In Part 1, "Closing the maintenance gap: A back-to-basics approach" we explored the many causes of the equipment maintenance gap—and their various impacts. We outlined a five-step, back-to-basics program that manufacturers can implement to stabilize their organization.
Here in Part 2, we describe predictive and prescriptive maintenance, along with the array of other smart-factory technologies that are changing the whole approach to maintenance. With a well-structured program, these technologies do more than help companies optimize maintenance: they deliver many consequential benefits – from minimizing downtime and extending the lifespan of equipment to improving training, lowering costs, and boosting sustainability.
Advances in AI, machine learning (ML) and smart factory technology are revolutionizing equipment maintenance. They have made it possible for companies to quickly identify—and in many cases anticipate—issues. They equip maintenance teams with critical information needed to accurately diagnose and address minor performance blips before these blips turn into larger, more disruptive breakdowns.
There are two aspects of this maintenance approach:
- Predictive maintenance, which combines real-time, condition-based monitoring (i.e., tracking vibration, temperature, pressure, and so on) with historical data to identify potential equipment issues prior to a failure. (The historical data could come from work-order and repair history as well as maintenance, repair, and operations [(MRO] records.) The continuous monitoring of equipment conditions, along with historical failure history, allows for real-time assessment of performance. Any deviations from normal operating conditions trigger alerts that prompt further investigation or predict potential failure. Ultimately, the goal is to reduce unplanned failures and the associated maintenance cost by predicting failures and acting on them preemptively while also preventing a potential major breakdown.
- Prescriptive maintenance takes the inputs of predictive maintenance programs and harnesses the power of AI and ML to analyze collected data, identify patterns, and forecast potential failures and propose solutions, down to even an individual component level. ML models consider various factors such as machine age, condition, location, and recent performance to suggest possible root-causes of the issues, thereby expediting the troubleshooting process. As ML models learn from historical data and real-time operational data, their prediction accuracy improves over time. When combined with condition-based monitoring, they set specific thresholds for different parameters, generating alerts when these thresholds are exceeded to indicate that maintenance is required in a specific area. Prescriptive maintenance further optimizes maintenance processes, in turn improving OEE and lowering costs.
Major benefits—but they're not for everyone (yet)
Beyond their monitoring, alerting, and diagnostic capabilities, predictive and prescriptive maintenance offer important benefits in maintenance planning and execution. Instead of relying on rigid or arbitrary time- or usage-based schedules, maintenance activities can be more precisely linked to the remaining useful life of equipment. The ML models can ensure that maintenance is performed only when necessary and can optimize maintenance planning by recommending the ideal times for maintenance activities based on predictive models and real-time data. By doing so, they help maximize maintenance effectiveness while minimizing costs.
Although their advantages are many, these maintenance methods are not necessarily for every company—or every piece of equipment. Not only do they require experienced maintenance personnel, they also rely on connected technologies that can access historical operating data (such as that compiled by applications like Historian) to integrate with production machinery. Companies that lack a repository of historical data or the ability to collect it can get support from third-party providers that compile databases with historical data on specific categories of equipment gathered from other organizations. This data can be used to quickly train a model with accuracy being improved as the company begins collecting and transitioning to their own data capture.
A predictive program is also costly to set-up properly; it entails developing the appropriate algorithms. Although technology advances have made predictive monitors cheaper, their blanket application, absent a data strategy and data management capability, is impractical. It will only create an avalanche of data that most organizations cannot efficiently use.
It makes little sense for companies to forge ahead with predictive maintenance without having in place the basics we outlined in Part 1: a strategy, a maintenance work management plan, storeroom and critical spare management, personnel management, and a maintenance system with KPIs. Without these basics, and a baseline starting point, organizations will struggle to reap any benefits.
Additional technologies that are pushing the maintenance envelope
AI and ML are not the only technologies transforming maintenance within the factory. A number of other technologies with monitoring, diagnostic, and repair capabilities can be incorporated into a robust maintenance management system.
Augmented reality (AR): Real-time assistance, expert guidance
AR overlays critical digital information, such as schematics, instructions, and part numbers, directly onto the equipment or through glasses, goggles, or a headset. AR helps technicians understand what needs to be done without having to refer to a manual. It presents step-by-step visual guides for complex maintenance tasks and can provide safety and compliance warnings.
In addition, AR enables technicians to connect with remote experts who are able to see exactly what the technician is seeing and can thus guide them in real time. By annotating the AR display, experts can highlight areas of interest or concern and speed up troubleshooting.
Virtual Reality (VR): Immersive training, skills assessment
VR simulates realistic, immersive (3D) training environments, allowing technicians to practice maintenance tasks without the risk of damaging equipment. Wearing goggles, headsets, or gloves, technicians interact with virtual machinery, gaining hands-on experience in a safe and controlled setting. VR advances scenario-based training by simulating various equipment failures; technicians can practice diagnosing and repairing different issues under different scenarios. In addition, it can train technicians how to respond to emergency situations, such as equipment malfunctions or safety incidents.
Finally, by monitoring their performance during virtual maintenance tasks, VR can assess a technician's skills. It identifies areas for improvement, ensuring that technicians are well-prepared for real-world maintenance challenges.
Drones: Accessing the inaccessible, assessing the operating environment
Initially drones were touted as a solution for geographically remote tasks. But their greater affordability and improved effectiveness makes them a practical option for inspecting hard-to-reach places. They provide high-resolution images and videos of such areas as roofs, pipelines, and power lines. Drones equipped with thermal cameras can detect heat anomalies, thus pinpointing leaks and electrical faults.
Drones are useful for the more routine monitoring of equipment, detecting changes over time and helping to predict maintenance needs. They can create detailed 3D maps and models of equipment and infrastructure that support planning and maintenance activities. Drones can also assess environmental conditions around equipment, such as monitoring air quality and even detecting gas leaks. Newer generations of drones can perform many precision maintenance tasks, such as tightening bolts, applying lubricants, or cleaning surfaces.
Robots: Going where humans shouldn't (or can't)
From the start, robots have been skilled at performing targeted repairs. They are also able to navigate and perform maintenance tasks in confined spaces where human access is limited or impractical. More important, they can perform specialized tasks, like welding and cutting, in hazardous or difficult-to-reach areas. They are particularly useful in environments that are dangerous for humans, such as toxic or radioactive areas, thus reducing risk to personnel.
Both robots and drones can collect real-time data during inspections and maintenance tasks, which can be analyzed to improve decision making and predictive maintenance strategies.
3D Printing: Cutting costs while improving efficiency
3D printing allows companies to rapidly create replacement parts onsite so that they can make repairs immediately, without having to wait for parts to be shipped or risking their lack of availability. By printing parts only as required, companies eliminate the need for large inventories of spare parts, minimizing storage expenses and reducing waste. 3D printing also lets companies create prototypes and custom parts that are tailored to specific needs, enabling quick testing and validation before final production. Companies can ensure that components are available when needed, while minimizing downtime.
With a digital inventory of parts that are made only when needed (or readily available for proactive maintenance), maintenance planning and logistics are streamlined.
Because advanced materials can be used in 3D printing, companies are able to improve the durability and performance of parts. They can create complex and optimized designs that actually enhance the functionality and extend the lifespan of components, a capability generally not feasible with traditional manufacturing methods.
Finally, 3D printing is environmentally friendly. It produces less waste than traditional fabrication methods, as it builds parts layer by layer using only the necessary material. Moreover, many 3D printing materials can be recycled, contributing further to more sustainable maintenance practices.
GenAI: Enhancing maintenance efficiency with a human-centric approach
GenAI technology is revolutionizing maintenance by enabling plants to build a robust digital infrastructure that integrates a wide range of critical information. A key element in this process is the creation of a comprehensive plant knowledge base, consolidating data from sources such as OEM manuals, preventive maintenance (PM) procedures, troubleshooting guides, and historical performance records. By organizing this information into an indexed and searchable format, GenAI can access and analyze relevant data in real-time to provide informed decisions. For example, indexing OEM manuals and PM procedures allows GenAI to quickly recommend best practices for routine maintenance or suggest troubleshooting steps based on similar past incidents. This capability reduces downtime, speeds up issue resolution, and enhances overall operational efficiency.
Machine Learning applications are also enabling step change improvements in predictive and prescriptive maintenance by allowing maintenance teams to quickly pinpoint top sources of operational loss and root causes. By integrating quantitative plant data—such as sensor readings, production metrics, and performance analytics—along with prescriptive maintenance algorithms, maintenance teams can now use GenAI-powered interfaces as dynamic Q&A portals. This allows the teams to not only see the sources of loss and root cause, but to access precise troubleshooting procedures and build out potential go-forward asset care recommendations. This data-driven approach simplifies decision-making, enables faster interventions, and optimizes plant performance by minimizing downtime and maximizing operational efficiency.
Are you ready for advanced maintenance methodologies?
The capabilities and benefits of these advanced methods and technologies are tantalizing.
Eager to capture their benefits, many companies jump the gun without having in place a mature foundational maintenance work-management process. You have to get the basics right first before you can start building predictive capabilities.
Before even thinking about a whole suite of predictive tools, it's essential to understand the equipment issues that are affecting plant performance—especially in critical or capacity- constrained lines—and consider how predictive tools could reduce the breakdowns and unplanned downtime.
Not every predictive tool needs to be sophisticated. You can equip maintenance personnel with simple vibration or thermal measurement tools. A temperature sensor might be enough to alert them to replace a motor before it breaks down. You can also identify critical equipment with simple "good/bad" gauges (such as temperature or vibration above a given threshold) to instigate a predictive maintenance action.
The following considerations can help you get started.
First, review and define your maintenance strategy. How critical is maintenance to your operations? Does unplanned machine downtime significantly drive productivity loss?
Is the organization willing to invest? Typical costs include hardware, software, model development, and potential additional repair costs.
Assess your current maintenance capabilities. How mature is your preventive maintenance capability?
- How robust are processes across different maintenance dimensions? (e.g., PM planning, spares, compliance, and so on).
- What level of machine connectivity do you have? How easy it is to integrate sensors, cameras, and data collection devices?
Next steps
If the answers suggest you're ready, you'll need to outline a plan for adopting an advanced tech maintenance program. The first step is to build out a roadmap and consider which technologies are most applicable to your situation. Once you have a plan in place, including where to pilot and then how to roll out more broadly, you can begin the process of testing and implementation.
Predictive and prescriptive maintenance is the future. We recognize that not every company is ready for this journey today, but for those who are, the benefits, which we've outlined here, can be truly transformative. Whether your interest is in a maintenance capability assessment or end-to-end supply chain transformation, our team is ready to help, regardless of where you are on your journey.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.