There is a difference between using a predictive technology for troubleshooting and using it in the context of a condition based maintenance program. In the latter, it is extremely important to define and standardize test procedures in order to insure that the machines are always tested in exactly the same way year after year. This allows data to be trended and analyzed in an objective fashion. This is a fundamental key to long term program success and viability. Defining standard test procedures also makes it easier to train new technicians to follow these procedures and keep the program running year after year independent of in house “experts”.
Test procedures should include a diagram of test locations, i.e. where to attach sensors to the machine. Sensor mounting pads or targets are highly recommended and a guide to their locations is required in order to replace them correctly when they fall off. The test procedures should also include instructions for how to test the machine, i.e. load, RPM and other external factors that may be controlled in order to achieve repeatable test conditions.
Good organization of a condition monitoring program is a key to success. Not only do there need to be precise instructions for how to test a machine, there also needs to be a road map that describes who collects the data and when, who analyzes it, who receives the reports and how the reports are acted on. In addition to this there should be a set of procedures designed to measure the accuracy of the diagnosis and the value to the plant of having that information. Oftentimes roles and responsibilities may span different departments in the plant and disconnects are more common than they should be. A good plan for running the program, acting on results and measuring the value of the program itself is extremely important if one wants to achieve good results.
In a condition monitoring program, repeatability and trending are the prime ingredients to success. If done correctly, the data analysis process should also become a procedure. Many people confuse condition monitoring with troubleshooting machines and they test the machine a different way each time, spend a great deal of time analyzing and playing with the data and never gain the benefits of trending. Those who plan their programs well and employ standard test processes and procedures will benefit from the ability to trend and their procedure for analyzing data will often be as simple as visually comparing new data to a baseline and noting any deviations. This process can even be mostly automated if good procedures are in place. In a well developed program, an analyst will approach the analysis process in a well defined and repeatable way that includes how he or she displays the data vs. alarms, how he goes about looking for mechanical fault indications and how he reports his findings.
