COVER SERIES | Pump System Optimization
Pump Fault Detection: Real-World Industrial IoT Case Studies
The benefits of continuous condition monitoring.
Jon Bernie | Petasense
Industrial pumps play a central role in maintaining uptime across critical infrastructure, from healthcare campuses and energy facilities to mining operations and public utilities. Despite their importance, pump failures remain a persistent source of inefficiency and unplanned downtime. In many cases, failures originate not from sudden mechanical breakdowns but from hydraulic conditions that develop gradually and remain hidden during routine inspections.
Issues such as flow turbulence and cavitation manifest subtly long before visible damage or performance loss occurs. Traditional condition monitoring programs, which rely on periodic data collection or visual inspection, frequently miss these early indicators.
Advances in industrial Internet of Things (IIoT)-enabled condition monitoring have changed this paradigm. Continuous vibration monitoring, combined with trend and spectrum analysis, enables maintenance teams to detect pump-specific defects early, understand their root causes and intervene before failures occur.
This article presents two real-world case studies illustrating how IIoT-based vibration monitoring was used to identify hydraulic degradation mechanisms in centrifugal pumps. Each example highlights pump-specific failure modes and demonstrates how early detection improved pump performance and avoided unplanned downtime.
Pump-Specific Faults & the Value of Spectral Analysis
Pump vibration behavior is influenced by both mechanical and hydraulic forces. While mechanical faults such as imbalance or bearing damage tend to produce discrete frequency components, hydraulic issues often appear as broadband energy, elevated noise floors or harmonics related to pump design parameters.
Flow disturbances, cavitation and unstable operating conditions can generate vibration signatures that vary with speed, load and process demand. Without continuous monitoring and operating context, these patterns are easily misinterpreted or overlooked.
IIoT-based vibration monitoring enables data to be collected across variable operating conditions. When trend data is paired with frequency spectrum data, analysts gain the ability to differentiate between normal process variation and developing hydraulic faults. This distinction is critical for pumps operating under variable speed or intermittent loading, where vibration levels may change significantly over time.
| IMAGE 1: Acceleration spectrum showing pronounced vane pass frequency harmonics and elevated noise floor (Images courtesy of Petasense)
Case Study 1: Identification & Resolution of Flow-Induced Vibration
Hospitals depend on reliable pumping systems to maintain continuous heating and cooling for patient care, operating rooms and critical support spaces. At a central energy facility supporting a large hospital and adjacent university campus, a variable speed centrifugal pump was monitored as part of an ongoing condition monitoring program. The pump operates intermittently, with frequent changes in speed and load to meet fluctuating thermal demand. Since interruptions to heating or cooling can directly impact patient safety and hospital operations, pump reliability is a highly critical requirement.
Observations
Following a change in operating conditions, maintenance personnel observed elevated vibration levels that deviated from historical norms. Although no immediate mechanical failure was evident, the increase raised concerns about long-term bearing life, hydraulic stability and overall system performance.
Analysis of vibration trends showed that overall vibration levels increased when the pump operated above approximately 1,200 rotations per minute (rpm) (20 hertz [Hz]). Frequency spectrum analysis provided additional insight, revealing prominent vane pass frequency harmonics and an elevated noise floor during higher speed operation.
These spectral characteristics are commonly associated with hydraulic excitation rather than discrete mechanical defects. In particular, elevated vane pass frequency components indicate increased hydraulic forces acting on the impeller and volute.
Root cause analysis
By correlating vibration behavior with pump speed and operating conditions, analysts determined that the pump was operating outside its recommended hydraulic range during certain demand scenarios. The best efficiency point (BEP) represents the operating condition at which a pump achieves optimal efficiency with stable flow patterns and minimal hydraulic stress. Operating away from BEP increases internal recirculation and turbulence, which amplifies hydraulic excitation forces within the pump.
In this case, variable speed operation combined with intermittent loading created unstable flow conditions. These conditions excited vane pass frequency components and increased overall vibration energy, even though the pump’s mechanical components remained intact.
| IMAGE 2: Vibration trend showing reduction in amplitude following corrective action
Corrective actions and results
Rather than pursuing mechanical repairs, corrective actions focused on optimizing pump operation. Control parameters were adjusted to keep the pump operating closer to BEP during normal demand cycles. Bearing lubrication health was also verified to ensure that mechanical components were not amplifying hydraulic vibration.
Following these adjustments, vibration trends showed a reduction in acceleration levels. Frequency spectra stabilized, vane pass frequency harmonics diminished and the noise floor returned to normal operating ranges. No further abnormal spectral patterns were observed.
Early detection and targeted corrective action resolved the issue without requiring mechanical intervention, extending pump life and maintaining the reliability required for hospital operations.
Case Study 2: Early Detection of Cavitation in a Barren Solution Pump
Site overview
In gold mining operations, barren solution pumps play a critical role in the gold recovery process. These pumps circulate the chemical solution used to extract gold from raw ore, returning it to the process so it can be reused for subsequent leaching cycles. Because this circuit directly supports gold production, unplanned pump downtime can disrupt processing operations and reduce overall plant throughput.
At a large gold mine and processing facility, a centrifugal barren solution pump was monitored as part of an ongoing condition monitoring program. The pump operated continuously and was supplied by upstream process pumps that had accumulated significant operating hours since their last internal inspection.
Observations
During routine vibration monitoring, analysts observed a subtle but meaningful change in the vibration signature of the barren solution pump. Although overall vibration amplitudes remained within acceptable limits and did not indicate an imminent failure, the frequency spectrum revealed a noticeable increase in broadband vibration energy compared to historical baseline data. While the condition did not appear severe on the monitored pump, the elevated noise floor raised concern due to its strong association with hydraulic instability rather than discrete mechanical faults. This prompted investigation into the broader process circuit supplying the pump.
| IMAGE 3: Vibration spectrum (left) showing elevated noise floor correlated with barren booster pump shown in flow diagram (right). This signal prompted inspection of the two unmonitored barren pumps upstream.
Root cause analysis
Based on the vibration findings, maintenance personnel expanded their inspection beyond the monitored pump to include upstream process equipment. Inspection of the two upstream pumps supplying the barren solution circuit revealed severe cavitation, with impeller deterioration observed on both units.
The deteriorated impellers reduced effective discharge pressure and flow, resulting in insufficient net positive suction head available (NPSHa) at the barren solution pump inlet. This hydraulic deficiency led to early-stage cavitation within the monitored pump, which manifested as the elevated broadband vibration noise floor observed in the vibration spectra.
Although the vibration response at the barren solution pump was relatively mild, it served as an early warning indicator of a much more severe hydraulic issue occurring upstream.
Corrective actions and results
Corrective actions focused on addressing the underlying hydraulic issue identified upstream rather than performing repairs on the monitored barren solution pump. The two upstream pumps supplying the circuit were taken out of service and inspected, at which point severe cavitation damage and advanced impeller deterioration were confirmed.
The damaged impellers were replaced, restoring proper discharge pressure and flow to the barren solution circuit. With adequate hydraulic conditions reestablished, net positive suction head (NPSH) at the barren solution pump inlet returned to acceptable levels.
Post-maintenance vibration monitoring showed that the elevated broadband noise floor at the barren solution pump returned to baseline levels. No further cavitation-related spectral characteristics were observed, and the pump continued operating without signs of accelerated wear.
By identifying the issue early and correcting the root cause upstream, the facility avoided severe damage to the barren solution pump, extended pump life across the circuit and restored reliable operation of a critical gold processing system.
| IMAGE 4: Physical inspection confirmed severe impeller damage on the upstream barren pumps
Failure Prevention Impact
In both examples, early detection enabled by IIoT-based vibration monitoring prevented hydraulic issues from escalating into more severe failures. Continuous data collection and spectral analysis provided visibility into developing conditions that would have been difficult to detect through periodic inspections alone.
- Accelerated impeller erosion and hydraulic degradation
- Mechanical seal failure
- Bearing damage due to sustained high-frequency excitation
- Unplanned pump outages and associated downtime
By identifying hydraulic issues early, maintenance teams avoided unnecessary mechanical repairs and minimized operational risk.
Jon Bernie is an enterprise account executive at Petasense with over a decade of experience supporting industrial reliability and predictive maintenance initiatives. For more information, visit petasense.com.
In This Issue
