In 1989, I was lucky enough to be part of a small pilot program of vibrational analysis and condition monitoring of rotational equipment in the power plants of Navy ships. I was a nuclear electrical operator on the USS Nimitz and was “volunteered” to spend several weeks of training on operating the data-acquisition equipment and interpreting the results of the processed data to implement a predictive-maintenance program.
The benefits of this new system were easy to identify and amazing at the time—instead of experiencing catastrophic bearing, motor, pump or gear failures that had to be reactively fixed while underway, we could monitor the tell-tale signs of upcoming issues and deal with them during routine maintenance periods, or at least address the issues prior to complete failure and prevent any secondary conditions that arise, such as a motor-bearing failure that leads to a fire. After all, a fire on a Navy ship is a very big deal.
Data acquisition was done by manually connecting a piezo assembly of the data-acquisition (DAQ) box to surface-mount pads on motors, pumps and gear boxes around the power plant, scanning the appropriate barcode representing the make and model of the part and waiting for 15 minutes or so for the data collection to finish and then moving on to the next sampling pad and repeating the process.
Then, the DAQ box was taken to our brand-new IBM XT computer, where vibrational analysis software processed the data and compared the results to an enormous database of each target’s bearing models, number of motor windings and pump impellers to give a rough diagnosis based on any peaks of concern for that particular component.
It was quite a cumbersome process, to say the least, but a game changer. It was straightforward to detect issues with bearings, motor windings, motor fan blades, centrifugal pump impellers, gear damage or wear, brush-to-commutator issues or general imbalances of rotating equipment.
A machine builder wishing to enable condition monitoring into current OEM equipment has the benefit of several generations of improvements in all aspects of the technology. Instead of manual sampling and off-line processing, the entire condition monitoring feature set can take place automatically and continually, with automated notifications to operators or engineering.
Just as they were made in the past, decisions about when corrective maintenance is required can be made by engineering, depending on the actual values detected, or on measured trends that indicate upcoming problems if not corrected.
Additionally, condition-monitoring technology is not limited to rotational equipment, although that is still a major focus and gives a very large benefit for the implementation effort. The benefit of more than 30 years of advancements in control hardware allows access to parameters such as motor and drive temperatures, currents and voltages, automation controller motherboard temperatures and voltages.
Even the central-processing-unit (CPU) fan speed can be monitored to determine if it is dirty. Strain gauges and accelerometers can determine if overall equipment is experiencing excessive jerk or frame vibration, which can be an indication of wear of slides or excessive stack tolerances of the mechanical assemblies.
To implement condition-monitoring (CM) features on new equipment, the OEM will need to plan the implementation as part of the overall control system, taking into account the architecture limitations on existing designs or change existing machine-control design to better facilitate the additional inputs and processing needs of the analytics.
1. If the existing control-system controller is scalable, and with a high-speed, low-latency fieldbus, there may not be a need to drastically change the control components. Numerous control hardware vendors make I/O specific for measuring vibration, stress, strain or other CM parameters. Using a high-speed fieldbus, the raw data can be sent directly to the main controller, either in the same fieldbus or a separate, dedicated fieldbus segment, where the data is processed in the primary automation controller. The controller obviously has to have enough spare processing power to handle this, which is where scalability of the controller comes in handy. Alternately, using multi-core processors and dedicating a core or two to the condition-monitoring analytics makes the task of crunching these highly complex algorithms invisible to the main controller’s loading.
2. Alternately, if the existing control system design doesn’t allow the above implementation for one reason or another—OEM standards, certification requirements—the use of dedicated condition-monitoring edge-computing devices installed closer to the devices being monitored can relieve the main controller and fieldbus from the task. Any issues or trends detected can be communicated via messages in the fieldbus, a separate network or wireless communication to the operator, maintenance department or supervisory system.
Giving the customer condition-monitoring features as part of the standard control system or even the option of a “condition-monitoring package” to the control system can be a big value-add for choosing one OEM over another. The end customer benefits, including reduced downtime, increased reliability and increased quality of ware, will likely be reflected to the machine builder as a happier customer and increased sales.