by Todd Reeves and Stewart Bowers, Emerson Process Management

Pumps & Systems, October 2007

A technology was introduced late last year to continuously monitor vibration levels of AC induction motor-pump machine trains along with automated analysis of each measurement. This technology provides operating personnel with frequent information for those situations when maintaining pump operation was critical to process integrity.

Utilizing the capabilities of smart, microprocessor-based instrumentation, one field-mounted machinery health transmitter receives inputs from six different vibration sensors on a motor-pump train, a tachometer reading for shaft speed, motor flux input from a flux coil, and temperatures at four different locations.   

Every 25 seconds, these parameters are scanned for evidence of common malfunctions, such as bearing or cavitation problems. Since motor-pump trains tend to have similar failure patterns, trends are used as the basis for automated analysis. This approach lends itself well to an embedded analysis engine capable of recognizing common motor-pump defects. The frequent examination of vibration data against known patterns provides current verification versus data obtained periodically, where a rapidly evolving problem might be missed.

An overall machinery health value based on various physical characteristics is calculated and assigned a numerical rank from 1-to-100, with the top number indicating perfect health. By checking periodically on machinery health values, operators get a quick fix on the condition of any monitored motor-pump train.

If the automatic analysis determines that something happening internally justifies a warning, an appropriate alert is communicated to the facility's control system or maintenance station. This immediately informs operators of a changing level of vibration on an essential pumping system. 

Operations personnel applaud this development because they receive up-to-date information on the condition of essential equipment. However, a simple warning of "pump cavitating," "motor overheating," or "bearing bad" is not enough to satisfy most maintenance personnel, especially the rotating equipment specialists trained in vibration analysis. They want access to the raw data - especially that related to operating incidents - to make a detailed evaluation of what caused an alert.

Conducting a personal assessment is very important to the vibration analyst charged with predicting when a machine may fail so it can be removed from service prior to that time, thereby preventing an unexpected stoppage.

New Developments

Recent advances in the realm of continuous vibration monitoring allow for storage of a vast amount of information within the field-mounted machinery health device when a problem serious enough to raise an alert is detected. The stored data can easily be retrieved later by a staff vibration analyst using a machinery health analyzer, commonly employed to measure and analyze machine vibration on a periodic basis.

The nearby sidebar outlines the high resolution data stored by a machinery health transmitter for each alert event. This data can be quickly downloaded to a machinery health analyzer via a serial cable. Back in the maintenance shop, advanced plotting tools enable the analyst to visualize patterns in the data and confirm detection of a problem. The analyst can then recommend corrective action, including removal of the machine from service if necessary to prevent an imminent breakdown.

Types of Event Data StoredHow It Works

Late on a Friday evening, after all the maintenance personnel leave for the weekend, a low level bearing problem develops in a monitored pump. An advisory alert message is sent to the control system where the operator may or may not take notice, as it is still a low level alert. By Saturday night, the condition worsens, the machinery health value drops to 60, and a maintenance level alert is generated and transmitted. 

The operator, now aware of a potential problem, sends a note to the maintenance department to check out that particular pump as soon as possible. On Monday morning, a maintenance technician goes out to the machine in question and retrieves the stored data from both alerts that reveal exactly what happened.

If the pump goes into failure mode during the weekend, the duty operator receives a failed alert message and immediately switches over to a backup pump. Under normal circumstances, maintenance can do little but go out and remove the pump that was shut down. But with access to all the data on conditions leading up to the failure, including the original baseline measurements for that pump and trend data for the previous 30 days, the vibration analyst can conduct a post-mortem examination to determine what needs to be done to fix the machine and possibly how to prevent such problems in other similar pumps on-site.

An Actual Case

A problem was developing on the outboard bearing of a digester surge pump in a large paper mill. The machinery health transmitter detected the degradation and sent an alert message to the process automation system, indicating the bearing problem had reached a point where the pump health was so low that immediate action was needed.

Since the data on the condition that caused the device to generate the alert message had been saved, the mill's vibration analyst was able to walk out to the device, connect his portable machinery health analyzer and retrieve the key information along with the historical trend data.

An examination of the velocity spectrum in the machinery health management software (shown in Figure 1) indicates the presence of a fundamental outer race bearing frequency at 5.4 orders of the shaft frequency.