Your maintenance strategy may not be the first thing that springs to mind when thinking about the bottom line. Yet, given that machinery, equipment and systems keep businesses running, maintenance strategies have a major role to play. Without due care and attention, things break—regardless of whether that’s a transformer in an electricity grid, an axle bearing on a train or a refrigerator in a restaurant.
When assets malfunction or aren’t performing optimally, there can be safety issues and financial implications – the average manufacturer reportedly loses about 800 hours a year in downtime (link resides outside ibm.com). Add to that aging infrastructures, workforce retention, budget constraints and sustainability pressures, and it’s easy to see why businesses need to find ever better ways to keep assets in good operating condition.
Understanding and planning for when your equipment is likely to fail can drive greater efficiency in production operations, but how do you decide which strategy is the most cost-effective one for you? The decision isn’t simple. Multiple factors must be considered, such as your industry, the type and usage of the asset, how expensive it is to replace, how much of the right kind of data you have, and how much impact failure would have on your business and customers. There is no one-size-fits-all solution, and most companies opt for a combination of different maintenance strategies across their asset portfolios.
Reactive, preventive and predictive maintenance strategies are the most commonly used maintenance approaches. Reactive maintenance (also called corrective maintenance) is exactly that—reacting to breakdowns when they occur. It is suited to low-cost, non-critical assets that don’t pose safety or operational risks if a run-to-failure strategy is deployed.
Preventive and predictive maintenance are proactive maintenance strategies that use connectivity and data to help engineers and planners to fix things before they break. Predictive strategies take this even further and use advanced data techniques to forecast when things are likely to go wrong in the future. Both strategies aim at reducing the risk of catastrophic or costly problems.
Let’s take a deeper look at these proactive approaches.
Preventive maintenance uses regular maintenance plans to reduce the chances of an asset breaking down by carrying out routine maintenance tasks at regular intervals. Using best practices and historical averages, such as mean-time-between-failure (MTBF), downtime is planned. Preventive maintenance strategies have been around since about 1900 and widely used since the late 1950s (link resides outside ibm.com).
Three major preventative maintenance types have developed that all involve carrying out maintenance on a regular basis but are scheduled differently and are tailored to different business operation purposes.
In all types of preventive maintenance, machine downtime is planned in advance, and technicians use checklists for checkups, repair, cleaning, adjustments, replacements and other maintenance activities.
Predictive maintenance builds on condition-based monitoring by continuously assessing an asset’s condition. Sensors collect data in real-time, and it is fed into AI-enabled enterprise asset management (EAM), computerized maintenance management systems (CMMS) and other maintenance software. Through these types of software, advanced data analysis tools and processes like machine learning (ML) can identify, detect and address issues as they occur. Algorithms are also used to build models that predict when future potential problems may arise, which mitigates the risk of the asset breaking down further down the line. This can result in lower maintenance costs, a reduction of some 35-50% in downtime and a 20-40% increase in lifespan (link resides outside ibm.com).
Various condition monitoring techniques are used to identify asset anomalies and provide advance warnings of potential problems, including sound (ultrasonic acoustics), temperature (thermal), lubrication (oil, fluids), vibration analysis and motor circuit analysis. A rise in temperature in a component, for example, could indicate a blockage in airflow or coolant; unusual vibrations could indicate misalignment of moving parts or wear and tear; changes in the sound can provide early warnings of defects that can’t be picked up by the human ear.
The oil and gas industry was a pioneering adopter of predictive maintenance as a way to lower the risk of environmental disasters, and other industries are also increasingly seeing the benefits. In the food and beverage industry, for example, undetected food storage issues could have major health consequences, and in shipping, anticipating and preventing equipment failures reduces the number of repairs that have to be made at sea, where it is harder and more expensive than in port.
Both types of maintenance strategies increase uptime and reduce unplanned downtime, improving the reliability and lifecycle of assets. The main differences are in timing and the ability to predict the future likely condition of an asset.
Preventive maintenance programs use historical data to anticipate the expected condition of an asset, and they schedule routine maintenance tasks at regular intervals in advance. While this is good for planning, assets may be under- or over-maintained, given that the vast majority of asset failures are unexpected. A problem might be diagnosed too late to prevent damage to an asset, for example, which will likely mean longer downtime while it’s fixed, or time and money may be spent when there’s no need.
Predictive maintenance avoids unnecessary maintenance by understanding the actual condition of the equipment. This means it can flag up and fix problems earlier than preventive maintenance and prevent more serious issues from developing.
Predictive maintenance leverages new technologies like artificial intelligence, machine learning and the Internet of Things (IoT) to generate insights. Maintenance management systems and software automatically create corrective maintenance work orders, enabling maintenance teams, data scientists and other employees to make smarter, faster and more financially sound decisions.
Inventory management workflows like labor and spare parts supply chains become more efficient and sustainable through minimizing energy usage and waste. Predictive maintenance can feed data into other maintenance practices based on real-time analytics like digital twins, which can be used to model scenarios and other maintenance options with no risk to production.
There are obstacles to overcome for predictive maintenance to be effective or even possible, such as complexity, training and data. Predictive maintenance requires a modern data and systems infrastructure that may make it costly to set up when compared with preventive maintenance. Training the workforce to use the new tools and processes and correctly interpret data can be expensive and time-consuming. Predictive maintenance also relies on the collection of substantial volumes of specific data. And lastly, implementing a predictive maintenance strategy requires a cultural change to accommodate the shift from predetermined to more flexible daily operations, which can be challenging.
In summary, although preventive and predictive maintenance strategies both focus on increasing asset reliability and reducing the risk of failures, they are very different. Preventive maintenance is regular and routine, whereas predictive maintenance focuses on providing the right information about specific assets at the right time. Preventive maintenance is suited to assets where failure patterns are predictable (e.g., recurring or frequent problems) and the impact of failure is comparatively low, whereas predictive maintenance may be more advantageous for strategic assets where failure is less predictable and the business impact of failures is high. Ultimately, if predictive maintenance strategies are successfully deployed and run, they will result in happier customers and substantial cost savings through optimized maintenance and asset performance.
The good news is IBM can help. IBM Maximo Application Suite is a set of applications that enables you to move maintenance planning beyond time schedules to condition-based predictive maintenance based on asset health insights.
Combining operational data, IoT, AI and analytics in a single, integrated cloud-based platform, Maximo will drive smarter, data-driven decisions that improve asset reliability, lengthen asset lifecycles, optimize performance and reduce operational downtime and costs.