Minimizing the cost of preventive maintenance is a must for maximizing its benefits.
Unexpected outages, a frequent problem for most companies, increase the cost and risk of doing business.
Maintenance, which is necessary to prevent such outages, comes in two types: reactive and preventive. Performed after an outage has happened, reactive maintenance involves lower upfront costs and fewer staff members. Reactive maintenance does not require planning. However, it makes budgeting, downtime, and overtime pay unpredictable. Energy costs increase, and equipment lifespan shrinks.
The aim of preventive maintenance is to increase production efficiency through regular maintenance of equipment and assets. Although preventive maintenance involves higher upfront costs, more staff members, and more planning, it reduces budgeting and downtime unpredictability. In addition, it reduces energy and payroll costs while increasing equipment life expectancy. According to a study by The New Building Institute, reactive maintenance can increase energy use by 30% to 60% and decrease equipment lifespan. In addition, a study by the U.S. Department of Energy Operations and Maintenance reports that preventive maintenance can result in energy savings of as much as 18%.
Preventive maintenance has broad applications in various sectors, including manufacturing, hospitality services, fleet management, oil and gas, and property management.
Successful preventive maintenance in smart manufacturing
Preventive maintenance can be time-based or condition-based. To increase equipment life expectancy, reduce safety risks and energy consumption, and avoid extra costs, time-based maintenance follows an established schedule — servicing a machine every quarter or doing an oil change on a vehicle every 5,000 miles, for example.
Preventive maintenance is essential in smart manufacturing. Some machines run 24 hours a day in three shifts.
Examples include auto assembly robotics, metal part stamping, and automatic component placement. Therefore, the equipment on a factory floor, including engines, motors, fans, transformers, heat exchangers, gearboxes, drives, pumps, valves, and circuit breakers, needs continuous monitoring. By using a scalable sensor-based solution to monitor equipment, electrical, mechanical, vibrational, and thermal status information from numerous machines can be collected.
For example, Silicon Labs, in partnership with Wirepas, offers battery-powered, scalable, and cost-effective IoT solutions that connect and localize sensors, tags, and luminaires. Its EFR32BG21 (BG21) and FR32BG22 (BG22) modules and SoCs make IoT products low-power, scalable, and secure. They solve large-scale preventive maintenance and high-density asset-tracking needs. The modules can be integral to a battery-operated infrastructure for monitoring buildings and tracking their assets. Moreover, the modules are highly reliable and interference-resistant in a variety of environments, contributing to network integrity. Suitable for a broad range of applications, the modules can be used in smart manufacturing (Industry 4.0), including smart HVAC controls, smart meters (submeters), human-machine interface, access control, commercial lighting control, and asset-tracking tags.
Still, preventive maintenance interrupts normal equipment operation and requires budgeting for planning and execution. Although time-based preventive maintenance is typically more cost-effective than reactive maintenance, it can be less efficient than condition-based strategies. Sometimes, preventive maintenance is unnecessary and wasteful, while at other times, warning signs during the intervals between service sessions go unnoticed.
Condition-based predictive maintenance
Condition-based preventive maintenance is more specific, efficient, and informative than time-based preventive maintenance. The IoT has enabled condition-based maintenance to become predictive rather than just preventive. IoT-based solutions enable the monitoring of an equipment’s use as well as its condition. By predicting when potential failures will occur, maintenance can be scheduled preemptively, so repairs take place only when necessary and downtime is minimized.
Predictive maintenance uses sensors that have been installed on equipment to capture parameters in real time. For example, obtaining temperature, humidity, vibration, noise, magnetic field, and energy consumption data supports optimal scheduling decisions. Analyzing the collected data can determine if the temperature, vibration, and energy consumption levels fluctuate within acceptable ranges. Unfavorable trends indicating potential problems can be spotted. In addition, machine learning can be implemented in analytics to help detect abnormalities. Smart sensor nodes are also key enablers of predictive analysis when equipment needs to operate without downtime. Gateways are used to collect and process the data from several smart sensor nodes or connect the nodes to the cloud. Edge processing can combine and distribute processing power among smart sensor nodes and gateways. Doing so makes it possible to send the right data at the right time to enterprise-level systems for more advanced analyses. This broadens the scope of anomaly detection and classification.
Predictive maintenance is especially relevant for manufacturing, food production plants, and the power and energy industry, wherein prioritizing tasks to perform maintenance and repairs within the shortest achievable time frame is critical. The objective is to prevent unplanned outages at the lowest possible operating costs.
A wide range of smart sensor nodes are available from manufacturers like STMicroelectronics (ST). Among these are environmental sensors for temperature, humidity, and pressure; inertial and motion sensors, including accelerometers, gyroscopes, inertial measurement units, and motion sensors; and ultrasound analog microphones. Smart sensor nodes containing these sensors can, for example, perform vibration analysis ranging from simple pass/fail monitoring to high-accuracy, frequency-based data analysis. In addition, ST makes microelectromechanical-system sensors containing digital logic for processing. The logic is optimized to run machine-learning algorithms that allow the developer to program the sensor so it can detect that the system sleeps unless the data it collects is appropriate to share with the host processor, by exceeding a threshold or matching a pattern.
Smart sensor nodes make good economic sense. Autonomous node operations cost significantly less than portable piezoelectric probes and the skilled technicians required to operate them. The data collected by the nodes is repeatable, reliable, and timely, and it can be analyzed locally to trigger immediate decision-making, reduce power, and contribute to longer equipment life. Alternatively, under programmed conditions, the data can be transferred to the ST microcontrollers and microprocessors or to the cloud for deeper analysis.
Therefore, although predictive maintenance’s requirement for more sensors on equipment increases the upfront costs in comparison with reactive or time-based preventive maintenance, it will be more economical in the long run, according to the U.S. Department of Energy Operations and Maintenance study referenced earlier.
As more factories implement preventive maintenance, the factories that do not will quickly become less competitive.
However, minimizing the cost of preventive maintenance is a must for maximizing its benefits. For example, preventive maintenance software is a great tool for creating an efficient maintenance schedule and helping managers make sure that resources are not spread too thinly on too many items at one time. In addition, prescriptive or customized maintenance, which uses advanced analytics, machine learning, and artificial intelligence to generate predictions about maintenance, can make recommendations to improve system operations. Finally, layering predictive maintenance onto prescriptive maintenance will produce a synergistic benefit.
This article was originally published on EE Times Europe.