Condition diagnosis
Units which fail to meet acceptable criteria should be given a complete frequency analysis. This applies to all weights and speeds of machine. Again readings should betaken at the various positions. A typical frequency analysis is shown in Figure
15.13.
So far we have discussed vibration in a general sense and indicated permissible overall limits. For important and/or arduous applications however, we need to be able to identify the causes of vibration and their likely effects on the machine which could be catastrophic in the event of a total breakdown.
FAN VIBRATION SIGNATURE
|
500 6 8 IK 2K 3K 4K 5K 6 0 !0K 20K 30K 40K 50K 6 8100K 200K 300K 400K 500K
|
The keys to the identification of the cause of a vibration are in its magnitude and frequency over most of the spectrum. Below about 10Hz displacement will be of primary importance, whilst above about 1kHz, acceleration is paramount. Over the remainder of the range, velocity measurements will be sufficient.
Different causes of vibration occur at different frequencies. For example, a faulty ball bearing would cause high frequency vibration at many times the rotational frequency whilst unbalance or misalignment produce vibration at the rotating speed frequency. To rely on an overall or linear response reading of vibration velocity could lead one to ignore a developing problem.
To obtain the installation’s “vibration signature”, a pick-up is used, which can either be handheld or more rigidly attached, feeding a meter giving a visual display. For analysis some type of tunable frequency analyser is necessary together with a stroboscopic light. The strobe permits rapid tuning to rotational speed when it “views” the rotating element and apparently sees a stationary image. From the analyser an X-Y recorder can be fed to show the magnitude of the vibration in narrow frequency bands right across the spectrum which the analyser can identify. By fixing the probe in either the vertical, horizontal or axial modes, different traces can be obtained which will inform the operator as to possible sources of vibration. Any “peaks” in the reading at specified frequencies will indicate the onset of certain troubles.
It is to these discrete frequencies we now look. Remember vibration severity is a quality judgement whilst frequency will indicate the cause. An increase in the severity of a particular frequency during inspection, commissioning or operation may indicate the onset of a particular problem. By referring to the initial signature and specifying that the particular frequency level should not increase by more than a predetermined amount, it is possible to construct planned maintenance procedures.
In the comments which follow fl7 is defined as rotational frequency and equals (rev/min + 60) Hz.
Unbalance As the heavy spot would give a “pulse” to the pick-up once every revolution, unbalance will be identified by
high readings in the horizontal and vertical directions at the rotational frequency, i. e. f|Hz. This is the most common cause of vibration.
High readings at commissioning can indicate residual unbalance in manufacture or “sag” if the rotating element has not been turned over regularly during storage on site. The build up of dust on a rotor, erosion or corrosion will also lead to increasing figures. Scaling at high temperature (above 400°C) or soaking in heat whilst stationary are other possible causes. 1
Misalignment This is almost as common as unbalance despite the use of self-aligning bearings, which should still be lined up as well as possible. Flexible couplings can be out of line both by height and angle. A bent shaft produces angular misalignment. Radial and axial forces are always produced, the size of such forces, and therefore the vibration which results being proportional to this misalignment.
As previously stated, the axial readings are usually 50% or more of the radial readings and again the frequency is normally f-) Hz. When the misalignment is severe however vibration at 2^Hz and even 3fiHz may be experienced. Misalignment can also occur where a machine has been distorted by tightening down onto foundations which themselves are not level. With sleeve bearings this will produce vibration according to the amount of residual unbalance, but with ball or roller bearings an axial vibration would be produced even if the unit were “perfectly” balanced, which is physically impossible.
Another very common fault is when pulleys and ropes of vee belt drives are not correctly aligned. This results not only in destructive vibration, but also leads to rapid wear of belts and production of frictional heat through to shafts and bearings.
Eccentricity An example of this could be where an impeller centre with excessive bore is “pushed over” by a taper key. The centre of rotation does not then coincide with the geometric centre. As far as the impeller is concerned, this leads to more mass being on one side of the rotational centre than the other,
I. e. unbalance.
It can therefore be corrected by rebalancing provided that the rebalancing takes place in its own shaft and bearings and that with ball/roller bearings the position of the inner race on the shaft also does not change. The predominant frequency is of course f|Hz. Where a fan is gear driven, eccentricity can produce reaction forces between pinions because of the cam-line action.
The largest vibration will occur along a line joining the centres of the two pinions at a frequency equal to (pinion rev/min h — 60) Hz of the one which is eccentric. It will look as if it is unbalanced but cannot be corrected by re-balancing. A similar situation can arise with vee belt pulleys which are eccentric. The largest vibration will be in the direction of belt tension at a frequency of (pulley rev/min ^ 60) Hz of the eccentric pulley. Again re-balancing cannot cure.
Looseness Common forms are loose foundation bolts and excessive bearing clearances. It will not be manifest unless there is some exciting force such as unbalance or misalignment to encourage it. Only small forces are necessary however to excite the looseness and produce large vibrations. Although rebalancing or realignment may therefore assist, extreme accuracy would be necessary which may be impossible to achieve. It is essential to tackle the problem at its source.
1, 2, |
Equ 15.3 |
Where |
Fp = |
To determine the characteristic frequency of looseness, let us consider an unbalanced rotor fitted to a shaft running in a bearing with loose holding down bolts. When the heavy spot is downward, the bearing will be forced against its pedestal. When the heavy spot is upward it will lift the bearing whilst at positions 90° away, the force will neither lift nor hold down, and the bearing will drop against the pedestal due to weight alone. Thus there are two applied forces each revolution of the shaft and the vibration frequency is 2fi Hz. This is the characteristic frequency of looseness.
Resonance The section on fan response (Section 15.4), showed that every object has a natural frequency at which it “likes” to vibrate. Should the forcing frequency coincide with the natural frequency of a part, then resonance will occur. To overcome the problem with a vee belt-driven unit is simple, a small change in rev/min will normally suffice. With a direct drive unit stiffening or a change in the design may be indicated, although unlikely.
Many of these problems can only be identified at the commissioning stage or during service. Figure 15.14 shows a technician taking readings on site.
I
Figure 15.14 Site testing with hand-held vibration analyser Courtesy of Schenck RoTec GmbH
Specific vee belt drive problems
Many of the problems found in impellers will also be present in vee belt drives. Often the balancing of pulleys has been overlooked and must be specified when ordering. Misalignment has already been mentioned.
In Chapter 14 a number of problems were identified, which could result in additional noise. However as these problems were essentially mechanical in origin, they are also manifest as vibration.
Such drives have good resistance to shock and vibration but may be blamed for causing trouble as they can be readily seen to whip and flutter especially when the belts are unmatched. Belts are often changed unnecessarily when the fault is really that of unbalance, misalignment etc. Nevertheless, the importance of using matched sets of belts cannot be emphasised enough.
Vibration from faults in the belts themselves occur at multiples of belt speed. The relevant frequencies are:
„ . pulley diameter, ..
3or4x—————- xf Hz
Belt length
Pulley rev/min 60
Likely faults are pieces broken off, hard or soft spots etc.
L |
Motor rev/min 60 |
Where |
F3=v |
1 + — cos A D |
Hz |
F2 — fi x ^ |
1 — cos A D |
Hz |
Hz |
1—r-cos A D2 |
Hz |
1 — cos A D |
Line frequency , slip frequency = 2 x————————————— r-i-:——fm Hz |
No. of poles |
|
||
![]() |
||
|
||
|
||
![]() |
|
|
|
||
![]() |
|
|
|
||
![]() |
|
|
|
||
|
||
![]() |
||
|
||
|
||
![]() |
||
|
||
B- |
T |
0.01 |
The time at which the first fatigue crater appears cannot be measured precisely. A batch of apparently identical bearings run under the same conditions of speed, load, lubrication and temperature will fail at different times. By using a statistical approach and analysing data accumulated over the years, it is possible for the manufacturer to quote the probability that a bearing running under the specified conditions will last for a given period of time, but this cannot be predicted with certainty.
The calculation of bearing life has now been made the subject of a Standard, namely ISO 281. The L10 life in hours or running is defined as that at which 10% of a group of apparently identical bearings can be expected to have failed by rolling fatigue. Being a statistical forecast, the result will be more accurate, the greater the numbers tried. Conversely 90% of all bearings can be expected to exceed their L10 lives, whilst the average life should be five times as great.
The need for early warning techniques
With such a wide spread of hours to failure, it has become desirable to monitor bearings, so that we may predict their lives much more accurately. When a bearing does fail, the damage to associated machine parts, and the production losses, are often far in excess of the actual cost of replacement.
The characteristic of fatigue failure is to increase the very high frequency vibrations at between 2.5 kHz and 80 kHz. Whilst these will always be present at a low level in nominally perfect bearings, they may be expected to increase by a factor of many hundreds before the onset of complete failure. During this time, the vibrations at the lower frequencies related to rotation and its multiples may not increase very much or may be attributed to other causes. Many vibration analyzers have a maximum cut-off frequency of about 1 kHz and by the time they are able to detect a significant increase in this vibration level, failure due to fatigue may be imminent.
Other techniques have therefore been developed, and these all monitor the high frequency vibrations in some form or another. It is to these methods which we now turn, describing the features and advantages of each. It is left to you to detect by inference or omission their respective disadvantages!
Posted in Fans Ventilation A Practical Guide