Reliable Predictive Maintenance systems

Conventional Predictive Maintenance (PdM) uses vibration analysis to provide early warning to predict imminent failure of machinery to maximize asset uptime, improve product quality and reduce cost. Each machine has a unique vibration signature based on its design. PdM is a process and uses analysis charts and graphic patterns like a heart EKG. A proper analysis of the signature provides a wealth of information. This is accomplished by installing vibration sensors at key locations on machines to monitor the condition. The data is analyzed using proprietary software to assess the machinery’s health. There is an inherent problem with this process in that it requires an experienced and costly analyst and considerable time in some instances to diagnose the issues. The process is based on thirty-year-old technology and is labor intensive.

There are two approaches currently being used. First is periodic monitoring (roughly 85%) using portable data collectors and software for analysis. And the second is online continuous monitoring of critical and important assets (about 15%). The result is trended over time to detect a significant change in asset performance. In the first case, the analysis is based on periodic snapshot of an asset. It is relatively inexpensive but unreliable. For continuous monitoring permanent sensors are installed at key locations. The information is continuously sent to a central server using a proprietary network. Continuous monitoring is obviously advantageous but prohibitively expensive due to the significant cost of sensors. In addition, the current analysis method (algorithms) is mathematically limited and cannot provide self-diagnostics.