Investigation of the failure of an alternator rotor shaft driven by a

Investigation of the failure of an alternator rotor shaft driven by a
Investigation of the
failure of an alternator
rotor shaft driven by a
reciprocating gas engine
An investigation of repeated failures of an alternator rotor drive
shaft is reported. Among other things it is shown that torsional
vibration was the major factor in the cause of these breakdowns and
that comparison of analytical prediction of such vibration with its
careful monitoring can be a valuable aid in deciding how to prevent
recurrence of the problem.
There are two main types of engine
torsional dampers (see Figure 1) Viscous
and Tuned, each having its particular
advantages. In brief; the Viscous Damper
tunes itself to the system torsional activities
with a seismic mass and specific silicone
fluid, while the Tuned Damper, as the
name implies, is tuned to the system;s
fundamental torsional vibration activity.
(It is outside the scope of this short study
to try to explain the advantages and
disadvantages of each)
Hamid R Malaki, VibraHiTec
Despite having worked for many years on
analysing the noise and vibration problems
of rotating machinery the author is still
surprised to see that in the 21st century
many industries still ignore vibration
monitoring and its potential benefits. Above
all, not many operators are willing to even
acknowledge that it will help to reduce
maintenance cost. Some operators believe
that by religiously following a maintenance
regime everything will be fine. They forget
that some parts may have extended life
but other parts may need replacing much
sooner, perhaps because of an unusual or
unforeseen incident. Constant vigilance,
coupled with a willingness to contemplate
a range of possible failure mechanisms
rather than grasping at the first that
comes to mind, may save a lot of time
and expense in the long run. The following
case study shows how the installation of
a simple monitoring system could have
given early warning of a first small incident
and would have had aided the avoidance
of two further catastrophic failures which
then ensued because the first went
Unfortunately, vibration is not of one kind;
there is linear vibration and also torsional
vibration. The latter may not be noticed
easily. However, most torsional vibration
problems do ultimately show themselves as
linear vibration, their indication and strength
depending on the supporting bearing and
structure. Few operators are familiar with
torsional vibration and its effect on rotating
parts, and many manufacturers marry
up the drive and the driven systems with
scant attention to linear vibration let alone
Usually, most turbine and reciprocating
engine manufacturers do carry out torsional
analysis when designing the drive and the
driven systems. It is the responsibility of the
manufacturer of the drive system to ensure
its compatibility with the driven system. In
doing so, they may install a heavy or light
flywheel and a specific torsional damper
to tune the torsional vibration behaviour to
ensure that the system is free from torsional
vibration within the running range.
Torsional analysis however is not easy, and
must be carried out by specialists. Many
manufacturers write their own software to
carry out the torsional vibration analysis,
while others consult specialised engineers
in this field to help to ensure the shaft
line is free from torsional vibration. Many
manufacturers of rotating systems ignore
the torsional compatibility of the drive and
driven system and assume everything
will be alright in the end. This wrong
assumption will haunt them when torsional
problems eventually occur.
46 | Mar/Apr 2012 | ME | maintenance & asset management vol 27 no 2
Figure 1 Torsional Dampers
This case study report aims to –
1. Demonstrate the significant damage
that can be caused by a small change
in the shaft line.
2. Highlight the effect of inattentiveness
to the mode of failure before any repair
work is carried out.
3. Emphasise that torsional vibration can
be a silent machine killer if ignored.
4. Recommend a full investigation and
verification by measurement before
putting the machine back in service.
With this in mind we were recently
consulted to investigate the failure of
an alternator rotor shaft driven by a
3MW reciprocating V-form 16-cylinder
gas engine installed in a power station
alongside other similar machines. It had
been reported that, on one of these sets,
the alternator shaft had failed inexplicably
after 64,000 hours and, to everyone’s
surprise, the replacement shaft had also
failed after only 800 hours operation – in
exactly the same manner and in exactly
Investigation of the failure of an alternator rotor shaft
driven by a reciprocating gas engine
stiffness value and/or damping are
changed (by any means) the result could
cause the stress in both crankshaft and
alternator shaft to increase. The maximum
stress in this situation would occur fairly
close to the running range with the flank of
the stress moving along into the running
speed which could cause the alternator
shaft to over-stress at the very point where
Figure 2 Failure of the shaft at the flange connection
the same position as the previous failure,
i.e. at the alternator shaft flange (see Figure
2). This suggested that some fundamental
problem had developed, possibly after
approximately 60,000 hours operation, to
cause such failures, one after the other.
The report for the first and second failures
had been studied and found inconclusive;
the failures being put down to the
possibility of alternator abnormal load and
shaft inferior material/process quality.
It was surprising to see that the report only
concentrated on the shaft material and
a possible overload. The short running
time before the second failure was put
down to the shaft material having lower
carbon content than the previous shaft.
This completely ignored the fact that in this
power station there were other generators
of exactly the same build, seemingly
without problems. But could they be on
the verge of failure too? Other possible
modes of failure did not seem to have been
considered at all.
Figure 3 Fracture faces
Looking at the fracture faces (see Figure 3)
it was fairly obvious that the failure mode
had resulted from torsional vibration and
torsional fatigue stress, which the report
had indeed acknowledged, but, clearly,
there had been no investigation of the
primary cause of this before re-assembling
and running the generator set after the first
Knowing the mode of failure, the
investigation was quite straight forward. It
was decided to first focus on the torsional
vibration behaviour of the drive and driven
system in order to obtain ‘analytically’
the shaft stress level during normal and
misfire (cylinder out) operation. The
torsional stress can rise significantly during
a misfire condition, and on a gas engine
misfire is common.
Upon determination
of the normal and
misfire condition,
other possibilities
that could change
the system torsional
vibration behaviour
characteristics could
be investigated.
The results of torsional
investigation showed
that the torsional
vibration (TV) stress for
both drive and driven
system were well within
(Standard running with misfire)
the allowable stress
level under both normal
Figure 4 Analysed torsional vibratory stress at the alternator shaft
running condition and
cylinder out (misfire).
fracture actually occurred – just before
the flange connecting it to the flywheel
and crankshaft. Bear in mind that the
crankshaft is of stronger material and the
maximum alternator shaft vibratory stress
occurs at system 2nd, 3rd and 4th modes
which are located between the alternator
shaft and flywheel.
Also, from the database it was possible
to find a set of TV measurements carried
out during the Factory Acceptance Test
(FAT) at full load and speed. The measured
values were lower than the predicted
values, so the system was torsionally
sound before leaving the factory.
Similar analysis was carried out for firing
order changes (see Figure 4) due to
possibly incorrect camshaft assembly at
some stage during maintenance work.
This also did not reveal the possibility of
causing any levels of stress in the shaft line
that could have been so high as to cause
the observed failure.
Finally, the damper sensitivity was
checked. The torsional analysis (see
Figures 5 and 6) showed that if the damper
The torsional vibratory stress can reach
its maximum at the points of twist which
are called nodes. The node location can
change if the rotating inertia and or shaft/
system stiffness changes. This change
would change fundamental torsional
natural frequencies and in turn the natural
mode. It should be noted that the mode
is the shape of the twist and the node
is the point at which the shaft will twist.
Therefore, the damper’s fundamental
characteristic is to reduce the amount of
twist in the shaft line hence the amount of
shaft stress at the nodes.
Based on this it was decided to focus on
the torsional damper as a primary suspect.
As it was a tuned damper, it had to be
maintained by the specialised engineer in
this field. The maintenance report revealed
that internal bolts had failed and also that
there was evidence of wear on the dynamic
maintenance & asset management vol 27 no 2 | ME | Mar/Apr 2012 | 47
a torsional failure would
normally be far greater
than just the failure of
the shaft.. Although
the drive system did
not see any damage,
all bearings and the
shaft line had to be
checked for cracks and
signs of consequential
damage. (It is important
to note that condition
monitoring would have
helped the situation
here – limiting the
extent of damage
if not preventing it
Figure 5 Analysed torsional vibratory stress at the alternator shaft
(Damper malfunction and misfire)
Unfortunately, many
operators incorrectly
conceive the damper to
 Linear vibration measurement
 Electrical excitation check/
The results of a set of torsional vibration
measurements taken at a suitable location
(see Figure 7) were analysed and the
performance compared with those of
similar measurements on one of the other
sets at the same station. In general, the
results for the re-assembled engine were
slightly lower, in all frequencies, than those
from the other generator set. Also, when
the measurement results were compared
with the torsional analysis results (see
Figures 8 and 9), the measured values
were slightly lower than the calculated
values at full load (hence the resulting
stress will also be lower). At no load up to
90% running speed the measured results
for the ½ and 1st orders were higher than
the calculated values. The ½ order higher
values are usually due to imbalance within
the engine cylinder pressure at no load.
The 1st order is usually due to the encoder
fixing to the crankshaft and tolerance on
eccentricity between the encoder and the
Although anything below ½ order is
ignored one needs to be aware of the
associated frequencies and their effect.
Figure 6 Torsional mode shapes (16 Vee 3MW generator shaft line)
seal face of the inner component of the
damper, although the damper unit was
still intact and showed no external signs of
The fractured bolts together with the seal
wear had most likely caused the change in
damper stiffness values and damping level.
This triggered a change in the damper
and its designed characteristic which
was tuned to overcome the fundamental
excitation frequencies within the shaft
line. Torsional vibration resonance and,
ultimately, shaft failure resulted. The
fact that the broken bolt had not been
found pointed the finger of suspicion
towards a failure in the damper inspection
regime – and irregular firing due to spark
plug problems probably exacerbated
the condition (misfire usually increases
torsional stresses).
It can also be appreciated that
consequential damage as a result of such
be similar to a flywheel (a lump of mass but
at the other end of the engine). Dampers
are a fundamental part of the engine, and
with engines working harder than ever,
are considered as critical components.
Advances in design and technology have
enabled many engine manufacturers to
increase power by changing the cylinder
pressure, material and a few associated
parts to deal with the power increase. But
one of the ways to reduce shaft line stress
is to employ a correctly designed torsional
damper tuned to the exact disturbing
excitation frequency.
Finally, after all repair work was complete,
the engine was started and - to ensure
that the machine was running as expected
and running similar to the other machines
– three tests, (which should have been
carried out after the first failure) were
carried out, viz.
 Torsional vibration measurement
48 | Mar/Apr 2012 | ME | maintenance & asset management vol 27 no 2
Figure 7 Measurement location and TV encoder
If they are significantly high then further
investigation would be needed. In this
case the identified frequencies were
not associated with the engine main
torsional excitation but mainly mechanical
Investigation of the failure of an alternator rotor shaft
driven by a reciprocating gas engine
Figure 8 Analysed vibratory angle at engine free end
Figure 10 Linear vibration measurements
Figure 9 Measured vibratory angle at engine free end
From the plotted results it can be seen that
the generator is operating as expected
where the measured values are almost in
line with the calculated values and better
than those from the comparison generator
When measuring torsional vibration on a
reciprocating engine attention should be
paid to a few areas such as –
 The levels of engine misfire and engine
imbalance. Generator sets are usually
tuned and balanced to run at full load
within 0.95 to 1.1 of the nominal speed.
It is known that at no load and low
speeds there is considerable variation
in cylinder pressures, especially in a
gas engine, hence the level of ½ order
activity is usually higher than calculated.
 The TV pick-up (encoder) and
adaptor piece. When connecting the
encoder adaptor to the free end of
the crankshaft care must be taken to
ensure that the adaptor is made with
a high tolerance and connected to the
crankshaft free end with high accuracy
in order to reduce
imbalance activities
and movement.
Controlling eccentricity
is paramount otherwise
the measured results
will show high levels of
first order activities.
 The engine overall
movement. Torsional
frequencies at low
engine speed can
coincide with a number
of engine natural
frequencies, such as that of the antivibration mount, hence some of these
movements might cause an indirect
engine torsional imbalance.
Due to time constraints only one set of
linear vibration measurements, in order to
establish the level of linear vibration at the
bearing, were carried out at the alternator
none-drive end. The vibration level for all
the generator sets was found to be within
an acceptable level. This measurement
(see Figure 10) was not intended to
substitute for a full vibration survey, but to
provide an indication, for future reference,
of vibration level at the alternator free-end
It should be remembered that linear
vibration measurements at only one
location can only be used as a general
guide, and cannot lead to a conclusion
that the machine is healthy. Iinstalling
a low cost vibration monitoring system
to monitor vibration level at a carefully
selected position at the alternator non-drive
end could give warning of catastrophic
failure. Such monitoring systems can be
obtained for as little as £2K to £3K.
Harmonic values and waveforms showed
that there were no electrical current issues
with the load.
 Shortcoming in the material is only one
of a number of possible reasons for
generator shaft failure.
 When two consecutive failures of a
similar nature occur, everything must be
scrutinised, from design to operation.
 When the mode of failure has been
determined look for the primary cause.
 If unexpected failure has occurred on
one generator and not on the others,
keep looking for the difference – it is
surely there. Don’t put the machine
back into full operation until that
difference has been isolated and its
effect quantified.
 Healthy operation of the other generator
sets and their engine dampers must
also be confirmed.
 Dampers must be regularly checked
at the manufacturer’s recommended
interval. Outward physical appearance
can be deceptive.
 Vibration monitoring will in most cases
help to spot failure before it actually
maintenance & asset management vol 27 no 2 | ME | Mar/Apr 2012 | 49
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF