Temporal resolved gas temperature measurement of 20 kHz
using 2.0 µm Semiconductor Diode Laser
Yuji IKEDA, Katsuhiko FUKUZATO, Keiichiro OKA and Tsuyoshi NAKAJIMA
Department of Mechanical Engineering, Kobe University
Rokkodai, Nada, Kobe 657-8501, JAPAN
Abstract
Tunable diode laser absorption spectroscopy (TDLAS) measurement of H2O gas temperatures was carried
out in a heated cell and a turbulent premixed burner at a repetition rate of up to 20 kHz. A near 2.0 µm
distributed feedback (DFB) laser operating at room temperature was developed and all optical components
were assembled in a box shown in Fig. 1. The absorption lines for CO2 and H2O gas were carefully
selected from the HITRAN / HITEMP database, considering its sensitivity. To measure high temperatures
over 1000 K, the ratio of the H2O absorption line strength at 5008.101 cm-1 and 4878.193 cm-1 was used. A
heated static cell filled with gas was used to evaluate the apparatus, the measured temperature agreed with
the preset gas temperature, and the accuracy was within 5 %. A premixed turbulent burner was selected for
actual measurements (Fig. 2). The gas velocity was measured using a particle image velocimetry (PIV)
system, which revealed the characteristics of turbulent flows throughout the flame region. The
measurement location was chosen carefully, based on this data. The temperature of the gas on the inner
cone of a turbulent premixed flame was measured at 20 kHz. The time-averaged measured gas
temperature was 1351 K and 37 K root mean square (RMS), for which the thermo-couple showed 1336 K
and 99 K RMS shown in Fig. 3. The profile of the measured temperature probability density function
(PDF) was almost Gaussian. There was good agreement with the thermocouple results for both
measurements. It proved the high temporal rate of temperature measurement could be used for real-time
combustion monitoring and extended to feedback control.
Figure 1 Developed system using 2 µm DFB lasers
Turbulent burner, (r, z)=(7, 56)
1500
Temperature(K)
ave.=1351.9 K
rms=36.7 K
1400
1300
1200
0
25
50
Time(ms)
75
100 0
5
10
PDF(%)
15
Figure 2 Measurement point Figure 3 Time series gas temperature measurement at 20 kHz at the inner
cone of turbulent premixed methane/air flame
-1-
1. INTRODUCTION
Combustion gas measurement has conventionally been done by gas sampling systems[1], which makes it
difficult to measure high-speed events (>kHz order). This makes gas sampling type systems unsuitable as
a real-time sensor for combustor control. And also, laser-based diagnostics, like Rayleigh and CARS
[2][3] has been restricted by its size, cost and time resolution. Using diode laser absorption spectroscopy,
the measurement of temperature and concentration of multiple gaseous species, with a high temporal
resolution has been proven [4-12], because a diode laser is small, robust, and has a high-speed wavelength
modulation capability. In addition, a diode laser can easily be interfaced with optical fibers to enable
optical access to many types of measurement systems.
Hanson et al. [13-24] reported research on a diode laser absorption spectroscopy system using the
wavelength of near infrared region. Their research initially focused on H2O absorption. Subsequently,
other species were measured, including CO [25], CH4 [26], and NO [27], followed by simultaneous
measurement of multiple gaseous molecule species (CO, CO2, and CH4) [28]. Allen et al. measured H2O
[29] and NO2 [30] using a diode laser at 1.3~1.5 µm. In order to evaluate the accuracy of these
measurements precisely, Tanahashi et al. [31] performed numerical simulations of diode laser H2O
absorption spectroscopy using DNS data for a turbulent flame. Aizawa et al. [32] measured OH radicals,
Iseki et al. [33] measured CH4, Tanoura et al. [34] measured NOx, and Torniainen et al. [35] conducted
tomographic research with 2-D measurements.
In order to achieve a high degree of measurement accuracy, a method using frequency modulation is better,
although high temporal resolution cannot be obtained. However, a scanned-area method exceeding kHz
order has not been reported. The resolution of the transmitted wavelength in an InGaAsP/InP DFB diode
laser is 0.01 nm; it is restricted by the resolution of the temperature controller of a DFB laser, which is 0.01
K. In other words, it is difficult to fix the wavelength of a diode laser because precise control of
temperature is too difficult.
To understand turbulent flame characteristics, measurements over kHz order of gas species concentration,
velocity and temperature are strongly required. Furthermore, time series measurements of these physical
parameters are essential to understand local flame front structure in turbulent flames. Gas velocity can be
easily measured using laser techniques [36], but it has been difficult to measure gas temperatures at high
data rates in a practical burner. Thus, we developed a TDLAS system using a 2.0 µm DFB laser to
measure gas temperatures with high temporal resolution (20 kHz) in a combustion system [37]. The
measurement accuracy was examined with a static cell, a heated cell and a flat flame burner. The purpose
of this study is to make time-series measurements of gas temperature in a practical turbulent burner at a
very high repetition rate exceeding 20 kHz using the TDLAS system. This paper describes the system
developed and the results of measurement of a premixed turbulent flame.
2. PRINCIPLE OF MEASUREMENT
Diode laser absorption spectroscopy utilizes the absorption characteristics of light, and follows the
Lambert-Beer Law. Webber et al. [38] describes this principle in detail. Several parameters must be
considered; we used a DFB type diode laser with a 2.0 µm absorption wavelength, and a sweep rate of 20
kHz.
Figure 4 shows a schematic of the layout used for the measurements, which was the same as reported in
references [13-30, 38-42]. The optical wavelength was controlled so that the diode laser current could be
swept (~150 mA) at a set frequency (20 kHz) to cover gas absorption lines. When an absorption line is
present, a ‘dip’ appears in the transmitted intensity signal. The wavelength of an absorption line and the
principle of gas temperature measurements are shown in Fig. 4. A second diode laser sweeps other
absorption lines, and measures the dips. The ratio of the areas of two dips gives the gas temperature. By
sweeping with two lasers, LD1 and LD2, at an optional frequency of 20 kHz, it becomes possible to
measure transient temperatures, as shown in the figure.
-2-
L
Io
Lambert–Beer Theory
Iν
T=
Iν
Io
=exp(-kνL)
6
LD1
2
4
3
LD2
Absorption
Wavelength sweep
Absorption
5
7
A1
20 kHz
A2
Time
Time
A1/A2 : Temperature
Temperature
Absorption spectrum (HITRAN)
Absorption
1
20 kHz resolution
Time
Fig. 4
Measurement principle
3. SELECTION OF THE ABSORPTION LINE
In considering conditions as follows, two lines (5008.101 cm-1 and 4878.193 cm-1) for H2O gas temperature
measurement were selected.
1.
2.
3.
A strong absorption line was required in the target temperature measurement range.
The strength ratio of the two absorption lines had to remain highly sensitive over a wide temperature
range.
The selected wavelength could not overlap other gas absorption lines, although there was little
interference in the baseline.
As an advantage of these lines, it is less sensitive at measuring temperatures below 600 K. This means that
no calibration for low temperature is required. For instance, if the laser passes through a region at room
temperature, the effect of the path length at room temperature is negligible.
Since most semiconductor diode lasers used for H2O gas measurements have been developed for
telecommunications purposes, they use the 1.3 µm band. The ratios of 4878.193 cm-1/5008.101 cm-1 in the
2.0 µm band and 7444.351 cm-1/7185.597 cm-1 in the 1.3 µm band were compared, and the results indicated
that there was a high S/N ratio for the 2.0 µm band. Obviously, the 2.0 µm band is very sensitive at high
temperatures, while the 1.3 µm band is very sensitive at low temperatures, and would require calibration at
room temperature. Therefore, the ratio of 4878.193 cm-1/5008.101 cm-1 in the 2.0 µm band is useful for
measuring H2O gas temperatures in actual combustion.
-3-
4. EXPERIMENTAL EQUIPMENT
We considered the following in order to improve the S/N ratio of the measured signal:
In DFB diode laser development, these subjects were considered and developed: longer wavelength relative
to the strong absorption line, stabilized temperature control, high beam power, high repetition rate, and
triangular signal for fast response. For the detector, these parameters were taken into account: high
quantum efficiency, low noise and effective response time.
Here, we used a low-noise AC amplifier instead of a DC amplifier in order to prevent drift in time-series
measurements. Figure 5 shows a schematic diagram of the experimental package for the TDLAS that we
developed. It contains two 2.0 µm DFB diode lasers (NTT Electronics: 1995.56 nm @298 K, 108.5 mA,
2049.54 nm @298 K, 108.5 mA), and all the necessary optics and electronics. This TDLAS can sweep at
frequencies of 100 Hz, 1 kHz, 10 kHz, and 20 kHz. In order to increase the temporal resolution and
stability of the DFB diode laser, a triangular wave sweep was selected, rather than a ramp wave sweep, at
high frequencies. Each frequency can be set for various temporal applications.
A heated static gas cell was filled with water vapor at 1000 K and an R-type thermocouple was used to
evaluate the measured temperature. A turbulent pre-mixed CH4/air burner was selected as a real
combustion application. Gas velocity and turbulence characteristics were measured using a PIV system
[43, 44]. The average velocity of the unburned gas in the center of the burner was ū = 3.18 m/s. The
Reynolds number was Re = 9345. The equivalence ratio was φ = 1.1. The point of measurement was X =
56 mm from the end of the burner; the radial distance was r = 7 mm from the center. The beam path was
270 mm, using a reflecting mirror. All temperature measurements were made at a high temporal resolution
of 20 kHz.
1.0 mm
PC
300 mm
20 mm
Optics
Turbulent burner
TDLAS system
Fig 5.
Developed TDLAS system and experimental setup of turbulent premixed burner, Re=9345
-4-
5. RESULTS AND DISCUSSION
H2O gas absorption lines were examined to evaluate the accuracy of measurements and operation frequency
width, as shown in Fig. 6. Here, the measured absorption, reference beam, and etalon signal are shown.
The rows in Fig. 6 show the signals for 100 Hz, 1 kHz, 10 kHz, and 20 kHz, respectively. Although it is
difficult to fill a heated static gas cell with water vapor at high concentration, a dip sufficient for analysis
was recognized.
1kHz
10kHz
20kHz
Etalon
Reference
Transmitted
100Hz
λ=1996 nm, FSR=1.09 GHz
Fig. 6
Measured absorption lines at different sweep frequencies
Heated static cell, H2O, T=1000 K
Temperature(K)
1200
ave.=1009.6 K
rms=18.5 K
1100
1000
900
800
0
25
50
75
Time(ms)
100 0
10
20
PDF(%)
Figure 7 Time series temperature measurement in a heated static cell at 20 kHz
-5-
30
Figure 7 shows the temporal results for a temperature of 1000 K in the heated cell. The data were plotted
separately, and not combined and averaged. The average temperature was 1011.2 K. The RMS error was
18.5 K. This result demonstrates the utility of measurements at a 20 kHz repetition rate.
Figure 8 shows the instantaneous gas velocity vector measured by PIV and the turbulence characteristics as
a velocity gradient. The temperature profile measured by the thermo-couple is also shown in the figure, in
which the radiation effect on the thermo-couple is taken into account. The measurement location of this
method is not a single point as shown in the figure due to use of a line of sight method. However, the
temporal resolution of the gas temperatures gives a very sophisticated picture of the characteristics of a
turbulent flame. This TDLAS system requires a long path, which makes the measurement volume 1.0 ×
2.0 mm. Here, the inner cone of the turbulent flame was selected for measurement. Figure 9 shows the
temperatures measured in a premixed turbulent burner. Each dot in the figure represents raw data recorded
at 20 kHz, not an averaged ensemble. The average temperature was 1351.9 K and the RMS was 36.7 K,
while the temperature as measured by the thermo-couple was 1336 K and the RMS was 99 K. This result
indicates that gas temperatures can be measured at 20 kHz in a turbulent flame. Although high temporal
resolution of measured gas temperatures was obtained, the spatial resolution was not sufficient to
understand the turbulent flame characteristics.
Instantaneous direct image and PIV data
4.5
(1/s)
460
460
x/D
Cone:
(r, z)=(7, 56)
d=10 mm
ū=3.18 m/s
Φ=1.1
Re=9345
3 m/s
0
110
du
dx
(K)
1873
357.778
230
Z (mm)
255.556
153.333
51.1111
0
-51.1111
1373
873
-153.333
-255.556
-230
-357.778
373
-460
-460
-25
0
r (mm)
25
-10
0
r (mm)
10
0
Figure 8 Turbulent flow characteristic by PIV and temperature profile by thermocouple
-6-
In this experiment, in order to improve the S/N ratio, the beam path had to be long, at 270 mm. A pair of
mirrors was used to increase the length of the beam path. Consequently, the beam had to pass through a
region at room-temperature, where it was not in the turbulent flame. As discussed above, no calibration or
correction was required, so this method has the advantage of being able to monitor H2O gas temperatures in
real time. However, ideally the beam path should be as short as possible, since the longer the beam path,
the greater the effects on line of sight measurement, such as beam steering effects. If the beam path is too
long, the spatial resolution decreases, which makes it difficult to specify the measuring point. Thus, there
is a trade-off between beam path length and spatial resolution, which is a disadvantage of line of sight
measurement. A longer relative wavelength was selected for these reasons.
Other sources of error were beam steering effects and base line matching caused by wing effects. To avoid
beam steering effects, which are expected to come from the flame edge, a large lens was selected for
detecting light. The angle of beam steering was less than 3 degrees. Base line matching with wing effects
is a more complicated process.
6. CONCLUSION
The measurements of H2O gas temperatures were made at 20 kHz using TDLAS techniques by 2.0 µm
semiconductor laser. The ratio of the 4687.193 cm-1 and 5008.101 cm-1 absorption lines around 2.0 µm
was chosen, and it was evident that it has advantages for the application of combustion measurement. The
newly developed TDLAS system was examined in a heated static gas cell, gas temperature measurements at
20 kHz were examined an actual turbulent premixed burner as well. There was good agreement with the
thermocouple results for both measurements. It proved the high temporal rate of temperature
measurement could be used for real-time combustion monitoring and extended to feedback control.
Turbulent burner, (r, z)=(7, 56)
1500
Temperature(K)
ave.=1351.9 K
rms=36.7 K
1400
1300
1200
0
25
50
Time(ms)
75
100 0
5
10
PDF(%)
15
Figure 9 Time series gas temperature measurement at 20 kHz at the inner cone of turbulent premixed
methane/air flame
-7-
REFERENCES
[1] Adachiet, M. al., SAE paper, 971018, pp.209-219 (1997)
[2] Eckbreth, A. C., Laser Diagnostics for Combustion Temperature and Species, Abacus Press, Energy &
Engineering Science Series Vol. 7 (1988)
[3] Chigier, N., Combustion Measurements, Hemisphere Publishing Corporation (1991)
[4] Philippe, L.C. et al., Applied Optics, Vol. 32, pp. 6090-6103 (1993)
[5] Roy, G., et al., Propulsion Combustion, Taylor & Francis, pp. 281-328 (1997)
[6] Sonnenfroh, D. M. et al., Applied Optics, Vol. 35, No. 21, pp. 4053-4058 (1996)
[7] Hanson, R. K., AIAA paper, 97-0115 (1997)
[8] Mihalcea, R.M. et al., Meas. Sci. Technol. 9, 3:327 (1998)
[9] Allen, M. G., Meas. Sci. Technol., Vol. 9, 542 (1998)
[10] Mihalcea, R .M. et al., Proc. Comb. Inst., 95 (1998)
[11] Webber, M.E. et al., App. Optics 40, 821-828 (2001)
[12] Webber, M. E., et al., Proc. Comb. Inst., Vol. 28, 2D02 (2000)
[13] Roy, G., Hanson, R. K. et al., Propulsion Combustion, Taylor & Francis, pp. 281-328 (1997)
[14] Wang, J., Maioriv, M., Jeffries, J.B., Garbuzov, D.Z. Connolly J.C. and Hanson, R.K., Meas. Sci.
Technol. 11, 11:1576 (2000)
[15] Mihalcea, R.M., Baer, D.S. and Hanson, R.K., Meas. Sci. Technol. 9, 3:327 (1998)
[16] Woiki, D., Votsmeier M., Davidson, D.F. and Hanson, R.K., Combust. Flame 113:624 (1998)
[17] Peterson, E.L., Davidson, D.F. and Hanson, R.K., Combust. Flame 117:272 (1999)
[18] Jenkins, T.P. and Hanson, R.K., Combust. Flame 126:1669 (2001)
[19] Nagali, V., Davidson, D.F. and Hanson, R.K., Journal of Quantitative Spectroscopy and Radiative
Transfer 64:651 (2000)
[20] Webber, M.E., Hanson, R. K. et al., Journal of Quantitative Spectroscopy and Radiative Transfer
62:511 (1999)
[21] Philippe, L.C. et al., Applied Optics, Vol. 32, pp. 6090-6103 (1993)
[22] Furlong, E. R., Baer, D. S., Hanson, R. K., AIAA paper, 96-2763 (1996)
[23] Mihalcea, R. M., Hanson, R. K. et al., International Symposium on Optical Sensing for
Environmental and Process Monitoring (1996)
[24] Petersen, L., Bates, R. W., Davidson, D. F., Hanson, R. K., AIAA paper, 97-0316 (1997)
[25] Wang, J., Maiorov, M., Baer, D. S., Garbuzov, D. Z., Connolly, J. C., Hanson, R. K., AIAA paper,
2000-0255 (2000)
[26] Nagali, V., Chou, S. I., Baer, D. S., Hanson, R. K., Segall, J., Applied Optics, Vol. 35, No. 21, pp.
4026-4032 (1996)
[27] Mihalcea, R. M., Baer, D. S., Hanson, R. K.,, AIAA paper, 96-0173 (1996)
[28] Mihalcea, R. M. et al., Applied Optics, Vol. 36, No. 33, pp. 8745-8752 (1997)
[29] Allen, M. G., Kessler, W. J., AIAA journal, Vol. 34, No. 3, pp. 483-488 (1996)
[30] Sonnenfroh, D. M., Allen, M. G., Applied Optics, Vol. 35, No. 21, pp. 4053-4058 (1996)
[31] Tanahashi, M. et al., Proceedings of Thermal Engineering Conference, No. 99-25, pp. 81 (1999)
[32] Aizawa, T., Kamimoto, T., Tamaru, T., Applied Optics, Vol. 38, No.9, pp. 1733-1741 (1999)
[33] Iseki, T. et al., Proceedings of International Conference on Trends in Optical Nondestructive Testing
(2000)
[34] Muta, K., Tanoura, M., Fujimoto, K., Tokuda, K., SPIE, Vol. 3108, pp. 64-71 (1997)
[35] Torniainen, E. D. et al., 26th International Symposium on Combustion (1998)
[36] Taylor, A. M. K. P., Instrumentation for Flows with Combustion, Academic Press (1993)
[37] Ikeda, Y et al., 10th International Symposium on Applications of Laser Techniques to Fluid Mechanics
at Lisbon (2000)
[38] Webber, M. E., et al., Proc. Comb. Inst., Vol. 28, 2D02 (2000)
[39] Allen, M. G., Meas. Sci. Technol., Vol. 9, 542 (1998)
[40] Mihalcea, R .M. et al., Proc. Comb. Inst., 95 (1998)
[41] Ebert, R .M. et al., Proc. Comb. Inst., Vol. 28, 2D04 (2000)
[42] Webber, M.E., Kim, S., Sanders, S.T., Baer, D.S., Hanson, R.K., Ikeda, Y., App. Optics 40,
821-828 (2001)
[43] Ikeda, Y. et al., Measurement Science and Technology, Volume 11, 6:617 (2000)
[44] Kojima, J. et al., T., Proc. Combust. Inst. 28 (2000)
-8-