T h e E m e rald R ese a rch R eg is te r fo r th is jo u rn a l is a v a ila b le a t T h e cu rre n t iss u e a n d fu ll te x t a rch iv e o f th is jo u rn a l is a v a ila b le a t

http://www.emeraldinsight.com/researchregister http://www.em eraldinsight.com/1355-2511.htm

JQME
9,2
Condition monitoring of steam
turbines by performance
analysis
102 Ray Beebe
Gippsland School of Engineering, Monash University, Churchill,
Victoria, Australia
Keywords Turbines, Testing, Condition monitoring, Preventive maintenance,
Performance measurement
Abstract Many power generation steam turbine generators today are required in service well
beyond their intended lifetimes. Dismantling for inspection is expensive, and owners need to
consider all relevant information in making the decision. Application of condition monitoring in all
the applicable methods is justified, with each showing different degradation modes. Performance
analysis is less well publicised, yet unlike vibration analysis and oil debris analysis, it will show
conditions which reduce machine efficiency and output, such as deposits on blades and erosion of
internal clearances. Data obtained from tests before and after overhaul also reveal whether any
restorative work achieved the expected improvements in performance. The paper outlines, with
examples, some condition monitoring techniques that have contributed to retaining some large
fossil machines in service for up to 17 years without opening high-pressure sections.

Practical implications
The condition monitoring technique of performance analysis is less well known
that the other methods, yet it is the only way of detecting and monitoring some
modes of machine degradation. Large steam turbines are expensive machines
of critical importance, and also expensive to maintain. Maintenance decision
makers, therefore, need to take into account the needs of their business,
recommendations from the original equipment manufacturer, their own
experience and that of other users of similar plant, and information on
condition available from the plant.
This paper contains examples of condition monitoring by performance
analysis on large steam turbines that are intended to help engineers working in
power generation in the application of condition monitoring. Although their
effect can be significant, many changes are small and detection requires special
test instrumentation. Much useful work can, however, be done with plant
instruments. The overall aim is to assist in making the decision to take these
machines from service only when justified technically and economically.

Introduction
Steam turbines are the mainstay of electricity production worldwide. Today’s
Journal of Quality in Maintenance
Engineering competitive electricity generation market has increased the pressure to keep
Vol. 9 No. 2, 2003
pp. 102-112
power generation plant online as and when required.
# MCB UP Limited
1355-2511
A contributing factor in providing ongoing assurance of acceptable plant
DOI 10.1108/13552510310482361 condition is the use of condition monitoring. Methods should be applied
according to the modes of degradation expected. Vibration analysis provides Condition
much of this assurance, and has developed such that access to on line vibration monitoring of
data is available to experts who may be located remote from the plant. Damage steam turbines
to blading should be detectable by vibration analysis, but other problems such
as deposition, erosion of blading and internal steam leakages require
performance analysis.
As other degradation modes can only be detected by visual inspection or 103
non-destructive testing, eventually casings do require to be opened for
inspection. This is particularly so for the many large power generation
machines continuing in service beyond their intended design life.
Performance analysis can be applied to most plant, rotating and stationary.
It is the one condition monitoring technique, which allows the optimum time for
restorative maintenance to be determined, where the deterioration results in
increased fuel consumption, or in reduced output, or both (Beebe, 1998a).
It should be accepted that a turbine outage after a long time in service would
probably take longer than if scheduled more frequently, as internal distortion is
likely to have occurred. Also, parts such as casing studs will probably need
replacement (Coade and Nowak, 1993). Once a casing is opened and clearance
measurements made, it is possible to estimate the performance improvements
achievable by refurbishment and so justify the expenditure (Kuehn, 1993;
Sanders, 2002). However, it is clearly preferable to try to determine the internal
condition first by testing.
The overhaul decision should not be made unless there is a compelling
technical or economic reason for opening a casing. A current EPRI project is
aimed at extending the accepted interval between overhauls (McCloskey et al.,
1995; Roemera et al., 1998; Roemer et al., 2000a, b). Condition monitoring by
performance testing has been used to extend time between opening of casings
to up to 17 years, making its cost/benefits very favourable (Beebe, 1995; Vetter
and Schwiemler, 1989).
Condition monitoring can also be used to evaluate the effect of maintenance
or modification work on the steam path. This is of great benefit when justifying
future work.
Table I gives some main degradation problems with steam turbine-
generators, together with an outline of how condition monitoring can detect
them.

Overall condition indicator

The basic method of monitoring steam turbine internal condition is the valves
wide open test (ASME, 1985). Essentially, the generator is used as a transducer
to measure the power output of the turbine at set datum conditions. For a
typical large fossil reheat condensing set:
The inlet area for steam flow is set to datum by opening the steam
control (i.e. governor) valves fully. This should be verified by direct
measurement at the valve power servos, rather than relying on control
room indicators. Fully open is the only truly repeatable setting.
JQME Comments, suitable condition
9,2 Part affected Mode of degradation monitoring

Blading Erosion by solid particles Usually occurs gradually, worst on inlet

(also erosion by water blading. Less usual on sets with drum
droplets on latter LP boilers and/or sub-critical inlet steam
blades) conditions, or with bypass systems.
104 Performance analysis detects.
Blading Parts breaking off Usually sudden. Vibration analysis and
performance analysis detects
Bearings Scoring damage to Performance analysis, vibration
whitemetal (i.e. Babbitt) analysis, wear particles in oil (but
representative sampling at each bearing
is difficult if not impractical)
Rotors Rubbing, temporary Vibration analysis, and off-line: some
unbalance, cracking, NDT (not detailed in this paper)
misalignment
Valve spindles Leakage due to wear, Likely to occur gradually, but can be
Shaft and interstage distortion, breakage sudden. Performance analysis detects
glands (seals, packing) Effect of seal wear is relatively greater
Casing joints for HP blading (Cotton, 1993). For
LP manhole gaskets impulse blading, the relative lost output
Internal steam piping for each 25·m increase above design
and fittings clearance of about 600·m is:
HP: blade tips, 5kW; interstage seals,
6kW per stage; end glands 15 to
25kW
IP: blade tips, 2.5kW; interstage 2kW
per stage; end glands 5kW
LP: blade tips and interstage, 1.5kW
per stage; end glands 2kW
For reaction blading, the effect will be
greater
Steam valve strainers Deposits (more prevalent Likely to occur gradually, mostly in
Valve spindles with base loaded sets as areas around 260ëC. Some on-load blade
Blading cyclic loading tends to washing occurs with forced steam
have a blade washing cooling. Performance analysis detects.
effect) Blade surface roughness has biggest
effect at higher steam pressures. One
case gave 17% drop in output from
deposits varying between 250 to
2,300·m in thickness (Cotton, 1993)
Permissible roughness for LP blading
can be 100 times coarser than for HP
blading. One test with surface finish
equivalent to 500 grit emery paper
Table I. caused 5% to 7% less efficiency in HP
Some modes of blading, about 2% in LP (Cotton, 1993,
degradation and how based on work by Forster)
detected (continued)
 Comments, suitable condition Condition
Part affected Mode of degradation monitoring monitoring of
steam turbines
Generator rotor, stator Insulation faults Electrical plant testing (several
techniques (not included in this paper)
Condenser Air inleakage Performance analysis (not included in
Tube fouling this paper)
105
Feedwater heaters Air inleakage, tube fouling Performance analysis (not included in
by scale or oil this paper)
Valves ± HP, IP Leakage Performance analysis. Acoustic leakage
bypass, etc. detection (not included in this paper) is
also possible Table I.

The temperatures of main inlet and hot reheat steam are set as close to
datum as can be achieved. This is usually the same as the rated values.
The inlet pressure is set to the datum value. As most turbines have
capacity beyond their nameplate rating, the standard inlet steam
pressure may need to be below the rated value if undesirably high
outputs would result (Beebe, 1998c).
Condenser pressure is largely a function of seasonal conditions and
weather, and is usually taken at the best attainable on the day.
Extractions to feedwater heaters should be all fully open. If feedwater
heater unreliability means that some heaters are out of service for long
periods, that condition may have to be used as datum, unless a method
of allowing for this effect on turbine output can be derived.
Test readings a test run of an hour or so during steady conditions are carefully
made using calibrated test instruments, with two separate measurements of
each point. Readings of test transducers can be made manually, but it is now
usual to use a data logger coupled with a computer. With the exception of some
minor flows read from plant instruments and used only in correction factors,
test measurements of flow are not required. This simplifies the test
considerably and minimises the cost considerably compared with the full heat
rate test used for the acceptance tests.
The generator MW output is corrected for any variations from the datum
terminal conditions. For example, if the condenser pressure on the test is higher
than the datum, then the turbine output will be less that expected at datum
condenser pressure. Correction data are usually provided by the manufacturer
for use in the acceptance tests but can be obtained using cycle modelling
programs or from special tests (Beebe, 1998c). With the instrument calibration
information available, the calculations can be performed immediately following
the tests.
JQME For some plant items, it is possible to use the normal plant instruments and
9,2 data processing system to determine condition parameters (Beebe, 1998b). In
the case of steam turbines, a more refined method using test quality
instruments is needed to give warning well in advance of changes evident from
permanent instrumentation systems (Groves, 1996). This may be possible with
highly stable transducers of recent design, or with adequate calibration
106 arrangements.

Example of VWO tests

350MW reheat turbine
Tests run some years apart and at different seasons gave the results in Table II.
From experience, the reduction observed is significant. Further tests would
be performed to ascertain parameters of condition of individual machine
components, which can be separately opened. Data for these are often gathered
concurrently with the VWO tests. (Note that an increase in VWO output would
indicate a different type of degradation.)

200MW non-reheat turbine

An operator of a non-reheat turbine generator normally run at 210MW output
on base load noticed that the control oil pressure was higher than usual,
indicating that the steam control valves were fully open. Because output was
not affected, others had not remarked on this.
A VWO test gave a corrected VWO output of 210MW, a reduction from
216MW at the previous test (new condition). As these monitoring tests were
being developed, it was decided to investigate further and perform a more
detailed test to include flow measurement.
Unfortunately, the connecting piping to the flow metering assembly leaked,
and it was five months before the detailed tests were able to be run to find that
the corrected VWO output was 216MW, the same as when new. Further
investigation revealed that the unit had been shut down (and started up) 11

Correction Correction
Test data Test A factor Test B factor

Generator output ± MW 355.8 349.7

Steam pressure ± main kPa 12,155 1.02285 12,255 1.02053
Steam temperature ± main ëC 529.5 0.99832 526.7 0.99773
Steam temperature ± reheat ëC 525.8 1.0101 539.5 0.99873
Reheater pressure drop % 6.76 0.99814 6.03 0.99633
Condenser pressure ± kPa 9.34 1.01225 12.44 1.03615
Generator power factor 0.923 1.00012 0.945 1.00064
Steam temperature control spray ± main kg/s 6.5 0.99889 24.6 0.99584
Table II. Steam temperature control spray ± reheater kg/s 0 1 0 1
Results from two series Final feedwater temperature ëC 234.9 1.0005 230.5 0.98957
of tests, 350MW Combined correction factor 1.04741 1.03521
turbine generator Corrected VWO output ± MW 372.7 362
times between the tests. It was known that blade deposition rarely occurs on Condition
turbines in cyclic loading due to the blade washing effect which results, so the monitoring of
load reducing effect was attributed to blade deposition. steam turbines
Another message here is the value of close operator surveillance, despite no
effect on production being evident in this case.

Section parameters 107

If routine tests from time to time indicate no change, the steam path can be
considered to be unchanged. However, when a changed corrected VWO output
is obtained, it is necessary to look further to localise the cause. For instance, a
full turbine overhaul is not necessary to open and clean main steam strainers,
which may be blocked with weld crud or other material sufficiently to restrict
steam flow and so reduce turbine output.
It is also possible that other changes in the steam path may have occurred
which do not affect output much, but may be otherwise undesirable. Therefore,
it is best to include condition parameters of sections of the turbine at little extra
cost during VWO tests.
Parameters obtained from temperature and pressure measurements are
shown in Table III.

Enthalpy drop efficiency tests

The enthalpy drop efficiency is the actual enthalpy drop divided by the
isentropic enthalpy drop. Figure 1 illustrates this parameter on a section of the
Mollier chart. It is usually between 85 per cent and 90 per cent, with typical
repeatability for HP casings ‹ 0.9 per cent; IP casings ‹ 1.0 per cent.
This can only be determined for the superheated steam sections. For
assessment across a turbine section within a casing, stage conditions are
usually available in steam extraction lines to feedwater heaters. Naturally,
these cannot be used for temperature measurement if the associated feedwater
heater is out of service.
The corresponding enthalpy drop efficiencies in the two test series given for
the 350MW machine mentioned earlier were:

Test A
High pressure casing (from main stop valve inlet) 85.5 per cent.
Intermediate pressure casing (from reheat stop valve inlet) 88.2 per cent.

Text B
High pressure casing (from main stop valve inlet) 83.8 per cent.
Intermediate pressure casing (from reheat stop valve inlet) 88.3 per cent.
A relative deterioration in the HP casing is evident. Further study would be
made of any other parameters available in this area.
JQME Parameter Comments
9,2
Steam strainer pressure drop Best measured with a differential pressure transducer
rather than an upstream and a downstream pair. An
increase indicates blockage, possibly from metal
particles from boiler tubing, or welding repairs
108 Corrected first stage pressure At VWO, proportional to steam flow through the
turbine, indicates first stage condition. An increase
points to upstream erosion, or downstream blockage,
and vice versa
Section enthalpy drop efficiency Calculated using steam tables computer program. A
(superheated steam sections) drop indicates blade fouling, or erosion damage
Section pressure ratios Stage pressures can be corrected to standard inlet
pressure (ASME, 1985), but any error in measuring it
is applied to all the stage pressures. Ratios use only
the outlet and inlet pressures of each section. Changes
show up erosion or deposition
Extraction temperatures to According to design, a higher than expected steam
feedheaters in superheated sections inlet temperature may indicate relative internal
bypassing leakage in the turbine upstream of the
extraction point
Extraction temperatures to Temperatures above saturation temperature indicate
feedheaters in saturated steam leakage of steam from a stage upstream of the
sections extraction point
Drain line temperatures from Where available, these may indicate relative leakage,
casings, or from shaft seal (gland, according to design. A similar approach can be used
packing) sections for points before and after pipe junctions of two
streams of different temperatures. Pipe surface
temperatures are often sufficient for repeatable
assessment
Table III.
Some parameters Estimated N2 packing leakage (on Test by varying relative inlet steam temperatures and
showing condition of turbines with combined HP-IP observing effect on IP enthalpy drop efficiency (Cotton,
turbine sections casings) 1993)

Uses of the expansion line

The plot of the expansion line as shown schematically in Figure 1 is also useful
in condition assessment. If a measured stage point does not fall on the line as
expected, bypassing of blading would be deduced. Here are two examples.

Reaction turbine HP casing dummy piston leakage

A 200MW machine with a single flow HP casing has a dummy piston on
which steam pressure is arranged to act to counteract the axial thrust
generated by steam forces on the reaction blading. Steam leaking past the
first section of the labyrinth seals around the piston circumference is led
through internal piping to join the steam flow from the blading at the
first extraction point.
 Condition
monitoring of
steam turbines

109

Figure 1.
Section of Mollier chart
showing expansion line

As this leakage steam is at a higher temperature then the extraction

steam, relative leakage is shown by the temperature difference between
the mixture and the normal extraction flow. On one of six machines of
this design, even when new this difference was as high as 60ëC. The
extraction steam temperature can be estimated closely enough from the
expansion line. To remove the effect of any blading deterioration of
stages upstream, a permanent thermocouple was arranged to fit through
the high pressure outer casing to measure the true extraction stage
temperature.

LP casing internal steam bypassing

A similar approach can be applied in the low pressure casings. Here the steam
at most, if not all, extractions to feedheaters is usually saturated, so the
temperature at a stage should be that of saturation for the corresponding
pressure. Superheated steam indicates internal steam leakage from an
upstream source bypassing the blading. On a 500MW turbine, a burnt area
observed on one LP hood was deduced to be due to failure of one of the large
expansion bellows in the inlet piping. As test results showed a 27MW drop in
VWO output, closer study revealed that steam was entering an LP feedheater
at 256ëC, rather than the expected 95ëC. From careful study of the complex
construction details from available drawings, it was deduced that the second
bellows in the inlet piping had failed.
JQME Use of stage pressures
9,2 Except for the first stage of nozzle-governed turbines, and the last stage,
at datum inlet and exhaust conditions, the pressures at stages along a
steam turbine are closely proportional to steam flow, and therefore to
output. Figure 2 sketches the relationship. Lines 1 to 4 represent stages
along the turbine.
110 During starting up of a 200MW turbine, great difficulty was
experienced with extremely high in-leakage into its condenser. Despite all
the air removal pumps running, condenser pressure was still well above
normal. As load was increased, the problem disappeared above 40 per
cent load (L in Figure 2).
By examining the stage pressure vs. load curves, it was noticed that the
pressure to LP2 feedheater was below atmospheric (A in Figure 2) until above
40 per cent load, when it became positive. The air leak was suspected to be
somewhere in the turbine system connected to that point. Meticulous study of
piping drawings and inspection of the machine revealed a sprung joint in a
flange joint in the leakoff piping from a shaft end gland. This piping led to the
LP2 feedheater.
Such investigations are not necessarily easy ± few drawings may be
available, the piping and connections may be under lagging, access is required
to areas of the machine which are probably dark, noisy, certainly hot, and
special staging may need to be built.

Figure 2.
Stage pressures along
turbine
Application of all paremeters Condition
A 350MW turbine as shown earlier above showed a significant reduction in monitoring of
corrected VWO output. As this was only the second series of tests on this new steam turbines
machine, further investigation proceeded (Beebe, 1978, 2001).
Calculation of enthalpy drop efficiency and pressure ratio for each section
localised the cause of the reduction as the intermediate pressure section. As this
base load machine had been in continuous service for six months, it was 111
postulated that blade deposition had occurred.
During coming off line for the next planned outage, a steam forced cool was
conducted. Here the steam temperatures at main inlet and reheat inlet are
reduced in steps as load is reduced over a few hours, taking care to observe
operating limits such as the margin above saturation temperature. This
practice is common where prompt access is required to the turbine, as it can
save many hours over cooling at the natural rate.
Tests repeated after the outage revealed that the corrected VWO output had
returned to its original value, and the section parameters had also reverted.
Learnt here is that steam forced cooling should be standard practice before
outages on machines which have long steady load service. The condensate
should be sampled and tested for signs of deposits from the blading.

Comparative tests
Where steam flows through blading sections can be found, further parameters
of condition are available.
Measurement of cycle flow is expensive, often requiring several flow
measurements in the cycle. Some plants have a high accuracy flow element for
final feedwater flow, with some having a removable inspection port, but these
are not common. In some situations, a comparative flow measurement is useful,
even without a high accuracy element.
Blades of a new design were installed in the last stage sections of a turbine.
Significant improvements in efficiency and output were promised.
Testing at VWO was run immediately before and after the outage. Flow was
calculated using the permanent feed flow element, which is welded in the line.
Test quality transducers were installed to measure its differential pressure
output. The condition of the flow element was assumed to be unchanged during
the outage period.
Test results showed that VWO output and efficiency had increased in the
order predicted, however, some of this may have been due to other work done
during the outage. Some casing joint distortion was found and the joint was
remachined on one casing. Some changes in section parameters in this casing
were observed (Beebe, 2000).
This case does highlight the difficulty of distinguishing between the effects
of two or more maintenance/modification jobs, both of which should have an
impact on performance. However, in the future, similar deterioration in this
area should be detectable in advance of proposed work, and the improvements
in performance used to justify maintenance action.
JQME Conclusion
9,2 From the cases described, it is concluded that condition monitoring by
performance analysis is a most valuable input in assuring continued operation,
and when deciding and justifying major maintenance actions.

References
112 ASME (1985), ``Simplified procedures for routine performance tests of steam turbines’’, ANSI
PTC 6S report 1974, reaffirmed 1985.
Beebe, R. (1978), ``Recent experience with condition monitoring of steam turbines by performance
analysis’’, Transactions of Mechanical Engineering, Institution of Engineers, Australia.
Beebe, R. (1998a), ``Condition monitoring by performance analysis to optimise time for overhaul
of centrifugal pumps’’, Proceedings of 52nd Meeting of Society for Machinery Failure
Prevention Technology, Virginia Beach, VA.
Beebe, R (1998b), ``Predictive maintenance by performance monitoring of plant’’, Proceedings of
52nd Meeting of Society for Machinery Failure Prevention Technology, Virginia Beach, VA.
Beebe, R. (1998c), ``Steam turbine generator condition monitoring by performance analysis:
derivation of correction factors by test’’, Proceedings of 11th COMADEM 98, Tasmania,
Vol. 1, pp. 87-92.
Beebe, R. (2000), ``Case studies in condition monitoring of low pressure steam turbines by
performance analysis’’, Proceedings of 13th COMADEM, Houston, TX.
Beebe, R. (2001), Machine Condition Monitoring, MCM Consultants.
Coade, R. and Nowak, S. (1993), ``Remaining life study of a 350MW HP/IP turbine’’, Conference on
Pressure Vessels and Pipework, Sydney.
Cotton, K.C. (1993), Evaluating and Improving Steam Turbine Performance, Cotton Fact Inc.
Groves, K. (1996), ``Turbine steam path monitoring using plant DPA system’’, unpublished
degree major project, Monash University, Melbourne.
Kuehn, S.E. (1993), ``Steam turbine technology keeps pace with demands’’, Power Engineering,
March, available at: www.encotech.com/software/eSTPE.html
McCloskey, T.H., Pollard, M.A. and Schimmels, J.N. (1995), ``Development and implementation of
a turbine-generator outage interval extension strategy’’, PWR-Vol 28 ASME/IEEE
International Joint Power Generation Conference, pp. 485-91.
Roemer, M.J., Kacprzynski, G.J. and McCloskey, T. (2000b), ``Advanced steam turbine-generator
maintenance planning using prognostic models’’, Proceedings of 54th Meeting of the
Society for Machinery Failure Prevention Technology.
Roemer, M.J., Orsagh, R. and Hesler, S. (2000a), ``Turbine-generator asset management using
EPRI’s Turbo-X program’’, EPRI Welding Conference, May.
Roemer, M.J., Dewey, R.P., Atkinson, B., Mauney, D.A. and McCloskey, T.H. (1998), ``Turbine-
generator maintenance outage optimisation: probability/risk assessment of net present
value’’, Proceedings of 52nd Meeting of Society for Machinery Failure Prevention
Technology, Virginia Beach, VA.
Sanders, W.P. (2002), Turbine Steam Path Maintenance and Repair, Vol. 1, PennWell, Tulsa, OK.
Vetter, H. and Schwiemler, G. (1989), ``First turbine inspection after a 15-year operating period’’,
VGB KRAFTWERKSTECHNIK, Vol. 69 No. 10, October, pp. 835-45.