Unless your company likes replacing perfectly good parts or experiencing preventable failures, your maintenance program is data-centric. And of course, a predictive maintenance program is data-centric. But can you trust your data?
One way that companies save money is to keep training costs within budget. But those budgets are typically not in line with actual training needs, so in reality there are cost overruns elsewhere.
Consider the case of a medium-sized manufacturing plant. They decided to bring their thermography in house, to “save money.” So they bought a thermographic camera that was a little above entry level (hey, if you’re going to save money, be sure to skimp on test equipment).
Then, to save even more money, they opted not to send anyone to training. It seemed obvious enough how to use the camera, and the previously engaged testing firm’s thermographer didn’t seem to have any problems using his camera. So why waste money on training?
The problem is the untrained user got bad data. Thermography on switchgear is hugely complicated by all that reflective metal. A thermographer with the right training can, however, get good data. Someone without the training is nearly guaranteed to fail. And that’s what happened. The “thermographer” failed to see the bad connection, and that connection subsequently failed—as in melted.
The plant was down for four days. How much money do you think they “saved” by not spending enough to ensure they have good maintenance data?
Training people to be able to correctly obtain good data is one part of the challenge. Another part is asking them for the correct data in the first place. Toward that end, it’s always better to ask for specific measurements (e.g., “What is the voltage?”) rather than to ask for a condition judgment (e.g., “Is the voltage OK?).