1989 Aker Predicting Gas Turbine Performance Degradation Due To Compressor Fouling Using Computer Simulation Techniques
1989 Aker Predicting Gas Turbine Performance Degradation Due To Compressor Fouling Using Computer Simulation Techniques
1989 Aker Predicting Gas Turbine Performance Degradation Due To Compressor Fouling Using Computer Simulation Techniques
Aker1
Research Assistant.
H. I. H. Saravanamuttoo
Professor,
Mem. ASME
Department of Mechanical and
Aeronautical Engineering,
Carleton University,
Ottawa, Ontario
Introduction
Background. The fouling of gas turbine compressors is recognized as one of the most common causes of engine performance deterioration facing users today (Upton, 1974). It is
now apparent that even units operating in the benign atmosphere of rural or nonindustrial settings can still become fouled.
Typically, dust, insects and pollen are the culprits, which when
mixed with engine exhaust and oil vapors from both internal
and external leaks, form a sticky mass that readily adheres to
the blading and annulus areas of the compressor. The principal
effects of compressor fouling are reduced power output and
increased heat rate. More insidious effects include an increase
in the turbine inlet temperature at a given power setting and
a decrease in the compressor surge margin.
However, because the principal effect of compressor fouling
is a reduction in power output and an increase in heat rate,
neither of which is usually measured, many engines operate
for significant periods of time in a fouled and thus uneconomic
condition. Initially, this was not a significant problem in that
fuel was relatively inexpensive and regularly scheduled maintenance took care of the problem before it reached the critical
point. Presently, oil and gas prices are such that previously
insignificant inefficiencies have been translated into considerable additional operating costs.
The solution to this problem was to perform regularly scheduled compressor washes. If, however, regularly scheduled
washes are performed too frequently, they can lead to unnecessary expense in terms of down time, increased maintenance
'Presently with TransCanada PipeLines Limited.
Contributed by the International Gas Turbine Institute and presented at the
33rd International Gas Turbine and Aeroengine Congress and Exhibition, Amsterdam, The Netherlands, June 5-9, 1988. Manuscript received by the International Gas Turbine Institute August 1, 1987. Paper No. 88-GT-206.
characteristics, consisting of pressure and temperature rise coefficients as functions of a flow coefficient, to define the performance of a rotor-stator pair or stage. The flow coefficient
is defined by the axial velocity and blade speed at any given
stage. The value of this flow function defines a stage pressure
ratio and temperature rise from the pressure and temperature
characteristics and the local flow properties. The stage inlet
pressure is then multiplied by this stage pressure ratio to provide the stage exit pressure. In a similar manner, the stage
temperature rise is added to the stage inlet temperature to
provide the stage exit temperature. These exit conditions are
then assumed to be the inlet conditions to the next stage where
the process is repeated. This process continues until the final
stage in the compressor where exit conditions determine the
compressor delivery pressure and temperature.
The selection of stage characteristics is an important aspect
of any stage stacking procedure. Obviously, the engine manufacturer's stage performance evaluations would be welcomed;
however such information is highly proprietary and is usually
unavailable. It is therefore necessary to obtain whatever information is required from data reported in the open literature.
From this information a generalized stage characteristic can
be derived by normalizing the coefficients to their values at
the maximum efficiency point and plotting them on a single
graph. The characteristics tend to collapse into a relatively
similar curve. The stage performance of any given compressor
can then be obtained from this generalized characteristic by
multiplying through by the appropriate coefficient values of
desired stage.
The performance of the clean compressor was obtained by
modifying the stator angles and blockage factors and restacking the compressor until the design speed line intersected the
mass flow rate and design pressure ratio specified by the manufacturer. A plot of the compressor maps generated by the
stage stacking model for the GE LM2500-30 and the Solar
Centaur are presented in Figs. 1 and 2, respectively.
Simulation of Compressor Fouling. In continuing their investigation into compressor fouling, Lakshminarasimha and
Saravanamuttoo (1985, 1986) simulated the fouling of individual compressor stages in the NACA five-stage compressor
described by Sandercock et al. (1954). This was done by adjusting the stage flow and efficiency characteristics to reflect
the loss in performance associated with fouling. While the level
of adjustment was somewhat arbitrary, this proved to be a
valid method of modeling compressor fouling.
While several authors have investigated the effects of fouling
on compressor and engine performance (Saravanamuttoo and
Maclsaac, 1983; Saravanamuttoo and Lakshminarasimha,
1985; Lakshminarasimha and Saravanamuttoo, 1986), most
of these investigations have dealt with the situation as a present
entity. Little work has been done in investigating the onset and
progression of fouling to the point where it severely affects
engine performance. Discussions with engine users revealed
that fouling can progress into 40 to 50 percent of the compressor stages. It therefore becomes necessary to develop a
model that can simulate this progression of deterioration.
In order to meet this requirement a linear progressive fouling
model was developed. In this model, fouling is assumed to
progress in steps, where each step increases the number of
stages affected by one and the level of flow reduction by one
percent. Therefore extremely light fouling is simulated by a
one percent decrease in the flow characteristic on the first stage
only. Similarly, the next progressive step in the fouling simulation would involve a two percent reduction in the flow
characteristic for the first stage with the one percent reduction
in flow being transferred to the second stage. This stepwise
progression is repeated until all of the stages in the compressor
have been affected.
It was reasoned that the level of the efficiency drop expeTransactions of the ASME
9-1
9 0 S Ngg
-
95X Ngg
100X
Ngg
US'
i<r.
QL7.
\cc
us-
90*
Ngg
95*
Ngg
100X
Ngg
5s
Ld
U
<r
u
PERCENT CHANGE
Fig. 13
PERCENT
CHANGE I N
CDP
a
a.
3
O
cn
cn
<x
CENTAUR
Ul
IS
z:
a
i
u
H
Z
Ul
u
Q:
u
a.
1
2
PERCENT CHANGE IN MASS FLOW RATE
Fig. 14
20
40
60
fouling can pose significant problems. In such cases, performing compressor washes using "on-condition" maintenance
procedures would appear to offer little advantage over regularly scheduled washing. For the user who requires maximum
efficiency from his units at all times, on-condition maintenance
does appear to offer several advantages over regularly scheduled compressor washes. These advantages include a reduction
in unnecessary compressor washes and/or limiting the amount
of time spent operating in an uneconomical condition.
PERCENT
Fig. 15
CHANGE I N
CDP
620
<L 9 5 0 - ,
0-
600
FIELD DATA
LU 9 0 0 580
3
in
i/i
u 850
SIMULATION
560
tx.
a.
540-
*800
520
>
_l
500
H 750-
480
O
W 700-|
u
CL
FIELD
DATA
460-
SIMULATION
650
I
13
12
14
i
15
(RPM)xl000
I
16
4401000
1500
2000
2500
OUTPUT POWER
Fig. 6
1
3000
(KW)
3000-
17. 5
2800-
17. 0
2600
16. 5-
2400
a.
f 16.03
FIELD DATA
u 2200o
- 2000-
- SIMULATION
1 5 . 515 0
i-
0. 1800-
14 5
1600-1
FIELD DATA
1400
SIMULATION
14 0
13. 5
I
13
1200
12
I
14
1
15
1
16
CRPM)xl000
Fig. 7
I
1
13
14
GAS GENERATOR SPEED
I
15
<RPM)xl000
I
16
620-,
u
3j 5 8 0
12
12
13
14
15
1-6
(RPM)xl000
o
in
cr
UJ
a
i
u
z
u
UJ
a.
UJ
a.
LEVEL OF DEGRADATION
Fig. 8
LEVEL OF DEGRADATION
Fig. 11
4.0(FLOW:EFFY)
1.00)
0.75)
0.50)
,_.0.25)
(1:0.00)
ul 3 . 5 -
or
3.02.5
Ul 2 . 0 UJ
z
CFLOW:EFFY)
(1:1.00)
(1:0.75)
(1:0.50)
51-5'
5i-
0-
u
ac
a. 0.
0
LEVEL OF DEGRADATION
LEVEL OF DEGRADATION
Fig. 9
CFLOW:EFFY)
(1:1.00)
/CI:0.75)
LEVEL OF DEGRADATION
Fig. 10
Fig. 12
9-1
9 0 S Ngg
-
95X Ngg
100X
Ngg
US'
i<r.
QL7.
\cc
us-
90*
Ngg
95*
Ngg
100X
Ngg
5s
Ld
U
<r
u
PERCENT CHANGE
Fig. 13
PERCENT
CHANGE I N
CDP
a
a.
3
O
cn
cn
<x
CENTAUR
Ul
IS
z:
a
i
u
H
Z
Ul
u
Q:
u
a.
1
2
PERCENT CHANGE IN MASS FLOW RATE
Fig. 14
20
40
60
fouling can pose significant problems. In such cases, performing compressor washes using "on-condition" maintenance
procedures would appear to offer little advantage over regularly scheduled washing. For the user who requires maximum
efficiency from his units at all times, on-condition maintenance
does appear to offer several advantages over regularly scheduled compressor washes. These advantages include a reduction
in unnecessary compressor washes and/or limiting the amount
of time spent operating in an uneconomical condition.
PERCENT
Fig. 15
CHANGE I N
CDP
DL
Q
U
z -2
CENTfiUR
UJ
CENTfiUR
u>
g-3
I
u
or
UJ
(L
LM2500
-6
I
20
I
40
I
60
0
20
40
60
PERCENT OF COMPRESSOR STAGES FOULED
Fig. 20
10-i
3.5-1
CENTfiUR
0
20
40
60
PERCENT OF COMPRESSOR STAGES FOULED
Fig. 19
20
40
60
405-414.