What is predictive maintenance?

By collecting data from sensors and applying advanced analytical tools and processes such as machine learning (ML). Predictive maintenance can identify, detect, and address issues as they occur, as well as predict the potential future state of equipment, and so reduce risk. The key is providing the right information at the right time to the right people.

Maintenance strategies and maturity depend on factors such as asset and replacement cost, criticality of asset, usage patterns and impact of failure on safety, environment, operations, finance, and public image. Predictive maintenance is one of three leading maintenance strategies that are used by businesses. The others are reactive maintenance, which fixes failures when they occur, and preventive maintenance, which relies on a predefined maintenance schedule to identify faults.

Because predictive maintenance is proactive it enhances preventive maintenance by providing continuous insights on the actual condition of the equipment. Rather than relying on the expected condition of the equipment based on a historical baseline. With predictive maintenance corrective maintenance is only carried out only when there is a need to do so, and so avoids incurring unnecessary maintenance costs and machine downtime.

Predictive maintenance uses time series historical and failure data to predict the future potential health of equipment and so anticipate problems in advance. This enables businesses to optimize maintenance scheduling and improve reliability.

Predictive maintenance also differs from preventive maintenance in the diversity and breadth of real-time data that is used in monitoring the equipment. Various condition monitoring techniques such as sound (ultrasonic acoustics), temperature (thermal), lubrication (oil, fluids) and vibration analysis can identify anomalies and provide advance warnings of potential problems. A rising temperature in a component, for example, might indicate airflow blockages or wear and tear. Unusual vibrations might indicate misalignment of moving parts. Changes in sound can provide early warnings of defects that can’t be picked up by the human ear.

Predictive maintenance relies on various technologies including the Internet of Things (IoT), predictive analytics, and artificial intelligence (AI). Connected sensors gather data from assets such as machinery and equipment. This is collected at the edge or in the cloud in an AI-enabled enterprise asset management (EAM) or computerized maintenance management system (CMMS). AI and machine learning are used to analyze the data in real time to build a picture of the current condition of the equipment. Thereafter, triggering an alert if any potential defect is identified and delivering it to the maintenance team.

As well as providing defect warnings, advances in machine learning algorithms enable predictive maintenance solutions to make predictions about the future condition of equipment. These can be used to drive greater efficiency in maintenance-related workflows and processes such as just-in-time work order scheduling and labor and parts supply chains. Furthermore, the more data collected the more insights are generated and the better the predictions become. This gives businesses the confidence that equipment is working optimally.

Benefits from a predictive maintenance strategy center around anticipating equipment faults and failures, reducing maintenance and operating costs by optimizing time and resources, and improving the performance and reliability of equipment. Deloitte reported in 2022 that predictive maintenance can result in a 5-15% reduction in facility downtime and a 5-20% increase in labor productivity.¹

Optimizing asset performance and uptime can reduce costs. Advance warning of potential faults results in fewer breakdowns as well as reduced planned maintenance or unplanned downtime. Greater continuous condition visibility enhances the lifetime reliability and durability of equipment. The use of AI can more accurately forecast future operations. This latter benefit is paramount in a world where rising prices and unpredictable events like the pandemic and climate-related natural disasters exposed the need for more predictable spare parts inventory and labor costs and a lower environmental impact from operations.

Productivity can be increased by reducing inefficient maintenance operations. Enabling a faster response to problems via intelligent workflows and automation, and equipping technicians, data scientists and employees across the value chain with better data with which to make decisions. The upshot is improved metrics such as mean time between failures (MTBF) and mean time to repair (MTTR), safer working conditions for employees, and revenue and profitability gains.

There are barriers to predictive maintenance, which can be costly, at least in the first instance.

• System infrastructure: startup costs that are associated with the complexity of the strategy are high. These often involve upgrading and integrating outdated technology and monitoring systems as well as investing in maintenance and data management tools and the data and systems infrastructure.
• Workforce training: training the workforce to use the new tools and processes and correctly interpret data can be expensive and time-consuming.
• Data requirements: the past is a predictor of future performance. For predictive maintenance to be effective, the availability of substantial volumes of time-series historical and failure (or proxy) data is vital. The ability to look at data correlations and analogies with similar equipment types in physical operating conditions is also essential and can also help improve the predictive nature of analytics.

Assessing the criticality and cost of failure of individual assets also takes time and money. But is fundamental in determining whether predictive maintenance is appropriate — low-cost assets with cheap readily available parts may be better served with other maintenance strategies. Predictive maintenance programs are hard but the competitive and financial advantages of a well-run strategy are significant.

Predictive maintenance technologies are already being adopted across industries for many assets whether that be cash points, wind turbines, heat exchangers or manufacturing robots. Asset-intensive industries such as Energy, Manufacturing, Telecommunications and Transportation, where unforeseen equipment failures might have widespread consequences, are increasingly turning to advanced technologies to improve equipment reliability and labor force productivity. Potential uses are many and varied:

The invention of the predictive maintenance technique is attributed by most to CH Waddington back in the second World War. He noticed that planned preventive maintenance appeared to be causing unplanned failures in the aircraft bombers.² This led to the emergence and development of condition-based maintenance but since most business systems have historically been siloed, adoption of predictive maintenance has been limited.

Technological advances in IoT sensors, big data collection and storage technologies have and will continue apace. The growth of data and accessibility of AI/ML are enhancing predictive maintenance models and promoting its adoption. The pandemic also accelerated digital transformation efforts, creating more integrated business environments and appetite for intelligence-based real-time insights. Finally, the soaring cost of unplanned downtime, which experts estimate is around 11% of turnover in Fortune Global 500 companies³, is also fueling the adoption of predictive maintenance within the market.

The following technologies are just some of those contributing to the ongoing evolution and value of predictive maintenance:

• Automated robotic inspection is making monitoring of equipment in remote or dangerous to reach locations such as in the Oil and Gas industry more efficient and cost effective. Robots act as roving sensors that monitor multiple assets and feed data into computerized maintenance management systems.
• Immersive technologies such as augmented reality (AR) and virtual reality (VR) are being developed to simplify inspections. AR can collect data and both technologies can enhance visual inspections and early fault detection.
• Digital twins can augment predictive maintenance by creating a virtual representation of a physical asset, which generates sensor data and simulates operational fault scenarios and solutions throughout an asset’s lifecycle with no risk to the asset.
• IoT-enabled predictive maintenance solutions are supplied as part of EAM/CMMS solutions and integrated with other enterprise applications.
• Predictive maintenance-as-a-service will make predictive maintenance more accessible and affordable. Delivered by partners it can be less disruptive than on-premise deployments, require less investment and training, and deliver faster time to value. It can also be tailored to individual environments and equipment.