Light detection and ranging (lidar) is a remote sensing technique [1] commonly used for the measurement of various atmospheric variables, including temperature, pressure, humidity, and trace gases concentration [1–6]. Lidar techniques have also been demonstrated for combustion diagnostics [7–10]. Typically, most of these methods are limited to the measurement of one species at a time as the laser wavelength is tuned to the absorption line of the investigated species. Recent advances in nonlinear fiber optics have led to the realization of spatially coherent and broadband light sources known as supercontinuum (SC) lasers. Under specific pumping conditions, the SC generation process exploits cascaded nonlinear dynamics resulting in highly directional broadband light [11]. This opens the door for simultaneous detection of multiple observables. Thus, supercontinuum light detection and ranging (SC-lidar) [12,13] and SC-based hyperspectral lidar [14,15] have shown great potential for simultaneous detection of multiple variables. To the best of our knowledge, an approach based on short-range ($\sim 1-10\mathrm{m}$) SC-lidar for temperature measurement in combustion power plants is yet to be reported in the literature.

Temperature distribution inside a furnace is one of the main factors affecting the performance of combustion units (CUs). The ability to accurately measure the temperature inside a CU would pave the way for precise control and management of the combustion processes. Conventionally, thermocouples are used for temperature measurement in CUs [16]; however, this approach is inefficient because of its point-wise nature and ineffective for dynamic temperature monitoring due to the poor response of the thermocouple to fluctuating temperatures [17]. Other prior art solutions besides the lidar approach in Refs. [7–10] include spectroscopic methods such as thermometry based on thermographic phosphors [18], two-line thermometry employing wavelength modulation spectroscopy [19–21] and atomic fluorescence [22]. Another important technique is collinear photo fragmentation and atomic absorption spectroscopy, which utilizes chemical reaction kinetics to measure temperature [23]. The aforementioned promising optical spectroscopic methods detect signal transmitted through the examined space, thus requiring at least two openings in the furnace walls, which limits their detection area to the line between those openings.

Herein, we present a new method for non-intrusive combustion diagnostics based on an SC-lidar using just one opening. The approach is robust as the stringent requirement of a narrowband laser linewidth being precisely in tune with the absorption line of the probe gas, particularly in the aforementioned techniques, is mitigated by the broadband SC laser. The technique exploits the gas absorption cross-sectional dependence on temperature between wavelength bands. Consequently, the absorption strength of water vapor (${\mathrm{H}}_2\mathrm{O}$) varies as a function of temperature, as shown in Fig. 1. We studied the temporal dynamics of ${\mathrm{H}}_2\mathrm{O}$ temperatures in a laboratory furnace using just two wavelength bands. By measuring the temporal intensities of the backscattered SC light, the corresponding transmittance of individual wavelength band can be deduced, hence providing information indicative of differential absorption between the two wavelength bands. ${\mathrm{H}}_2\mathrm{O}$temperatures in the range of 400°C–900°C were inferred by comparing the transmittance ratio between these two wavelength bands.

 

Fig. 1. Modeled ${\mathrm{H}}_2\mathrm{O}$ transmittance spectra for two temperatures and an overlay of filter transmissions. Blue (—) and orange (– - -) are the ${\mathrm{H}}_2\mathrm{O}$ transmittance spectra at 400°C and 900°C, respectively. Green (– –) and purple (.) are the corresponding filter transmissions for Ch1 and Ch2, respectively.

Download Full Size | PDF

The SC-lidar signal (i.e., the detected back scattered signal) can be expressed as

When two wavelength bands are used to probe flue gas parameters (as in Fig. 2), the corresponding signals detected at an individual wavelength band at a given time can be written as

 

Fig. 2. Schematic diagram of the experimental setup. BS: beam splitter.

Download Full Size | PDF

The transmission spectrum of ${\mathrm{H}}_2\mathrm{O}$, one of the main combustion products with a relatively high concentration and a rich absorption spectrum in the entire infrared, was thoroughly investigated. Specifically, the transmission spectrum of ${\mathrm{H}}_2\mathrm{O}$ corresponding to a 100% concentration, a pressure of 1 atm, and an interaction length of about 1 m was modeled using the HITRAN 2012 database [24], as shown in Fig. 1. Note that the aforementioned ${\mathrm{H}}_2\mathrm{O}$ parameters correspond to an equivalent optical thickness of the ideal power plant conditions. Typically 10% ${\mathrm{H}}_2\mathrm{O}$ concentration, 10 m interaction length and a pressure of 1 atm. Based on the data presented in Fig. 1, two wavelength bands, namely, channel one (Ch1) and two (Ch2), were carefully selected. They are characterized by different responses of absorption to the temperature variation which allows for differential absorption measurement.

Figure 2 depicts the schematic of the experimental setup. The SC is generated by a cascaded Raman process in a 500 m long SMF-28 fiber (Thorlabs), pumped by KEOPSYS PYFL-KULT laser operating at 1064 nm and generating 2 ns long, 1 kW peak power pulses with a repetition rate of 280 kHz. The output spectrum of the source is presented in Fig. 3. SC light is collimated with a reflective collimator and guided towards $\sim 1\mathrm{m}$ long quartz tube furnace containing 100% ${\mathrm{H}}_2\mathrm{O}$ concentration at atmospheric pressure. Using a 50:50 beam splitter (BPD508-G, Thorlabs), part of the incident light is directed towards the 1st scatterer/diffuser (DG10-1500-P01, Thorlabs), placed just before the furnace. The rest of the beam traverses through the furnace undergoing absorption and, subsequently, scattered by the 2nd scatterer behind the furnace. In an ideal combustion environment, the role of the scatterers is played by the naturally present aerosol particles which also provides sufficient scattering that enables signal detection over relatively long distances within the CU. Note that aerosol scattering is weak compared to that of the scatterers used in this experiment; hence, additional electronic pre-amplification of the signal will be required. The signal from the 2nd scatterer passes through the furnace again on the way back. The backscattered light from both scatterers is collected by two lens sets (AC254-030-C & AC254-075-C, Thorlabs) placed adjacent to each other, filtered by Ch1 and Ch2 filters (BP-1375-085 & BP-1240-050, Spectrogon) and focused on to the active area of two biased photodetectors (DET08C, Thorlabs). The temporarily resolved signal is recorded by a 12-bit oscilloscope (HDO6054, Lecroy). An example recorded signal is shown in Fig. 4.

 

Fig. 3. SC spectrum and the corresponding FWHM position of the filters (Ch1 and Ch2) in the spectrum.

Download Full Size | PDF

 

Fig. 4. Measured backscattered signals with no ${\mathrm{H}}_2\mathrm{O}$ present in the furnace. The first and second peaks are the signals from the 1st and 2nd scatterers, respectively. The dotted line represents the Ch1 signal, while the solid line represents Ch2 signal.

Download Full Size | PDF

The two distinct peaks separated in time represent the signals from the 1st and 2nd scatterers, respectively. The signal plotted in the figure is an average of 10,000 backscattered pulses, and the corresponding measurement time is about 40 ms. Furthermore, the measured temporal intensities of both channels are converted into individual channel transmittance using Eqs. (4) and (5). Figure 5 shows the measured transmittance of Ch1 and Ch2 at varying furnace temperatures of 400°C–900°C. The measurement at a given temperature composes 20 sets of measurement, with each measurement further composed of 10,000. Ch1 shows no change in transmittance with furnace temperature, while Ch2 shows an increase in transmittance with increasing furnace temperature, which is in good agreement with the modeled ${\mathrm{H}}_2\mathrm{O}$ transmittance in Fig. 1. The standard deviation of the transmittance measurement, calculated based on 20 measurement repetitions of 10,0000 pulses, presented in Fig. 5 is 0.0024, with the mean value of transmittance ranging from 0.4704 to 0.5837, depending on the channel and temperature.

 

Fig. 5. Measured Ch1 (blue line) and Ch2 (orange line) transmittance with respect to furnace temperature. The measurement was carried out for a temperature range of 400°C–900°C, with 20 sets of measurement per temperature and each single measurement composing of 10,000 pulses.

Download Full Size | PDF

In order to infer the temperature of ${\mathrm{H}}_2\mathrm{O}$, the transmittance ratio $R$ was calculated using Eq. (6). The mean value of all 20 measurements, which composes 200,000 pulses at a given temperature, was used in the calculation. The simulated and experimentally measured $R$ as a function of the furnace temperature are compared in Fig. 6. We can see very good agreement between the simulated and experimentally measured ratios with a 50°C accuracy in the range 600°C–900°C. However, at lower furnace temperatures (400°C–500°C), the discrepancy between the simulated and measured ratios is more pronounced. This is because the quartz tube (i.e., the gas cell) extends a few centimeters outside the furnace heating zone on either sides of the furnace, thus creating a colder interface (i.e., temperature gradient) with respect to the internal part of the furnace (i.e., the heating zone). The colder region was measured to be less than 100°C. As a result, ${\mathrm{H}}_2\mathrm{O}$ in these regions begins to condensate which yields additional loses on the beam path that are not accounted for in our model. Moreover, the temperature difference between the colder region and the heating zone was persistent, even at elevated furnace temperatures. Hence, the assumption of uniform ${\mathrm{H}}_2\mathrm{O}$ temperature and concentration inside the furnace is partially violated by the presence of the colder regions in the test tube outside the furnace heating zone. Therefore, condensations of ${\mathrm{H}}_2\mathrm{O}$ at lower furnace temperatures (400°C–500°C) coupled with the alteration of ${\mathrm{H}}_2\mathrm{O}$ concentration due to interference from the colder regions at all furnace temperatures, have led to the observed discrepancies between the simulated and experimentally measured transmittance ratios in Fig. 6.

 

Fig. 6. Experimentally measured (circles) and simulated (solid line) transmittance ratios as a function of temperature.

Download Full Size | PDF

In summary, we have reported a novel technique based on a short-range SC-lidar for temperature measurement in CUs via a single opening. Water vapor with optical thickness equivalent to combustion power plant conditions was mimicked in a laboratory quartz tube furnace. ${\mathrm{H}}_2\mathrm{O}$ temperatures were inferred from the transmittance ratio between two distinct wavelength bands. The measured and simulated ${\mathrm{H}}_2\mathrm{O}$ temperatures were observed to be in a good agreement. The 2 ns duration of the spectrally filtered SC pulses should allow for a spatial resolution of about 30 cm with this technique in a case where scatterers are present all along the measurement path, as is the case in a combustion furnace. In addition, the SC pulses have significantly high peak power with a duty cycle of 0.0005, thus preventing any potential interference by thermal emissions such as black body radiations and chemiluminescence, from the furnace environment. Moreover, the same technique can be employed for simultaneous detection of temperature and concentration if multiple (i.e., more than two) wavelength bands of the SC spectrum are utilized in the combustion diagnosis, e.g., by adding an additional bandpass filter in the 1450–1550 nm spectral region. Finally, we emphasize that the technique possesses a great potential for simultaneous 3D mapping of both temperature and concentration, which can be achieved by varying the direction of the probe beam in a non-parallel plane.