Carl does this by visiting each pump once a week, clipboard in hand, recording data from gauges and comparing it to past readings. He augments this data with aural input gathered by listening, and with vibration input acquired by laying hands on each pump. He combines this input with knowledge of plant operations, including expected day to day variances.
The result is an extremely accurate picture of pump health, from which Carl creates a maintenance schedule. His actions ensure each pump is serviced just when it needs to be, and not too frequently as this would drive up costs, but in all cases before failure.
Carl’s activities are critically important because studies show pumps are very maintenance-intensive equipment, suffering failures or some level of degraded operation about once a year. Fixing these problems before they occur is extremely important because reactive maintenance, where it is necessary to fix something after failure, costs 50% more than predictive maintenance, which detects and addresses problems prior to failure.
Perhaps your plant or facility has its own Carl, or maybe quite a few of them. These employees are vital to efficient operation, and you may spend quite a bit of time wondering just how you can capture their knowledge, store it and transfer it to other employees when the time comes.
And just when is this time? In the past, many of these workers would retire on a fairly predictable schedule driven by their defined benefit pension plan. Their replacements would be waiting in the wings, with staffing levels sufficient to allow training and knowledge transfer over a predictable period of time.
Nowadays, with most workers relying on 401(k) and other defined contribution plans, many are working longer, and as these workers age their retirement dates become more uncertain. Compounding this issue, staffing levels at most every industrial plant or facility are below what they once were, making it more difficult for junior workers to cross-train side by side with experienced employees. This crucial overlap time is now often an unknown interval, which could be years, or maybe just a few months.
There are two main ways to address this issue. The first is to keep doing things as they’ve always been done, but with a new person. Carl trains his replacement, bestowing a clipboard and the requisite training.
The second approach is to automate some or all of Carl’s activities by collecting data with instruments instead of manually reading gauges, transmitting this data to analytics software to generate results and optimise maintenance, and having Carl train his replacement to make final decisions.
The first method, continuing to work as before but with a new person, presents several issues. First, it’s hard to find a Carl, and it can take many years to train him or her. Second, the new employee will probably be much younger than Carl and will naturally balk at using antiquated methods of data collection such as a clipboard. Third, the required training time will be quite unpredictable as it will depend heavily on the new employee’s aptitude and experience. Finally, Carl’s actual retirement date will probably be unknown, making it impossible to judge the best time to hire and fully train Carl’s eventual successor.
The second technique is automating the activity to the greatest extent possible and practical. For pump predictive maintenance, this would mean installing instruments to monitor condition and performance, ensuring problems are detected as soon as possible so appropriate action can be taken. Typical pump parameters monitored would include flow, vibration, bearing temperature, inlet/outlet pressure, power consumption, seal fluid pressure/level, etc.
Many of these condition monitoring instruments are now available in wireless versions, which can cut the cost and time of installation by up to 50 percent. Whether wired or wireless, these instruments provide raw data to analytics software, which can be used to predict problems before they occur. Some vendors offer analytics designed specifically for evaluating pump performance, and this special-purpose software provides superior ease of use as compared to more general-purpose platforms.
Results from analytics software can then be correlated with expert opinion to indicate whether a problem exists or not, along with the recommended course of action. For example, if the pump flow decreases on a Tuesday, it’s not an issue because it’s the result of an expected regular weekly change to the product mix. But if the same condition occurs on a Thursday, it’s an issue. This type of information can become part of the analytics software configuration, increasing its intelligence and effectiveness.
Companies establishing automated equipment monitoring systems will find that new employees can be brought up to speed in weeks or months instead of years. And by applying this approach to other types of assets and equipment, an employee can become extremely productive in many areas as he or she combines the power of automation with their growing knowledge of plant operations.
