Professional Documents
Culture Documents
ABSTRACT
INTRODUCTION
The ability to model the behavior of gas turbine engines is
critical in economic, thermal and condition monitoring studies. Since the gas turbine is the prime mover in cogeneration cycles,
This is especially true for applications involving alternative fuels, an accurate performance estimate of the turbine is essential in
such as biomass and coal derived fuels, where the engine operates evaluating the economics of a proposed plant. This is especially
far from its ISO performance ratings found in publications such true in plants that will run far from ISO rated conditions, as in
as Gas Turbine World. This paper presents an approach taken for integrated gasification combined-cycles. Thus, a systematic
modeling a Frame 6 engine and its control system given a fairly method of predicting engine behavior given limited data is of use
limited set of available data. The approach and model are tested for preliminary powerplant studies [Benvenuti et al., 1993].
against vendor cycle deck data for both natural gas and a low-Btu
biogas fuel. The model is then used to predict behavior of the To Gasifier
FUEL-1
FUEL-2
GASIFC
FUEL
NOMENCLATURE
A Effective turbine flow area Figure 1: GateCycle Frame 6 Model
CMV Compressor map variable
CS Corrected speed This paper details a method for modeling a GE PG6541B Frame
MW Molecular weight 6 single-shaft engine given standard (though limited) vendor data.
P Absolute pressure The model was produced by connecting individual compressor,
R Gas Constant combustor and expander modules together in the commercial
T Absolute temperature GateCycle powerplant analysis program (Figure 1) [Erbes and
m Mass flow Gay, 1989] [Palmer et al., 1993]. The model results are verified
Ratio of specific heats by comparison with cycle deck data for two low-Btu fuels, as
des Design-point conditions well as steam and water injection with natural gas. Then the
model is used to predict the behavior of the Frame 6 run on two
1.02 for the whole engine is built into this map. This implies that
overall gas turbine performance variations that may actually be
1 due to turbine effects (such as variation in efficiency) or other
94 96 98 causes are attributed to the compressor. This assumption has
0.98 100 Corrected been made in the absence of detailed information, such as a
0.96 102 Speed turbine map. The authors are aware that this method wrongly
104% attributes performance deviation to the compressor. At two
0.94
Compressor Efficiency
0.92 Normalized Efficiency 106
0.9
1.3 0.89 .
Normalized PR .
CMV=1.0 0.88 . ... .. . .. .. . .................. .......
..........................
1.2 . . ....
. ...................................................
0.87 ..........................................
.
Surge Line CMV=0.75 ... .................................... ...... ...
1.1 0.86 ....... ... ........ ...
. Plant Data .... . ..... .
0.85 ..
1 0.84 'Best Fit' to Data
CMV=0.5
0.9 0.83 Frame 6 Map
106
104 0.82
0.8 102
100 94 96 98 100 102 104 106
0.7 98
96 Corrected Speed
0.6 94
0.85 0.9 0.95 1 1.05 1.1 1.15 Figure 3: Comparison of Frame 6 Map to Actual Data
Frame 6 plants where the company has on-line monitoring
Normalized Corrected Flow installed (EfficiencyMap) [Gay et al, 1992], the data indicates
that compressor efficiency does not drop off nearly as steeply
Figure 2: Frame 6 Compressor Map with increasing corrected speed as the map indicates (Figure 3).
Expander
manufacturers typical correction factor plot as a function of
ambient temperature [Data Source 7](Figure 6). The compressor In absence of a true 'turbine map' which gives the relationship
map's flow was adjusted to match the given exhaust flow rate. between expansion ratio, corrected turbine speed, flow rate, inlet
The pressure ratio was calculated using the method detailed in the temperature, and gas properties for a given expander, a
Expander section below. The compressors efficiency was relationship is needed to estimate the turbine flow function. In
calculated by matching the overall power. The corrected speed at this model, the turbine flow function is estimated using the
every ambient temperature is known, since the Frame 6 is a following equation relating the inlet flow parameters to an
constant speed machine, and the inlet temperature and gas effective turbine nozzle area
molecular weight are known. The formula for corrected speed is:
m T
100 100 Constant =
CS = = AP Nozzle Inlet
RT / RTdes MWdes * T / MW * Tdes
Here, the value for is[Reynolds and Perkins, 1977]:
+ 1
The above method only provides values for the compressor map
2 1
at one point per speed line (at CMV=0.75), so a method must be =
formulated to generate the remainder of the compressor map. R + 1
The pressure-flow relationship along any speed line is assumed to
be almost vertical, which is typical of industrial engines, as in the when above the critical expansion ratio (choked), and [Streeter
map for a Westinghouse engine [Diakunchak, 1992]. The and Wylie, 1979]:
efficiency lines are set to slope off evenly on either side of the
2 1
running line at any given speed. Note that these are very gross
2 P4 P4
assumptions of the actual turbomachinery behavior. = 1
The compressor map generated from the GE Specification data R 1 P3 P3
using this method is presented in Figure 2. It is important to
note that the performance variation versus ambient temperature
accuracy. In the design point model of the Frame 6 engine, the 990
980
total cooling air to the nozzles is varied until a 2020F temperature
970
to the first rotating stage inlet is reached (GE's definition of firing
960
temperature for Frame turbines). The resulting cooling flow
950
equals 15.3% of the total inlet gas flow to the expander. The 140 150 160 170 180 190 200 210
turbine consists of 3 nozzle/rotor pairs. Seventy percent of the Compressor Discharge Pressure (psia)
total cooling flow to the nozzles is assumed to go to the first
stage; the remainder cools the second stage. The cooling air to Figure 4: Frame 6 Control Curve Generation
the rotors is assumed to be 4% of the total expander inlet gas
flow. Seventy percent of this rotor cooling flow is assumed to go
Using the control curve derived as described above, the model
to the turbine first stage rotor, the remainder cools the second
was run over a range of ambient temperatures (0-120F) to
stage rotor. From examining photographs of the Frame 6 buckets
evaluate the variation of the actual rotor inlet temperature with
in marketing literature, [Lavin, 1984] it was assumed that most
ambient temperature. For low inlet temperatures (0-70F), the
(90%) of the air to each rotor stage is mixed after the rotating
control system holds the rotor inlet temperature almost constant.
stage, and thus does not produce work in that stage (i.e. is not
At higher inlet temperatures, the rotor inlet temperature starts to
"chargeable"), as the location of the cooling holes on the buckets
fall from the 2020F value (Figure 5). When estimating the
was at the tip and trailing edge of the blade.
control system behavior on other fuels, we would like to
The off-design cooling flow rate to both the nozzles and rotors
duplicate this firing temperature behavior, as this is what the
is assumed to keep the same 'friction factor' along the cooling
control system algorithm was designed to achieve. Again, note
flow paths as in the design case [Consonni et al, 1988]. Thus the
that the control settings for the gas turbine are determined solely
off-design flow can be calculated with the following equation:
by the engine manufacturer.
P T
m=m des
des P T
des Cooling Flow
2020 1.1
2015 1
2010 0.9
2005 0.8
2000
0.7
0 20 40 60 80 100 120 140
1995
0 20 40 60 80 100 120 140 Compressor Inlet T (Deg. F)
Compressor Inlet Temperature (Deg. F) GE Spec. Power GE Spec. Exh. Flow GE Spec. Heat Rate
Model
Figure 5: Rotor Temperature as Set by Control Curve
for Methane Fuel
Figure 6: Frame 6 Model Run on Methane Fuel: Power,
Tuning to GE Specification Data Exhaust Flow, Heat Rate
The Frame 6 engine model using the natural gas control system
was run over a range of ambient temperatures and the results
compared with GE Specification data from Data Source 7. The
model results are in good agreement with the specifications. The VERIFICATION OF MODEL RESULTS
model over-predicts fuel flow at the higher ambient temperatures
(over 90F). Figures 6-7 show the results of this comparison. It is To verify that the method of formulating a Frame 6 engine model
important to note, however, that the compressor map and control is appropriate, the model was tested against data from GE cycle
system algorithm were tuned so that the off-design GateCycle deck runs as shown in the following sections.
engine model provided the best fit to this data, so the fact that the Water Injected Cases
model matches well is deceptive.
Table 2 shows good agreement between the GateCycle engine
model results and the data from two GE cycle deck runs for a
water-injected gas turbine (Data Source 9). Note that the
operating point on the compressor map is near the point where
the model was tuned (as indicated by the CMV value).
Case 1 Case 2
Inlet T, Relative Humidity 44 71
Inlet Relative Humidity 60 90
Water Injection Flow, lb/hr 19400 14720
Cycle Deck Model % Difference Cycle Deck Model % Difference
Power, MW 42.010 42.062 0.1 37.550 37.552 0.0005
Heat Rate, Btu/kW-hr 11170 11031 0.05* 11420 11278 0.06*
Exhaust Flow, lb/hr 1,160,000 1,159,300 0.06 1,081,000 1,082,70 0.2
0
Exhaust Temp, F 993 992.7 0.03 1012 1009 0.3
Rotor Inlet Temp, F 2008 1997
Compressor Map Variable 0.776 0.765
* After removing the assumed margin of the design case
Using control curve from actual site.
Constants: Inlet loss 4", Exhaust loss 12", LHV=20422 Btu/lb, 20' elevation
discrepancy may also be attributed to the fact that the gas turbine TABLE 6: VTT BIOMASS-DERIVED FUEL
model is based on 'total' conditions whereas the GE cycle deck CHARACTERISTICS
data may be based on static conditions (i.e. kinetic energy is not
included). Regardless of cause of the discrepancy, one must Composition (mole %)
make sure when viewing the performance estimates of this model Fuel #1 Fuel #2
for other fuels in an integrated system that these losses or H2 11.86 12.42
discrepancies are accounted for in the extraction air for CO 17.69 18.55
gasification. CO2 11.00 11.62
It must be noted, however, that in these two cases the large CH4 4.22 4.42
required air extraction rate required causes the gas turbine C2H4 0.59 0.62
compressor to operate near its normal operating line. Thus, the H2O 15.43 11.29
model is already 'tuned' to give the proper results for power, flow N2 39.16 41.08
and exhaust temperature. NH3 0.05 2.9e-7
Temperature (F) 491 896
PERFORMANCE ESTIMATES ON OTHER LOW-BTU Calculated LHV (Btu/lb) 2104 2172
FUELS Required Compressor Air
Extraction 0.5797 0.5986
Using the gas turbine engine model, the performance of the
(lb air/lb fuel)
Frame 6 engine was then estimated for operation on two low-Btu
biomass-derived fuels as specified by the Technical Research
Centre of Finland (VTT). The fuel characteristics are shown in
Table 6. Note that the air extraction rate is lower than that for the Variation with Ambient Temperatures
fuels in Table 4. To estimate the performance of the Frame 6 over a range of
ISO Rating ambient conditions, the engine control system must be simulated.
The actual Texh-CDP control system relationship needs to be
The GateCycle Frame 6 model was run at ISO conditions with designed for these conditions, since the pressure ratio increases
the two fuels specified by VTT. Because the water fraction of the due to the higher pressures associated with the increased fuel
combustion products is less than that for the SFA fuels, the mass flow rate of the low-Btu fuel. To generate the control
engine firing temperature was assumed to stay at 2020F. The curve, the model was run using Fuel #2 fired at a constant rotor
results are presented in Table 7. As the operating point of the temperature over a range of ambient temperatures. The resulting
compressor map (CMV=0.81, which is close to the operating exhaust temperature was plotted versus the compressor discharge
point of the STIG case) has shifted further away from the normal pressure (Figure 8). Then, a straight line (with a maximum
running line (CMV=0.75: where the model was tuned), the temperature limit) was drawn through the points, taking care to
results are certain to be less accurate than those for the fuels in match at the lower ambient temperatures (higher CDP) and
Table 4. intentionally under-estimating at the higher ambient temperatures.
Description Net Net Fuel Exhaust Exhaust CDP gasif. Comp. CDT Rotor T
Power Effic Flow Flow T flow Map
Var
MW % lb/hr lb/hr deg F psia lb/hr deg F deg F
kg/hr kg/hr deg C bars kg/hr deg C deg C
Fuel 1 42.257 32.9 208,260 1,168,700 1017 181.3 120,740 0.814 669 2021
94467 530122 547 12.50 54,768 354 1105
Fuel 2 40.838 34.5 186,131 1,156,700 1015 179.0 111,420 0.795 664 2019
84429 524679 546 12.34 50540 351 1104
This is done to mimic the qualitative behavior of the Frame 6 Correction Factor
control scheme for standard operation on methane (Figure 5). 1.3