by Chris Armitage
When we think of Reliability and ensuring that assets perform when required, the tendency is to migrate towards those technologies and software programs that promise early detection, efficiency of data collection and robustness of data analysis. Just as we periodically evaluate Preventive Maintenance (PM) procedures for applicability to current asset conditions as well as proper frequency of execution, we should also apply the same principles to the judicious use of monitoring technologies.
When a vibration program is initially established, most experts recommend collecting data on a monthly frequency to rapidly gather trend information to establish asset conditional behavior and to determine defect propagation rates. However, since vibration analysis is passive in nature the belief is that more is better. We get lulled into a sense of security that we will always know the material status of assets and will have ample time to plan, schedule, and execute preemptive actions to avoid a functional failure. Therefore, we take data because we can but not necessarily because we should. Again, since it is passive, collecting more data does not do any harm. Wrong! Any scheduled action that is performed that has no benefit to the asset is a waste of manpower that could be used elsewhere.
The use of vibration analysis is nothing more than a much more detailed inspection of an asset’s health. I can detect issues at very early stages thus allowing for early planning and scheduling of repair actions. However, just like a visual inspection, the frequency that it is performed should be based on the observed asset behavior and the rate of change in degradation. For example, a paper dryer roll on a paper machine turns at approximately 200 rpm. The bearings are double row spherical bearings with oil flow lubrication. Barring any outside events, the bearing will last 50+ years. Even when a defect has occurred, it may take 6-18 months before propagation of the defect requires bearing replacement. Thus, if initial program setup starts out at a monthly frequency for data collection, once performance trends are established, an extension to quarterly collection can be implemented with little increase in risk.
Vibration analysis is designed to identify wear related conditions such as defects, imbalance, misalignment, etc. It is not designed for transient or event related (random) occurrences. The only way to successfully safeguard against random events is through the use of on-line continuous monitoring.
Therefore, just as we optimize PM procedures based on operational experience, we should optimize data collection frequencies based on observed conditions and trends.
Recently, we evaluated a client’s vibration program that was presently set on all data collected monthly. After looking at asset trends over a year’s time, the resulting adjustments to extending some frequencies to every 2 months or quarterly, resulted in a 49% decrease in obligated labor to execute the coverage of the subject assets. These extensions did not incur unmanageable risks as they were based on established trends and well documented data. This allowed the client to free up manpower to collect data on additional assets in the plant, thereby, expanding the overall vibration program which was previously limited because of resources.