**5. Results**

To validate the performance of the wavelength-multiplexed NIR/MIR absorption sensor, measurements are conducted in the burned gases above the flat flame at conditions identical to measurements presented by Weigand et al. [34], where gas-phase temperatures were obtained by coherent anti-Stokes Raman scattering (CARS) of molecular nitrogen, and CO, CO2, and H2O concentrations are calculated assuming local thermodynamic equilibrium; three stoichiometric and slightly fuel-rich flames labeled No. 17, 18, 19, and 20 ae studied and provide validation data for measurements with the sensor developed here. Table 2 lists flow rates (in standard liters per minute, slm) of CH4, air, and co-flow N2 and the corresponding equivalence ratio *φ* for the premixed CH4/air flames. Also listed are the adiabatic flame temperatures (*T*ad) and the measured N2 CARS temperatures, with the latter varying in the range 1883–2100 K. The three flames have equilibrium concentrations of CO ranging from 0.31 to 4.3%, CO2 varying between 6.5 and 9.1%, and water vapor at approximately 19% (Table 2).

**Table 2.** Operating conditions of the four premixed CH4/air flames from Weigand et al. [34] examined in this work. Adiabatic flame temperatures and species mole fractions from equilibrium calculations (EQ) as well as measured CARS temperatures as listed from [34]. CARS temperatures were measured at 15 mm HAB. Measured laser absorption temperature and species mole fractions from this work are listed below the reference values.


The measurement line-of-sight is parallel to the burner surface and crosses the center of the burner at a height of 15 mm (matching the height above the burner for the CARS measurements). At this height, temperature gradients in the exhaust above the flame front are negligible [35]. The four lasers are tuned at 10 Hz with individual inverse sawtooth current ramps to cover each of the respective absorption line shapes from which integrated line areas are calculated. The signal voltages from each detector are acquired at 10 MHz and averaged for 30 s.

Temperatures are measured with two-line thermometry using water vapor transitions with center wavenumber positions 7185.59 and 6806.03 cm−<sup>1</sup> from Table 1. Figure 8 shows the transmitted signal (upper row) and the corresponding absorbance (lower row) over the relative wavenumber for these two lines for flame no. 17 in Table 2. As can be seen from Figure 8a,c, the "corrected" baseline that is scaled by measured reference baseline without absorption matches very well in the far wings of the lines due to the large tuning range of the NIR lasers (>1 cm−1) and the fact that the transitions are isolated from interference of neighboring lines from other species. The absorption line at 7185.59 cm−<sup>1</sup> has significant absorption at room temperature from ambient water vapor. Although the sender unit is also covered by an acrylic glass box on the optical table, the observation of a small absorbance indicates imperfect purging of the sensor unit. In contrast, the high energy of the lower state of the transition near 6806.03 cm−<sup>1</sup> reduces the sensitivity to interference from ambient water vapor, and its baseline scan is free of the corresponding interference absorbance (Figure 8c). As can be seen from Figure 8b,d, the fitted absorbance line shape agrees with the measured one fairly well, with a maximum residual of 0.5% of the peak absorbance.

For CO detection in the MIR, interference by CO2 and water vapor present a challenge. Since the water absorption lines are not significantly disturbed by other interfering species and temperature is measured using two-line thermometry, any water vapor interference can be calculated from the corresponding absorbance of one of the water transitions.

**Figure 8.** Water vapor transmitted signal (black lines in (**a**,**c**)) and calculated absorbance (black lines in (**b**,**d**)) over relative wavenumber at the H2O center wavelengths 7185.59 and 6806.03 cm<sup>−</sup>1, respectively, for the premixed flame no. 17 (Table 2). The separately measured baseline (blue lines in (**a**,**c**)) and fitted Voigt functions (dashed red lines) are also shown. The residuals shown in the lower section of each figure are between original data and fitted line shapes.

For the MIR scan of the CO lines, the interference of CO2 must be considered. Using the temperature determined from the water vapor, the CO and CO2 concentrations are simultaneously fit to the two MIR scanned-wavelength absorbances. The CO line near 2059.91 cm−<sup>1</sup> experiences only slight interference from CO2, while the spectral region around the CO line near 2190.02 cm−<sup>1</sup> at elevated temperatures is dominated by CO2 absorbance. Figure 9b and d show that the Voigt fits match the measurements well, with maximum residuals below 2% of the absorbance. Although there is a significant residual in the baseline in Figure 9d, the best-fit spectral profile matches well with the measured data suggesting a fairly complete CO2 spectroscopic model extracted from the HITEMP data base in this spectral range. However, the residual between measurement and modeled spectrum still is not "flat", signifying discrepancies in CO2 spectral line positions and relative intensities in the data base.

**Figure 9.** Transmitted signal and calculated absorbance originating from CO, CO2, and H2O over the relative wavenumber at the CO center wavelengths 2059.91 cm−<sup>1</sup> (**a**,**b**) and 2190.02 cm−<sup>1</sup> (**c**,**d**) for the premixed flame no. 17.

Table 2 and Figure 10 compare the measurements of the laser absorption sensor with the CARS temperatures and equilibrium concentrations for CO, CO2, and H2O [34]. Laser absorption recovers line-of-sight averaged temperature and concentration values, while the CARS technique provides spatially resolved measurements (within a roughly cylindrical probe volume of a few millimeter in length and a diameter of approx. 500 μm). Laser absorption temperature measurements are higher than the reported CARS measurements for lower flow rates and lower than CARS temperature for the high-flow-rate flame (No. 22). These statements lose their significance when the measurement uncertainties (plotted in Figure 10) are considered; note the error bars of CARS and the absorption measurements (for details, see figure caption) strongly overlap. Nerveless, it can be concluded that the temperature can be successfully measured using the sensor within the measurement uncertainties.

**Figure 10.** Measured temperatures (**a**) and concentration of CO2 (**b**), H2O (**c**) and CO (**d**) in comparison to the temperatures measured by CARS and the simulated equilibrium concentrations from Ref. [34]. The measurement uncertainties for the CARS measurements were assumed to be 2.5% according to Ref. [34]. The measurement uncertainty for temperature measurements using two-line absorption thermometry with 1.4-μm lasers is discussed in the relevant literature as being between 2.8% [7] and 6.7% [36]. In Ref. [29], the temperature deviation of identical burners is estimated at 2.5% as an additional source of error. Therefore, and as no detailed accuracy analysis was performed as part of this study, we estimated the error for our temperature measurements to be ~5%. Mole fraction measurements using laser absorption methods under best conditions (only CO2 in N2 at room temperature) have been reported in the literature with accuracies up to 4% [37]. In Ref. [36], an accuracy of 10% was achieved for the determination of water vapor mole fraction in the exhaust gas of a flat Mckenna-type burner. In Ref. [7], an accuracy of 4.7% for the determination of shock-tube-heated water vapor was reported. Since the conditions and the used setup in Ref. [36] best match ours, we estimated the error for our mole fraction measurements to be ~10%.

Laser absorption measurements of the concentrations of CO, CO2, and H2O shown in Figure 10b–d agree well with calculated equilibrium values from Gaseq [35]. For only the two rich operating points (flame #18 and #19), strong deviations between the equilibrium calculations and the CO concentration results are observed, which are larger than can be explained by the measurement uncertainty. In this case, there are apparently unconsidered systematic errors. Since the dynamic range of the CO signal is very large, small nonlinearities in the detector could cause these systematic errors at higher concentrations. Another explanation could be inaccuracies in the spectroscopic line parameters for the selected CO transitions or deficiencies in the determination of the CO concentration via the equilibrium

calculation using GasEq. Since these deviations occur at concentrations of CO that are irrelevant for the operation of gas turbines (only a few ppm CO may be generated) and the chosen laser was designed for this low-CO range, this deviation does not represent a limitation for the applicability of the sensor.

#### **6. Summary**

The design and demonstration of a compact, modular fiber-coupled laser absorption instrument for simultaneous remote measurement of temperature, concentrations of water vapor, carbon dioxide, and carbon monoxide is reported. The instrument is designed for remote operation in test facilities with extremely harsh ambient conditions. The assembly of the sender and receiver units on water-cooled breadboards allows them to be enclosed with a purged atmosphere to protect against the high-temperature humid gases and the extreme acoustic noise of the gas turbine combustion test facility. The gas-phase temperature is determined by two-line thermometry using two well-known NIR transitions near 7185.59 cm−<sup>1</sup> (1.3917 μm) and 6806.03 cm−<sup>1</sup> (1469.3 nm) without spectral interference from other primary product species; the H2O concentration is then inferred from the total absorbance of either transition. The concentrations of CO and CO2 are detected using two MIR laser wavelengths near 2059.91 cm−<sup>1</sup> (4.8546 μm) and 2190.02 cm−<sup>1</sup> (4.5562 μm). These two spectral regions have quite different amounts of CO2 interference, enabling simultaneous modeling of the absorbance measured by both MIR lasers using the temperature determined by water vapor absorbance to infer the CO and CO2 concentrations.

The two NIR and two MIR lasers are multiplexed into an optical fiber (single-mode in the MIR) for delivery from the sending unit to the measurement line-of-sight. The four laser beams are combined into an indium trifluoride (InF3) single-mode optical fiber, delivering all lasers colinearly overlapped along the absorption path length. We believe the efficient, low-noise laser-multiplexing into the InF3 fiber is unique to this sensor.

An algorithm is developed for finding the baseline for wavelength-scanned direct absorption spectroscopy; even when within the scan range of each laser, no true zeroabsorption baseline can be recovered.

The sensor operation is validated by measurements in the burned gases in the exhaust of a premixed flat flame previously studied by CARS [34]. The best-fit laser absorption measurements of gas temperature and CO, CO2, and H2O concentration agree well with the CARS measurements and equilibrium calculated concentrations. The validation experiment illustrates that the combined NIR/MIR laser absorption sensor is suitable and ready for field applications.

**Author Contributions:** Conceptualization, T.E., T.D.; methodology, L.S., T.E.; software, L.S.; validation, T.E., J.B.J.; formal analysis, L.S.; investigation, L.S.; resources, C.S.; data curation, L.S.; writing—original draft preparation, L.S.; writing—review and editing, T.E., T.D., J.B.J., C.S.; visualization, L.S., T.E.; supervision, T.E., J.B.J., T.D., C.S.; project administration, T.E.; funding acquisition, C.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded by the Federal Ministry for Economic Affairs and Energy of Germany, project 03ET7011L, and Siemens Energy AG. JBJ's visit to Duisburg was sponsored by the German Research Foundation (DFG) in the context of a Mercator Fellowship within project 229633504.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** We acknowledge support by the Open Access Publication Fund of the University of Duisburg-Essen.

**Conflicts of Interest:** The authors declare no conflict of interest.
