value to their customers.