Analysis of Compressor On-Line

Washing to Optimize Gas Turbine

Power Plant Performance
Due to compressor fouling, gas turbine efficiency decreases over time, resulting in de-
1 creased power output of the plant. To counteract the effects of compressor fouling, com-
Ernst Schneider pressor on-line and off-line washing procedures are used. The effectiveness of compres-
e-mail: ernst.schneider@power.alstom.com sor off-line washing is enhanced if combined with the cleaning of the VIGVs and the first
compressor blade row by hand. This paper presents a thorough analysis of the effects of
Saba Demircioglu compressor on-line washing on the gas turbine performance. The analysis is based on the
Bussjaeger measured data of six gas turbines operated at two different plants. Different washing
schedules and washing fluids are analyzed and compared. Furthermore, the effects of
Susana Franco compressor on-line washing on the load distribution within the compressor are analyzed.
The performance benefit of daily compressor on-line washing compared with weekly
Dirk Therkorn compressor on-line washing is quantified. As expected, daily compressor on-line washing
yields the lowest power degradation caused by compressor fouling. Also, the effect of
Alstom (Switzerland) Ltd., washing additives is analyzed. It is shown with long term data that compressor on-line
Brown-Boveri-Strasse 7, 5401 Baden, washing cleans up to the first 11 compressor stages, as can be detected well in the
Switzerland compressor. With a view to gas turbine performance optimization, the recommendation is
to perform compressor off-line washing at regular intervals and to take advantage of
occasions such as inspections, when the gas turbine is cooled down anyhow. Especially
for gas turbines with a high fouling rate, a daily compressor on-line washing schedule
should be considered to reduce the power loss. For gas turbines operating with high
fogging, compressor on-line washing has no added benefit. To determine the optimal
compressor washing schedule, compressor blade erosion also has to be considered. A
reasonable balance between compressor on-line washing and off-line washing improves
the gas turbine performance and optimizes the gas turbine availability.
关DOI: 10.1115/1.4000133兴

1 Introduction caused by small solid or liquid particles starting to adhere to the

compressor blades and annulus surfaces 关2–4兴. Due to their higher
The condition of a gas turbine compressor is affected by the
mass and higher inertia, these small particles can deviate from the
environmental conditions of the site. With increasing operating
streamline trajectory of the air and impinge the material surfaces.
time degradation of the compressor can be registered in terms of Depending on the flow conditions, the particle mass, and the ma-
reduced performance. Apart from abrasion and erosion of the terials, the small particles may remain stuck to the surfaces after
blade surfaces, compressor fouling is the main cause of the reduc- impingement. Oil vapors, high humidity, desert environments,
tion in compressor efficiency and compressor inlet air mass flow. etc., accelerate the fouling process. The result of fouling is an
The degradation of the gas turbine compressor has a direct influ- increased surface roughness with negative effects on the flow and
ence on the gas turbine power plant efficiency and power. a change in the effective flow area. Especially the first compressor
To prevent degradation, industrial gas turbines are equipped stages are affected by the fouling process. Compressor fouling
with sophisticated air filter systems. These air filter systems sig- decreases compressor efficiency and the compressor flow capacity
nificantly reduce the amount of contaminants the gas turbine is 关4兴.
subjected to, but cannot filter out the contaminant particles alto-
gether. Most of the contaminants are of small size; normally 80%
are below 2 ␮m 关1兴. However, even small concentrations of for-
eign particles can accumulate to several tons of contaminants that
are ingested by the gas turbine during the operating time between
two major inspections 关1兴. 共A contaminant concentration of 1 ppm 2 Compressor Efficiency
exposes the gas turbine to 12,500 kg of impurities assuming an air The gas turbine compressor is the greatest consumer of turbine
mass flow rate of 450 kg/s and 8000 OH.兲 power. The power consumption of the compressor amounts to
The source and the properties of these contaminants can differ 60–65% of the turbine power 关5兴. Hence, even small changes in
greatly. Not only sand, dust, or dirt are registered, but also chemi- the compressor efficiency have a significant effect on the overall
cal substances such as oil or salt and natural contaminants such as gas turbine performance and efficiency.
plant spores or insects 关1兴. The contaminated air is the cause for The efficiency of an axial compressor is defined as the ratio
compressor fouling. Compressor fouling is the phenomenon between ideal and actual enthalpy increases of the gas due to the
compression from the pressure level p1 up to the pressure level p2
共see Fig. 1兲. For an ideal process the compression is isentropic.
1
Corresponding author. However, a real process also involves an entropy increase. Hence,
Contributed by the International Gas Turbine Institute 共IGTI兲 of ASME for pub-
lication in the JOURNAL OF ENGINEERING FOR GAS TURBINES AND POWER. Manuscript
the registered enthalpy increase is higher than for the ideal pro-
received April 14, 2009; final manuscript received April 28, 2009; published online cess. If the actual process is approached by infinitesimal isentropic
March 18, 2010. Editor: Dilip R. Ballal. steps the isentropic compressor efficiency is obtained as

Journal of Engineering for Gas Turbines and Power JUNE 2010, Vol. 132 / 062001-1
Copyright © 2010 by Alstom

Downloaded 13 Jun 2011 to 138.250.8.97. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 1 Enthalpy versus entropy map of a compression

h2s − h1
␩s = 共1兲
h2 − h1 Fig. 2 Turbotect MK1 on-line compressor washing injection
nozzle
Modern axial gas turbine compressors have efficiencies of about
87% 关5兴. Assuming the air as an ideal gas, the enthalpies can be
expressed as functions of the temperature h = f共T兲. Hence, the en-
thalpy differences can be expressed as temperature differences 关6兴 which are repeated until the water that pours out of the compres-
⌬hs = c p共T2,s − T1兲 共2兲 sor through the opened blow-off valves does not indicate anymore
signs of contamination 关5,7兴. With compressor off-line washing,
⌬hn = c p共T2 − T1兲 共3兲 the effects of compressor fouling can be removed virtually com-
pletely. Compressor off-line washing is recommended at the nor-
and considering that for isentropic processes 关6兴

冋冉 冊 册
mal inspection intervals to avoid nonavailability of the gas turbine
R/c p because of the necessary cool-down periods.
p2
⌬hn = c pT1 −1 共4兲
p1 3.3 Compressor Hand Cleaning. Cleaning the compressor
the polytropic compressor efficiency can be expressed as by hand involves cleaning the VIGVs and the blades of the first
p2
compressor row with brushes and a detergent. Cleaning the com-
R ln p1 pressor by hand is very effective for removing particles sticking to
␩n = 共5兲 the blade surface, but has the drawback that it is very time con-
c p ln T2
T1 suming and requires the shutdown and cooling of the gas turbine
Hence, the polytropic compressor efficiency can be calculated beforehand 关7兴. Since cleaning the compressor by hand affects
based on the measured flow conditions at the compressor inlet and only the VIGVs and the blades of the first compressor row, it
at the compressor exit. should be always done in combination with a compressor off-line
washing.
3 Compressor Cleaning 3.4 Abrasive Cleaning. The use of abrasive materials for
3.1 Compressor On-Line Washing. Compressor on-line compressor cleaning, e.g., the injection of rice or walnut-shells
washing is normally done during gas turbine base load operation into the compressor, is suspended in most plants 关5兴, since the
with the VIGVs in the fully open condition. For compressor on- erosion effects are too high and there is the danger that these
line washing, an on-line washing system has to be installed at the nonliquid materials collect in the gas turbine and clog the pas-
air inlet of the gas turbine. Generally demineralized water is in- sages of the secondary air system. Particles may enter seals and
jected through a configuration of small nozzles in the air flow the lubrication system, or they may damage the coating of the
before the first compressor stage. Examples of such on-line com- compressor blades 关5,7兴. Nonetheless, abrasive compressor clean-
pressor washing injection nozzles are shown in Figs. 2 and 3. In ing may be performed on a removed rotor during a major over-
addition to demineralized water, a detergent can be used to in- haul.
crease the cleaning effect of compressor on-line washing. Since a
cleaning effect can only be obtained as long as the injected water
is liquid and has not evaporated, the cleaning effect of compressor
on-line washing is limited to the VIGVs and the first stages of a
compressor 关7–9兴. Compressor on-line washing can diminish the
degree of compressor fouling, but it cannot prevent it completely.
Hence, from the point of view of performance recovery, the rec-
ommendation is to carry out a compressor off-line washing from
time to time, because this shows a higher effectiveness for perfor-
mance recovery.
3.2 Compressor Off-Line Washing. For compressor off-line
washing, the gas turbine has to be shut down and cooled. A com-
pressor off-line washing procedure normally consists of several
steps. First the gas turbine compressor is flushed with demineral-
ized water. A detergent can also be used. Using the start-up system
the gas turbine is accelerated to 20–50% of nominal speed and
then allowed to coast to a stop. The flushing may be repeated Fig. 3 Turbotect MK3 on-line compressor washing injection
several times. The procedure ends with several rinse-cycles, nozzle

062001-2 / Vol. 132, JUNE 2010 Transactions of the ASME

Downloaded 13 Jun 2011 to 138.250.8.97. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Table 1 Environmental conditions at Plant A and Plant B

Plant A

• Industrialized area
• Petrochemical industry
• Maritime conditions

Plant B

• Industrialized area
• High air humidity
• High concentration of particles ⬍10 ␮m
• Maritime conditions

4 Compressor Washing Schedule Analysis

Different on-line washing schedules were investigated to deter-
mine the effects on the compressor efficiency and the power plant
power output. Measured data of six gas turbines were used. Only
operating points at base load conditions were considered. The
measured power output was corrected for ambient temperature,
ambient pressure, relative humidity, generator power factor, rotor
speed, and condenser vacuum. The compressor efficiency was cal-
culated based on the measured data. It was assumed that the com-
pressor inlet conditions correspond to the ambient conditions. Fig. 4 Relative corrected compressor efficiency Plant A
Hence, the pressure drop over the air inlet filters is considered in
the compressor efficiency value. The compressor efficiency was
corrected for ambient conditions and rotor speed. mulates on the blades only slowly. Influenced by the dirt already
Three of the gas turbines analyzed are operated at Plant A. The sticking to the compressor blades, the degradation rate increases.
three Alstom GT26 gas turbines will be named PA1, PA2, and After several weeks, a new equilibrium is found, the increased
PA3. PA1, PA2, and PA3 are three identical single-shaft, roughness of the blade surfaces abets the further accumulation of
combined-cycle power plant units. Plant A is situated in the mod- dirt, but at the same time dirt particles are transported downstream
erate climate zone. At Plant B, situated in the tropical climate by the air mass flow. A stable degradation level may be reached
zone, three Alstom GT13E2 gas turbines are operated in a multi- after a certain time, as it is also confirmed by the tests of Tarabrin
shaft arrangement for a triple pressure water steam cycle. They et al. 关4兴 and Stalder 关11兴. The compressor off-line washing of
will be named PB1, PB2, and PB3. In Table 1 is given an over- PA1 performed during the test campaign was without additional
view of the environmental conditions of Plants A and B. compressor hand cleaning. It shows only a small effect on com-
4.1 Analysis Plant A. At Plant A different compressor on-line pressor efficiency recovery.
washing schedules were compared over a period of 7 months. The small compressor efficiency degradation rate at the begin-
During this time PA1 had no compressor on-line washing, PA2 ning of the test campaign is the reason why, during the first weeks,
had a weekly compressor on-line washing schedule with deminer-
alized water and PA3 had a daily compressor on-line washing
schedule with demineralized water, too. No detergents were used
and the water was not heated.
In Fig. 4 a comparison of the compressor efficiencies of the gas
turbines PA1, PA2, and PA3 is given for the period of the com-
pressor on-line washing. Before the start of the test campaign, the
compressors of the three gas turbines were cleaned thoroughly
with compressor off-line washing.
The general trend of the compressor efficiency degradation is
similar for all three units. During the first weeks of the test cam-
paign, the units show no signs of degradation. This phase of rela-
tive constant compressor efficiencies is followed by a phase with
gradually increasing degradation up to a level where the compres-
sor efficiency values stabilize again at a lower level.
Similar overall trends can also be seen in Fig. 5, where the
relative power output of the three units PA1, PA2, and PA3 is
plotted for the period of the compressor on-line washing. The
exchange rates between compressor efficiency and power, i.e., the
effect of a compressor efficiency change on the power output, are
known from performance calculations. Based on the exchange
rates it is possible to deduce that the power output degradation is
caused mainly by the compressor efficiency degradation. At the
beginning, the power output remains relatively constant, then it
decreases gradually and stabilizes again at a lower level. The rela-
tive constant compressor efficiency and power trends during the
first weeks after the compressor off-line washing may be ex-
plained by nonlinear fouling effects. In the beginning, dirt accu- Fig. 5 Relative power output comparison Plant A

Journal of Engineering for Gas Turbines and Power JUNE 2010, Vol. 132 / 062001-3

Downloaded 13 Jun 2011 to 138.250.8.97. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 the effects of compressor efficiency degradation disappear within
the scatter of the compressor efficiency values. The same can be
said for the power plant power output. Hence, the first signs of
compressor efficiency degradation can be registered for all units
only after about 6 weeks. The first signs of power output degra-
dation can be registered for PA1 after 6 weeks, for PA2 after 9
weeks, and for PA3 after 12 weeks following the beginning of the
test campaign. The gas turbine without compressor washing
shows the highest degradation rate. For the gas turbine with
weekly compressor on-line washing, the degradation rate is lower,
and the gas turbine with a daily compressor washing schedule has
the lowest degradation rate. Compressor on-line washing helps to
keep the power output of the plant at a higher level for a longer
time. In other words, compressor fouling progresses at a slower
rate as a result of compressor on-line washing 共see Fig. 4兲. As can
be seen in Figs. 4 and 5, the degradation rate for PA3 increased
during the time span when no compressor on-line washing was
performed.
At the end of the compressor on-line washing test period, the
registered total corrected relative compressor efficiency degrada-
tions were about 1.4% for PA1, 1.2% for PA2, and 1.0% for PA3.
Based on the empirical exchange rate of 2 for the dependency of
power output degradation on compressor efficiency degradation,
which is consistent with the values published in literature 关2兴,
power output degradations of 2.8% for PA1, 2.4% for PA2, and
2.0% for PA3 were to be expected. The actual registered power
output degradations were 3.6% for PA1, 3.4% for PA2, and 2.0%
for PA3. Therefore, although compressor efficiency degradation is
the main contributor to the degradation of the total power output, Fig. 6 Power output of the units of Plant A during the period
it is not the only effect. with the same daily washing schedule
The effects of power recovery due to air inlet filter changes are
relatively small and have about the same magnitude as the scatter.
Nonetheless, it is necessary to change the air inlet filters regularly, In contrast to the compressor efficiency trends observed at Plant
since new filters that are not soiled and are undamaged are more A, the compressor efficiency degradation progresses faster at
effective in preventing the fouling of the compressor. Hence, they Plant B without a relatively stable phase directly after the com-
reduce the degradation rate of the compressor. pressor off-line washing at the beginning of the test period. Also,
After identifying the daily compressor on-line washing sched- the degradation does not reach a stable level as observed at Plant
ule as the most effective from the point of view of power output, A. The probable reason for the different degradation trends is that
further compressor on-line washing tests were performed at PA1, the environment at Plant B is more highly polluted, so the rate of
PA2, and PA3. The three units were subsequently operated during compressor fouling is greater than at Plant A.
a period of 13 months with a daily compressor on-line washing
schedule, and the power outputs were compared 共see Fig. 6兲. As
expected, the degradation behavior of the three units is compa-
rable, since they are identical and are operating under the same
atmospheric conditions and with similar load schedules. After 9
months of the test campaign there is a period of about 2 months
during which PA3 accumulated less operating hours than PA1 and
PA2. During this period the degradation rate of PA3 is lower.
Nonetheless, the results for PA3 obtained during the second test
phase confirm the ones obtained during the first phase with daily
compressor on-line washing.
4.2 Analysis Plant B. At Plant B different compressor on-line
washing schedules and washing fluids were compared during a
period of 6 months. During this time PB1 had a twice-a-week
compressor on-line washing with a detergent, PB2 had a daily
compressor on-line washing with demineralized water plus a
weekly compressor on-line washing with a detergent, and PB3
had no compressor on-line washing. The water used was not
heated.
In Fig. 7 the relative corrected compressor efficiencies of PB1,
PB2, and PB3 are plotted for the time span of the compressor
on-line washing tests. In Fig. 8 the relative corrected power output
of the units PB1, PB2, and P3 is also given for the time period of
the test campaign. As can be seen, the trends of the compressor
efficiencies and the power output for each unit are similar. For
about 2 months during the test campaign, not all data needed for
the calculation of the compressor efficiency for PB1 were avail-
able. Fig. 7 Relative corrected compressor efficiency Plant B

062001-4 / Vol. 132, JUNE 2010 Transactions of the ASME

Downloaded 13 Jun 2011 to 138.250.8.97. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Based on the known exchange rates between compressor effi-
ciency and power output, the conclusion can be drawn that the
power output degradation is primarily caused by the compressor
efficiency degradation. Hence, compressor fouling is also the
main reason for power output degradation at Plant B.

5 Effect of a Single Compressor On-Line Washing

The effect of a single compressor on-line washing on the com-
pressor efficiency and the power output of the gas turbine was
analyzed. Averaged data of the time period 2 h before the com-
pressor on-line washing were compared with the data of the time
period 2 h after the compressor on-line washing. The data were
only considered, if the gas turbine had no load changes during the
4 h period.
The corrected relative compressor efficiency of PA2 before,
during, and after one of the weekly compressor on-line washing
routines of the test campaign is plotted in Fig. 9. The compressor
on-line washing starts at 15:06:00 and ends at 16:06:00. Before
the compressor on-line washing, the corrected relative compressor
efficiency has an averaged value of about 0.9973. After the com-
pressor on-line washing the averaged value is 0.9980. Hence, the
compressor on-line washing produces an increase of 0.07%. The
power output of PA2 shows a similar development. Before the
compressor on-line washing, the corrected power output level is at
0.9945 and after the compressor on-line washing, the power out-
Fig. 8 Relative power output comparison Plant B
put increases at 0.9960. Hence, as a result of compressor on-line
washing, the power output increases by 0.15%.
During the 60 min of the compressor on-line washing proce-
As can be seen in Fig. 7, the gas turbine PB2 has the lowest dure, an increase in the power output up to a value of 0.9995 is
compressor efficiency degradation rate. The effect is also a rela- recorded. This increase results from the injected water for the
tively flat power output degradation trend. The compressor effi- compressor washing, which has a similar effect on the gas turbine
ciency and power output degradation rate of PB3 are significantly as high fogging.
higher. During the test period analyzed, gas turbine PB3 had a The analysis of numerous weekly and daily compressor wash-
power output degradation of about 4% in 6 weeks, whereas PB2 ing has shown that the averaged value of compressor efficiency
lost only about 1.5% in 11 weeks as shown in Fig. 8. recovery due to a weekly compressor on-line washing is 0.112%,
The data of Plant B allow a comparison of the effectiveness of whereas the averaged compressor efficiency recovery due to a
compressor on-line washing with water and compressor on-line daily compressor on-line washing is 0.031%. A daily compressor
washing with a detergent. As can be seen in Fig. 8, the power on-line washing has an averaged power recovery of 0.065%, a
output trend of PB1, the compressor of which is washed twice-a- weekly one of 0.21%.
week with a detergent, first follows the power output trend of As expected, the weekly compressor on-line washing has a
PB2. Then the PB1 degradation rate becomes higher and is more higher impact compared with a single daily compressor on-line
like that of PB3, the unit without compressor on-line washing. washing procedure, but after 1 week the fouling process is more
Hence, the effectiveness of the compressor cleaning with a deter- progressed than after 1 day and more dirt has accumulated on the
gent diminishes over time, rendering a daily compressor on-line surfaces that can be washed away. If the effect of compressor
washing routine with water more effective. But during the first on-line washing is compared based on the time interval, the effec-
weeks after a compressor off-line washing, the effectiveness of a tiveness of daily compressor on-line washing is about twice as
detergent is higher than that of water only. This is consistent with high as that of weekly compressor on-line washing—e.g., the
the results obtained by Boyce and Gonzales 关5兴. compressor efficiency recovery 7 ⫻ 0.031% = 0.187% ⬎ 0.112%.

Fig. 9 Effect of a weekly compressor on-line washing

Journal of Engineering for Gas Turbines and Power JUNE 2010, Vol. 132 / 062001-5

Downloaded 13 Jun 2011 to 138.250.8.97. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Table 2 Degradation trends of the compressor section pres- Table 3 Changes in compressor section pressureratios and
sure ratios for periodswith and without compressor on-line compressor section efficiencies
washing
Compressor Sec I Sec II Sec III Sec IV
Compressor section pressure ratio
Section pressure ratio

↓ ↓
Sec I Sec II Sec III Sec IV Section efficiency 
↓ ↓
No compressor washing

→ →
Daily compressor washing
↑  →

the compressor on-line washing period were analyzed and com-

pared with the changes in the total compressor efficiency and
compressor pressure ratio during the same time period.
6 Compressor On-Line Washing Stage Cleaning Ca- For the high pressure Section IV of the compressor, both the
pability compressor section efficiency and the compressor section pressure
As stated, compressor on-line washing can only clean compres- ratio decrease more than the overall values. This is an indication
sor stages where the water used has not yet evaporated. One aim of the compressor fouling degree in Section IV.
of the analysis was also to determine up to which compressor For Section III the relative decrease in compressor section pres-
stage the compressor on-line washing still has a cleaning capabil- sure ratio is higher than for Section IV. The relative decrease in
ity and what the effects of compressor on-line washing are for the compressor section efficiency is elevated, too. Hence, Section III
different compressor sections. has higher relative fouling than Section IV and higher relative
The analysis was based on measured data of the secondary air fouling than the overall compressor.
system. The pressures and temperatures were measured in the ex- Due to the fouling, the loading within the compressor sections
ternal cooling air pipes of the gas turbine. The cooling air extrac- changes. The compressor section pressure ratio of Section II in-
tions are after the 5th, 11th, 16th, and 22nd stages of the compres- creases. The increase in the section efficiency normally indicates a
sor. These data were used for the calculation of the compressor decrease in the reduced mass flow for the same characteristic line
section efficiencies 共see Eq. 共5兲兲 and the compressor pressure ra- of the compressor.
tios of the compressor stages 1–5 共Section I兲, 6–11 共Section II兲, In Section I a degradation of the compressor section efficiency
12–16 共Section III兲, and 17–22 共Section IV兲. can be observed. At the same time, the compressor section pres-
Before the analysis the determined compressor efficiencies and sure ratio increases. Hence, the load of Section I increases and
compressor pressure ratios were corrected for ambient effects. The partly compensates the fouling of the rear stages. A part of the
analysis was based only on stable base load operating points. The degradation of Section I might be related to nonrecoverable deg-
data were measured at Plant A unit PA1. In a first step the aim was radation aspects such as blade erosion as a result of the compres-
to determine qualitatively the differences in the compressor foul- sor on-line washing regime.
ing behavior with and without compressor on-line washing. The analysis has shown that because of the fouling, a tendency
The results show that the compressor pressure ratio of the high exists for the stage loadings to decrease for the rear stages and
pressure Section IV is in line with the overall compressor ratio increase for the front stages. The cleaning effects of compressor
over time. That signifies that the loading of Section IV is indepen- on-line washing can be seen up to stage 11. After stage 11 the
dent of the washing schedule and follows the overall characteristic injected water droplets have evaporated and no cleaning effect can
of the gas turbine compressor. be obtained. To clean the rear stages, compressor off-line washing
For Section III the compressor section efficiency and the com- has to be performed.
pressor section pressure ratio are also in line with the overall In Table 3 are indicated with arrows the changes in the com-
compressor values if no compressor on-line washing is performed. pressor section pressure ratios and the compressor section effi-
Hence, without compressor on-line washing, the degradation of ciencies of a fouled compressor with daily compressor on-line
Section III progresses similar to the overall degradation of the washing regime. The relative changes in section pressure ratios
compressor. However, for a daily compressor washing schedule, and efficiencies are higher for the rear stages than for the front
the degradation of Section III is higher than the overall degrada- stages.
tion of the compressor. This suggests that dirt particles are flushed A similar analysis was also performed for PA2 that had a
from the compressor front to the rear stages. weekly compressor on-line washing regime. The obtained results
The fouling process increases the loading of the stages of Sec- were comparable, but the fact that daily compressor on-line wash-
tion II for both a compressor with and without daily compressor ing is more effective must be kept in mind.
on-line washing. The compressor section efficiency and the com-
pressor section pressure ratio increase whereas the extent of the
increase is higher for the compressor with daily compressor on- 7 Conclusions
line washing. The increasing stage efficiency values indicate a Compressor on-line washing can reduce the compressor effi-
compressor rematching due to the fouling. ciency degradation rate. Compressor on-line washing is no substi-
A similar effect can also be observed for the low pressure Sec- tute for compressor off-line washing, but it lengthens the time
tion I. The compressor section efficiency and the compressor sec- between two compressor off-line washing procedures, if these are
tion pressure ratio increase because of the fouling effects. Again, performed at fixed power degradation levels. Hence, compressor
the increase in the compressor section pressure ratio is higher for on-line washing increases the availability of the plant, if a certain
the period with daily compressor on-line washing. level of performance degradation must not be exceeded.
For the period without compressor washing the degradation is The absolute power recovery of a single compressor on-line
more uniformly distributed throughout the compressor in com- washing procedure that is embedded in a daily washing schedule
parison to the period with compressor on-line washing. In Table 2 is about a fourth of the absolute power recovery that can be ob-
the degradation trends of the compressor section pressure ratios tained with a single compressor on-line washing procedure that is
over time are displayed with arrows, indicating increasing, con- embedded in a weekly schedule. But analyzing the same time
stant, or decreasing trends compared with the total compressor frames and not the single compressor on-line washing procedures,
pressure degradation. it becomes apparent that a daily compressor on-line washing
In a second step, the changes in the averaged section efficien- schedule is more effective by about a factor of 2 than a weekly
cies and the averaged section pressure ratio values of PA1 during compressor on-line washing schedule.

062001-6 / Vol. 132, JUNE 2010 Transactions of the ASME

Downloaded 13 Jun 2011 to 138.250.8.97. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 For a certain time period, weekly compressor on-line washing PA ⫽ Plant A
with a detergent can be as effective as daily compressor on-line PB ⫽ Plant B
washing with water only. Increasing the compressor on-line wash- p ⫽ pressure 共Pa兲
ing frequency with a detergent may provide better results than the R ⫽ gas constant 共J / kg K兲
daily compressor on-line washing with water. Sec ⫽ section
The highest compressor degradation during the period with T ⫽ temperature 共K兲
daily compressor on-line washing is observed in the rear stages of W ⫽ work 共J/kg兲
the compressor, especially between stages 11 and 16. Hence, the VIGV ⫽ variable inlet guide vane
loading distribution is shifted to the front stages. The degradation
without on-line washing is more equally distributed throughout Greek
the compressor. At the first stages compressor rematching over ⌬ ⫽ difference
time has been observed with and without on-line washing. The ␩ ⫽ efficiency
cleaning effects of compressor on-line washing can be seen up to ⌸ ⫽ pressure ratio
the middle of the compressor. Compressor off-line washing can
Indices
also reach the stages that are more to the rear of the compressor,
1 ⫽ compressor inlet
although it could not be proven that a cleaning effect is given up
2 ⫽ compressor exit
to the last stage.
comp ⫽ compression
To keep the gas turbine performance at a high level, the recom-
n ⫽ polytropic
mendation is to perform regular compressor off-line washing dur-
p ⫽ isobar
ing shutdowns, e.g., boroscope inspections, etc., of the gas tur-
bine, since the gas turbine has to be cooled down anyhow on such s ⫽ isentropic
occasions.
From the performance point of view, compressor off-line wash- References
ing should be performed if the gas turbine has more than 2% 关1兴 Hamed, A., Tabakoff, W., and Wenglarz, R., 2006, “Erosion and Deposition in
power loss. If the gas turbine has a high fouling rate, a daily Turbomachinery,” J. Propul. Power, 22共2兲, pp. 350–360.
关2兴 Diakunchak, I. S., 1992, “Performance Deterioration in Industrial Gas Tur-
compressor on-line washing schedule should be considered to re- bines,” ASME J. Eng. Gas Turbines Power, 114, pp. 161–168.
duce the power degradation in the interval between inspections. 关3兴 Meher-Homji, C. B., and Bromley, A., 2004, “Gas Turbine Axial Compressor
Compressor on-line washing has no additional benefit during Fouling and Washing,” Proceedings of 33rd Turbomachinery Symposium,
high fogging operations of the gas turbine. High humidity in the Texas A&M University, Houston, Texas.
关4兴 Tarabrin, A. P., Schurovsky, V. A., Bodrov, A. I., and Stalder, J.-P., 1998, “An
inlet air mass flow can have a certain self-cleaning effect on the Analysis of Axial Compressor Fouling and a Blade Cleaning Method,” ASME
compressor because of condensing and coalescing water droplets J. Turbomach., 120, pp. 256–261.
关10兴. As was observed in different power plants, compressor on- 关5兴 Boyce, M. P., and Gonzales, F., 2007, “A Study of On-Line and Off-Line
line washing may increase the erosion of the compressor blades, Turbine Washing to Optimize the Operation of a Gas Turbine,” ASME J. Eng.
Gas Turbines Power, 129, pp. 114–129.
especially of compressor blade 1 共see also results published by 关6兴 Saravanamuttoo, H. I. H., Rogers, G. F. C., and Cohen, H., 2001, Gas Turbine
Oosting et al. 关10兴兲. The erosion is dependent on blade surface, Theory, 5th ed., Pearson Education Ltd., Harlow, England.
material, and the design of the compressor on-line washing 关7兴 Demircioglu, S., 2008, “Detail Analysis of Compressor On-Line Wash Impact
nozzles. A new generation of compressor on-line washing nozzles in the Gas Turbine Performance,” Master thesis, FH Offenburg, Offenburg,
Germany.
that reduces the erosion effects has been developed and is cur- 关8兴 Bromley, A. F., and Meher-Homji, C. B., 2004, “Gain a Competitive Edge
rently undergoing testing. The overall cost benefit optimum needs With a Better Understanding of GT Compressor Fouling, Washing,” Comb.
to be investigated further unless the erosion problem is solved. Cycl. J. 共CCJ兲, PSI Media; Fourth Quarter, pp. 37–41.
关9兴 Stalder, J. P., and van Oosten, P., 1994, “Compressor Washing Maintains Plant
Performance and Reduces Cost of Energy Production,” ASME Paper No. 94-
Nomenclature GT-436.
关10兴 Oosting, J., Stalder, J., Boonstra, K., Eicher, U., de Haan, A., and van der
Latin Vecht, D., 2007, “On Line Compressor Washing on Large Frame 9-FA Gas
c ⫽ specific heat capacity 共J / kg K兲 Turbines Erosion on R0 Compressor Blade Leading Edge Field Performance
With a Novel on Line Wash System,” Paper No. GT2007-28227.
h ⫽ enthalpy 共J/kg兲 关11兴 Stalder, J. P., 2001, “Gas Turbine Compressor Washing State of the Art: Field
OH ⫽ operating hour 共h兲 Experiences,” ASME J. Eng. Gas Turbines Power, 123, pp. 363–370.

Journal of Engineering for Gas Turbines and Power JUNE 2010, Vol. 132 / 062001-7

Downloaded 13 Jun 2011 to 138.250.8.97. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm