Professional Documents
Culture Documents
V003t06a011 94 GT 253
V003t06a011 94 GT 253
V003t06a011 94 GT 253
Presented at the International Gas Turbine and Aeroengine Congress and Exposition
The Hague, Netherlands — June 13-16, 1994
OUTER
SIMRLER
MIXING
INNER DUCT PREFER
SIMRLER
CENTERBODY
FUEL
the combustor. With the use of inlet guide vanes, VBV and The work of Roffe and Venkataramani (1978) suggested that
bleeds it is possible to cover the entire operating range of the it was indeed feasible to get low NOx emissions at high
engine with this configuration. pressures and temperatures representative of modern aero-
derivative gas turbine engines. These researchers have used a
The combustor/diffuser length had to be limited by the perforated plate to conduct their experiments. Their work
locations of the compressor exit guide vanes and the turbine showed that the development of the premixer was challenging.
nozzle. A multi-passage diffuser design allows the diffuser to Early groundwork on lean premixed combustion for gas
be shortened. The maintainability constraint, that the fuel turbines was laid in the works of Fenimore (1971), Leonard
injection system be externally removable, requires that free and Correa (1990 and Correa (1992). The development of
space be maintained between the dome and the pre-diffuser, at heavy duty gas turbine combustion chambers using lean
least as much as required by the engagement of the fuel premixed combustion technology has been described by
injection system (premixers). In order to maximize the Davis (1992).
residence time in the combustor for CO oxidation at part
power conditions, the size of the premixer was apriori set at
0.05 m (2" ) for the mixing duct and 0.0175m (0.7") for the DACRS I DESIGN
swirl vanes.
The Double Annular Counter-Rotating Swirler (DACRS)
A cross-section of the LM6000 Dry Low Emissions premixer, shown in Figure 2, was concieved to satisfy the
combustor is shown in Figure 1. The size constraints become constraints of auto-igntion, flashback and size. The duct
clear when the premixers must pass about 80% of the diameter is reduced at the exit (in comparison to its inlet ) to
combustor air flow to keep the flame temperatures within continuously accelerate the air flow in an effort to reduce
limits where low NOx emissions can be maintained (Leonard boundary layer thickness and limit growth of boundary layers
and Stegmaier 1993), unlike conventional aero-engines where to prevent flashback in the boundary layers. The area
between 10 to 20% of the combustor air goes through the reduction is about 2:1. The exit velocity is set by design to
dome and the swirlers (Ekstedt 1987). The conventional be significantly greater than the turbulent flame speed in lean
radial inflow swirlers have been replaced by axial inflow mixtures (Abdel-Gayed et.al . 1987). The mass averaged
premixers. In addition, to satisfy operability requirements the forward velocity is set to be greater than 84 m/sec (275 fps)
number of premixers has been increased to 75 (Leonard and for the prototye design. A conical centerbody along the
Stegmaier 1993) in the triple dome arrangement. This reduces centerline of the premixer can be utilized to supply liquid fuel
the air flow to each premixer to manageable levels. This to an atomizer at its tip and gas passages for diffusion
arrangement, though very tight, has mechanical design burning at low power conditions.
advantages of having smaller liner surface area to cool. A
general description of the combustion system can be found in Swirling flows in the duct give rise to free vortex like flows
Leonard and Stegmaier (1993). Since the low emissions with a low static pressure region along the duct's centerline.
combustor must fit inside the length restrictions of the Such a situation typically leads to a recirculatory flow into
existing turbomachinery, an additional length restriction is the duct. This situation is clearly unacceptable. A counter-
imposed on the combustion system design. The flame rotating inner swirler in the design helps increase the axial
temperature in a low NOx emissions combustor is expected to velocity along the centerbody and keeps the flow attached to
be lower than that of conventional diffusion flame the centerbody as well. The counter-rotating swirler also
combustors. This together with wall quenching can result in helps increase shear within the premixing duct. The
2
0.9
0.88
0.7 ■ Axial
Vebcity
OUTER
SMRLER
INNER
SLER
MIXING
DUCT
PREMD^R
-/CENTERBODY
A
0.6
0.5 —A Tangential
OUTER
SWIRIER MDUNG
DUCT
INNER
SWIRLER
COOLED
CENTERBODY
3
SILANE IGNITION PORT
AIR --
I
8FUE SPOKES
CERAMIC LINER
60-60 DEG .089 M (3.5")
SWIRLER
DESIGN 1 ,
FUEL TRAVERSING
NOZZLE EMISSIONS
SAMPLING
.05M (2 ") 0.15M PROBE
MIXING
DUCT
AIR `
premixer is cooled by small amounts of purge air channelled 10 to 20 minute long tests. Air temperature and pressure is
through a small hole through the center. In addition it is measured just upstream of the test combustion chamber. A
shorter than the mixing duct which helps to keep it away from pressure feedback control system is used to maintain
the flame. The basic design has been scaled to accomodate combustion chamber pressure during each test. This pressure
different air and fuel flows in the inner, the middle and the system permits the heat release to be modulated in the
outer domes of the triple dome combustor. combustion chamber without changing its pressure. The
choked flow inlet venturi keeps the air flow unchanged which
maintains the dome velocity constant during the test.
1 CUP TEST APPARATUS
The single cup test combustor is shown in Figure 7. The 2 CUP TEST APPARATUS
premixer is attached to the upstream end of the 0.089m (3.5")
nominal diameter cylindrical combustion chamber 0.15m (6") A two cup test apparatus has been designed to fit within the
long. The length of the 1 cup test combustion chamber test rig shown in Figure 8. This test combustor has two
simulates the residence time of the LM6000 DLN combustor. premixers side by side and thus permits the evaluation of cup-
The dome of the combustor is back-side cooled by convection cup interactions. The first design of this 2 cup test combustor
using the premixer air. The cylindrical section is made of has been implemented with ceramic uncooled sidewalls. The
high temperature ceramic and is uncooled. Emissions subsequent implementation has back-side cooled metal liner
measurements were made using an 'Aero-quench" probe along with heat-shields on the dome which are similar those
designed per guidelines of Roffe and Venkataramani (1978). designed for the LM6000 combustor. The heatshields have
Care was exercised in the design of the gas sampling probe to been manufactured with a removable wall separating adjacent
minimize the backpressure at the gas sampling tip to ensure primary combustion zones. This additional feature permits
that a reasonable expansion pressure ratio (>5) was always the evaluation of interactions between adjacent primary
maintained. Gas samples were carefully dried before injection combustion zones.
into the NOx analyser. CO2, CO and UHC were also measured
along with 02. The gas sampling system provides a backup An elaborate checklist was used to ensure that all systems
method of calculating fuel air ratio. High pressure tests were would function properly during the test. The air flow was set
performed using the blowdown facilities at General Applied using a critical flow venturi and by regulating the pressure
Science Labs, Lake Ronkonkoma New York. Air flow for all feeding the venturi. The fuel flow was measured using critical
tests is measured by a critical flow venturi before it passes flow venturi flowmeters. Emissions were measured by aero-
through a pebble bed heater. A temperature feedback control quench gas sampling probes. Two sets of probes were
is used to maintain constant air temperature throughout the manufactured for the test. A radial traversing probe was used
4
Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1994/78859/V003T06A011/2404772/v003t06a011-94-gt-253.pdf by Mr. Aaron Frost on 07 June 2021
SECT X—X
for most tests to generate emissions profiles across the exit not out of reach of engine type hardware. The measurements
plane of the combustion chamber. The axial probe was used of Leonard and Correa (1990) suggest that further
for some tests to establish the axial emissions profile. Water improvements in design could lead to even lower NOx
was sprayed into the combustion chamber exit flows to cool emissions.
the flow down to less than 533 K (500 F) for the backpressure
valve. The interaction of adjacent premixers has been studied in
the 2 cup combustor apparatus described earlier. Figure 13
shows an emissions profile measured by a traversing gas
DACRS II TEST RESULTS sampling probe that travels across the 2 cup combustor with 1
premixer unfueled and the other premixer fully fueled. In the
The DACRS II premixer was tested at a variety of conditions Figure 13a both premixers are fueled at the same equivalence
representative of the LM series of gas turbine engines in the 1 ratio, the centerbody separating the primary combustion
cup test apparatus described above. The dome reference zones of the two premixers has been removed for this test.
velocity (cold flow velocity) for most tests has been This configuration represents adjacent premixers within a
maintained close to 12 m/sec (40 fps). The radial profiles of dome of the triple dome combustor shown in Figure 1. In the
emissions are measured with a single point radial traversing Figure 13b, data shown correspond to the condition where an
probe as shown in Figure 7. Representative emissions profile adjacent premixer within a dome has to be either switched on
of NOx and CO are shown in Figure 9. The fact that the or has just been switched off. This configuration is transient
profiles are flat suggests that a representative sample could be in the combustor and occurs when the mode of combustor is
taken along the centerline for gathering data. An axial being staged, i.e. another dome is being lit or switched off.
emissions profile measured with an axial traversing sampling The complete switching of the peak equivalence ratio
probe is shown in Figure 10. This measurement is restricted measurement to the centerline of the non-burning cup
to the centerline of the 1 cup combustor due to rig limitations. suggests that tremendous mixing occurs within the 2 cup
It appears that the CO oxidation goes to completion about 0.1 combustor. A strong secondary flow set up within the
m (4") downstream of the dome. Most of the heat release confines of the 2 cup liner transports the hot burning gases to
appears to have occured in the first 0.05m (2") of the test the centerliner of the non-burning premixer. Clearly there is
combustor based on the combustion efficiency calculated significant mixing between the flows emanating from the two
using CO and CO2 emissions. Emissions data are shown in premixers. Figure 13c shows the dramatic impact of the
Figures 11 and 12. The NOx emissions, corrected to 15% centerbody separating the primary reaction zones of the two
oxygen dry, correlate over the test conditions to the flame premixers. This configuration represents adjacent domes of
temperature as seen in Figure 11. CO emissions have been the combustor. With the separation of the primary
measured close to the calculated equilibrium values. This is combustion zones, the strong secondary flow seen in Figure
not surprising since the liner is not cooled. These 13b is absent as seen by the significantly smaller equivalence
measurements show that single digit NOx emissions goals are ratio peak at the centerline of the non-burning premixer.
120
12
10- 100 ® 620 K, 5 ATM •
a 8
❑
n 80 ❑ 620 K. 17 ATM
^ r ■4
^ r
ir^.^, r
c
6 60 • 720 K, 21 ATM
0
.^
u
Figure 9. A representative radial profile of emissions from 1 Figure 12. CO emissions from the DACRS II premixer at a
cup test combustor. P3 = 30 atm, T3 = 755 K variety of test conditons.
1000
2500 1 CUP #1 ER = 0.65
E5
Fu I Fbw
U 800
2000
Combustion Efficiency 0 ' 8 -2
0.6 w s 600
Q 1500 0
U
° 1000 0.4 - - 400 ■
0
Figure 10. A representative axial profile of emissions from the Figure 14. CO emissions from 2 cup test combustor with one
1 cup test combustor. P3 = 30 atm, T3 = 755 K premixer fueled at lower equivalence ratio and the other
maintained at a constant equivalence ratio. P3 - 10 atm, T3 -
610 K.
35 ® 620 K, 5 ATM
30
0620K, 17 ATM ® 0.8 T
25 0 0.7
= • 720K,21ATM
20 444 0.6
15 © 760 K, 30 ATM 44 o: 0.5
w w
4
10 • Z >< 04LBO
5
Q^0.3
a 0.2
o 0.1
1800 1850 1900 1950 2000
w
1700 1750 0
Flame Temperature, K 0.2 0.4 0.6 0.8
0
EQUIVALENCE RATIO IN PREMIXER #2
Figure 11. NOx emissions from the DACRS II premixer at a
variety of test conditions.
Figure 15. Lean Blowout equivalence ratio in each of 2
premixers showing impact of adjacent premixers. P3 10 atm,
T3 -610K.
6
20 ■■■ 0.7
18 r U.■■
0.6 LA4® g& MAbgA
■ 0.5
E 14
A®® N Both premhcers fueled,
.9 0.4 no cente tody
F5 10 pFi
(A) w 0.3
4• NOx 0.1
2 On On On On
0 0
0 0.03 0.06 0.09 0.12 0 0.03 0.06 0.09 0.12
distance, m Distance, m
Figure 13. Emissions profile measured across a 2 cup test combustor showing the impact on emissions of adjacent non burning preen
the 2 primary combustion zones. P3- 18 atm, T3-710 K.
7
35
30
25
E 20
Figure 16. NOx Emissions from DACRS III premixer in the 2 Figure 17. CO Emissions from DACRS III premixer in the 2
cup test combustor cup test combustor. P3 —18 atm, T3-710 K.
The DACRS III configuration has been tested in the 2 cup Dynamic pressures up to 5 psid pp were measured with the
test combustor. The NOx emissions obtained with the DACRS metal walled 2 cup test combustor. These dynamics did not
III premixer design generally are similar to the DACRS II adversely affect the operation of the tests though some
design and are shown for a variety of test conditions in Figure hardware damage was encountered. The dynamic pressures
16. As noted before this data is also correlated to the flame could be reduced by increasing the dome pressure drop or
temperature in the two cup combustor. The data obtained by changing the inlet air temperature. Little effort was expended
Leonard and Correa (1990) is significantly lower than the in the conduct of these tests to suppress the dynamic
DACRS III data suggesting that further mixing improvemente pressures, however considerable effort was expended in
are possible. CO emissions data is shown in Figure 17. A recording the dynamic pressures.
typical radial traverse is shown in Figure 18. The quenching
of CO oxidation reactions at the walls is clearly seen in the
data. This type of quenching has been observed with the 2
cup test combustor with backside cooled sidewalls. The
8
The development plans for the LM6000 DLN combustor
include 15 cup sector tests and full annular combustor tests Becker, B., Berenbrink, P., and Brandner H., 'Premixing Gas
before engine tests of the combustor. These tests will be and Air to Reduce NOx Emissions With Existing Proven Gas
conducted through the first quarter of 1994. Turbine Combustion Chambers", ASME 86-GT-157.
Significantly higher CO emissions occur close to the walls Leonard, G., and Stegmaier, J., "Development of An
of the combustor apparantly due to the quenching effect of Aeroderivative Gas Turbine Dry Low Emissions Combustion
backside cooled metal walls. System", ASME 93-GT-288, Paper presented at the IGTI
conference in Cincinnati, June 1993.
The DACRS II and DACRS III designs can be applied to all
engines including the LM1600, the LM2500 and the LM6000, Roffe, G., and Venkataramani, K.S., "Emissions
since single digit NOx emissions have been obtained at test Measurements for a Lean Premixed Propane/Air System at
conditions encompassing operating regions of these engines. Pressures upto 30 Atmospheres", NASA CR 15921 (1978).