by Bill Morosan, CET

   When you're standing in the lobby of your office building, can you feel the floor shake? Do you have any idea how much energy it takes to shake that floor and at what cost? A predictive maintenance program provides an alternative to guessing.
   Predictive maintenance through vibration analysis assesses a machine's condition to show when repairs are needed. Known also as condition based monitoring, it measures vibration and other parameters on a machine, and over time, indicates specific faults to predict when they'll become unacceptable.
   Predictive maintenance uses many measurements to monitor a machine's condition. Measurements must be good indicators of condition and be cost-effective to obtain on a regular basis. Timing is paramount. The first few sets of measurements can help with the diagnosis of pre-existing faults. An effective predictive maintenance program is based on consistent monitoring of trends in those measurements. Results must be repeated to avoid confusion between variations in machine measurement conditions.
   The usual approach is to monitor a broad range of machinery faults through measurements of mechanical vibrations. Vibration is simple to measure and provides information regarding the overall machinery condition and specific defects.

Identify problems early with predictive maintenance

Vibration measurements are normally made using a hand-held data collector with a vibration transducer. The predictive maintenance software loads a list of locations that are due for measurement into the data collector. The operator then moves around these locations taking measurements at each point. Once completed, the measurements are transferred to the predictive maintenance database.
   Then the predictive maintenance software compares all measurements against a set of alarm specifications. The development of a fault is indicated when current measurements exceed alarm specifications. Locations where alarm levels have been exceeded are grouped together into an Exception Report, which forms the basis for all further action. Plots of measurements showing alarm conditions are also produced to support this report.
   Starting a predictive maintenance program requires management commitment, otherwise resources are unlikely

to be maintained, hurting the chances for success. You can purchase the required hardware and software then learn how to do the program in-house, a major commitment in both manpower and dollars. The alternative is to hire a consulting firm that sets up and operates the predictive maintenance program for you.
  Implementing a predictive maintenance program can maximize equipment potential. Companies can realize a higher return on invested capital by increasing the running time. Problem machines are identified early without major overhauls. In addition, maintenance costs are lower but productivity is higher. Companies may be able to reduce inventories of spare parts because projected trends can predict failure and parts can be ordered as required. Predictive maintenance programs can improve safety because equipment is repaired before reaching dangerous conditions.

Bill Morosan, CET, is president and owner of Predictive Analysis Ltd, 587 Country Club Mews, London, Ontario,
N6C 5R1, (519) 685-1213