Reduce dreaded pump problems or failures with condition monitoring
To avoid costly, unplanned downtime, leveraging condition monitoring in a reliability plan is essential
- Learn about condition monitoring and its techniques and benefits.
- Understand the most common pump failure causes and how condition monitoring can help prevent them.
- Determine which technologies are most effective for specific failure modes.
Condition monitoring insights
- Condition monitoring is part of a larger proactive maintenance strategy to identify potential problems while still manageable.
- As the heart of a system, pumps are critical for efficient operation and a smart application for this cost-saving tool.
Pumps perform critical liquid transfer applications in many industrial facilities, from heavy industry like mining and pulp and paper to lighter industry, such as municipal recreation. An unplanned critical pump outage can be costly in resources and downtime.
Failure pattern studies show that industrial equipment, including pumps, have only a small percentage of equipment failures due to aging — the vast majority occur early in the asset’s life. Thus, time-based maintenance will rarely be the most effective method for improving plant reliability.
Plant managers can use condition monitoring to help identify precursors to common pump failures and allow a maintenance program to move from preventive maintenance to proactive maintenance.
What is condition monitoring?
Condition monitoring is the process of observing machinery, such as pumps, to detect any signs of potential failure before it occurs. This is done by collecting the machinery’s performance data and analyzing the trends to detect anomalies. It has been in practice since the 1850s when railway engineers would inspect and monitor the conditions of locomotive wheels using a small hammer. Compromised wheels have a different sound from intact wheels, so by tapping the wheels with the hammer, engineers could analyze the state of each wheel and provide corrective action, such as replacing a wheel.
Throughout industrial history, there have been several evolutions of condition monitoring, beginning with walkaround routes that depended heavily on resources and having the same people working at the same facility for years to gain familiarity with the equipment.
Facilities now take advantage of the decreasing cost of smart sensors and the improved connectivity within plants. The current goal is to capitalize on plant staffing to make the best use of resources while working to minimize downtime. Combining more modern techniques — such as oil analysis, vibration analysis and thermography — will maximize the ability to identify abnormalities well before problems occur. The potential failure (P-F) curve (see Figure 1) shows the P-F interval, where various condition monitoring techniques are performed early to maintain the equipment in the “P” portions of the curve.
Common condition monitoring techniques
Condition monitoring is practiced at various levels at most industrial facilities in North America. Its power comes from how the data is trended and analyzed. One can employ a route-based maintenance system to monitor the health of critical assets on an annual/quarterly/monthly basis.
However, that requires consistency and dedicated, well-trained resources to ensure that data is collected properly. This method identifies trends as a snapshot in time.
For more powerful trending, plants have begun to employ online condition monitoring. Online monitoring programs collect data at certain intervals throughout the day and can have alarms to alert the appropriate personnel when the asset is operating outside of preferred conditions. These programs can be set up completely inside existing plant data infrastructure by connecting to a human-machine interface or a programmable logic controller (PLC) to push data into a plant distributed control system or historian. The data can then be sent to the cloud via Ethernet or cellular. Many companies have analysts to provide insights on the data; this can be included as part of the program.
Prevalent online monitoring methods used in centrifugal pump preventive maintenance include:
Vibration analysis detects abnormal vibration patterns in machinery, indicating problems such as misalignment, unbalanced components, lubrication issues and bearing failure. Vibration analysts can review vibration trend plots to pinpoint the problems by comparing the patterns that develop in the frequencies captured against known patterns associated with common failure modes (see Figure 2). The most common techniques for vibration analysis are route-based data collection using handheld vibration collectors and online monitoring, either wired or wireless.
Thermography is a technique that uses specialized cameras to detect upset temperature patterns in machinery, such as bearing failure, lubrication issues (i.e., over or underlubricating), seal flushing issues and overloading. Thermography can also identify process issues like plugging in piping, performed with handheld or online monitoring.
Oil analysis identifies abnormalities in the oil, indicating problems such as contamination, wear and water ingress. Tools include handheld oil analysis kits (to be used in-house or by a reputable third-party lab), handheld oil analysis devices (such as particle counters) and online monitoring systems.
Ultrasonic data identify early signs of wear and abnormalities in slow-moving equipment. Ultrasonic greasing is invaluable in determining the correct amount of grease to apply — increasing the asset’s life. Ultrasound can be performed with various handheld devices (depending on the application) or online monitoring.
Electrical monitors, such as current transmitters, identify how much load the motor is pulling. If a motor is working too hard or not hard enough, there could be a performance issue with the pump or the motor may be sized incorrectly. Electrical monitoring can be performed via handheld clamp-style devices, inline measurement devices or online monitoring.
Typical pump failures
Common failures among centrifugal pumps include:
Seals are critical to the performance of many centrifugal pumps. Depending on the pump type and application, various seals can be chosen, such as packing and mechanical seals, including component and cartridge types. Seals can fail for many reasons, including improper installation (such as misalignment), upset conditions in the pump and contamination; another factor could be improper application of a seal, such as the seal not being designed for abrasive applications.
The more sophisticated the seal, the more likely condition monitoring will help us identify anomalies. While severe seal failures can be identified visually if leakage comes from the pump, online monitoring can identify anomalies before they become full-on failures. Leaks from seal failures can damage the pump and surrounding equipment.
Vibration analysis and thermography can identify seal failures. A compromised seal will have an identifiable vibration pattern detectable by vibration analysis and could cause the pump to overheat, identified by thermography.
For seals with an auxiliary flush system, oil analysis can ensure the new oil is clean and dry before application. This will provide optimal life for the seal faces. Oil analysis can also identify wear on the seal faces by monitoring for wear metals.
Bearing failure is another common pump problem stemming from roots like mechanical seals, including assembly misalignment, improper lubrication and upset process conditions. Pump bearings provide the primary support to position the impeller and maintain the concentricity of the running clearances. Centrifugal pumps typically have two bearings but may have more. Journal, rolling element, tilting pad, magnetic and hydrostatic bearings are all used in various scenarios. Depending on the size and speed of the pump, bearings may be oil lubricated or greased.
For greased bearings, ultrasonic greasing is an excellent tool to ensure the correct amount of grease is applied. Many companies offer grease guns with integrated ultrasonic testing. It is a relatively simple process for technicians to be trained to use properly and can dramatically increase the life of greased bearings.
Regardless of the lubrication method, vibration analysis and thermography are instrumental in identifying bearing faults. By monitoring the vibration spectrum, bearing faults can be identified well before needing to replace the bearings. A skilled vibration analyst can often catch situations like misalignment or looseness that demand correction to maintain bearing integrity and ensure the equipment continues running.
Corrective actions include tightening baseplate fasteners, adding/reducing lubricants and realigning the shaft. If corrective actions cannot be applied, the advanced notice from monitoring allows for scheduling the repair during a convenient planned outage.
Cavitation is a common problem when the pressure in the pump drops below the vapor pressure of the fluid being pumped. This causes the fluid to vaporize, creating bubbles that collapse and damage the impeller and other pump components; this could be either suction cavitation, where there is not enough net positive suction head for the application, so the pump is starving for water or discharge cavitation, where the pump is not designed to meet the actual discharge conditions
Aggressive cavitation often sounds like rocks moving through the pump, and vibration analysis can identify collapsing air bubbles before one can hear it with the naked ear. Corrective actions for cavitation can include lowering the pump suction or raising the feeding tank.
Impeller damage can be caused by cavitation, corrosion and erosion. Impeller damage can result in reduced pump performance and increased vibration, damaging the surrounding piping. While performance loss is a notable indicator of impeller damage, it is a lagging indicator. Leveraging online vibration analysis gives a leading indicator, allowing us to correct impeller damage before it causes performance loss.
Motor failure resulting from overheating, overload and bearing failure can reduce pump performance and increase downtime. It can be detected using a current transformer to identify anomalies in the motor draw. Vibration analysis and thermography can also identify motor failure.
Placement of online sensors
If there is interest in monitoring pumps continuously, deciding where to place the sensors depends on the size and configuration of the pump and motor combination.
Each motor and pump bearing should have at least one vibration sensor, potentially more depending on the size, speed, criticality and maintenance philosophy. For example, for smaller motors (below 50–75 horsepower) that don’t get rebuilt, a single sensor could identify when to start planning for a replacement. Motors sent out to rewind shops for complete rebuild should have a sensor on each end to detect failures as soon as possible.
In rare cases, such as where there are redundant lines, there may only be a single sensor on the motor. More data points are better for vibration analysis, so having sensors at each bearing will allow experienced analysts to identify concerns more accurately.
Auxiliary lubrication systems benefit from lubrication monitoring (particle counter, moisture sensor, oil quality sensor) placed upstream of the filter and a vibration monitor each on the pump and motor. An additional layer of protection is flow control connected to a PLC to shut down the main pump if the seal isn’t getting fluid. Auxiliary cooling systems would be like auxiliary lubrication systems, except the fluid monitoring may change depending on the fluid in the system.
Condition monitoring is an essential tool to include in a reliability plan. It can increase operating profits by decreasing repair and downtime costs. Given the criticality of pumps in industrial settings and companies’ continuous supply chain problems, unplanned downtime can prove very costly. Many cases show that the return on investment of a well-designed condition monitoring system is as short as four months. That means that the cost of not using condition monitoring could be three times (or more) than the startup costs of the system.