https://doi.org/10.3390/s23020622
”
A variety of techniques have been developed and used for temperature measurements in combustion systems [1,2,3,4]. There are non-optical (thermocouples) and optical diagnostics (absorption spectroscopy, imaging, laser-induced fluorescence, and coherent anti-Stokes–Raman scattering CARS). The results of such diagnostics of a hot zone are used both in fundamental investigations and technological applications.
The decisive advantage of optical methods is their “non-invasiveness”. The intensity of the probing radiation is so low (on the order of a few mW) that it does not affect the processes in the probed zone. In contrast, the use of thermocouples to determine T is fundamentally impossible, since their insertion into the test zone significantly affects the values and spatial distribution of the parameters of the hot jets or internal combustion chambers with temperatures from some hundreds to some thousands K. In many cases the diagnostic of such types of media needs temporal resolution in the ms-µs range, which is impossible for thermocouples but can be realized by laser-based spectroscopic techniques.
Due to the relative simplicity and high sensitivity in detecting atoms and molecules, tunable diode laser absorption spectroscopy (TDLAS) has received special attention. In hot zones, the TDLAS technique allows for the measurement of temperature, total pressure of the gas mixture, and partial pressures of the main molecular components with a time resolution in the micro–millisecond range [5,6,7,8]. The main advantages of the TDLAS technique include non-intrusive, non-perturbing measurements, the relative simplicity of design, the relatively low cost of the main components, and the possibility of delivering the probing radiation of a diode laser (DL) to the hot zone via optical fiber. This last advantage makes it possible to locate the sensitive recording part of the TDLAS sensor away from the testing hot zone with high acoustic and electrical noise [9].
The method of determining the temperature of a gas medium using TDLAS is based on the measurement of the integrated absorbance on several lines of a test molecule having different lower energy levels [10]. The lines’ integrated absorbances, in turn, are determined through a process of fitting a simulated spectrum, constructed using spectroscopic databases, to the measured one. Having determined the temperature of the medium in this way, it is also possible to find the concentration of the molecular components of a mixture by measuring the absorption of the probing DL radiation in the medium.
To date, various versions of TDLAS have been developed for determining the temperature: direct absorption spectroscopy (DAS) and several variants of wavelength-modulation spectroscopy (WMS). It should be noted that in these methods, the absorption of a test molecule is measured against a baseline, determined by different sources. The recorded signal is proportional to DL intensity, and, thus, depends on the time-varying DL intensity and its fluctuations.
DAS is the simplest method for determining the parameters of a medium [11,12,13]. It is easily interpreted, and works well in cases when the noise is small compared to the magnitude of the absorption signal. In industrial applications, especially in combustion installations with supersonic jets, DAS has limited use due to strong noise distorting the absorption signal. Such noises are both additive in nature (broadband radiation of a hot zone, electromagnetic pick-ups, etc.), and multiplicative, caused by fluctuations in the intensity of laser radiation transmitted through the measuring region. This situation is typical for strong turbulence in the gas zone, in the presence of scattering particles in the laser beam path, and high levels of acoustic and electromagnetic noise.
To significantly reduce the effect of additive noise on the accuracy of determining the intensity of absorption lines, WMS (wavelength modulation spectroscopy) is used [14,15,16,17,18,19]. In this technique, in addition to slow scanning of the DL wavelength across the absorption line (with frequencies of the order of 1 kHz), a fast modulation of the wavelength is applied with a modulation amplitude comparable to the width of the absorption line and with frequencies f of the order of 10–100 kHz. The absorption signal is detected at harmonics of kf (usually 2f) using a lock-in-amplifier (LIA). If the frequency of kf is outside the spectrum of additive noise, then the latter can be greatly attenuated, which leads to an increase in the signal-to-noise ratio compared to DAS. Measurements at higher (k > 1) harmonics also make it possible to reduce the influence of the laser radiation residual amplitude modulation (RAM) accompanying the wavelength modulation. However, the WMS signal when registered at one harmonic (as well as the DAS signal), due to its multiplicative nature, depends on the intensity of the laser radiation being recorded.
This difficulty was overcome by normalizing the kf harmonic of the signal to the first harmonic [20,21,22,23]. The signal received as a result of normalization does not depend on the intensity of laser radiation. The signal processing, taking into account the tuning characteristics of the DL, improves the accuracy of a temperature determination in a hot zone. It should be noted that in any version of the WMS method, data on the parameters of the gas under test can be obtained only after determination of the tuning and modulation characteristics of the DLs. The algorithms for processing raw data in the WMS method are much more complex than in the DAS method.
One of the radical ways to eliminate the dependence of the sensor output signal on non-selective variations in the intensity of the detected laser radiation is the use of a logarithmic conversion. Several applications of logarithmic conversion in conjunction with WMS log-WMS have been described. In [24,25], when detecting Cl atoms in a glow discharge, the log-WMS technique provided suppression of excessive laser noise. The increase in linear dynamic range of the output signal using log-WMS was demonstrated in [26]. The log-WMS technique was used to determine the concentration of absorbing molecules [27,28,29,30]. However, all studies used the registration of the second harmonic at a relatively low frequency of laser modulation, and measurements were carried out on separate isolated absorption lines to determine the concentration of the tested atoms or molecules.
In our work [31], a version of the log-WMS technique was proposed using a combination of logarithmic conversion of the experimental data and modulation spectroscopy with the registration of the first harmonic of the absorption signal of the test molecule (log-WMS-1f). The proposed version of the TDLAS technique and the processing algorithm significantly simplify the procedure for the evaluation of the experimental data in comparison with the published versions of the WMS. In [31], the proposed technique was demonstrated by determining the integrated absorbance of a single absorption line of an H2O molecule at atmospheric pressure and room temperature.
In this paper, the possibility of using the technique developed in [31] to determine the temperature of a hot zone by measuring the first harmonic of the signal of two absorption lines of a test molecule was investigated. Two DLs were used in the range of 1.3–1.4 µm. Scanning of the laser wavelengths in the vicinity of the selected absorption lines of H2O was carried out alternately (time multiplexing) with a frequency of 122 Hz. At the same time, the wavelengths of the DLs were modulated with a frequency of ~50 kHz. The operation of the proposed technique under conditions of strong multiplicative noise and broadband thermal radiation was investigated in detail. Data collection and signal processing algorithms were based on the logarithmic conversion of DL signals, a differential registration scheme, and LIA at the first harmonic of the modulation frequency. The temperatures inferred from absorption measurements were compared with thermocouple data in the presence of multiplicative noises and without them. The influence of background radiation, which after logarithmic conversion causes an additive signal, was analyzed in detail. The advantages and limitations of the proposed technique are discussed.
“