May the force stay with you

Why monitor the condition of pumps? The simple answer is that any piece of equipment is worth monitoring if the ability to predict its failure gives greater financial benefits than the cost of monitoring. And the same criteria should be applied to pumps as to any other piece of equipment. Methods of justification as to whether the pump should be monitored range from the simple `will it get me out of bed if it fails?' to the more sophisticated `failure mode, effect and cause analysis' (FMECA) types.

At ICI Runcorn, the Halochemicals Reliability Support Group has been monitoring the many pumps on site for some years and we have drawn several conclusions about procedures. Pumps lend themselves to monitoring. If we consider their main functions and operations, we observe that they rotate or reciprocate, they have bearings and fluid flow and, above all, they fail.

Established systems can monitor the condition of the bearings, rotating elements, hydraulics and seals. Various vibration-based techniques are available to keep an eye on the almost universal rolling element bearing.

Methods such as shock pulse, spike energy or the envelope technique give good results. All are based on the processing of an acceleration signal from the vibration transducer. At Runcorn we employ the envelope technique.

In Figure 1 is the trace of a bearing in good condition, in Figure 2 the same bearing in poor condition and the photograph shows the actual damage. Both overall levels (in dB) and the peak-to-peak heights give clear indication of the amount of damage the bearing is suffering. The sensitive technique and our experience of the equipment helps us to determine whether total replacement or just lubrication will remedy the situation.

The above technique is not suitable for plain bearings and slow speeds - below 300rpm. For badly worn plain bearings the appearance of the half speed whirl (Figure 4) is a key indicator of the (usually terminal) condition of the bearing. We measure this by velocity readings from the transducer.

Slow speed rolling element bearings (REBs) occur on few pump types and are typically difficult to monitor. The development of portable acoustic emission devices appears to present a way forward in the assessment of slow speed bearings, both REB and plain bearings. Figures 5 and 6 show a good and a bad REB respectively. Similar profiles are obtained from plain bearings.

This capability is proving useful for the increasing number of canned pumps being installed, where plain bearings are at the core of the design. The device uses a tuned sensor to detect a particular high range of frequencies (60kHz plus) to generate a reasonable output (in volts) which standard vibration meters can register, then process and display the results. Reading and diagnosing these results is still at an early stage, at least for us at ICI Runcorn.

Rotating element condition

The standard techniques for vibration monitoring give a good indication of two main parameters; balance and alignment. The use of velocity readings, correctly applied and interpreted will give adequate indication of these faults. A more precise diagnosis can be achieved if phase data are derived. If vibration levels are high enough to cause concern and could be attributable to the system being either out of balance or mis-aligned, then the pump should be attended to anyway.

To verify alignment, an in situ alignment check takes little time (see overleaf), particularly if one of the proprietary laser systems is employed. All centrifugal pumps should be balanced during overhaul. To find by means of vibration analysis that a newly installed pump is out of balance and mis-aligned would imply that one's condition monitoring is better than one's maintenance.

Centrifugal pumps generate unnecessary vibration and noise if running in poor hydraulic conditions. Increased velocity readings from running at either of the `wrong' ends of the pumps tend to curve, indicating cavitation.

The forces generated in a single discharge centrifugal pump running off curve give rise to the so-called vane pass. While this level of vibration detected externally is low, the internal forces are quite high and are detrimental to good seal life, particularly on weak pump designs.

Cavitation is difficult to identify readily with standard vibration transducers because the high frequency generated (10kHz plus) gives low output from the accelerometers used. Acoustic emission readily detects the small implosions created by cavitation and which can also be sensed by the human ear.

Seal condition is probably the `philosopher's stone' of mechanical sealed centrifugal pump monitoring. Most operators appear to wait until leakage occurs at an unacceptable rate, but it would be a breakthrough to have a means of detecting imminent failure.

Some research by seal manufacturers and a research establishment suggests that some sense can be made of seal face condition. This is investigated using the acoustic emission technique. Whether this can be translated into an on-site procedure remains to be seen.

In conclusion, I would say that the changes to monitoring equipment and techniques aided by computers are changing condition based monitoring into a cheap and reliable tool. If you work with pumps, can you really afford to be without it?

The article is adapted from a presentation by Steve Moore (site machines engineer at ICI Runcorn) to the Focus on Pumping conference at the Pump Centre, Warrington, on 20 March 1997.

ICI tests show the cost of pump misalignment

Poor shaft alignment between rotating machines could be adding millions of pounds each year to power bills, claims condition monitoring company Pruftechnik. A company survey of 160 random machines found that only 10 per cent were aligned within `acceptable' limits.

The link between shaft misalignment and excessive vibration has been well documented but less research had been done on power losses.

Help is at hand, in the form of Pruftechnik's Rotalign alignment calculator. The company carried out work with it on a test pump at ICI Runcorn, to determine the effect of progressive misalignment on power consumption. The rig consisted of a 7.5kW pump operating through a closed loop. Horizontal offsets and angular misalignments were introduced. Different couplings, tyre and pin and bush, were investigated.

For both couplings offset alignment was more influential than was angular displacement. An offset of 1.25mm increased power consumption by more than 5 per cent.