Professional Documents
Culture Documents
Turek2005 PDF
Turek2005 PDF
ASME Turbo Expo 2005: Power for Land, Sea and Air
June 6-9, 2005, Reno-Tahoe, Nevada, USA
GT2005-68152
ABSTRACT present in the near-field region close to the swirl vanes. The
The flow fields of two different Siemens-Westinghouse data obtained indicates that the flow behavior in the region near
gas turbine mixers were studied experimentally in an effort to the swirl vanes strongly influences the mixing of the fuel and
better understand fuel-air mixing in confined swirling flows air. Frequency analysis of the fuel concentration data shows
found in industrial applications. The mixers consist of an that some turbulent structures prevail throughout the mixing
annular flow region and mixing is achieved using swirl vanes, region in both mixers, revealing that some large scale flow
the pressure side of which is used to inject the fuel. The features emanating from the swirl vanes are not dissipated even
difference between the two mixers studied is the degree of swirl in the high degree of swirl hardware. Lastly, unmixedness
imparted on the flow by the swirl vanes (45o vs. 55o). Velocity levels in both mixers tested are calculated and compared with a
(both axial and azimuthal) and fuel concentration profiles were discussion on how they might impact NOx emission levels.
obtained for non-reacting, atmospheric pressure flows at
several axial and radial locations downstream of the swirl vanes Keywords: fuel-air mixing, unmixedness, swirl vanes,
by the use of LDV and infrared laser light absorption turbulence, LDV, infrared absorption.
techniques, respectively. The fuel used in this work was a
methane/air mixture, which was injected at a momentum flux INTRODUCTION
ratio comparable to that under operational conditions. Results The need for reduction of NOx emissions from gas turbines
show that flow uniformity, as evidenced by velocity and fuel is due to the detrimental effect these chemical species have on
concentration profiles, is reached further downstream of the ozone molecules in the upper atmosphere [1]. A great deal of
swirl vanes for the 45o mixer than for the 55o mixer. This work has focused on reduction of NOx formation and emission
indicates a lesser mixing performance in the 45o mixer. The in practical combustion systems. This work has largely focused
axial and azimuthal RMS velocities were consistently higher on the reduction of NOx during the combustion of lean
for the 55o degree mixer and this was a likely contributor to its premixed flows, which are often used in gas turbine
superior mixing performance. High velocity and fuel applications. Under such conditions, fuel-air mixing has a
concentration gradients are common for both mixers and significant effect on the NOx formed [2, 3]. This is due to the
Figure 1. Diagram of swirl vane geometric parameters. Fuel/Air (RFA) analyzer [17, 18]. As shown in Fig. 3, a
vacuum pump extracts a sample from the test section through a
small (1/16 in. diameter) probe, which passes through a 50-mm
The swirl vanes in the two premixers had dimensions as long sample cell. The cell is capped by sapphire windows and
shown Fig. 1, and as listed in Table 1 (normalized by the hub infrared (IR) radiation at approximately 3.4 µm from a He-Ne
diameter of the premixers). The premixers are referred to laser is directed through it. As seen in Fig. 4, this wavelength
simply as the 45o premixer and the 55o premixer. As shown in coincides well with the strong transitions in the P-branch of the
Fig. 2, room air entered the premixers through the inlet, and ν3 vibrational band of methane [19, 20]. An IR detector
was drawn by a fan motor located at the downstream end of the measures the amount of absorbed laser light by the gas present
test rig. The overall mass flowrate through the test hardware in the sample cell. Right after passing through the cell, a
was on the order of 0.45 kg/s. The fan motor was located well thermocouple and pressure transducer monitor the temperature
downstream of the exit of the premixers and was connected and pressure, respectively, of the sample gas. The entire setup
through a sealed plexiglass tube containing flow conditioners was mounted on the same 3-axis traversing table, as was the
(i.e. flow straighteners) so that any flow instabilities introduced LDV.
by the motor would not be felt by the flow through the The concentration of methane present in the sample can be
premixers. Axial and azimuthal velocity measurements in the related to the absorption measured by the IR detector using Eq.
premixing region were performed using a two-component LDV (2):
operated in the backscatter mode. Frequency shifting (i.e.
Bragg cell) was used to provide directional sensitivity. The
LDV probe volume dimensions were 655 µm length and 131 ⎛ I ⎞ RT
C = − Log ⎜⎜ ⎟⎟ u (2)
µm diameter with a fringe spacing of 1.38 µm. The flow ⎝ I o ⎠ εPtotal L
through the premixers was seeded at the inlet with 1 µm TiO2
particles in order to generate the Doppler signals. To allow Since the absorption spectrum (Fig. 4) suffers from
optical access to the test hardware, grooves approximately 15 pressure broadening as well as being temperature dependent,
mm tall were machined along the length of each of the
premixers and were fitted with antireflective fused silica
windows. The entire LDV system was mounted on a 3-axis 1
traversing table with a resolution of 2.5 µm and positioned as
Absorbance (a.u.)
0.8
shown in Fig. 2.
Methane concentration was obtained using a Real-time 0.6
0.4
Figure 8. Normalized axial RMS velocity for the 45o Figure 9. Normalized axial RMS velocity for the 55o
premixer. premixer.
to 1.906). For the 45o premixer, at r/R = 0.2, there was only a Mean Fuel Concentration
4% decrease from V/Vbulk= 1.491 at x/Dh = 0.89 to V/Vbulk= The mean fuel concentration data from both premixers are
1.429 at x/Dh = 3.02. shown in Fig. 14 and Fig. 15, respectively. It is easy to see by
inspection that the variations across the annular profile are
Azimuthal RMS Velocity larger for the 45o premixer (Fig. 14) than for the 55o premixer
Figure 12 and Fig. 13 show the normalized azimuthal RMS (Fig. 15). All methane concentration measurements were
velocity for the two premixers and, as in Fig. 8 and Fig. 9, show normalized by the methane concentration at the point of fuel
that the average value for azimuthal RMS velocity for the 55o injection (i.e. 5,000 ppm), which was on the pressure side of the
premixer is higher than that for the 45o premixer (typically by a swirl vanes. Thus, for example, a value of C/Cf of 0.1 indicates
value of 8% to 20%). This, again, is significant since higher that the measured value of methane was 10% of the methane
RMS velocity values will be beneficial to the mixing of fuel concentration at the point of fuel injection, and the fuel has
and air. Another factor which likely contributes to the growth therefore been diluted by the bulk air flow by a factor of 10.
of the boundary layer in the inner annular region of the 55o In Fig. 14, the normalized concentration in the 45o
premixer is the fact that with a greater swirl angle, there will be premixer varies between 0.025 and 0.11 at the most upstream
a greater centripetal force on the flow molecules due to the location and between 0.047 and 0.064 at the most downstream
higher azimuthal velocity. This larger centripetal force will location. In Fig. 15 the normalized concentration in the 55o
tend to pull the air molecules away from the hub more in the premixer varies between 0.044 and 0.077 at the most upstream
55o premixer than the 45o premixer, promoting the boundary location and between 0.046 and 0.050 at the most downstream
layer growth that was observed. Also, the RMS velocity at the location. Thus, the fuel concentration profiles for the 55o
inner annular region in Fig. 13 can be seen to increase as the premixer are more spatially uniform and indicate a more
flow moves towards the downstream end of the premixer, homogeneous fuel-air mixture generated by this hardware. In
which again reinforces the fact the 55o premixer has a strong addition, Fig. 14 and Fig. 15 show that even for the most
trend of inner boundary layer growth. upstream axial location, the variations in the mean
concentration value are smaller for the 55o premixer than for
Figure 12. Normalized azimuthal RMS velocity for the 45o Figure 13. Normalized azimuthal RMS velocity for the 55o
premixer. premixer.
Figure 16. Normalized RMS fuel concentration for the 45o Figure 17. Normalized RMS fuel concentration for the 55o
premixer. premixer.
Concentration Spectrum
Figure 19. Unmixedness at x/Dh= 1.74 for the 45 premixer o The power spectrum of the fuel concentration was
and x/Dh= 1.63 for the 55o premixer. examined in an effort to gain understanding of the flow as fuel
and air were passing through the premixers. The spectra of the
concentration at selected axial locations for both premixers are
shown in Fig. 21 and Fig. 22, respectively. For ease of
comparison, a line with a slope of -5/3 is placed next to each
power spectrum so that its slope of decay may be compared to
this value, since this slope is indicative that the flow is
receiving turbulent energy at large scales and dissipating it at
small scales.
r/R