The Industrial Internet of Things (IIoT) includes, among other capabilities, monitoring equipment and process parameters using sensors connected to networks. In addition, some maintenance systems using IIoT can now use artificial intelligence and machine learning to increase process uptime and reliability to previously unheard-of levels.
THE EVOLUTION OF MAINTENANCE STRATEGIES
Over time, companies have improved their maintenance strategies to increasingly prevent unexpected downtime, an important advance. During downtime, overhead costs such as payroll and infrastructure, continue, but no value is produced. Downtime that is unplanned only aggravates the situation. Unplanned downtime has been estimated to cost ten times as much as planned downtime.
A white paper from Siemens described this evolution of maintenance practices from fixing things only when they break to anticipating equipment degradation and taking steps to avert the problem.
Reactive maintenance is basically waiting until a failure occurs and then fixing it. Most companies try to avoid this situation, as it causes unplanned downtime and may incur additional delays and expense in sourcing needed replacement parts on short notice.
Preventive (scheduled) maintenance provides for repair, adjustment or replacement on a schedule. This is the traditional approach to maintenance in many or most companies. The frequency of maintenance is based on experience with the equipment, manufacturers recommendations and other factors. The work is done by the schedule, regardless of the condition of the equipment. This might mean that parts are sometimes replaced when they are nowhere near failure, resulting in unnecessary costs. Also, this approach to maintenance likely requires more frequent downtime than might otherwise be necessary.
Some companies implementing more advanced versions of scheduled maintenance track equipment and systems with enterprise asset management or computerized maintenance management systems. These collect information about the company’s operations, which can help fine tune the preventive maintenance program. However, compiling and analyzing the data in order to use it for maintenance optimization is a daunting and labor-intensive task.
Predictive maintenance uses the ability to monitor the condition of equipment using sensors and collect the data through an IIoT system. The predictive maintenance system analyzes the collected data for parameters that are trending toward the limits of their normal range. For example, the system might detect a trend in the data from a sensor on a valve that indicates increasing vibration or acoustic noise. The change could be detected long before failure occurs. The predictive system could flag a technician to inspect the valve and determine what repair or adjustment is necessary and when. Prediction allows companies to run equipment as long as possible before replacement or repair, and lets maintenance be performed before the equipment fails and causes unplanned downtime.
Moving from preventive to predictive maintenance can provide significant improvements. According to a Department of Energy report, using predictive maintenance could reduce maintenance costs by 25% to 30% compared to a conventional preventive maintenance program. The predictive approach could reduce downtime 35% or more, with an accompanying increase in production.
Prescriptive maintenance not only detects when equipment or process parameters are heading out of range, but suggests or implements solutions. A prescriptive maintenance system uses artificial intelligence to provide a more automated approach. In addition to monitoring the equipment or process, it uses an IIoT-based operating system to deliver not only readings of equipment health, but also solutions to detected problems or ways to manage them.
For example, if a piece of equipment displays increasing bearing temperature, a predictive system would detect and track the temperature trend, forecasting when failure is likely to occur. A prescriptive system would, in addition, offer via its AI capabilities, alternatives for mitigating the situation. It might suggest reducing the speed of the equipment, say 3%, if compatible with the process, to enable it to continue operation until the next scheduled shutdown.
Such a prescriptive maintenance system can be designed to perform a wide variety of maintenance-related tasks. It could order the needed parts, schedule the service, log when the service is complete and account for labor and parts costs, as well as downtime impacts. In some cases, the needed adjustment or fix can be completed by the software itself. In the example above, the prescriptive system might interact with the process automation system to reduce the equipment speed the necessary few percent.
The white paper offered an example of a company that used an IIoT prescriptive maintenance package to detect the initial stages of asset failure. This helped prevent downtime. The prescriptive system was able to produce asset reliability of greater than 99%. The monitoring and AI capability of the system also improved root-cause analysis and reduced complex-fault resolution times by more than 20%.
Prescriptive maintenance systems are provided by a number of vendors that can customize them as needed to provide the desired data and analytics for a company’s specific equipment and processes. Reports and/or a dashboard display may present many kinds of data, such as key performance indicators. Asset management capability or the ability to connect with an asset management system and other features may be available.
The deep understanding of the health or condition of equipment and processes enabled by a predictive or prescriptive maintenance system’s IIoT sensors and systems allows management to make informed decisions about maintenance that align with the company’s operational and business needs. A prescriptive maintenance system can also provide alternatives for meeting these needs and, in some cases, automatically take action to help support uptime and maximize reliability and production.