Adding Continuous Monitoring
To understand the cause of the vibration spikes, a continuous vibration monitoring system was installed by the consultant. The system consisted of the data module (DM) that was installed near the motor and pump, with combination triaxial vibration/temperature sensors connected by long extension cables to the DM. The DM unit transmits a radio signal to the communication module (CM), which sends this signal to the tower and from there to the network.Plant technicians, maintenance personnel and other operators can view the data directly on their computers via the Internet from any location. As seen in Figure 2, vibrations spiked from a low level of approximately 0.05 inch per second (which is what the monthly single-measurement data tracked) to a high level acceding the warning and close to the 0.5-inch-per-second alarm level.
The vibration and temperature sensors were installed at several locations on the motor and the pump, but the high level of vibration was recorded only at the top of the motor. The lower part of the motor and the pump experienced no spikes and had low vibration readings, close to 0.05 inch per second. The full-spectrum analysis signature is shown in Figure 3.
From the FFT plot, the main high spike frequency is at 196 hertz, 44 times the running speed value of 267 rpm (4.4 hertz) at the time of the test. While the problem appeared to be located in the upper part of the motor, the specific source of its excitation was unclear. Possible culprits were:
- Motor electric (rotor bars)
- Variable frequency drive (VFD) harmonics and noise filtering
To determine the specific cause, an additional bump test was performed on the idling pump (see Figure 4). This test identified the natural resonance frequency of 196 Hz as the root cause. However, which component caused the resonance was unclear.
Additional tests at different positions were conducted along the entire length of the unit. These tests indicated that the highest vibrations emanated from area around the motor’s lifting lugs. There are 11 short steel ribs at the upper part of the motor housing, and two longer lugs that are used to lift the motor. The dimensions of the lifting lugs are 18 inches by 12 inches by 1 inch.
Finite element analysis (FEA) of the lifting lugs was performed to see the natural modes of frequency resonance. The first three shapes of the resonance modes are shown in Figure 5. In a simplified model, the lifting plate was estimated to be a 2-inch thick plate that was 18 inches long by 12 inches high. The exact method of the plate’s attachment to the motor frame was difficult to see without disassembly.
If most of the attachment boundary were along the short edge of the plate, the natural frequency of resonance would be somewhat less than if the plate were attached mostly along the wide edge. The true resonance was probably in-between these constraints. The mostly short-edge FEA analysis predicted 102 hertz as the first harmonic and 345 hertz harmonic later.
The bump test value of 196 Hz fell within this range, indicating a strong possibility of the predicted structural resonance of the upper side of the motor lifting lug structure, although a more accurate model would need to be constructed to get a more accurate correlation.
Figure 5. FEA results
Vibration Trending & Monitoring Benefits
Vibration trending analysis aided by the FFT spectral analysis provide a valuable tool for predicting and anticipating developing pump issues before they evolve into problems.
Such predictive maintenance also helps identify system-related deficiencies and allows for proper adjustments to fine tune and optimize overall equipment operation.
Collectively, these methods result in significant savings because of the avoided costs of catastrophic failures and results in a more informed and educated team to maintain and operate critical equipment.
Pumps & Systems, April 2012