Summary

Centrifugal pumps often have vibration problems. Even brand new pumps suffer from this malady.

Details

Several conditions tend to aggravate the vibration problems in pumps. These include:

•     Vertical installation
•     Running under conditions that vary from the pump’s design point
•     Running at variable speed

Vibrations can occur in the radial, axial and even torsional modes.  When a pump has a vibration problem, the frequencies of that vibration are important. It is not enough to just consider the overall vibration level.

Looking at the frequency distribution of the vibration can help us determine whether a problem may be due to:
•     Resonance (critical speed)
•     Piping induced vibration (including vibration transmitted from another machine)
•     Bearings
•     Seals
•     Lubrication
•     Fluid flow through the pump
•     Mechanical unbalance
•     Hydrodynamic unbalance
•     Cavitation
•     The drive motor

If the pump is properly leveled, aligned and connected, it should run dry with almost no vibration. Under these conditions, high vibration at 1 X rpm would indicate a problem with unbalance or a bent shaft.

Impeller and Shaft Unbalance

One thing often forgotten is that many pump impellers are cast and are not uniform from vane to vane. This gives rise to the effect that the pump is in balance when run dry but vibrates under load due to this “hydraulic” unbalance (the mass of the pumped fluid is not uniform from chamber to chamber). This is the reason to run the pump dry first — to determine if there is an impeller unbalance or bent shaft. If the pump is OK dry but not wet then we can suspect that the impeller is not uniform.

Misalignment between pump and motor is usually characterized by vibration at 2X rpm and is normally higher in the axial direction. The 2X vibration may be accompanied by 1X and sometimes 3X vibrations.

When the gap between the impeller and housing is very small (for efficiency) the dominant vibration can be the vane passing frequency (RPM x number of vanes). Any runout in the impeller rotation would add a 1X vibration to that.

The close gap mentioned above can also result in rubbing which produces both noise and vibration. There will normally be a harmonic train of frequencies (1X, 2X, 3X, 4X…) since the rub generates a broadband noise once per revolution.

Looseness can also result in a rub but the effect is unstable. Looseness will often show up as an unstable 1X vibration. There will often be a non-synchronous vibration as well.

Another cause of vibration a mounting foot that is not naturally touching the mounting surface. If this is not shimmed with a rigid spacer, the foot will bend and this will cause a condition known as “soft foot”.

Vibration problems on larger pumps with plain bearings can get to be complex. Measuring vibration in two directions simultaneously so as to determine the actual shaft orbit may be required. This is not the type of application for an inexperienced person.

If unbalance or misalignment are not present, where do we look for the problem?

Resonance. The next step is make sure that various natural resonant frequencies of the pump and various things attached to it do not coincide with the operating speed of the pump.

Natural Resonant Frequencies and Pumps

A good starting point for diagnosis would be to check for resonances with everything connected but the pump not running. By measuring the vibration while the pump is being excited by tapping with a rubber mallet or similar nonmetallic tool, we will excite natural frequencies from the piping, foundation, pump casing, and even the impeller on the shaft.

Such resonance vibrations may show up in one, two, or all three axes.

There are many remedies for resonance problems. For example, if the output piping was resonating due to excitation by the pump, we could change the stiffness of the pipe mounting or add a flexible piece of pipe to fix the problem.

Resonance often plays a significant part in a new installation because of many unknown degrees of stiffness. These usually get modified empirically during startup and the pump will run fine until there is an overhaul and suddenly resonance is back. Often a pipe is out of position — perhaps the pump was replaced with another that was almost the same but not identical, and pipes became stressed in the process of coupling.

Flow-Related Problems

Often a pump is run at a reduced flow by throttling the inlet. This creates low-pressure areas where cavitation can occur. Cavitation is characterized by high frequency vibration (up to 20 – 70 K Hz) and below 1 X vibration. If the pickup (sensor) is not mounted very rigidly to the housing it will not be able to “see” the high frequencies involved. A bolted connection to a machined surface (with a thin layer of grease to fill any voids) is the best. With a lightweight pickup, using instant glue or epoxy may be adequate. A handheld pickup will not transmit over 1 KHz and a magnetic base will not normally couple to more than 2K Hz.

A pump is designed for a specific flow rate and pressure. This is called Best Efficiency Point or BEP. Running below 50% BEP will result in audible rumbling and low frequency vibration, with peaks near, but not at 1X rpm. Under these conditions a pump is not energy-efficient and consideration should be given to using a smaller pump. The pictures illustrate BEP and variation of vibration correlated to BEP point of the pump.

Summary

When approaching a vibration problem, investigate unbalance, misalignment and resonance in that order. Run the pump dry and under load. Look at alignment of couplings, rigidity of mounting and alignment of inlet and output connections. Check that the pump is running close to BEP and whether it can be controlled to produce the required flow with a different setup (change RPM or output flow valve). Ensure that the inlet valve is allowing adequate flow to avoid cavitation. Since there are many possible causes of vibration, it is often a matter of careful checking to eliminate the potential problems one by one and not jump to any hasty conclusions.

Derek Norfield
Director of Applications
Datastick Systems, Inc.