Predictive maintenance leverages data to revolutionize operations
Industrial facility managers are under continuous pressure to improve maintenance processes in plants and operating environments. According to a recent McKinsey report on big data, manufacturing processes produce more data than any other source. That data is a largely untapped but powerful tool facility managers can leverage to implement a highly efficient maintenance regime across their plant, from an individual machine to an entire facility.
When data is used strategically as a foundational pillar of a maintenance program, facility managers can achieve what’s known as predictive maintenance—a maintenance technique that correlates information from different devices and machines in real time, to enable maintenance to be conducted on an as-needed basis—minimizing downtime, increasing productivity, and eliminating time and resources spent conducting unnecessary repairs.
Most manufacturing facilities have an opportunity to dramatically improve their bottom line by implementing predictive—as opposed to reactive—maintenance to reduce the total cost of ownership throughout equipment lifecycles. This helps to facilitate a capital expenditure (CapEx) vs. operational expenditure (OpEx) financial model, optimizing energy use, reducing equipment downtime, and much more.
Predictive maintenance competitive advantage
Facilities with legacy, or even outdated, maintenance practices often incur unnecessary costs in the form of operational downtime, wasted energy, and human capital. With traditional maintenance practices, routine maintenance is performed on a schedule, meaning operators are quite possibly wasting time and resources maintaining a piece of equipment that may not need any attention, or are switching out pieces of equipment that may still have useful life. With traditional maintenance practices, signs that a piece of equipment is about to fail are often overlooked if a machine is not due for scheduled maintenance.
On the other hand, facilities that have implemented a predictive maintenance solution conduct machine and equipment maintenance on an as-needed basis, which may be more or less frequently than prescribed by a routine scheduled maintenance program. By leveraging an infrastructure of networked, connected equipment that produces data detailing operational parameters such as energy use, temperature, output speed, and quantity, operators and plant managers can identify equipment that is functioning optimally or is about to fail. Operators and facility managers can then make informed and targeted decisions about when to conduct maintenance, take a machine offline, or allow a piece of equipment to continue to operate in its current condition.
With predictive maintenance, facility managers can avoid "virtual downtime," when a piece of equipment is not operating to its fullest potential, but its output remains within normal operating variability. Consider, for example, a piece of battery manufacturing equipment that produces units at a tremendous rate, spitting batteries out faster than one can see. Three machines may have a variability of 10%-15%, which can fall under normal operating output range. However when additional data points such as energy use, run time, and temperature are monitored, an operator can improve the machine’s output by 10% and achieve significant cost savings.
Data: Foundation of predictive maintenance
Networks, connected devices, and the collection, monitoring, and analysis of data—also known as Big Data—are the foundation of a predictive maintenance program. This infrastructure of data and data-driven intelligence is known as the Internet of Things (IoT). Defined by Gartner as the network of physical objects that contain embedded technology to communicate and sense or interact with their internal states or the external environment, an IoT infrastructure monitors equipment across a plant. Data and information can then be fed to plant managers and operators to shift to a predictive maintenance paradigm in their plants.
Predictive maintenance systems can leverage many types of data including equipment run time, temperature, energy use, output, and more to improve decision making and operations. For example, a piece of equipment in a consumer packaged goods facility may continue to maintain a steady output of paper towels, but just before it fails its energy use spikes. By monitoring the energy use data generated by the machine, operators can intervene as soon as they see the energy use beginning to spike, but before a machine fails. With routine maintenance, the machine would go offline, resulting in unplanned interruptions to the production cycle. By leveraging current data about a machine’s operations, as well as historical data about operational conditions leading up to past failure, operators can reduce disruption to a facility’s broader operations.
From data to predictive maintenance
The path to implementing predictive maintenance is iterative and multifaceted. Below are three key steps for getting started with implementing predictive maintenance in a facility.
- Shift procurement priorities: Leveraging data, big data, and the IoT for predictive maintenance requires equipment with capabilities to generate operational data. Connected devices are increasingly becoming the norm, but procurement processes must shift to prioritize acquisition of connected, network-enabled machines over legacy equipment. This shift may present organizational challenges as a legacy piece of equipment with no networking capabilities may mean a lower upfront cost as compared to a connected, smart machine. However, the additional cost of the networked machine can potentially be offset by leveraging the data it produces to avoid a single failure event and accompanying production downtime. Procurement decisions must be made based on total cost of ownership as opposed to simply looking at initial CapEx.
- Leverage data experts: Once a facility’s equipment is networked to measure and monitor data, facility managers can partner with data experts to ensure they are collecting and using data optimally. Data experts can improve data operations through on-site assessments, as well as virtually. Data collected by networked equipment can be stored in the cloud and monitored virtually through a service-based model. When data is stored virtually, it can then be accessed, analyzed, and used to direct and implement predictive maintenance through the help and guidance of data experts. This virtual, as-a-service partnership with data experts can accelerate the implementation of predictive maintenance programs in a facility.
- Push the right data to the right people: A key aspect of leveraging data-driven intelligence to achieve predictive maintenance involves pushing data across a plant’s organizational structure so it can have the highest impact on the decision-making process. Data must not remain siloed at certain organizational levels but must be pushed to the factory floor where it can be leveraged by individual machine operators. In an effort to ensure data spans an organization and reaches all the way to the factory floor, facility managers should think of making data pervasive, similar to how it they might receive data through push notifications on their smartphones. For example, in the mining, minerals, and metals industry, weather is a key factor in implementing predictive maintenance. If there is a data collection infrastructure in place, as well as a system to optimally distribute the data, a facility manager can alert staff and operators across an entire mine site that inclement weather is coming without requiring individuals to monitor weather forecasts. Smart data infrastructures can indicate which pieces of equipment may be most degraded by inclement weather, the current condition of that equipment, and the specific maintenance processes that should be executed ahead of the weather event. In any industry, facility managers should ensure data is reaching the ground level or factory floor so people act on it. It is not always about calling in a maintenance specialist, but about empowering individual equipment operators with the data they need to conduct predictive maintenance to optimize performance.
A comprehensive predictive maintenance program can lead to significant operational benefits. Plants and facility managers that effectively implement predictive maintenance can achieve significant operational advantages and emerge as leaders with a competitive advantage in their field. Once a facility’s equipment is networked, stakeholders across a facility must trust the insights generated by the data. Unexpected insights may challenge assumptions about optimal production parameters, such as times of day, performance of shifts operators, and much more. However, it is crucial to act on the insights to yield the full benefits of data-enabled predictive maintenance.
– Andy Roxburgh is vice president systems and services, industry business at Schneider Electric. Edited by Mark T. Hoske, content manager, CFE Media, Control Engineering, email@example.com.
- Networks, connected devices, and the collection, monitoring, and analysis of data-also known as Big Data—are the foundation of a predictive maintenance program.
- Plant data is a largely untapped but powerful tool that facility managers can leverage to implement a highly efficient maintenance regime across a plant, from an individual machine to an entire facility.
What’s the cost of "run to stop" maintenance? Wouldn’t that pay for predictive maintenance tools?
This article at www.controleng.com/archives October has links to related information.