Академический Документы
Профессиональный Документы
Культура Документы
GT2009
June 8-12, 2009, Orlando, Florida, USA
GT2009-59239
THE FOULlNG OFAXIAL FLOW COMPRESSORS - CAUSES, EFFECTS,
SUSCEPTIBILlTY ANO SENSITIVITY
ABSTRACT ,. INTRODUCTION
Incrca<;ed fu~1 COSIS have crealcd a slrong incentive for ga, The use of ga, rurbir.es in power generaion and mechan ical drive
turbine operaton t[) understand, minimize !!Jld oor.trol performance applications has ;rowD signihcantly in !he past IWO decad~s
delerioranoD. l1le mosl prevaknt d~leri or"tion problcm faccd by ga, Compressor f(Juhng and its control is. hy faro me most importanl
turbine operato .. io compressor fouling. Fouling ~auses a drop in "'pie in t he ",ea of gas rurbine performance deteriorat i"n. An overall
airflow, P",,"urc: ratio and compre,"or efticic""y, ","ulong in ~ ~rc trealmcnt of gas tnrbine perf(Jffi,an= dct::rioflltion i. prc,.ented in
malclliog of the ga~ turbine and ::ompress[)r and a drop in powcr
M
Mchet"-Hornj i el al [1]. Kurz and Bnm [2], Kurz el al [3] and
oulput and Ihermal effic ieney. Th is papcr addre,ses Ihe causes and Oihmchal: [4]. An oyerview of fouJi ng detcriormion may be fuund
effects of fouling and proviru,s a eon::prehen.i\"e. tre.alment of the i~ Meher_Homj i [5] and M ~her_ Homji and Bromley [6]
impaot of "alicn t ga.~ rurbine dc"ign pararnetc.. on the msccptibility Thc fon lins ofaxial flow comp~~",o," i. a "~rion" operating
and IiCnsitivity lO oomprc.<.sor fonling. Simnlation analysis of ninety problem and its control is of grcal importanee lO gas rurbine opemlors
IWO (92) g; Imbine, of ranging from a few kW t(J large engine, espccially in the dere~a1ed and highly competilive power m",ket
rated at grcaler than 300 MW has bccn condu~tcd. It is hoped that 11 i, also signiiicam in he mechanical drive mar~et wherc a l(Jss in
lhis paper wil l provide pramical information 10 gas turoine opemtorR. gas tnrbi n~ OUlput can direcl ly affeet plant throllghput. Fou lants in
Ihe l'pu, m"g~ GilL ~ali'>~ uq!USJt" UIl blauin;, rt:,ul lill; in se v~J"
NOMENCLATURE performance detcrioralion. TIte effecl of compressor fooling is a drop
iJ airtlow and compressor i;entropic efficiency. which mm lts in a
B = blooe chord "re_matehing" of the sas lU1bin., and compre.oor eau.ing a drop in
Bit = bladc solidiry powcr oUlput and mcrn::al cffic icncy. In c~{[{:mc ca",,". f(Jul",; can
B I = Tnlet fl [)w angle also lesult in .u rge problems, as il tends lO move Ihe compressor
COP - Compre,"or Dis~hlll"8e Presmre surge line lo the right; i.t. toward the qx:raling line.
COT = Compre,,,,, Di .. ~h"'ge Tcmperalure Estimatcs have placed fOllllng a~ being rcsponsibk for 70 to 85
c;, = spccific heat OlkgK) pc",ent of aH ga. lurbine. perforrnnncc lo",es llCCumulotcd d\lring
Ilpnliliuu. Outl'ut I",,,,s l!t: IW<:t:lI 2 l'eru;lIt (\luJer rliVlll"lib l ~
Dc = Tip Diameler ofAx ial Comprc.~s[)r firsl stage, m.
He = Coeftieiem of Cascade.Entrai nrn~nt oonditioos) and 15 to 2U pen:em (under advcrse condition s) han
GT _ Gu. Turbine been expcrienced
TSF = Index of Sensitiyity lO Foulin ~
ISO = Int"mational Standani, Organi2aoon Thi, paper addres<;c, 1"""0 key issucs:
LHV = Lowcr lleaong Valuc Th~. """""pf;hility "f tiiff~re.nl e'~ ",rhinr. .i..,~i ens to ("ulin!,::
N\VR = ~et work rmio = rutputl W, i.e. th, axio l compre",or's plOpen. ity to foul.
PR = Prc.~sure Ratio The sensithitJ of different gas rurbine designs lo lhe irnpact
t"), - hub/tip rui" forthe fi"t ,tase of fonlin:: : i.c. Ihe effecl offoulin:: on Iheir performance.
Slk = Stokc" Numbe,
T= blade pitch In the pn'l, lhe,e two i,"He, llave ofien b::en wmngly analyzed or
TIT = Turbine Inlet Ttmperalure confmcd. Fouling susceptibility and lite cffe;:t of fouling are quite
\\', , W, - compressor ar,d lurb in" work, kW differenl and are namined in Ihis papero Wh!le sev~ral previous
liD _ Flow Tuming angle studie. haye bcen made of a few relce! S'" turoi ~e thi, is, 10 the
liT"", = Average Total TeJl1lCfatme Ri.,e'stage, oC autliOl'" knowledge, the ti ..t time that 92 difieren! en;incs 01 varyiug
'1. = Corapre,,,,,r EfTieaey sizc and configuration haye becn evalualcd - covering heavy-duty
"frarne'" Iype modcls, aerod~rivatives and hybrids - and includin::
m = mas~ flow rate (kgfste) botli older lechnology unilll as wd l a. ,,,:h-anced m<><lel. with a wide
V = Ax ial Comp"'''Of Volumcl,ic I'low Rate span (Jf characlcri.tic,," Thc ran;e of .alienl parameters u,ed rOllhe
Zc = Fonling Influence Coefficlent analysio is shown in Figure J
The salient relationsh ips of this Figure may be convenicntly
summarizcd as follows:
"" O
O O
~"" l. For a givcn TIT, gas rurb inc spccific work increascs with
pressure ratio, rcaching a maximum and then decreasiog with
fmthcr pressure ratio in crease.
"Il>
.
.' '
2. The spccific work increascs with increas ing rurb ioe iolct
tcmpcrature.
J
~
-. .'# ~ ~
'"
3. The maximum spccific work as the TIT is increased occurs at
incrcasing pressure ratios.
Qldcr heavy duty gas turhine designs tcn ded to have low turbine
"
100 I~ ~ ~ ~ ~ ~ ~ ~ io let tcmpcratures and low cycle pressurc ratios (around 6- 10:1).
SMc/fk: I'obrk, kW'kO'."" Modcm F-class turh ines havc modcst pressure ratios of
approximately 14 to 16:1 but opcratc at high turbinc inlct
tcmpcratures. Thc newcr cngincs have hi gh spccific outputs
compare<! to most commOn acrodcrivative eng in es, but opcrate at
lower thermal efficiencies. Aeroderivative engines tend to operatc at
ROO 45
higher pressure ratios but lower spccific wo rk and at higher thermal
700
1'hIIrm llll l!l
10 cfficicncics.
O
O ".0;>
-
35 ._ The ga~ turhine Net Work Ratio (NWR) is define<! a~ the useful
It!i' ,., O ~. O
omput of the engine d ivided by the lotal turh in e work l . As the
" treatment in scetion 5 will show, the high pressure ratio un its tend to
~
'" W
O
M __ Row 25 be more sensitive to compresoor efficiency deteriorat ion as they
,Ji.'
'" ~ opcrate at lower net work ratios. with the compressor con suming a
greate r pemem (approximateJy 60-65% of the turbine work co mpared
"
W
to approximateJy 50% for a modero heavy duty gas turbine).
A plot of Specific Work vs. Effici enc.v for a number of F-as
100,000 200,000 300,000 400,000
turbine.~ is shown in Fi gure 3. Different models are shown with their
_ . r,kW
pressure ratios an d llTs. Modem acroderivat ive engine.~ (yeJlow
triangles) are c lustered in the spccific work region of 260-350
Fig. 1: Range of saJiellt parameten for 92 gas turbinl'S; kWlkgfsec. Modem F c lass machines exhibit higher spccific work.
Evaluated - Power, Mass F1ow, TIT, PR, Therm.a1 efliciency aud but lower thermal efficiencies Ihan aeroderivat i ve.~. TIlc intem ofthis
Specilk work. figure is to indicate the general reJationship betwcen spccific work.
efficiency and O!her cycl e pararneters. This graph is nO! intended as a
screcning tool for differem engines as il contains engines of differem
1.1 Gas Turbine Specific Work, Efficiency, aud Net Work Ratio. villlages. Its iment is to provide a quantitative feel for the cyde
parameters that were quaJitatively depicled in Figure 2.
To undcntand he trcatrncnt made of fouJing susccptibility and 1.2 Gas Turbine AirOow Ingestion aud Fouling
sensitivity in this papero il is instruct ivc lo vicw differclII cJasr.es of
gas turbines in terms of th eir spcc ific work and cfficicncy and Gas turbines ingest extremeJy large quamities of air, with larger
examine basic design parametcrs that can be chosen by a designer lo machines having airflow rates as hlgh a~ 700 kglsec (l,544 1bs/sec).
mece operational objcctiycs. A schematic of specific work: vs. The linkage between a irtlow rate and compressor fouJing is
hermal efficiency is givcn in Figure 2. on which lines of in creasing ind icated in Table 1. The data was derived by GTPRO' simulation s
TIT are shown running al an ambielll tempcrarure of 15C (59F) and with typical
inlet and out let losses. This table ind icates the ingested amount of
fou lant assuming an ambient loadiog of 10 ppm for a variety of gas
turbines. To help v isual iz.e the huge airtlow, the volume of air
consumed per year is presented in terms of miles aboye a traditio nal
360 x 160 ft (110 x 49 m) football field. As an example, he Frame
935 1FA gas rurbine . (ISO airflow of 14291bs/sec), would i"gest, in a
ycar of opcration, a column of ai r over a footbalJ fi eld 1,950 miles
high (3 120 kM). At a 10 ppm foulalll loading rate, 4.'50,650 lbs or 204
tonnes of foulam wou ld be ingestcd. Axial compres sor work [W, ]
and total rurb in e work [WJ is also shown io Table 1, and it can be
secn that he compres sor scetion COnSumes a large port ion of the total
turbine work .. To quantify this important fact, the la~t COIUDUl of
Tab le I prov ides the WJW, ratio and it can be seen that
approximltely 50 to 60 pcrcelll of the total work produced in the
Speclllc Wcrk, kW ' Kglsec turbine is consumed by its axial compressor. Comequently,
maintaining high compressor efficiency is importalll for ihe plant's
Fig. 2: Effect of pressure ratio aud turbine inJet temperatnn on revenue strCam.
specific work and thernm] efficiency.
1::;!:a ~~~~.
",,~ ~::;=\::\~'\
50
,:: ","= ","'M
~ ~ 11~\ t:i~=
45
40
: 35
,.
~~\"~ ~O.0\O~~. ~ -,;~~.,.
~
o
e
o
E 30
w
11 .7/1104
25 O~ L1I 2500.
PGT2
""" '"'''''''
V94.2 l,,'.~!',E
15.511327
6591C
'000 ..., uu. 12.7/1038 "',., 15.1/1172 14.2I1 1n 11.811149 ".... lt11327
'5
'00 '50 200 250 300 350 400 450 500
Speciflc Work (kWlkg/sec)
Flg. 3: Specific work l'S. elf"tciency for seJttted engines with a.-.sociated PRfflTeql
Paras itic losses and mcchan ical efficic ncy have also beco co ns idcrcd
in lhe model. which accou lltS rOl' Ihe o utput being somewhal le.~s tha n change in the acrodynamie profil e, reduc ing Ihe compressor ma~~
lhe diffcrence bclwecn Ihe turbinc and compressor wo rk. 1\ is useful flow-rate and affccling lhe fl ow coeffi cie nt and efficiency; thus
10 pre.~ent Ihis type of data 10 operators 10 highlight Ihe importance of redu cing the uni t's ovcrall performance. Further. contaminated air can
maintaining clcan axial compressors. cause a hOSI of olher problems Ihal inc ludc crosion, fouling, COlTOllio n
and, in sorne cases. plugging of hOI seclion cooling passages. There is
Table 1: Example uf airflow ingestion uf diffe rent gas turbines also a close co-relalion between mechan ical reliabililY and foul ing
and r'oltio of Cllmpressor work 10 lowl lurbine work.
. -
dClcrioralion, sueh as lhe damaging cffccLS of fouling on bladi ng
-
- -- - -
..::.. ...,-
- -- ~ ~
~
~- ~-
,:::,.
.., ...,
inlcgrity as discussed in Ihe fo llowing sections. This is an()(her
importa nl reason 10 keep lhe compressor clean .
Typieal pholos of fou Jed compressorn are sl'lown in Figure 4.
- ":"'"..= ~
~
M
-......
n"PA -_.
,,,.Jo
..
10.0"
N ,",'
'>.O,,
M,_
.- .
~.,
~"
~
- .-.-... .
" O'?
'"
m ~m
" ,.1.\,
m_
,.=
".'"
M'.
M,m
~
,~.
~
--
..... A
.,-....,..
" ' ' ' U
n~
' o.U1
'J&.'"
"'.1'1" ~ ~
.. ---"
~. ,.~ lt.... ,
""LII
--
n,."'A
onorA
'
m_
"1.'"
,.,n
."
".
""
""
.
' '"
""'>'.
-
~"
'.... ,
,. ~
'''.1<1'! ~.
u,
u,
".
!
M~
An axial comp res sor is a machine wherc the aerodynamic
perfonnancc of cach stage depcnds on that of the previous stages. Fig.5: Cbange in eompressor efficieney and beat rate due to
Thus, when foul in g occurs at the inlet gu ide vanes and the first few compressor fouJing.
stages, there may be a dramat ie drop in compressor performance.
This can often occur when oil and industrial smog or pollen are
present and fonn adherent depos it~. The forward compressor stages
are usually foulcd the worst. If the rear stages foul. thi s scems to 2. UNDERL VING CAUSES OF FOULlNG
have less of an impact on perfonnancc; but due to higher
temperatures. deposits can become bakcd and difficult to remove.
2.1 F ouJants and Contaminants
This h~king "ff~c.r i. mn,..., "".v"re nn rh" high f'I"'""_~sl1re r~tin
compressors of aeroderivative machines ranging from 18:1 to 35: 1
Experience ha~ shown {hat axial compressors will fou l in most
pressure ratio. a~ opposed to the typical \0:1 or 14:1 prcssun:: ratios
operating environments - be they industrial, rural or marine. Thcre is
found on the heavy duty industrial ga~ turbines.
a wide variety of industrial po lJutants and a range of environmental
An axial fl ow compressor is aerodynamicaJl y very sensit ive to a
oonditions (fog. rain , humidity) that playa pan in the fou ling process.
chan;e in airfoil shape and bladc roughr.ess, with surface quality
Compressor fonling is typicaJly ealL~cd by:
affecting the efficiency of energy conversion as discusscd by
Bammert and Woelk [7]. Fouling causes an increa~e in blade surface
Airborne salt
roughness which increases the boun dry layer thickness and increases
Industrial poJlmion - hydrocarbons. fly ash, smog, traffic
thc ri,k of flow separat ion. Blade roughness also cffect~ the profile
exlJaust emissions cte.
losse8.
Ingestion of ga< rurbine exhaust gases.
Zaba's cla~sic work publishcd in 1980 [8] pointed out that the
Minerals such as sand, lime,tone, cement and coal dusts
rcduction in tlow and efficicncy due i() compressor fouling is
Cbemieal fertilizcrs, insecticides, and plan! matter.
dcpendant on the locat ion of fouling. He exam in cd scenarios of
Intemal gas rurb ine oil leaks - leakage from the front
fouling location and dcrived a fouling influenee coeffieient z.c
bcaring of the axial compres sor is a common cause. (see
defined as:
Figure 6) Oilleaks eombined with din in gestion causes
hcavy fouling problcms.
Mi Impure water from evaporalive ooo lers (carryover).
--
~
~~
~.
Output and heat rate variation with the deterioration steps is
i -
shown in Figure 9. The output at the end of the 7'" deterioration step H
r"6
-
has dropped 5.5 MW whiJe the heat rate has increascd by 850 Compre$sor Work
Bru/kW hr. The change in ma~s flow rate, compressor discharge
! -
pressure and compressor discharge temperature corresponding to the
-
s imulatcd deterioration steps is shown in Figure lO.
Output Work
, , , ,
" ,000
,",000 ""'"
,,.00 ~
Oele rl oratlon Ste p'"
31,000 ~ Fig. 11: Chan ge in compressor work, turbine work and overaJl
~
" ,000 ""'" 3 ~
output witb compressor d egradatio n stl'pS imposed on a 40 J\.f\V
~
,
" ,000
" ,000
""" m
''''00 i
~ gas turbine.
Even wilh good air filtration, sal! can colleet in Ihe compres sor
~
O " ,000
32,000
""'"
10 ,600
~
~
scelion, and will continue 10 aecumulate (together with other
foulants) umi l an equilibrium condition is reached. Al this point,
~
large particles will start to break away and are ingestcd into the
31,000
""'" combustion scetion in re lativeJy high concemrations. This in gestion
,",000
2 3 4 5 , 7 """ has to be prcvented by the removal of salt from the compressor prior
to saturation. The rate al which saturation occurs is highly dependem
Oegradalicn Steps ( 1:: tew a rd Clean) on filter performance.
The effcet of hot scetion fou lin g is thal Ihe nozzle Ihroal arca is
F ig. 9: Qutput and heat rate change~ witb compressor reduced. As this comrols the compressor-turbine match, it causes a
d egra dati on steps im posed on a 40 J\.fW gas turbine. movemem away from the design match poim and results in a
corrcsponding loss of performance. Deposits will aJso form on the
;; = =
---
~ rotating blades causing a further loss in pe rformance.
~
~
Also, as the turbine section of a ga~ turbine fouls, there will be a
, drop in Ihe turbine flow coefficiem and the compression ralio of the
&
-
compressor will increase a~ the turbine "swallowin g capacity is
o = ~ reduccd. Note Ihat in sorne OEM control systems, the compression
o = <o
ratio and exhaust gas temperature are uscd to determine the turbioe
! c"' ~ c inlet temperature; Zaba [8]. This a lgo rithm is based on an
---
~
~ assumption of constam turbine effic iency, and if this efficiency
"
!
,~
,~
CDP
,i. dcereases due to turbine fouling, then the comro l system will indicate
a higher turbine inlel temperature than is really present. Therefore the
,.
-8 output of Ihe turbine is funher reduced.
! ,ro
"
ca 0 .6
t7.2:1
0.480
.. Il.S:t
"
0.422
t2.9 ,t
"
0 .422
.
tO.6:1
0520
in power will be noted (as the ga~ generator can spccd up) than if the l!.T por . ... 30, K 28 .20 2S.90 24.60 23.34 19.2S
eng ino:: is on tempcrarure control. ~d" . oft " ' '''ie , rn 0.4400 0.73S6 0.S7t2 2.36S0 2 .t730
Nomina l mul<>n 'p"0d of tS,OtS 9,160 t2,420 '.~ '.~
,a<
4.1 Gas Turbioe Desigo Paramete rs
,se
,~n a nUof, rpm
8." ..., $ .72 1.36 U,
Computational results prov ided by Tarabrin for a gas turbine
Tarabrin et al [10,11] and Song et al [16, 17] concluded tha! the operating under ISO conditions and having a mass flow rate of 35.47
rate of partide deposition on blades increases with grow in g angle of kglsee at a speed of 12420 rpm, a pressure ratio of 12.5:1, a O, of
attack, and that smaller eng ines exhibit a higher susceptibility to 0.57 12, a hubltip ratio of 0.422 and an ISF of 5.72 are presented in
foul in g than larger engines. Funhermore, the susccptibility to fouling Figure 12. This figure indicates the ehange in adiabatie effieieney
increases with increasing compressor stage head (CpI'l.T). Dcsign and pressure ratio as sueccssive stages foul for the first 6 stages. As
pararneters that will determine the inherent susceptibility to fouling can be seen, after the siuh progressive stage of foul in g, there is a
for a specific compressor des ign indude air inlet ve loc ity at the inlet 4.5% reduction in mass flow. a 4% reduction in pressure ratio and a
guide vanes. compressor pressure ratio, aerodynamic and geometri cal 2% drop in compressor efficiency.
characteristics.
'.
4.2 Gas Turbioe Fouling Susceptibility
~m
Tarabin et al [11] have proposed that an "Index of Sensitivity to
Fouling" (lSF) can be calculated for axial compressors as follows: /
mCp~T,,,,g~ ><10-6 n ~ p:
""
p.
ISF = (2)
(1
2
rh )Dc
3
,
In examining this cquation, high loadin gs and small channel I'~'~',
""
dimensions increase fouling susceptibility, Accordin g to Tarabrin, , .... . Of"
and a~sum in g operation under s imilar env ironmental conditions and ~ " t~'
with the same level and quality of air filtration, engines with higher "",
ISF values would exhibit a greater reduction in airl10w compared to ,
...... t!tl. '.--"""
eng ines with lower ISF. Note that although this is really an index of " ,
"susceptibility" to fouling, it a loo influences engine performance.
,
Thus, rcduction in discharge pressure and efficiency will aloo be
~" ..I I I - 'f1!~
greater for engines with higher ISF values. Tarabrin' s ISF factor is
ba~ed on the fo ll ow in g underlying principIes:
,
0.1 0.15 O. ' D.I, 1.0 1.15 G
The dcposition of partides on the surface of bladcs takes place
under the action of inertia forces acting on the partide an d F ig. 12: M odificaon of Co mpresso r Adia bac E ffi cieney a nd
forcing thcm to move al on g the curved strearnlines. Parti dc.~ of PTessure Ratio with Simula ted Foulin g Stages (Tarabrio et al,
din that oollidc with the blade can stick to the blade surfacc. [996)
Sticking tendency is increased by thc prescncc of o ily
hydrocarbons tha! may have deposited on the blades. The The relationship betwcen the ISF factor and key parameters such
coefficient of cascade entrainment as dcfincd by Tarabrin is: as temperarure rise per stage. hubftip ratio and tip diameter are shown
in Figure 13. Newer gas turbines tend to operate with high work
E<= (0.08855 Stk - 0.0055). (bit) . Sin(ABI2)f Sin B I (3) stages and inerea~ed Mach numbers, and several of the modern high
performance machines have tran sonic early stages. It is important to
Where ll.B =fl ow turning angle note, therefore, that the Delta T pcr stage is not uniform in modem
As the design compressor head (worklstage) increases, the flow axial flow eompressors as the blade loading may vary throu gh the
tuming angle AB and the ool idity of the cascadc typicall y compressor.
increa~e. By using corrclations for the sol idity and axial flow
vclocity, and as the head per stage is Cp(AT), thi s resu lts in this
"~-------------------------------,
o 400 800
'.-
1200 1100 2000 2400
'" 1 1 1
-0.005
la 6 ..0.0\ 1\ ......., I
, :.0.015 \\ IG-k"-:.
, g -0.02 \'
,, +-------~------~----------------'
"
,,, ,,,
!.(J.025
.0.01
-"-" r----
CcmpresSOl' Delta T/Slage -0.035
-0.04
0.6
r- -+- Concave side
-o- Convexside
r
turbioe parameters. 0 .5
"
~
e
o
~ 0.4
4.3 FouJing Degradatio n Rate
ExpcricnCl': has shown tha! fou ling tends lo occur during nitia]
~
c.
0 0.3 \\
0
\.\
..
opcration and roughly fo llows an exponcmial Jaw, stabiJizing aflcr m
abom 10CJ0-2000 hours. An emp iricaJ formula proposed by Tarabrin E 0.2
el al [11] is
~Power = a [ l -e--l<] (4)
0
o 0. 1
'\
c.
Whcre,
a= 0.07, b=O.OO5.
is showII in Figure 14; (Tarabrin el al., [1 0]). Rotor Blades in a two-shaft Fmm e 5 gas turbine. (Tarabrin et aJ,
[10])
0.4
-,- --- _._-- -
1.4
1 _ Concave side ~
-8 1.2 " ,
"
1 -:- Convex s ide
~ 0.3
/ . /
41 1.0
e ,"
:.
8. 08 "- ,,,
o 1\
o 0.2
w
g 0.6
1\\ , - - - cylin der model
m
E 0.4 \ 0. 1
,I
I
._.- by Tara brin et a1.(1998a) _
,,
model1
~ --- model2
IGV 2 '"
3
Stage number
4 5 6
O 10 20 30
Particle d iam eter, d p (p.m)
40 50
0.5 Pcrcent change in mass flow is approx imateJy cqual to the pcrccm
change in compressor efficicncy if aH the stages are equa[[y
o. fouled.
.
T ABLE 3: Coefficient of lnfluence ofGas Turhine Parameter:s
the fronl stages and influence downstream compres sor stagcs.
......
ConcIiIiono
... ~ n
T 1223"K n. conll
~_'2.5 T _coros, 2.12 '4' _1.25 1.25
U'G
,~ T, _ 122301{ T, . toril!
" !.SS
207 2.07 . 1.'
~nll
12.5
.,, 1
E "Comp output, 2-shafl .2
...;; 0.98
"
..
~~
F :: Comp oulpul, 3shaft 0.96
~
0.94
2l
" 0.92
/t.
Fig. 21: Influl'nce ofaxial now compressor fouliog on efficiency
0.90
and power for different shafl configurations (Tarabrin [10]).
",-,-:: ~ A
"llN 0.88
"".
~ f::!c
Note Iha! in Ihe case of a two-sh aft turbinc wilh variable powcr
rurbinc nozzle.~ (as may be foulld on mechanical drive sp li t-shaft
.. 0.86
"
machines), Ihe compressor spccd can be kept dose 10 constan! by D
E 0.84
varying Ihe powcr tllrbine area - hus rcducing Ihe powcr drop lO O ~E
2.82%, as calculated for Ihe single-shaft eng inc. Kurz el al [3] have Z 0.82
dcveloped sim ilar concJusions bascd 00 simulation studics. F
0.80
~
C. Conventional "E-Class" medium oulpUl, sin gle-shaft GT
D. Conventional "E-Class" small Olllpul, sing le-shaft GT ~ 0.99
E. Aeroderivali ve wilh 2 compres sor spools and a power turbine ie 0.98
~~ -...;
F. Aeroderivali ve wilh Ihree compressor spools
..
0.97
~ ~
(Nole Ihal B is a sca led-up vers ion of C. and D is approx. half Ihe o 0.96
power of C) i;::...
As before, lhe imposed degradalion sleps were: E
W
0.95
0.94
,"'; '0 A
Slep 1: New an d Clean,
~ ""
1"-' -
"- 0 Eg
0.93
Mass flow drop = 0%, Compres sor effic iency drop = 0%
Slep 2: Mass tlow drop = 1%, Comp. effic iency drop = 0.833 %
~
1-
0.92
0.91
Slep 3: Mass tlow drop = 2%, Comp. effic iency drop = 1.67 %
Slep 4: Mass tlow drop = 3%, Comp. effic iency drop = 2.5 % g 0.90
'F
Slep 5: Mass tlow drop = 4 %, Comp. efficiency drop = 3.33 %
~ 0.89
Slep 6: Mass tlow drop = 5 %, Comp. efficiency drop = 4.167 %
234567
Slep 7: Mass tlow drop = 6 %, Comp. efficiency drop = 5%
Degradatlon Steps (1 = New & Clean)
The fo ll owing assumplions were also made for all simulalions:
2 Runs as defincd in point (1) aboye were then made at an ambient "0.25
+-----------------------------"
o.", 0.35 o.", o." 0.55 o .~
tempemture of 40C to examine the impact of amb ient
tempcrature. GT Nel Work Rallo
3 AH gas turbine salient operating parameters were loggcd into a
spreadshcct for analysis.
4 Using the spreadshcct, computations were then made of Fig,25: Gas turbine Det work ratio vs. GT thermal efficiency.
deterioration levcls in terms of power and heat mte, and other
parameters such a~ the temperature differemial per stage and
fouling factors as defined by Aker and Saravanamuttoo and
Tarabrin. 5.3 Power, Real Rate and Fuel Consumption Cbanges
Outputi n kW : ~i '
~~~ . ... ~ ,
,
, , ,
NWR (Net Work Ratio) (6) I ,\
W, ,~ ,
'"
Another way of expressing th is rat io is:
NWR = W, - Wc = 1- Wc (7)
,
W, W,
50,000 100,000 150,000 200,000 2SO,000 3OOpoo 350,000
NV/R rcquires knowlcdge of the turbine scetion total work and tan [Oot. r lor _ d Brtp"', kW
the work consumed in the compresscr, an d these are factors that are
not eas il y available to most users. The qualitative rcJationship of
NV/R with the ga~ turbine specifie work is shown in Fi gure 24, and it Fig. 26: Per~nl Power Reduction (due to imposed fouling) vs.
can be seen that high specifie work turbines tend to have higher net gas turbine Non-Deteriorated Qutput in kW (- size).
work ratios. The rcJationship betwccn NV/R and thermal efficiency is
also depicted in Figure 25. As discusscd in scction 1.1, there are low The pereent powcr reduct ion can also be secn 10 be invcrsely
correlated 10 the nel work ral io as shown in Fi gure 27, with engines
having low net work ral ios lendin g lO be more sensitive to a given
amount of imposed fouli ng.
,I 0.8 , -- - - - - - - - - - - - - - - - - - ,
. ,. ... . w......\".. ..
~ 0.7
" 0.&
"w ,8. .....
"' ...."
0.5
:", "", i .. ""t"
,. '. ~ 0.3
0.4
,
,;
o
~
0.6
O,
.' ~ :tf~~
,~ 0.2 '--_--_-_--_--_-_-~
50,000 100,000 150,000 200,000 250,000 300,000 350,000
~
"-o
O.,
. :~.\i.. Non..(Jelenorated Outpul, kW
1":
J"'
Saravanamuttoo [13] was suggestcd a~ an indieator of foulin g
scnsitivity; i.e. with higher indexes imply in g a greater scnsitivity to
foulin g. However, in examini ng Figure 34, thi s faetor does not sccm
to be strongly indieative of deterioration in terms of compressor
t 0.4 diseharge pressure. Aloo, when viewcd with respcct to power
rcduction (Figure 35) the eorrelation seems to be oppos ite to what
"' wa~ stated in the Scddigh and Saravanamuttoo paper (i.e. that a
higher fouling index implies a greater sensiti vity to fouling).
0.30 ()3~ 0 .40 0.45 O..'iO 0.55
In examining the Fo ulin g Index, it is esscntially the specific work
ar_WorkRotIo divided by the eompressor stage enthalpy risc, and in examin in g
Figure, 3S one ean "ee that the power roduction (sen" it ivity) BOOm" to
Fig.31: Power drop per oC for 92 gas turbioe, as a function uf be less fo r the high spccific work machines; i.e. those that com:spond
GT Del work ratio. to higher NWRs.
Tho:: powcr rcduction wilh fouling detcrioration vs. Itas rurb in e nel
16 ,
work ratio for two ambicnt tcmperaturcs of 15 and 40 is shown in e , ,
.-
Figure 32. TI can be secn Iha! me effcct of fouJing is a function oflhe
nc! work ratio and ha! jI is aloo more sevcrc al high temperarurcs . A
sim ilar plo! showing Ihe heat rate changc is providcd in Figure 33.
, ' .". -:,~~
~ . '~ '-'Y.'.
, .. .'
-
A
A
,
"
""
I "",
""
,."
"
~
e
2
O
O
4. A Ai t : t t
,
"
Fouling ndex Par.meter
A
"
..'"
20
"
4
I
Flg.34: CDP reduction % and power reduction % \"s.. Foulin g
lnd ex Parameter.
0.20 0.25 0.30 0.35 DAD DAS 0.50 0.55
"
.~
~ ~
.; \ .... ..;.:.. . .
A
+ "" ,
....
Fig. 32: Powl'r reduction wilh compressor deterioration (ur 92
, -
.
.~
gas turbioe as a functioo uf GT oet work ratio fur two dilTerent
ambient temperatures (15 and 40C).
~
~
~ ~\ r..:~ ....( .
.. :G
,
i
1 ~~)\w:t'P:". "~"
~
I "
.~
l
" ,-- - - - - --cc-- - - - - - - - ,
.~
.~
,. ~ ~
-
, . . . _ Work, kW"'G'IO.
~ ~ "
7
r
'80
0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5
I I
Tarabrin ISF Faclor
'70 - Clean c o m pr...
~ Fo uled compro
~
f--
~ ,.o '--,
F ig.36: Power R eductio n % due to fouling (green tria ngles) a nd
N\VR (yellow squa r e:s) vs. Tarabrin ISF fa ctor .
,
~
r----, r----, ~r----,
'50
~
..
' .00
~
0.99
0.98
......
.~ 7.,,1: ~
~~-
-..--
~
~
~
'40
~r----,
Q
~
0.97
0.96
'\
..at.' ..
. ..a........
-
-
'lO
'20
z .o
".
0.95 40 50 70 80 90
~
- 30
0.94 Comprellor Inlel lemperllure , F
0.93 Fig. 39: D rop in Comp resso r Discbarge Pressure befo re a nd
0.25 0.30 0.3 5 OAO OA5 0 .50 0 .55 aner fo uling.
Deslgn NW R
! ,
dCleriornlion.
A dctailcd Irc31mcnl of losses is prov ided in Koch and Smith (23] .
A trealmenl of roughncss cffects on !he acrodynam ics is provided in
i 'tt
3 - Barnmen an d Woelk [7]. An interesting lreatrncm of lhe effcct of
" Reyoold~ number and blade surfaec rollghness is provided in
I 2
Sehafflcr [24]. Ao interesting sel of papcrs ha~ becn provided by
Syvenud et al [25. 26J and Millsaps ct al [27].
0.5 1.5 2..5 3..5 4..5 5.5 6..5 7..5 8..5 9..5
Fig. 38: Heat Rale increase due lo fouJins lS. Tarabrin ISF
factor.
6.2 Fouling, AirOow Dislomon and Compressor Su rge compressor, foulants can enter the coo lin g system. Cement
dust, coal dust and fly a~h can be responsible for this problem.
As fouling reduces the mass flow (flow coefficient) in the first The effects can be improper cooling an d acceleratcd thermal
compressor stage, the performance of the laller stages is affeeted and fatigue, though typ ically the effeets are gradual in nature.
the opcrating point on thc first-stage characteristic moves towards the
left on the compressor map specd line, thus increasing the first-stage Foulant buildup on compressor blad in g can lead to serious
pressure ratio. This causes a higher air density at the in let to the corrosion problems, especially when humidity is high.
sccond stage, with a eorrcsponding funher reduction in the second Airbome salt may be ingested as dry pan icles or as droplets
stage flow coefficienl. This effect progresses through successive of brine, and moisture wiJl be absorbed during high humidity
stages until a rear stage stall occurs, and surge is triggercd. Dctails on operating conditions. Salts, mineral acids and aggressive
the effccts of stage characteristics on axial flow compressor gases (e.g. SOx, NOx, Ch, etc.) alo ng with water can cause
performance may be found in Stone [28]. pitting of compressor blades due to elcctrochemical cOTTosion
Dundas [29] has eonductcd a detailed analytical investigation into meehanisms. This may lead to local stress raisers Ihat can
the deterioration of turbine operation inelud in g drop in eompressor dim inish blade fati gue Jife. Under Ihese typc condition s it is
effic iency, fou ling, first stage nozzle d iston ion, intemal bleed seal important to perform frequent compressor wash in g (on-l ine
deterioration, tumine effieieney reduction, inlet filter foulin g and low and off-li ne) 10 prevent Ihe accumulat ion of eorrosive
fuel heating value. These pararneters were examined to study the materials. Compressor blade coatings can also be of value
effeet on the tumine opcrating lin e, and his invest igation eoncluded here. Note, however, Ihat on rclalive ly small ga~ turbines a
that compressor fouling had a pronounccd effeet. 0.1 mm coatin g applicd 10 the blading can cause a flow
Whereas fouling causes the operatin g line to move towards the reduclion of 10 percent and a reduction of compressor
surge line, there are other factors that can calL~e movement of the effic iency of 5 percent.
surge line itself. Eros ion of compressor blading can affect boundary
layer development and increase the tendency toward separation. Stall 5-10 micron panicle siz.e rcpresents the tran sil ion zone
can therefore occur at a lower in cidence angle than with smooth bctwecn fou lin g and erosion, and particles causin g eros ion are
compressor blading. Heavy erosion can aloo reduce blade tip chords: normally 10 microns or greater. (Note that 10 microns = 1115
thereby reducing blade tip so lid ity, wh ich would adversely affect diameler of human hair). Erosion impairs aerodynamic
stage stability. performance and can affeel the mechan ical strenglh of blades.
The importance of considering foulin g effects on surge becomes Eros ion fi rst incrca~es blade surface roughness, thus lowe rin g
more important with the use of gas turbines in comb incd cycle effic iency slightly. As erosion progresses, airfoil contour
cogeneration app li cations, and also wilh IGV contro l and slcam changes occur at the leading and trailing cdges, as well a~ at
injeclion appl ication s. Closure of !he IGVs durin g part load blade tips. Severe erosion has alm bccn known to cause
opcration restri cts airtlow and this, in conjunclion w ilh severe changes in blade natural frequency.
fouling, can pro mote surge. Steam injection results in a higher back
pressure on Ihe compressor, thus movin g the opcrating point closer to A detailcd treatment of gas turbine blade failures includ in g
Ihe surge line. underlying causes and troubl eshooting is provided in Meher-Homji
There have becn several ca~es where excessive distort ion of the and Gabriles [3 1].
in let airtlow ha~ triggered a surge event resulting in compressor
damage. Icing, for example, can cause un eve n inlet circumferential
distonio n. or uneven clogging of filters - poss ibly due to a bend in 6.4 Foreign Object Damage (FOD)
the air inlct duct before the filter or impropcr inlct system des ign o
Studies and resu llS rclating to the flow in in lct ducts are given by Although nOI d ireelly linked to fou ling, this subjeet is mentioncd
Manfrida, el. al. [30]. bccause it can be caused by a loss of fi ller integrity. Damage is
typ ically caused to Ihe forward compressor litages, although in sorne
cases Ihe foreign object can also work its way 10 rcar stages. Damage
is a function of foreign object size and composition, blade
6.3 Effects of Fouling on Compressor 81ading Inlegrity construclion and impact localion. 1I can lead 10 dirccl or sccondary
failure. Foreign object damage can be caused by ice, failcd intake
Wh ile fouling cannO! be said to be a major cause of blade failure, il scclion componcnts, or by materials and tools left in the inlel plcnum.
can contribute to various blading prob lems a~ indicatcd below:
6.5 Corrusion Effects
By promoting surge o r rotating slall wh ich may have a
dangerous effect on blades. In coastal and marine environments the ingestion of airbome sea
salt (primarily NaCl) is oflen a major cause of compressor blade
In sorne cases. blading natural frequencies can be affectcd by corrosion. Even wilh effective a ir fillration, Ihe conditions of fog,
the increa.sc in mass due to din buildup. Excessive d in on the humidity or rain can cause migration of salls through the in let filter
blades can also cause imbalance, and a cons.equent increase in (lecching) an d into the compressor. A detai led study of salt
nmning spccd vibrat ion. pcrcolation through gas turbine air filtralion syslems has becn
reponed by Stalder & Sire [32].
Di n accu mu lalion belwecn bearing surfaces of the blade root Water evaporates as it moves through the compressor (causing an
may cause blades lO operate in an abnormal pos ition - thus increase in chloride concentralion), an d in sorne cases crysta lline sa lt
adding to the Slresses. If the root conslraint is changcd due 10 deposit~ can be found on compressor blading, as shown in Figure 40.
build-up in the fir tree region, a change in natural frequency NOle that dry salt is nol corrosive to the compressor, and Haskell [33]
eould also resu ll as Ihe boundary cond ilion changes. states Ihat cOTTos ion is rarely observcd beyond the 81h compressor
stage, as no moisture wi ll survive at the operating temperatures
Panial blockage of cooJing passages of hot scction stators and beyond Ihis point. However. rehydralion of sall deposit~ can occur
blades can be caused by fine particle size foulants (typicall y durin g shut-down o r standstill pc riods, a~ a result of water absorption
less than 5 microns). As Ihe cool in g air is bled from the by ambient humidity. This causes the re-format ion of aqueous salt
solm ions. th us in itiating e lcctrochemieal eorros ion and resulting in Table 4: Effect of ambient gases on acidity at tbe com pressor
local izcd pittin g of oompressor blades as shown in Figure 41. Note inlet (Based on Haskell, [33]).
that in salty environ ments, the fati gue strength of steel can drop 50-
60 pereent, an d this situation is worsened when notches due to Sulfurou. acid
oorrosion pitting are present.
Am blent SO. IPpb) DIssolvll'C! SO,IPpm) p.
An additional and potentiall y se riou s oonsequence of salt
in gestion is its effcct on turbine hot section componenK As 1 02
memioned aboye, dry salt is not corros ive to the compressor, blll if
salt deposits break-away during opcration and pass imo the
oombust ion section, catastrophi c high-temperature corros ion fa ilures
10
100
1,000
O."
...
2.0 ...
' .0
'.0
of turbine b ladin g can occur. This type of hot corrosion is known as 10,000 19.8 3.'
"sulfi dation". and results from the combined effect of fuel sulfur and Hydrochlorlc acid
trace quantities of alkali metals such as sodium. Normally it is Amblent HCt IPpb) DIssolved HCI (ppm ) p.
oontrolled via the OEM fucl spec ification; Le. by limiti ng tota l alkali
1 1.... 1."
metals in the fuel to a maximum, wh ich is typicall y in the range O. 1
to 1.0 ppm depending on the particular ga~ turbine model. However, 10 5,500 O."
if a irborne sodium en ters from the compressor. even trace amounts 100 17,600 0.44
can cause the OEM contaminant limit to be excecded. For examp le,
at a typi cal air/fuel mass ratio of 4Ofl, an airborne sodium
ooncentration of 0.OO5ppm (5ppb) en teri ng from the compressor is 7. DETECTION OF FOULlNG
equ ivalem to 0.2ppm sod ium in the fuel; whi ch is a typical maximum
li mit for most aeroderivative and advancc-c lass gas turbines. Gas turbine manufacturers an d operators usually devclop
Additional in format ion on gas turbine hot section cofTosion is guidelines to define whcn fou lin g deterioration call s for corrective
prov ided by Bromlcy [34]. action. This is based on a combination of load and exhaust J!as
In polluted urban and industrial environments, the ingestion of temperaturcs (EGT). Users also monitor compres sor discharge
acidic gases suc h a~ SOx, NOx and HC l can a lso create very prcssure and compressor effic iency. Graphs can be plotted to show
aggressive condi tions on compressor blading - e ven if the airtome expccted (c lean) versus mea~ured parameters. However, it is the
ooncentration of these pollmants is in the "parts per bilJion" (Ppb) opinion of sorne operators that the on ly way to detect a fouled
range. Again. the effect of even modcrate relat ive humidity can comprc.~sor is by visual inspection. Unfortunatcly. though. with most
create extrcmcly low pH con dition s (Le. ac idic) resu lting in turb ine designs thi s means shutting down the unit. removing the in let
comprcssor blade corrosion; sce Tab le 4, which is based on wOJk by plenum hatch an d visually inspccting the compressor in let,
Haskell [33]. bellmouth, in let guide vanes (IGVs) and visible early stage b ladi ng.
7. 1 lndicators of Fouliog
The most sensiti ve parameter of the aboye factors is the mass flow
rateo The real prob lem is to detect fouling in time to prevent a
sign ificant power drop. Several phi losophies are in use. Sorne
operators believe in regular periodic wash in g of the machine,
whereas others base the washing requirement on a ccrtain set of
performance parameters. The phi losophy uti li zcd is a function of the
normally expccted fou lin g levcls and thcir severity. washin g
effcctiveness and plant operation criteria. Measurement of air-intake
dep ression is also a practical an d cconom ical mcthod for fixed
gcometry machines. The tcchnique involves measuring intake
Fig. 40: CrystaIline sea salt on com pres..'lOr blading.
dcp ression as an ana log of airtl ow rate. In thi s approach, the gas
turb ine in let bellmouth is uti lized as a flow meter. This approac h has
bcen suecessfu ll y uscd by Scott [35, 36].
By mcans of suitab le software, data availablc in the ga~ turbine
control system can often be used lo monitor compres sor deteriorat ion.
An application of such an analysis is p rovided in Dusatko [37]. where
the general re lat ion betwccn compressor discharge pressure CDP) vs.
compressor in let temperature (CIT) can be found. This data was
taken over several hundred hours of opcration at full load con dition s.
For a Frarne 7EA gas turbine Dusatko noted that a 2 psig loss of
compressor disch arge pressure is cquivalcm to one gross MW of
power. The effect of a crank wash on the CDP vs. CIT plot is also
shown in this papero
A common approach to trend ga~ turbine olllput is to use a power
capacity factor defined a~
Fig. 41 : Pitting coITus ion on compres.'iOr blading.
Power Capacity Factor'" MW Acw.VMW furtkd (8)
1be MW expecled i5 liJe ~Ied "new and d ean" performance I)' pe of servicc !hal liJe ga~ lurbi nes are in musl also be oonsidered.
les! OUlpUI. which is typically 3-5 percc nl higher man the "guarnntcc" For example. by nOI having lhe abil ity to shUl down for erank WIl5he....
ompul. Corrcctions shotild be made for IPP opc:ralors and merchant power plant~ may neal 10 be more
aggress ive in eontrolliog fou ling via on lin e wash progra ms. A
Inl et tempcr~ture valu able and useful set of papc:rs rcl ating lO eom pre.~sor wa.~hing llave
Inlet pressure bec n provided in Bagshaw (45]. SIa!der and van Ooslen (46]. and
Specific humidity Sla lder (1 5].
NO, watu injection rate.s Severa! different methods of gas rurbine compre.~sor c leaning
Inlet and outlet Pre.<;sure drops ha ve been applicd ovcr (he ycars. bU I ''Wet cleaning" ha~ becn found
Specd oorrcctions (thi s is nOl: a major iss ue for power lO be by far the mosl appropriate and effective tcchn ique. Ho wever.
geoenllion applicalions wim single shaft ga.~ rurbines uole.~s loday's sophisticatcd large industri al engines and bladc coolings
$ign ificaIII off. frcqueocy opemtion is experienced. This rcquire carefull y dcsignc:d washing syslems 10 ensure opet"lIlional
happens in sorne couIllrie.s due 10 grid ''unde r.frequency'' safely. rc:liabili ty and optimum efficielley. T wo differc:nt wet cleaning
prob lems). tcehniques are ge neralJy appli ed. known a~ off- line (cran k) washing
a nd o n-Jine was hing. Under cxtreme fou li ng cond itions. hand
A 1m.'iCJine developmenl proce.s.~ and the use of scatter pl<MS for cleanin g o f !he IG Vs should also be peormed if time permil~.
deterior,lIion analysi~ of advllIKnl F dass rurbines is providcd in During overbauls. hand cleaning of lhe fu lJ axial compressor is mOSI
Meher-Homj i el al, (38 ] and a detailed analysis of actua! data on a effective
Frame 7EA instalJation is provided by Gulen et al [39].