Fluorescence spectroscopy of kerosene vapour at high temperatures and pressures: potential for gas turbines measurements

Abstract

Laser-induced fluorescence spectroscopy of kerosene vapour was performed in a heated test cell operating between 450 and 900 K, at pressure from 0.1 to 3.0 MPa, for oxygen molar fraction between 0 and 21 %, with different laser excitation wavelengths (248, 266, 282 and 308 nm). Results show that, depending on the laser excitation scheme, kerosene fluorescence spectrum exhibits one or two fluorescence bands in the UV–visible range (attributed to aromatics naturally present in kerosene fuel). Fluorescence intensity of these bands decreases with increasing temperature, pressure and oxygen molar fraction. Different imaging strategies were derived from spectroscopic findings to simultaneously measure temperature and equivalence ratio fields in kerosene/air sprays, or flame structure and fuel spatial distribution in kerosene/air aeronautical combustors, by means of planar laser-induced fluorescence on kerosene vapour (K-PLIF).

1 Introduction

Recent research on civil gas turbine engines has shown that new injector concepts such as RQL (Rich burn, Quick quench, Lean burn), LPP (Lean, Premixed, Prevaporised) and multi-point offer some advantages to obtain high combustion efficiency and ultra-low pollutant emissions (NO_x, CO_x and soot), despite possible combustion instabilities. Further optimisation of gas turbines operation (to cope with increasingly stringent regulations on pollutant emissions) requires the improvement of these injection systems to better evaporate liquid fuel and homogenise air/fuel vapour mixture in the combustion chamber. It is, therefore, necessary to be able to monitor local equivalence ratio of the air/fuel mixture in the combustor in order to determine combustion parameters, which can yield engine operation at optimum efficiency. Measurement of fuel vapour concentration requires the use of instantaneous and non-intrusive optical diagnostics in order not to disturb the flow and to gather information on physical parameters. In particular, laser-based diagnostics are well suited for this type of measurements and have been largely developed over the last two decades. The most common techniques include Rayleigh scattering [1–3], laser-induced fluorescence (LIF) [4–9], absorption [10–12], spontaneous Raman scattering [13–16] and laser-induced breakdown spectroscopy (LIBS) [17–20]. LIF is particularly attractive because it allows to probe the liquid and vapour phases of the fuel and thus to determine the spatial distribution of each phase, which cannot be done easily with Rayleigh scattering. LIF can also be used to image the flow as opposed to line-of-sight absorption measurements. In addition, LIF delivers intense signals compared to those obtained with spontaneous Raman scattering, which leads to better sensitivity. Until now, the use of these laser diagnostics for equivalence ratio measurements generally required surrogate fuels or real fuels seeded with fluorescent tracers. This is not totally satisfactory because surrogate fuels or added tracers may have different combustion behaviour compared to real fuels, which may lead to possible misinterpretations of the combustor operation. Therefore, it is desirable for gas turbine manufacturers to have data with real fuels like kerosene at realistic operating conditions (i.e. with a preheated air/fuel mixture at T = 900 K and P = 3.0 MPa). Indeed, the use of a fluorescent tracer naturally present in the fuel is the key to accurate and relevant diagnostics. Nonetheless, distillation effects must be considered carefully, and one has to make sure that these tracers evaporate in the same way as the main aliphatic compounds present in the fuel. In the past, few studies have applied LIF to kerosene vapour detection at high temperature and pressure (T ~ 800 K and P = 2.0–4.0 MPa) in order to determine fuel distribution for several injectors [21–23]. Unfortunately, these results were only qualitative and did not provide information on which species fluoresced. Arnold et al. [24] also applied planar laser-induced fluorescence (PLIF) to fuel–oil detection in gas turbines for a limited temperature range (360 ≤ T ≤ 486 °C) at atmospheric pressure. The authors have indicated that PLIF could enable equivalence ratio measurements within a 25 % error. To our knowledge, Fujiwara et al. [25] were the first to observe laser-induced fluorescence spectrum of kerosene in the 310–400 nm range in kerosene/air flames, with a laser excitation at 290 nm. The authors could not assign the observed fluorescence, but mentioned it may be due to polyaromatic compounds present in the kerosene or produced in the combustion process. Later on, Löfström et al. [26] have performed spectroscopic studies of kerosene vapour fluorescence for temperature ranging from 300 to 900 K, with a laser excitation at 266 nm. The authors mentioned the existence of two separate structures in the kerosene fluorescence spectrum, which was found to vary with temperature and oxygen molar fraction. However, the dependence of kerosene fluorescence on pressure was not studied, and the molecular species responsible for these spectral structures were not identified. This crucial information is needed in order to understand properly kerosene fluorescence and to derive a strategy for measuring equivalence ratio in gas turbines. It is well known that the composition of kerosene commonly used in civil engines (i.e. Jet A1) consists of a wide range of hydrocarbon molecules. Among these, aromatic molecules are known to have UV and visible transitions exhibiting strong fluorescence [27–29]. In the literature, many articles on aromatics have provided detailed spectroscopic information required for the quantitative analysis of these spectra [30–38]. Nonetheless, these studies only concerned pure aromatic molecules, but did not attempt to consider them altogether, for instance to simulate a surrogate fuel with spectroscopic properties similar to that of kerosene. In addition, most of these studies were performed at low temperature and low pressure, which are not representative of operating conditions in gas turbines. Only a few studies have performed LIF measurements on mono-aromatics [39–43], di-aromatics [44–48] or tri-aromatics [49] at high temperature or pressure for concentration or temperature measurements in automotive engines or flames. These results are useful to better understand fluorescence signals from kerosene vapour in aeronautical gas turbines, as well as fluorescence signals from gasoline or diesel fuels in internal combustion engines.

In the present study, detailed spectroscopic measurements were carried out in order to identify the main fluorescent species in kerosene vapour and to determine the photophysics of kerosene (Jet A1) vapour fluorescence under various conditions of temperature, pressure and oxygen molar fraction. From the analysis of kerosene fluorescence spectra, strategies were derived to perform planar laser-induced fluorescence measurements on kerosene vapour (K-PLIF) with different applications in real gas turbines environments.

2 Experimental set-up

Description of the experimental apparatus used in the current experiments is divided into two parts. The first part details the set-up and procedures used for spectroscopic point measurements performed in a test cell for various thermodynamic conditions. The second part describes the laser and detection systems used for 2D imaging of kerosene fluorescence in reactive and non-reactive flows.

2.1 Spectroscopic measurements

An optical accessible gas cell allowed measurements of kerosene vapour fluorescence at temperatures between 450 and 900 K, pressures from 0.1 to 3.0 MPa and different excitation wavelengths (248, 266, 282 and 308 nm). The set-up used for experiments is represented in Fig. 1. Jet A1 kerosene was pressurised by nitrogen in a 1-litre reservoir, and nitrogen and oxygen were supplied by pressurised bottles. It is noted that N₂ has been slowly bubbled for an hour through the liquid fuel to remove possibly dissolved oxygen from the liquid prior to fluorescence measurements. Liquid and gas supply lines were connected to a Controlled Evaporator and Mixer (CEM, Bronkhorst), which heated and mixed kerosene vapour with N₂ and O₂ at controlled mass flowrate (0.81 g/h for the kerosene mixture, 0.75 l/min for the N₂/O₂ mixture), temperature and oxygen molar fraction (from 0 to 21 %). The molar fraction of kerosene in the mixture exiting the CEM was fixed to 2.6E−3. The outlet of the CEM was connected to a stainless steel cell preheated at temperature between 450 and 900 K, where the gas mixture was injected. In order to avoid fuel pyrolysis in the test cell (especially at high temperature under N₂/O₂ atmosphere), residence time of the gas mixture in the cell was <0.5 s. Absence of pyrolysis was verified by changing the gas flowrate and ensuring that absorption and fluorescence remained constant. Temperature of the gas inside the cell was controlled during the fluorescence measurements by means of a type K thermocouple placed a few millimetres aside of the laser probe volume. Temperature values mentioned in Figs. 2, 4, 5, 6, 7, 8, 9, 10, 11 correspond to the gas temperature measured by this thermocouple. Pressure between 0.1 and 3.0 MPa could be obtained inside the cell and was measured with a pressure transducer (Tb244, JPB). Optical access was provided by three UV-silica windows. Two windows were used for laser access, while fluorescence emitted by kerosene vapour was recorded through a third window placed at 90° from the propagation of the laser beam.

Fig. 1

Set-up for test cell experiments

Fig. 2

Influence of laser energy on kerosene fluorescence spectrum at 450 K, 0.1 MPa, in N₂, with 266 nm excitation wavelength

Some comments are required on the use of the CEM for vaporising multi-component mixtures. Indeed, multi-component fuels may lead to differential evaporation: high-volatile compounds of the fuel may vaporise before its low-volatile compounds. Therefore, at a given time of the evaporation process, the molar composition of fuel vapour may be different from that of the initial liquid fuel (see, for example, the article of Arnold et al. [24]). This feature is further enhanced if the distance and/or time allowed for fuel evaporation are limited. However, there is little chance that this could happen in the current experiments. Indeed, the CEM was operated at steady state conditions (spectroscopic measurements were performed 15 min after starting feeding the test cell with a mixture of fuel vapour and carrier gas), and the test cell was continuously fed with the mixture (this allowed perpetual replacement and mixing of the molecules present in the cell). Additionally, the length of the CEM capillary, where the liquid fuel vaporises and mixes with the carrier gas, was 185 cm, which ensured efficient mixing between molecules. Nonetheless, gas chromatography–mass spectroscopy (GCMS) measurements were performed by colleagues from Chemistry Department at ONERA on the “optical fuel” used to simulate kerosene fluorescence (see Sect. 3.1.4). GCMS was carried out on the initial liquid mixture and on mixtures recondensed at the outlet of the CEM 15 and 30 min after the CEM operation had started. No composition difference was observed between the three samples, which indicate no differential evaporation. Although such measurements were not performed on kerosene fuel, similar results are expected because the distillation curve of the “optical fuel” is close to that of kerosene.

In each experiment, the kerosene/N₂/O₂ gas mixture was excited by UV pulses from four different laser sources. Excitation at 248 nm was obtained with an excimer laser (PSX-501, OSYRIS) using a KrF gas mixture as the lasing medium. Excitation at 266 nm was provided with a frequency-quadrupled Nd:YAG laser (YG781, Quantel). Excitation at 282 nm consisted of using the second harmonic of the Nd:YAG laser (YG781, Quantel) at 532 nm to pump a dye laser (TDL50, Quantel) using Rhodamine 590, leading to a 282 nm laser beam after frequency-doubling with a KD*P crystal. Finally, an excimer laser (EMG150, Lambda Physik) delivered a laser beam at 308 nm (using a XeCl gas mixture as the lasing medium). For all the laser sources used in the current experiments, a repetition rate of 10 Hz was used and laser energy was kept below 1 mJ with beam diameter of 1 mm to ensure weak (linear) excitation. Note that laser energy was limited to this low level because preliminary tests showed that higher energy values (>3.0 mJ/pulse) lead to the photolytic decomposition of hydrocarbon molecules into other compounds (e.g. CH and C₂ radicals) resulting in systematic measurement error, as displayed in Fig. 2. Indeed, as can be seen, additional spectral structures due to the chemiluminescence of CH*, C₂* and other radicals (consecutive to hydrocarbons breakdown) are superimposed onto the fluorescence spectrum of kerosene as laser energy increases.

Fluorescence from part of the illuminated kerosene was collected at right angle by a spectrograph (SPEX 270M, Jobin–Yvon) with an entrance slit of 100 μm and a 600 groves/mm grating blazed at 400 nm. The light dispersed by the grating was recorded using a 16-bit intensified CCD camera (Princeton Instrument) with an intensifier gate width of 1 μs. This gate width is more than an order of magnitude longer than kerosene fluorescence lifetime. The CCD array was 576 by 384 pixels, and the framing rate of the system was fixed at 10 Hz. The ICCD camera was interfaced to a personal computer, which was used to control the camera and acquire the spectra. The spectral resolution of the detection system was about 1 nm, which was not enough to distinguish the fine structures in the fluorescence spectrum. However, this resolution allowed detection of changes or shifts in the fluorescence spectra, for example with increased temperature. Measurements of the absorption cross section at 266 nm of kerosene vapour were also performed using two photodiodes (DET210, Thorlabs) which signal was time-integrated by a boxcar (SR250, Stanford Research Systems) and recorded via a personal computer. Laser energy was recorded for each laser shot by the photodiodes and was calibrated against a power meter (Nova II, Ophir) in order to derive absolute values of energy. Linear response of the photodiodes with laser energy was also verified.

For each experimental condition, fluorescence and absorption measurements were averaged over 300 laser shots.

Absorption cross section was classically calculated using the Beer–Lambert’s law:

$$I = I_{0} \exp ( - \sigma N_{\text{abs}} L)$$

(1)

where I is the laser intensity at the probe volume, I ₀ the laser intensity at the entrance window of the test cell, N _abs the number density of absorbing molecules (cm⁻³) calculated from the ideal gas law, σ the molecular absorption cross section of the molecule (cm²), and L the distance between the entrance window of the test cell and the probe volume (cm). Absorption measurements were performed with and without kerosene flowing through the test cell in order to take into account the effect of losses through the cell windows. The experimental procedure used to obtain absorption cross section was first checked against values published in the literature for toluene [40], 3-pentanone [50] and acetone [51]. Good agreement was found, which gave us confidence on the method used and on the values proposed in the present article. Typically, at 450 K and 0.1 MPa, laser absorption by kerosene molecules present in the test cell was <10 % between the entrance window and the probe volume.

Each fluorescence spectrum was corrected for beam attenuation (due to absorption between the entrance window of the test cell and the probe volume) and absorption cross section and was normalised by laser energy. In addition, all the data shown in the article are referenced to the same number density. Indeed, despite the flowrates of liquid kerosene and carrier gas (N₂ or N₂/O₂) were kept constant for all experimental conditions, the spectra of kerosene have been corrected for decrease in number density when increasing temperature at fixed pressure or for increase in number density when increasing pressure at fixed temperature. An equivalent set of measurements was also acquired immediately afterwards with the same laser energy in the cell evacuated from any kerosene vapour to allow subtraction of background light from fluorescence. Spectra were corrected for the spectral response of the detection system using calibration from the emission spectrum of a deuterium lamp. Temperature (±1 K tolerance), pressure (±0.005 MPa tolerance) and kerosene vapour molar fraction (±2 % tolerance) were thoroughly controlled and particular attention was paid to perform measurements at steady state conditions. Accounting for all these sources, uncertainty on fluorescence signals was estimated to be <10 %.

2.2 PLIF imaging measurements

The first K-PLIF set-up aimed at measuring local equivalence ratio in kerosene/air flows using a data processing that will be described later. Experiments used a single-excitation scheme, which comprises a frequency-quadrupled Nd:YAG laser (Brilliant B, Quantel) generating 8 ns, 50 mJ pulses at 266 nm. The laser beam was transformed into a collimated sheet using a combination of cylindrical and spherical lenses. Two cylindrical lenses, −20 and 300 mm focal lengths, formed a cylindrical telescope which spread the beam into a collimated, 50 mm high sheet. The spherical lens, 1,000 mm focal length, focused the sheet to a 130-μm waist (at 1/e intensity). Kerosene fluorescence was recorded with two facing 16-bit intensified CCD cameras (IMAX-512, Roper Scientific) with an intensifier gate width of 100 ns. The CCD array was 512 × 512 pixels, and the framing rate of the system was fixed at 3 Hz. Both ICCD cameras were equipped with a f/4.1, f = 94 mm, achromatic UV lens (Cerco). Each camera was equipped with custom-made band-pass filters to collect separately fluorescence over different spectral ranges. The first filter was centred at 290 nm (FWHM = 40 nm, transmission of 80 %), and the second filter was centred at 360 nm (FWHM = 100 nm, transmission of 90 %). The choice of this detection scheme aimed at increasing the signal-to-noise ratio on each fluorescence image. A 50 × 50 mm² area of the flow was imaged by the ICCD cameras, so that the spatial resolution was about 100 μm per pixel along the dimensions of the image and 130 μm in the third dimension (i.e. perpendicular to the laser sheet).

The second K-PLIF set-up allowed simultaneous probing of spatial distributions of kerosene and OH radical in two-phase reactive flows. In these experiments, the laser source (TDL50 dye laser pumped by the second harmonic of the YG781 Nd:YAG laser) was tuned to a transition of the OH radical (e.g. the Q₁(5) spectral line at 282.75 nm, in the (1,0) band of the OH (X²Π–A²Σ⁺) system), which also allows to promote kerosene fluorescence. Laser energy was about 14 mJ/pulse, with a pulse duration of 8 ns. Fluorescence signals were recorded using the same intensified CCD cameras as with the first K-PLIF set-up described above. The first camera was equipped with a high-pass filter (ZUL0325, Asahi), which collects kerosene fluorescence above 325 nm. The second camera was equipped with a custom-made narrow band-pass filter centred at 308 nm (FWHM = 5.4 nm, transmission about 60 %), which detects OH fluorescence. Spectral domain for the detection of kerosene is redshifted compared that of OH radical, and therefore, fluorescence detected by the first camera results only from kerosene without any interference from OH. Crosstalk between kerosene and OH fluorescence is also limited on the second camera thanks to the use of the narrow band-pass filter. The approximate level of interference between OH and kerosene fluorescence in that narrow spectral range was estimated from measurements of (1) OH fluorescence in a reference CH₄/air premixed flame (developing over a porous burner) and (2) kerosene fluorescence in a heated kerosene vapour/air jet at 650 K (with known composition, see ref [43]), using the same camera and narrow band-pass filter. For an equal molar fraction of kerosene vapour and OH radical, results showed that the kerosene fluorescence signal represents only 1 % of the OH signal over the spectral range of the narrow band-pass filter.

3 Results

This section presents absorption and fluorescence measurements of kerosene (liquid and vapour phases) for various experimental conditions. The first part of the section describes detailed spectroscopic measurements performed to provide a better insight into the photophysics of kerosene fuel. LIF measurements are carried out in the heated cell in order to determine the influence of the laser wavelength on kerosene fluorescence spectrum and which excitation scheme has the best potential. The main fluorescent species in kerosene vapour are also identified. Systematic measurements are performed to investigate the influence of temperature, pressure and oxygen molar fraction on kerosene fluorescence and absorption cross section. All the measurements presented in this part were carried out with N₂ as carrier gas, except for experiments used to determine the effect of the collisional quenching of oxygen on kerosene fluorescence. In the second part of the section, K-PLIF is applied to gas turbine environments, in order to demonstrate the potential of the technique for measuring fuel vapour spatial distribution or to probe multiple species (fuel and OH radical) under various experimental conditions.

3.1 Spectroscopic measurements

3.1.1 Absorption spectrum of liquid kerosene

Figure 3 represents the absorption spectrum of liquid kerosene (displayed as optical density) recorded with a monochromator (Lambda 18, Perkin Elmer) between 200 and 900 nm, with a spectral resolution of 0.1 nm, at 293 K and 0.1 MPa. A zoom of the spectrum between 230 and 330 nm is also displayed on the top right corner, in order to better visualise this region of particular interest. Very strong absorption is noticed below 230 nm and in fact exceeds the capabilities of the monochromator (i.e. optical density of 4 is the maximum that can be detected). Nonetheless, probing kerosene at such low wavelengths would probably be of limited interest due to strong absorption by molecular oxygen. Between 245 and 330 nm, a second absorption band is present with a maximum in the range 266–274 nm. Due to the limited resolution power of the current monochromator, fine structures (e.g. rotational spectral lines) that might exist in the absorption spectrum of liquid kerosene cannot be observed in Fig. 3. The shape of this absorption spectrum indicates that fluorescence measurements on kerosene fuel (in the spectral range considered here) should preferably be performed with lasers in the UV range. This feature guided us for the choice of the laser sources used in the laser-based spectroscopic experiments. Compared to absorption at a wavelength of 266 nm, absorption is 37 % lower at 248 nm, 22 % lower at 282 nm, 88 % lower at 308 nm. This corresponds to wavelength of typical commercial lasers and helps selecting which laser may have the best potential for K-PLIF measurements.

Fig. 3

Absorption spectrum of liquid kerosene at 293 K, 0.1 MPa

3.1.2 Absorption experiments in kerosene vapour

Unfortunately, with the current monochromator and test cell geometry, it was not possible to record the absorption spectrum of kerosene vapour at high temperature. Therefore, only values of the absorption cross section of kerosene vapour obtained in the test cell, with a laser excitation at 266 nm, are presented here. Figure 4 shows that it remains almost constant with temperature between 450 and 900 K, with a mean value equal to 5.75E−19 cm² and a standard deviation of 0.20E−19 cm². By comparison, in the same range of temperature, the absorption cross section of 1,2,4-trimethylbenzene varies between 1.15E−18 and 1.35E−18 cm², whereas that of naphthalene lies between 1.25E−17 and 1.35E−17 cm² [43, 47]. Additionally, measurements at high pressure showed that pressure has no influence on the absorption cross section of kerosene. The fact that the absorption cross section of kerosene is independent of temperature and pressure helps simplifying data analysis of kerosene fluorescence signals.

Fig. 4

Kerosene absorption cross section as a function of temperature at 0.1 MPa, in N₂, with 266 nm excitation wavelength

3.1.3 Influence of excitation wavelength on kerosene vapour fluorescence

Figure 5 shows the typical resultant fluorescence spectra recorded for the four excitation wavelengths (248, 266, 282 and 308 nm). Spectra have been normalised by their respective maximum intensity. As can be seen, kerosene vapour fluorescence spectrum is mainly composed of two fluorescence bands, which strongly vary with excitation wavelength. The spectral band ranging from 310 nm to 420 nm can be observed with the four excitation schemes. It exhibits a broadband profile, which displays little changes (slight red-shift) with excitation wavelength. Note that small structures are also present in the 320–340 nm spectral range. The peak observed at 308 nm (with the excimer laser excitation) corresponds to laser beam reflections on optical windows and stray light. Another spectral band located in the 270–310 nm region shows only limited structures. It can be observed with 248, 266 and 282 nm excitation schemes only. However, fluorescence intensity of this band is clearly stronger for 248 and 266 nm excitation, which results from a strong absorption of aromatics at those wavelengths. From findings in Figs. 3 and 5, excitation at 266 nm seemed to be the most promising scheme; therefore, it was decided to systematically use it for all the spectroscopic experiments reported in the following.

Fig. 5

Influence of laser excitation wavelength on kerosene fluorescence spectrum at 450 K, 0.1 MPa, in N₂

3.1.4 Identification of fluorescent species in kerosene vapour

In order to identify the molecular species, which are responsible for kerosene fluorescence, fluorescence spectra for different pure aromatics diluted in n-undecane were systematically performed at various conditions of temperature, pressure and oxygen molar fraction [52, 53]. Only a few results of this work are recalled in this paper, and the reader is invited to refer to the articles of Orain et al. [43, 47] for full spectroscopic results on the pure aromatic compounds considered in this study. Selection of these molecular compounds has been guided by the results of the chemical analysis of the Jet A1 fuel by gas chromatography (GC). Examination of the species composition (given in Table 1) shows the presence of aliphatics, which are not considered as fluorescent species and two main families of aromatics: the single-ring aromatics such as alkylbenzenes and the two-ring polycyclic aromatic hydrocarbons such as naphthalenes. Using this chemical composition and following the work of Guéret et al. [54], several molecular species were then chosen to represent these aromatic compounds. 1,2,4-trimethylbenzene was used to simulate the mono-aromatic compounds and naphthalene, 1-methylnaphthalene and 1,3-dimethylnaphthalene were selected to represent the two-ring aromatics. The choice of these compounds is detailed in the following paragraph. Finally, n-undecane was selected to represent the major hydrocarbon compound of kerosene that does not fluoresce. Thermodynamic properties of 1,2,4-trimethylbenzene, naphthalene 1-methylnaphthalene, 1,3-dimethylnaphthalene and n-undecane are summarised in Table 2.

Table 1 Chemical composition of the kerosene used in the experiments

Table 2 Thermodynamic properties of 1,2,4-trimethylbenzene, naphthalene, 1-methylnaphthalene, 1,3-dimethylnaphthalene and n-undecane at 0.1 MPa

Figure 6 shows a typical comparison between the fluorescence spectrum of kerosene and that of the four pure aromatics diluted separately in n-undecane. Measurements were performed under N₂ atmosphere at 450 K and atmospheric pressure. Slight discrepancy observed on the left side of the first band (270–310 nm) suggests that, in addition to 1,2,4-trimethylbenzene, it may be necessary to consider another mono-aromatic species (with very small molar fraction compared to that of 1,2,4-trimethylbenzene) which fluorescence spectrum is slightly blue-shifted compared to that of 1,2,4-trimethylbenzene. Indeed, it is well known that mono-aromatic compounds exhibit similar fluorescence spectra whatever the number of CH₃ substitutions. Increasing the substitution number is also known to slightly shift the fluorescence spectrum to the red [29]. Potential candidates include for example toluene (mono-substituted mono-aromatic) and xylene (di-substituted mono-aromatic) [40, 43, 55]. However, fluorescence from those compounds was not investigated in the current study. Nonetheless, the omitted signal is fractional to the total fluorescence intensity of the kerosene first band or that of 1,2,4-trimethylbenzene (i.e. <7 % of the intensity integrated between 270 and 310 nm). Unlike the first fluorescence band, the second fluorescence band located between 310 and 420 nm cannot be reproduced from the fluorescence of a single two-ring aromatic compound. Indeed, as shown in Fig. 6, while the left part of the naphthalene and kerosene fluorescence spectra match well, the right part of naphthalene spectrum extends to smaller wavelength compared to kerosene spectrum. Therefore, this means that naphthalene only is not enough to well reproduce the second fluorescence band of kerosene spectrum. Additional two-ring aromatics belonging to the naphthalene family (i.e. the simplest two-ring polyaromatic hydrocarbon) are indeed required. Among these molecules, selection of mono- or poly-substituted two-ring aromatics was performed relative to their fluorescence spectra, which are slightly red-shifted compared to that of non-substituted di-aromatics [29]. It was observed that, despite the right part of 1-methylnaphthalene and kerosene fluorescence spectra match well (see Fig. 6), a mixture (naphthalene/1-methylnaphthalene) is still not enough to well reproduce the second band of kerosene fluorescence spectrum (slight discrepancies were still observed at wavelengths above 350 nm). This leads to add 1,3-dimethylnaphthalene (which fluorescence spectrum is further red-shifted compared to that of 1-methylnaphthalene) as a third compound of the mixture. As a result, similarly to the choice of 1,2,4-trimethylbenzene to represent the first fluorescence band of kerosene, a mixture of naphthalene/1-methylnaphthalene/1,3-dimethylnaphthalene was used to represent the second kerosene fluorescence band.

Fig. 6

Comparison between the fluorescence spectrum of kerosene and the spectrum of the four aromatics at 450 K, 0.1 MPa, in N₂, with 266 nm excitation wavelength

In order to obtain the best agreement between the fluorescence spectra of kerosene and the mixture of the four aromatic compounds diluted in n-undecane, whatever the conditions of temperature, pressure and carrier gas, a specific composition of aromatics has been determined from several adjustments based on results from chemical analysis of kerosene. The respective proportions of the different compounds are given in Table 3. This allows to define an “optical fuel” which fluorescence properties are equivalent to those of the jet A1 fuel considered in the present experiments. Figure 7 shows that the fluorescence spectra of kerosene vapour and of the mixture of aromatics with n-undecane are similar, in terms of spectral shape as well as in terms of intensity, for all the thermodynamic conditions investigated (from 450 to 650 K, between 0.1 and 1 MPa, under N₂ or air atmosphere). All these results generate confidence that the fluorescence spectrum of kerosene (Jet A1 considered in the present experiments) can be properly represented by the fluorescence of the mixture composed of the four aromatics considered in this study, over a large range of temperatures, pressures and oxygen molar fractions. In addition, comparison between fluorescence spectrum of the mixture and an analytical spectrum calculated from individual spectra of the four aromatics (weighted by their proportions in the mixture, given in Table 3) was performed for different temperatures, pressures and oxygen molar fractions. Results showed that the mixture spectrum and the analytical spectrum are in good agreement for all conditions; in particular, no spectral distortion is observed between both spectra for a given condition. This suggests that there is no energy transfer between the different aromatics after laser excitation and that the fluorescence of the mixture is the sum of the individual fluorescence of each aromatic compound (weighted by its proportion in the mixture). As a consequence, the fluorescence of kerosene considered in the present experiments could be modelled by the individual fluorescence of these four aromatics, and there is no need to consider interactions between the energy levels of two (or more) different aromatics. Finally, it shall be noticed that the mixture composed of the four aromatics has also the potential to define a surrogate fuel for kerosene because of its chemical and thermodynamic properties, which are similar to those of Jet A1. This mixture may be then used in model gas turbines experiments because the spatial distribution of mixture vapour in the combustor may be representative of the vapour repartition obtained with the real fuel.

Table 3 Chemical composition of the “optical fuel” used to simulate kerosene fluorescence

Fig. 7

Comparison between the fluorescence spectrum of kerosene and the spectrum of the aromatics mixture at a 0.1 MPa, in N₂, for 450 and 650 K and b 450 K, 0.1 MPa, for various oxygen molar fractions, with 266 nm excitation wavelength

3.1.5 Temperature dependence of kerosene vapour fluorescence

Figure 8a shows the evolution of kerosene fluorescence spectrum with temperature at 0.1 MPa under N₂ atmosphere. Each spectrum is normalised by the maximum value of the fluorescence spectrum recorded at 450 K. Fluorescence intensity decreases as temperature increases. However, each fluorescence band identified to represent, respectively, the fluorescence of mono-aromatics and di-aromatics exhibits different behaviour. The intensity of the first fluorescence band strongly decreases by about two orders of magnitude with increasing temperature, similarly to results with 1,2,4-trimethylbenzene [43]. On the opposite, the intensity of the second fluorescence band decreases only by about twenty with increasing temperature. Additionally, the profile of this band exhibits a clear red-shift when increasing temperature (about 5 nm per 100 K at wavelength larger than 340 nm), which is similar to observations with naphthalene and derivatives [44, 47, 48]. Moreover, the spectral structures (around 325 and 335 nm) observed in the profile of the second fluorescence band at 450 K progressively vanish with temperature. As temperature increases, population of the strongly coupled vibrational energy levels (presumably from naphthalene) responsible for these well-defined structures possibly decreases [45], which leads to reduced fluorescence from these levels.

Fig. 8

Temperature dependence of a kerosene fluorescence spectrum and b integrated fluorescence of kerosene, at 0.1 MPa, in N₂, with 266 nm excitation wavelength

Differences between the temperature dependence of the two fluorescence bands can be more conveniently shown in Fig. 8b where the temperature evolution of the integrated fluorescence of kerosene together with that of each spectral band is displayed. Each value is normalised by the integrated fluorescence of kerosene at 450 K. The first band is integrated between 270 and 310 nm (except for temperatures above 850 K where integration starts at 275 nm), and the second band is integrated between 310 and 420 nm. In Fig. 8b (and later in Figs. 9b, 11b and 12), dashed lines are plotted to guide the eye of the reader. Different evolution with temperature is observed between the two bands, which may be due to the effect of non-radiative deactivation processes which significantly differs with temperature between mono- and di-aromatics. Such processes include inter-system crossing rate between the first electronic state S₁ and the triplet states, internal conversion (i.e. radiationless transition from S₁ to the ground state) and vibrational relaxation. However, it is difficult to determine if a deactivation process is dominant over the others, and additionally, it may be that the extent to which each process reduces fluorescence depends upon temperature. Radiative lifetime measurements could be useful to better understand the kinetics of the excited electronic state and may help determining the strength of each deactivation process.

Fig. 9

Influence of oxygen molar fraction on a kerosene fluorescence spectrum at 450 K, 0.1 MPa and b integrated fluorescence of kerosene at 450 K, 0.1 MPa, with 266 nm excitation wavelength

3.1.6 Influence of collisional quenching by O₂ on kerosene vapour fluorescence

Figure 9a displays the evolution of kerosene fluorescence spectrum with oxygen molar fraction at 450 K and 0.1 MPa. Each spectrum is normalised by the maximum value of the fluorescence spectrum recorded for oxygen molar fraction equal to 0. The intensity of each spectral band continuously decreases as $X_{{{\text{O}}_{ 2} }}$ increases, due to increasing collisional quenching of kerosene fluorescence by O₂ molecules. This decrease, somewhat stronger for the di-aromatics fluorescence (60 vs. 20), suggests that oxygen quenches the fluorescence of di-aromatics more efficiently than that of mono-aromatics. This is inherent to the longer lifetime of the electronic state S₁ of di-aromatics (e.g. ~100 ns for naphthalene) compared to that of mono-aromatics (e.g. ~40 ns for 1,2,4-trimethylbenzene), which increases probability of collisions with O₂. Similarly to findings with temperature, the spectral profile of the second fluorescence band exhibits a red-shift as $X_{{{\text{O}}_{ 2} }}$ increases (about 10 nm when $X_{{{\text{O}}_{ 2} }}$ increases from 0 to 21 %), which is in agreement with results from naphthalene and 1-methylnaphthalene [44, 47]. In addition, the vibrational structures occurring at 325 and 335 nm disappear with increase in oxygen molar fraction. These structures are believed to exhibit long lifetime and are, therefore, strongly quenched by O₂ [45].

Differences between the influence of oxygen molar fraction on the fluorescence bands are evidenced in Fig. 9b where the evolution of the integrated fluorescence of kerosene together with that of each spectral band is displayed as function of oxygen molar fraction. Each value is normalised by the integrated fluorescence of kerosene at 450 K. Large decrease is observed for oxygen molar fraction <0.04, confirming that oxygen is a very strong quencher for aromatics, even at low concentration. For higher oxygen molar fraction, the decrease is lot smoother.

For a weak laser excitation, the fluorescence signal intensity may be written as:

$$I = \eta_{\text{opt}} \frac{E}{{{\text{hc}}/\lambda }}V_{\text{c}} N_{\text{abs}} \sigma \frac{{k_{\text{f}} }}{{k_{\text{f}} + k_{\text{nr}} + \sum\limits_{\text{q}} {k_{\text{q}} } X_{\text{q}} }}$$

(2)

where η _opt is the overall efficiency of the collection optics, E the laser fluence (J/cm²), hc/λ the energy (J) of a photon at wavelength λ, V _c the collection volume (cm³), N _abs the number density of absorbing molecules (cm⁻³), σ the molecular absorption cross section of the molecule (cm²), k _f the spontaneous emission rate, k _nr the upper-level decay rate owing to collisionless processes (such as intersystem crossing, photodissociation or vibrational relaxation), k _q the collisional-quenching rate and X _q the molar fraction of the quenching species.

Considering that oxygen is the main quencher for aromatics fluorescence [27, 28], the influence of oxygen on kerosene fluorescence intensity may be simplified using the Stern–Volmer formalism expressed as:

$$\frac{{I_{0} }}{I} = 1 + \alpha X_{{{\text{O}}_{ 2} }}$$

(3)

where I ₀ is the kerosene fluorescence signal for $X_{{{\text{O}}_{ 2} }}$ = 0 and α = $k_{{{\text{O}}_{ 2} }}$/(k _f + k _nr) the Stern–Volmer factor.

Figure 10 represents the integrated intensity of kerosene fluorescence spectrum displayed as Stern–Volmer plots for various temperatures (the dashed lines represent the linear fit from Eq. 3). Unlike experiments under N₂ atmosphere, in the presence of oxygen, measurements were restricted to temperatures up to 750 K due to safety regulations in our lab. As can be seen, Stern–Volmer plots exhibit linearity with oxygen molar fraction, whatever the temperature between 450 and 750 K, and the Stern–Volmer coefficient is equal to 190 bar⁻¹ at 450 K.

Fig. 10

Stern–Volmer plot of the kerosene fluorescence for different temperatures at 0.1 MPa, with 266 nm excitation wavelength

3.1.7 Pressure dependence of kerosene vapour fluorescence

Figure 11a shows the evolution of kerosene fluorescence spectrum with pressure at 450 K under N₂ atmosphere. Each spectrum is normalised by the maximum value of the fluorescence spectrum recorded at 0.1 MPa. As observed in this figure, the intensity of the first spectral band continuously decreases by about 40 % with increasing pressure, similarly to findings with 1,2,4-trimethylbenzene [43]. By contrast, the fluorescence of the second band first increases by 4 % (for pressure up to 0.5 MPa) and then gradually decreases by a mere 9 %. In a previous article, it was observed that naphthalene integrated fluorescence increases for pressure up to 0.7 MPa and is almost independent of pressure above 2.0 MPa [47]. No spectral deformation of both fluorescence bands is observed; in particular, the vibrational structures between 325 and 335 nm remain visible for all the pressures considered here, similarly to observations with naphthalene [47]. This suggests that population density in the first excited energy state S₁ may not be modified with pressure for di-aromatics, indicating that a fast thermalisation of the population on the different energy states of S₁ is already achieved before fluorescence occurs.

Fig. 11

Pressure dependence of a kerosene fluorescence spectrum and b integrated fluorescence of kerosene, at 450 K, in N₂, with 266 nm excitation wavelength

The pressure evolution of the integrated fluorescence of kerosene together with that of each spectral band is displayed in Fig. 11b. Each value is normalised by the integrated fluorescence of kerosene at 0.1 MPa. The two bands exhibit different evolutions: a monotonous decrease with pressure is observed for the mono-aromatics fluorescence, whereas di-aromatics fluorescence initially slightly increases with pressure, followed by a smooth small decrease. It is noted that the overall kerosene fluorescence decreases only by 16 % as pressure increases from 0.1 to 3.0 MPa. This indicates that kerosene fluorescence measurements can easily be carried out at high pressure, without large signal reduction due to non-radiative deactivation processes (e.g. collisions).

3.2 Potential for measurements in gas turbines

Spectroscopic results presented in the previous section have practical implications for optical diagnostics in kerosene/air flows. A first outcome is that, owing to the large fluorescence signal of aromatics naturally present in kerosene fuel, very low concentrations of these compounds can be detected with the K-PLIF technique. For example, the sensitivity of the first K-PLIF set-up described in the apparatus section was estimated using the heated kerosene vapour/air jet (with well-known composition) described in ref [43]. Results showed that it is typically about 1 ppm at 700 K, 0.1 MPa, in air, with a signal-to-noise ratio equal to 5. Second, although kerosene vapour cannot mark flow regions where temperature is above ~1,100–1,300 K (due to aromatics pyrolysis [56, 57]), measuring kerosene fluorescence may yield information on fuel vaporisation and mixing in kerosene/air sprays (upstream from the flame front). Third, the four aromatics selected in this study are representative of kerosene not only in terms of fluorescence, but also from a thermodynamic point of view. Indeed, their boiling points cover the whole distillation curve of kerosene (from 440 to 540 K), and therefore, these compounds are representative of kerosene evaporation (hence fuel vapour concentration). Finally, different possibilities for laser excitation scheme of kerosene fluorescence can be derived from the previous spectroscopic measurements depending on the desired application. This will now be discussed.

3.2.1 Excitation at 266 nm for temperature and equivalence ratio measurements

3.2.1.1 Rationale

A widely used optical method for temperature and concentration measurements [42–44, 47, 49] consists on using a single-excitation laser source and a dual-channel detection scheme (two intensified CCD cameras equipped with optical filters to detect fluorescence over two spectral domains). From the ratio of the two fluorescence images, it is usually possible to derive the temperature field. Once the temperature field is obtained, it can be retrofitted into the analysis of the fluorescence images in order to derive species concentration field, using the temperature dependence of fluorescence signals together with a calibration point in a reference cell where thermodynamic parameters (temperature, pressure, gas composition) are well known. Finally, equivalence ratio can be obtained from the simultaneously measured fuel and oxygen concentration fields [6]. However, while measurement of oxygen concentration is critical to determine equivalence ratio in dense automotive sprays [5, 41], it is of minor importance in dilute aeronautical sprays. Indeed, kerosene/air stoichiometric combustion occurs for kerosene molar fraction of only 0.0126, and therefore, the kerosene molar fraction in the flow region upstream from the flame front is always very low, which ensures that oxygen molar fraction remains of the order of 0.21 at that location. Additionally, Fig. 9b shows that the integrated fluorescence of kerosene (and also that of mono- and di-aromatics) displays low dependence upon oxygen molar fraction for 0.16 ≤ $X_{{{\text{O}}_{ 2} }}$ ≤ 0.21. Therefore, in flows where $X_{{{\text{O}}_{ 2} }}$ is in that range, measurements of kerosene concentration directly yield equivalence ratio without the need to measure simultaneously oxygen concentration (which is assumed to be 0.21), if the temperature field is known. The error on the measured equivalence ratio resulting from the unknown exact value of $X_{{{\text{O}}_{ 2} }}$ remains negligible, and the evolution of kerosene fluorescence with temperature under air atmosphere (i.e. $X_{{{\text{O}}_{ 2} }}$ = 0.21) is enough to analyse kerosene fluorescence images and derive equivalence ratio.

3.2.1.2 Selected measurement scheme

Spectroscopic results showed that with excitation at 266 nm, kerosene fluorescence spectrum exhibits two bands which have different behaviour with temperature and oxygen molar fraction. The most obvious choice for the two spectral domains considered for temperature measurements is to detect fluorescence from each band on a separate camera. Figure 12 represents the evolution with temperature of kerosene fluorescence and of the ratio between the fluorescence intensity from di-aromatics (from 310 to 420 nm) and that from mono-aromatics (from 270 to 310 nm), at 0.1 MPa under air atmosphere. Data were obtained in a heated kerosene vapour/air jet (with known composition, see ref [43] ) using the first K-PLIF set-up described in the apparatus section. In Fig. 12a, intensities are normalised by the maximum value of kerosene integrated fluorescence at 450 K, whereas in Fig. 12b, the ratio is normalised by its value at 450 K. The fluorescence intensity ratio (di-aromatics/mono-aromatics) exhibits strong dependence upon temperature in the range 450–750 K (this corresponds to typical air inlet temperatures of aeronautical gas turbines), which suggests that it is suitable to measure fuel vapour temperature in kerosene/air flows. Once temperature is measured, it is used to analyse mono- and di-aromatics fluorescence images and derive equivalence ratio, using a calibration point in a reference cell. Equivalence ratio is determined on each camera (with a difference <2 %), and the resulting equivalence ratio corresponds to the mean value of both cameras. The uncertainty of the technique was determined in the reference cell by performing statistical analysis. For single-shot images, it is estimated to be about 40 K for temperature and 0.05 for equivalence ratio, at temperatures above 550 K and atmospheric pressure.

Fig. 12

Temperature dependence of a integrated fluorescence of kerosene and b the ratio between the integrated fluorescence from di-aromatics and that from mono-aromatics, at 0.1 MPa, in air ($X_{{{\text{O}}_{ 2} }}$ = 0.21), with 266 nm excitation wavelength

3.2.1.3 Illustrative result

Example of mean temperature and equivalence ratio measurements (average over 100 single-shot images) using the method described above is shown in Fig. 13. The colorscale ranges from black to red, where black colour corresponds to temperature below 521 K in Fig. 13a and equivalence ratio equal to zero in Fig. 13b. Experiments were performed on a helicopter lean premixed prevaporised (LPP) swirled injector (schematically represented on the left-hand side of the images) operating with jet A1 kerosene fuel, with air inlet temperature of 550 K, at atmospheric pressure and overall equivalence ratio of 0.1. K-PLIF measurements were carried out at evaporation conditions (i.e. no combustion), and the fuel spray was fully vaporised at the outlet of the premixing duct. Figure 13 shows that fuel remains located close to the LPP injector axis: the initial radial extension of the kerosene/air jet does not exceed half a diameter of the premixing duct. Temperature field of the kerosene/air jet is homogeneous along its longitudinal axis: the mean temperature is equal to 555 K with a standard deviation of 6 K. Local equivalence ratio is homogeneously distributed up to an axial distance equal to 5/4 of the premixing duct diameter: the mean equivalence ratio is equal to 0.3 with a standard deviation of 0.04. At larger axial distance, equivalence ratio decreases due to the radial expansion of the jet, which spreads fuel over a larger area. Equivalence ratio at the outlet of the premixing duct is larger (i.e. ~0.4) than overall equivalence ratio, due to confinement of the kerosene/air jet close to the injector axis. However, the equivalence ratio of the jet weighted by the ratio (jet section/injector section) is similar to the overall equivalence ratio.

Fig. 13

a Mean temperature field (in K) and b equivalence ratio field in a helicopter injector operating with air inlet temperature of 550 K, at atmospheric pressure, using 266 nm laser excitation

3.2.1.4 Additional prospects

In methane/air flames, Vermuth and Sick [58] demonstrated the use of a tunable Nd:YAG laser to probe simultaneously acetone and OH radical at 266.188 nm (i.e. the P₁(10) spectral line in the (2,0) band of the OH (X²Π–A²Σ⁺) system). Together with our results, this indicates that excitation at this particular wavelength can potentially be used to probe both kerosene and OH radical at the same time. Using this excitation scheme, OH fluorescence signals can be collected around 290 or 310 nm. The first possibility should be avoided because there may be large crosstalk with the fluorescence from mono-aromatics, whereas the detection scheme around 310 nm is located in a spectral region where kerosene displays a local minimum and is, therefore, more suitable. Although, we do not know accurately the wavelength of the Nd:YAG laser used in the first K-PLIF set-up, it is believed that results in Fig. 12 should be marginally affected if the laser was tuned to 266.188 nm. Consequently, using a single laser excitation at 266.188 nm may potentially yield simultaneous measurements of equivalence ratio (with the data processing described above) and flame front structure by means of a three-camera detection system (the two cameras from the first K-PLIF set-up and the OH camera from the second K-PLIF set-up).

3.2.2 Excitation at 282 nm for probing multiple species

Excitation at ~282 nm with a tunable dye laser can be used to simultaneously measure kerosene spatial distribution at the outlet of an injector (by probing aromatics present in the fuel spray) and the flame front structure by recording fluorescence from OH radical (formed in the flame). Example of simultaneous detection of kerosene fuel and OH radical, using the second K-PLIF set-up described in the apparatus section, is shown in Fig. 14. Fluorescence images from kerosene and OH radical are superimposed to better visualise the spatial correlation between the fuel and the flame. In the kerosene fluorescence image, the colorscale ranges from black to red. The OH radical fluorescence image is represented on a greyscale, which ranges from black to white. In both images, the black colour corresponds to a fluorescence signal equal to zero and the respective colorscales are displayed on the right-hand side of the figure. Experiments were performed on a fired aircraft swirled injector (schematically represented in red colour on the left-hand side of the figure) operating with jet A1 kerosene fuel, with air inlet temperature of 500 K, at atmospheric pressure and overall equivalence ratio of 0.76. Please note that only half of the hollow-cone kerosene spray and half of the flame front are visualised in Fig. 14. A good spatial correlation is observed between the distribution of kerosene vapour and the location of OH radical: the flame front occurs at the position where kerosene vapour disappears. Unlike ketones (e.g. acetone), which pyrolyse significantly upstream from the flame front [58, 59], aromatics are consumed within the flame front and no “dark zone” is visible between the fuel and the flame front. Additionally, small-scale flow structures can be observed at the same position in each image (a zoom of a region of interest is displayed on the right-hand side of Fig. 14 to illustrate this). Finally, while with this injector the droplet spray is located on the outer side of the flame front, large droplets may also sometimes cross the flame front and be transported further downstream by the flow (as indicated by the red arrows).

Fig. 14

Simultaneous spatial distribution of kerosene fuel and OH radical in an aircraft injector operating with air inlet temperature of 500 K, at atmospheric pressure, using 282 nm laser excitation. The colorscale (kerosene fuel) and the greyscale (OH radical) are represented in terms of counts (ICCD camera dynamic range 0–65,535 counts)

In a recent article, Mosburger and Sick [60] suggested the use of laser excitation at about 283 nm to simultaneously probe OH radical and CO molecule in flames (and possibly fluorescent tracers like ketones). Together with our findings, this suggests that excitation at ~283 nm may have the potential to probe, at the same time, kerosene, OH radical and CO using a single laser excitation scheme and a three-camera detection system. Such measurements would be very useful for gas turbine manufacturers and also for the validation of numerical models in reactive fluid dynamics simulations, because they could provide simultaneous information on fuel spatial distribution (e.g. equivalence ratio), local flame front structure (e.g. heat release) and local pollutant emissions (e.g. CO formation) in kerosene/air combustors.

3.2.3 Excitation at 308 nm

Excitation at 308 nm can be used to probe di-aromatics naturally present in kerosene fuel, which can yield information (spatial distribution, fuel evaporation, etc.) on medium-to-low volatility compounds of the fuel. Additionally, using a single-excitation scheme at 308 nm and a dual-channel detection scheme (to collect kerosene fluorescence over two spectral domains), it may be possible to measure the temperature field from the ratio of the two fluorescence images. However, due to the low sensitivity of di-aromatics fluorescence to temperature [47], the accuracy of temperature single-shot measurements using this method may be limited (i.e. it is estimated to be about 50–70 K for temperature below 700 K).

In principle, excitation at 308 nm has also the potential for simultaneous measurements of kerosene fuel and OH radical (by probing spectral lines in the (0,0) band of the OH (X²Π–A²Σ⁺) system). Nonetheless, with this excitation scheme, OH fluorescence intensity is significantly lower compared to excitation at 282 nm, and it is usually detected around 347 nm where kerosene fluorescence intensity is high. This is likely to yield large crosstalk between fluorescence signals from kerosene and OH radical. Therefore, simultaneous detection of those species using laser excitation at 308 nm may be difficult.

3.2.4 Excitation at 248 nm

With excitation at 248 nm, both mono- and di-aromatics are probed and kerosene fluorescence spectrum exhibits two bands. Therefore, considering an optical set-up and data processing similar to that used with excitation at 266 nm, it might be possible to measure equivalence ratio and temperature. It is yet necessary to first perform detailed spectroscopic fluorescence measurements under various conditions of temperature, pressure and oxygen molar fractions (such as those presented with excitation at 266 nm in this article), in order to estimate the capabilities of 248 nm excitation scheme for future measurements in gas turbines.

Potentially, excitation at 248 nm may also provide simultaneous information on kerosene fuel and OH radical (by probing transitions in the (3,0) band of the OH (X²Π–A²Σ⁺) system). Nonetheless, absorption of OH radical at 248 nm is much reduced compared to excitation at 282 nm and, therefore, requires high-energy lasers. Additionally, with this excitation scheme, fluorescence from OH radical is classically collected around 296 nm where the fluorescence from mono-aromatics is maximum. This may lead to important crosstalk between fluorescence signals from the fuel and OH radical and make the simultaneous detection of those species quite challenging (using excitation at 248 nm).

4 Concluding remarks

Photophysics of Jet A1 vapour has been investigated for temperatures between 450 and 900 K, pressures between 0.1 and 3.0 MPa, oxygen molar fraction between 0 and 21 %, with four laser excitation schemes (248, 266, 282 and 308 nm). Using a 248, 266 and 282 nm excitation wavelength, kerosene fluorescence spectrum exhibits two bands which come from the fluorescence of aromatics naturally present in the fuel. By contrast, fluorescence spectrum obtained with excitation at 308 nm contains less information, as only one fluorescence band can be detected. Identification of the main species responsible for this fluorescence was achieved by comparison with the reference spectra of different preselected single-ring and two-ring aromatics. The first fluorescence band can be attributed to mono-aromatics, while di-aromatics are responsible for the second fluorescence band. Intensive spectroscopic measurements were carried out with excitation at 266 nm, results showed that both fluorescence bands decrease as temperature, pressure and oxygen molar fraction increase. However, the first band is more sensitive to temperature, whereas oxygen molar fraction has stronger influence on the second band. Pressure effect on both bands is limited. The large dataset of spectroscopic results obtained in this study allows to propose different strategies for K-PLIF measurements in reactive and non-reactive kerosene/air flows, which may yield multi-species probing, temperature and equivalence ratio fields.