Predictive maintenance (PdM) is a proactive, data-driven maintenance approach to monitoring real-time equipment operating conditions, to predict possible equipment failure and establish efficient maintenance schedules. It continues to gain popularity over traditional methods, such as six sigma and lean management, as a tool for driving industrial efficiency. PdM systems are capable of reducing maintenance costs by 12%, improving equipment uptime by 9%, extending equipment lifetime by over 20% and minimizing health, environmental and quality risks by 14%.
Here are some factors that plant managers and technicians should keep in mind when switching to predictive maintenance.
1) Infrastructure requirements. Predictive maintenance relies on sensors to collect real-time data from plant equipment. The sensors are capable of monitoring and detecting the slightest changes in equipment operation. Condition-based monitoring utilities collect data like temperature, pressure, vibration, rotational speeds, etc. Adjustments then can be made to prevent small errors from expanding into major failures capable of crippling plant processes.
Converting sensor data into alert signals and useful information requires robust infrastructure that incorporates software and hardware platforms. These systems should be able to track and analyze data in real-time, managing data from the multitude of sensors.
Before switching to a PdM system, identify plant needs and select appropriate sensors and system infrastructure that will be able to collect, convert, and manage critical machine data.
2) Data management. Huge amounts of data can be collected at any given instant by sensors fitted on critical plant equipment. This data can be overwhelming in its raw form. When switching to a PdM system, it is important that the system is capable of filtering through and mapping out critical data from all sensors.
The PdM algorithms and predictive models should be able to predict failures and failure modes from analyzed data. The data-management platform is expected to provide a user-friendly interface, through which complex data is displayed in simple and understandable forms (graphs and pictorials), which can be acted upon to perform proactive maintenance.
PdM systems should be capable of collecting data on critical assets continuously, without affecting plant operation or the technical capabilities of the data-management platform.
3) User adoption and training. Switching to predictive maintenance means that the maintenance staff will be exposed to new systems, software, and procedures. A PdM system is bound to disrupt existing maintenance practices, and it is important that staff understand additional software needs and changes that arise from its implementation.
Smooth transition to a PdM maintenance approach will depend on user acceptability and clear understanding of systems by maintenance teams. Additionally, creating data models that will be used to predict equipment breakdowns is a complicated task and often will call for employing additional staff, such as data scientists and reliability engineers.
4) Ease of monitoring and alert generation. Before you switch to predictive maintenance it is critical that you adopt a system that is easy to monitor. Apart from breaking down complex data sets into understandable analytics, detection of errors and generation of appropriate alerts is equally important. PdM systems use alerts as feedback systems for effective communication between machines and maintenance teams.
A PdM system has to be capable of detecting and averting false alerts, generating structured alerts from specific data points. Accurate alerts that pinpoint location or root sources of errors are preferable to generic alerts. Creating a database for frequent alerts and error sources forms a basis for companies to develop a functional failure mode and root cause analysis model.
5) Dealing with security issues. Predictive maintenance ties industrial assets to various cloud computing platforms, enable by IoT connectivity. These platforms are necessary for improving real-time data analysis and storage of equipment performance metrics.
Once plant equipment is connected to these platforms, maintenance moves beyond physical protection, bringing on board cybersecurity strategies toward management of industrial assets.
Priority should be given to safety of IoT and cloud computing platforms adopted for PdM systems. There are several instances of attacks and security breaches on critical infrastructure facilities across the world. There have been reports of severe cyberattacks that have predominantly targeted the Industrial IoT technology. It is important that a PdM system incorporates robust security features against possible internal and external attacks.
6) Starting with a pilot project. Before you switch to a predictive maintenance model, it is important to lay down a strategic implementation plan. Implementation planning identifies critical plant assets and maps out possible implementation bottlenecks that may arise from adopting a predictive maintenance system.
Transitioning to a PdM system will take less time if a pilot project is implemented in advance. The pilot project can be used as a tool to train staff, test system security, and optimize predictive algorithms.
Many industries are shifting from reactive maintenance strategies to predictive maintenance. This has been hastened by development of IoT technology, IoT-enabled CMMS solutions, and the fact that the cost of implementation is constantly dropping down.
It takes time for PdM strategy to be fully implemented. Challenges associated with implementation can be significantly reduced by performing prior, in-depth analysis of plant requirements, proper financial and material planning, and sufficient training of maintenance teams.
Bryan Christiansen is the founder and CEO of Limble CMMS. – an easy-to-use mobile CMMS software that takes the stress and chaos out of maintenance by helping managers organize, automate, and streamline maintenance operations.
