Study of soot formation during the combustion of Diesel, rapeseed methyl ester and their surrogates in turbulent spray flames

Fuel 107 (2013) 147–161

Contents lists available at SciVerse ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Study of soot formation during the combustion of Diesel, rapeseed methyl ester and their surrogates in turbulent spray flames R. Lemaire a,b,⇑, S. Bejaoui a,c, E. Therssen a,c a

Université Lille Nord de France, F-59000 Lille, France Mines Douai, EI, F-59500 Douai, France c Laboratoire PC2A, UMR CNRS 8522, F-59655 Villeneuve d’Ascq, France b

h i g h l i g h t s " The sooting behaviour of rapeseed methyl ester (RME) has been studied. " The addition of RME to Diesel reduces the quantities of soot and soot precursors. " Different RME surrogates (2 alkanes, 1 alkene and 1 ester) have been tested. " The effects involved in soot reductions have been identified and quantified. " Biodiesel soot are bigger than Diesel ones but oxidize faster.

a r t i c l e

i n f o

Article history: Received 20 March 2012 Received in revised form 18 December 2012 Accepted 18 December 2012 Available online 7 January 2013 Keywords: Diesel Ester Surrogate Soot TSI

a b s t r a c t Effects induced by the use of rapeseed methyl ester (RME) as additive or Diesel substitute on the soot formation process have been studied in turbulent spray flames. Investigations have been carried out by coupling Laser-Induced Incandescence and Fluorescence (LII/LIF) at 1064, 532 and 266 nm. LII and LIF profiles obtained with fuels containing various amounts of ester (from 10 to 100 vol.%) showed that the addition of RME to a European low-sulphur Diesel or to a Diesel surrogate (a n-decane/1-methylnaphthalene blend derived from the ‘IDEA’ fuel) induces significant reductions of the quantities of soot and soot precursors (including high-number ring aromatic species and light soot precursors). The study of different RME surrogates (n-decane, n-hexadecane, 1-octadecene and methyl oleate) also revealed that the details of the oxidation of biodiesels could be mimicked only using large methyl esters as surrogates. N-alkanes and n-alkenes were found to be unable to reproduce the soot formation process occurring during the combustion of large fatty acid methyl esters (FAME) such as those contained in RME. The analysis of the correlation existing between the threshold soot index (TSI) and the peak soot volume fraction measured in flames of m-IDEA/RME blends containing up to 80% of ester allowed the identification of the different effects involved in the soot reduction (i.e. the dilution and the ester functional group effects). The estimation of their relative contribution has also been investigated. Finally, the LII fluence curves and time decays obtained at different heights above the burner in flames burning Diesel and a Diesel/RME mixture have been compared. By this way, it has been demonstrated that biodiesel soot are bigger than Diesel ones in the soot formation region. On the other hand, particles oxidize much faster when RME is added to Diesel. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction The growing use of oxygenated biofuels appears to be an interesting means to reduce the energetic dependence on petroleum as a fuel source. In addition, biofuels also contribute to reduce net ⇑ Corresponding author. Address: Ecole des Mines de Douai, Département Energétique Industrielle, 941 Rue Charles Bourseul, CS 10838, 59508 Douai Cedex, France. Tel.: +33 (0)3 27 71 21 29; fax: +33 (0)3 27 71 29 15. E-mail address: [email protected] (R. Lemaire). 0016-2361/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.fuel.2012.12.072

emissions of greenhouse gases since these alternative fuels are derived from renewable sources [1,2]. Among the wide variety of alternative fuels currently studied, fatty acid methyl esters (FAME) also called biodiesels have received a particular attention since these monoalkyl esters of animal or vegetable derived long chain fatty acids are quite similar to conventional Diesel in its main characteristics [2,3]. FAME can thus be used pure and more generally in blends with diesel fuel in Compression Ignition engine with no major modifications in the engine hardware or performance [2–4]. Moreover, positive effects are globally associated to the addition

148

R. Lemaire et al. / Fuel 107 (2013) 147–161

of FAME to conventional Diesel in terms of pollutant emissions. This includes lower emissions of carbon monoxide (CO), unburned hydrocarbons and particulate matter (PM). Increases of NOx concentrations in exhaust gases are generally reported on the other hand [1–5]. Many studies have thus been carried out during the past two decades to analyse and better understand the effects associated to the use of FAME on Diesel engines performance and emissions. Among the wide variety of works conducted in this field, the following literature survey will focus on studies that deal with the impact of methyl esters on the formation of regulated pollutants. It will especially focus on the formation of PM and hence soot which are generally the main constituents of Diesel exhaust particulates. 1.1. Literature survey In a 1996 paper, Schmidt and Van Gerpen reported significant reductions of PM and hydrocarbon emissions at the exhaust of a Diesel engine when adding up to 50% of various pure esters to a standard diesel fuel [6]. Additional experiments were performed by adding octadecane (a long chain paraffin that does not contain oxygen) to Diesel in order to investigate effects of cetane number, long chain hydrocarbons and oxygen on emission levels. The obtained results lead to the conclusion that the particulate reducing effect of biodiesels was due to a combination of their oxygen content and to the displacement of aromatic species with long-chain esters (a phenomenon called ‘‘dilution effect’’ [7]). The same year, Krahl et al. summarized the published emissions measurements issued from different experimental studies carried out at the exhaust of Diesel engines fuelled by rapeseed methyl ester (RME) [8]. Significant reductions of the emissions of hydrocarbons ( 20% to 30%), CO ( 10% to 30%), PM ( 20% to 40%) and aromatic compounds (up to 40%) were thus reported. Reductions of PM ( 20% to 53%) and CO ( 2.4% to 48%) were also observed by McCormick et al. during a series of engine tests based on the use of various biodiesels and esters produced from pure fatty acids [9–11]. In 1997, Choi et al. lead experiments in a Direct Injection Diesel engine fuelled with blends composed of 80 vol.% Diesel and 20 vol.% ester or octadecene (a good representative of the soyate hydrocarbon composition with one double bound but without oxygen) [12]. By this way, the authors wanted to isolate the effect of oxygenates on the reduction of emissions such as PM. The obtained results showed that the 20% biodiesel blend reduced PM emissions an additional 36% versus the 20% octadecene blend. The authors thus concluded that the presence of oxygen in the ester was the main factor inducing the reduction of PM as opposed to the dilution of aromatics and sulphur in the baseline diesel fuel. This result contrasts somewhat with the conclusions from Schmidt and Van Gerpen who adopted a similar experimental strategy using an alkane as additive as seen before [6]. In 2002, the US Environmental Protection Agency (EPA) published a report including results issued from 39 experimental studies conducted with heavy-duty engines [13]. The processing of all the data presented in this reference report lead to correlate the concentration of biodiesel added in conventional diesel with changes in regulated pollutants. The obtained results revealed that the use of biodiesel generates significant reductions of PM (up to 47% when using pure biodiesel), hydrocarbons (up to 67%) and CO (up to 48%) while slightly increasing NOx emissions (up to 10%). Many works have then confirmed these trends with reduction of pollutant emissions sometimes far more important [14–17]. Knothe et al. especially compared the regulated exhaust emissions issued from a Diesel engine fuelled by three fatty acid methyl esters (methyl laurate, methyl palmitate and methyl oleate), biodiesel, low sulphur petrodiesel and alkanes (dodecane and hexadecane) [17]. They concluded that the use of a commercial biodiesel as well as pure fatty acid methyl esters sig-

nificantly reduces PM emissions (from 75% to 83%) compared to petrodiesel. The tested alkanes achieved lower PM reductions (from 45% to 50%) suggesting that the methyl ester moiety influences PM emissions considerably more than neat straight-chain hydrocarbons. Szybist et al. then conducted motored engine experiments and noted that significant amounts of CO2 were produced during the low-temperature heat release part of the ignition process when using an ester (the methyl decanoate) instead of Diesel. The authors argued that this CO2 production was the result of the decarboxylation of the ester group and not the product of an oxidation process [18]. This phenomenon also pointed out by Westbrook et al. in a previous modelling work is of interest as it eliminates C atoms from the product pool of species that can subsequently make soot in fuel-rich Diesel ignition zone [19]. On the other hand, Zhang et al. showed that this early CO2 production could be significantly influenced by the presence and the position of double bounds in the aliphatic chain of fatty acid esters as observed in the case of various C9 FAME [20]. In an extensive literature survey devoted to the analysis of the effect of biodiesels on Diesel engine emissions, Lapuerta et al. noted that a global consensus was found on the main trends presented in the EPA report even if a wide disparity of results was observed due to the number of different engine technologies used or to the operating conditions varying from test to test [21]. Based on the analysis of numerous experimental works, the authors concluded that the addition of FAME to Diesel induces significant reductions of PM emissions with a global decrease of the size of soot particles. Presence of oxygen as well as absence of aromatic compounds in biodiesels have been pointed out as being the main reasons explaining this trend. The authors explained that the oxygen contained in the ester molecule enables a more complete combustion even in regions of the combustion chamber with fuel-rich diffusion flames. This explains the observed reductions of hydrocarbons, carbon monoxide and PM emissions. Lapuerta et al. also argued that the oxygen content of biodiesels promotes the oxidation of soot. This process that is well correlated with the local flame temperature [22] would be influenced by modifications in the structure of biodiesel soot as observed in various works from Boehman et al. [23,24,26] and Jung et al. [25]. Finally, the authors noted that the concentrations of aromatic compounds such as Polycyclic Aromatic Hydrocarbons (PAHs) in exhaust gases were generally lower during the combustion of biodiesel. This observation is of interest since these species are directly involved in the soot formation process. Since then, many recent engine-based experimental works have confirmed these trends especially in the case of RME [27–29]. Nevertheless, there are a multitude of factors that contribute to a wide variability in the obtained results during engine studies [21]. Moreover, experiments conducted in engines do not allow a direct analysis of pollutants formation mechanisms and kinetics as pointed out in a recent paper from Maricq [30]. Well defined and reproducible laboratory scale experiments have thus been conducted to better understand fundamental effects induced by the use of esters on the formation mechanisms of pollutants such as soot. In a computational study from 2000, Fisher et al. developed detailed chemical kinetic models for the combustion of two small methyl esters (methyl butanoate and methyl formate). These models have been tested against data obtained in static vessels. They were found to overestimate the overall reactivity of the two esters prompting demand for more complete and well-characterized experimental databases to test the developed kinetic codes thoroughly [31]. In 2007, Dagaut et al. proposed for the first time an experimental study of the oxidation behaviour of RME in jet-stirred reactor (JSR) [32]. The oxidation of the long saturated alkyl chains present in RME was found to be very similar to the oxidation of large alkanes. The JSR results obtained with RME were thus compared with the predictions issued from a kinetic scheme

R. Lemaire et al. / Fuel 107 (2013) 147–161

developed for the oxidation of n-hexadecane. The kinetic modelling gave a good description of the experimental data suggesting that n-hexadecane could be a convenient surrogate for modelling the combustion of RME. Nevertheless, the authors pointed out that the early formation of CO2 was not reproduced by the model since this phenomena is due to the decarboxylation of the ester function as reported in the engine studies cited before [18–20]. The same year, Sarathy et al. showed that the oxidation of an unsaturated C4 methyl ester (the methyl crotanoate) generates much more soot precursors (such as benzene) than the oxidation of a saturated one (the methyl butanoate) [33]. These authors then studied the oxidation of another unsaturated ester (the methyl (E)-2-butenoate) to analyse this trend [34] that has been observed since then during engine tests [35]. In parallel, Herbinet et al. developed a detailed chemical kinetic mechanism for the oxidation of methyl decanoate (a small saturated methyl ester used as a biodiesel simplified model-fuel) [36]. The computed results were then compared with those obtained in engine and JSR with methyl decanoate and RME, respectively. The developed mechanism was found to reproduce well the early CO2 production associated to the combustion of biodiesels. The authors thus concluded that kinetic details of the oxidation of biodiesels (including the early CO2 formation) could be predicted only with mechanisms developed using methyl esters as surrogates while large n-alkanes could be good model fuels to predict the overall reactivity of FAME of similar sizes. Later, Ramirez et al. compared the oxidation of a commercial Diesel/RME blend (70/30 vol.%) with a mixture of n-decane, 1-methylnaphtalene and methyl octanoate [37]. Methyl octanoate was used as a RME model fuel while the n-decane and 1-methylnaphtalene blend corresponds to the ‘IDEA’ fuel [38]. This is a well known Diesel surrogate that we previously studied in turbulent spray flames [39]. The experimental and modelling results obtained by Ramirez et al. gave reasonable agreements in terms of concentration profiles of reactants, stable intermediates and products. Nevertheless, the need of extensive experimental databases was pointed out (as in a recent review from Lai et al. [40]) especially for the oxidation of large methyl esters which are more representatives of commercial biodiesels. Various JSR and computational studies have been performed lately with a view to find more convenient biodiesel surrogates for soy or rapeseed methyl ester (SME/RME). This includes the study of large molecules including methyl palmitate [41,42], methyl stearate [42,44], methyl oleate [42–44], methyl linoleate and methyl linolenate [42] that are the five major components of SME and RME. Very few fundamental works really focused on the effect of FAME addition in Diesel on soot formation, however. The major part of the works conducted in this field has been carried out in engines as seen in the first part of this literature review. Investigations performed in flames then offer attractive complements since they focus on combustion generated soot that are the main constituents of Diesel exhaust PM. In a paper from Pespiot-Desjardins et al., a structural group contribution approach has been proposed to interpret experimental observations concerning the effect of various oxygenated additives on the sooting propensity of hydrocarbon fuels [7]. To do so, the authors used threshold sooting index values (TSI [45]) derived from an extensive set of smoke point measurements that were carried out for mixtures of a given fuel with several oxygenated molecules (alcohol, ether, ester, ketone and aldehyde). It was found that oxygenated groups reduced the sooting tendency of the base fuel in which they were added with an efficiency which was strongly dependent of their nature and molecular structure (a phenomenon also pointed out by Westbrook et al. [19]). For example, FAME were found to be less efficient than alcohols to reduce the formation of soot particles since two O atoms are bounded to a single C atom in esters. Consequently, when these CO2 moieties lead to the formation of CO2 molecules, the oxygen initially present in

149

the oxygenated additive is less efficient in eliminating C atoms from the product pool of species that can subsequently make soot. The analysis conducted by Pespiot-Desjardins et al. also lead to the identification and quantification of two different effects induced by oxygenated compounds: the chemical effect due to the presence of oxygen atoms in the additives and the dilution effect that corresponds to the replacement of highly sooting components of the base fuel by cleaner hydrocarbons or inversely to the replacement of clean-burning fuels by more complex sooting molecules. Such an analysis showed that when small FAME like methyl butanoate are used as additives in low concentration (less than 20 vol.%), the dilution effect is the main factor inducing a decrease in the sooting tendency of the fuel compared to the effect of the oxygen content. More recently, Maricq compared the soot formation in diffusion flames of Diesel, SME, hexadecane and methyl butanoate [30]. He concluded that the observed soot reductions were due to both the absence of aromatic species (dilution effect) and to the presence of oxygen (chemical effect) in esters without quantifying the relative importance of each effect. He also indicated that hexadecane was a better surrogate to mimic soot size distribution and composition of SME than methyl butanoate due to the use of a small methyl ester that is not representative of the large alkyl esters (mainly methyl linoleate and methyl oleate) met in SME. Further experimental works using larger methyl esters are thus necessary. The use of realistic spray flames can moreover be of interest to better reproduce and understand the combustion and emission characteristics of automotive fuels like biodiesels. Such a strategy has been recommended especially by Wang et al. in a recent paper devoted to the analysis of the effect of injection pressure on the formation and combustion of biodiesel sprays [46]. 1.2. Experimental strategy adopted in this work We recently studied the effect of ethanol addition in gasoline in terms of soot propensity [47]. To do so, we used a hybrid spray burner that allows the ignition of high-speed sprays of micrometric fuels droplets (D32 < 10 lm) whose characteristics are similar for a wide range of fuel. This burner has been characterized in detail in [48]. The turbulent diffusion flames obtained with this system have identical dimensions (18-cm height and 2-cm width) and comparable sooting flame patterns. This characteristic eases the analysis of the impact of the fuel composition on the formation of pollutants. Consequently, this burner has been used here to study the formation of soot particles in flames burning a European low-sulphur Diesel, RME and a Diesel/RME mixture (B30) containing 30 vol.% of ester (this type of blend being commonly used in engines [14]). In a previous work related to the study of ethanol [47], we used a two-excitation wavelength Laser Induced Incandescence/Laser Induced Fluorescence technique (LII/LIF at 1064 and 532 nm) to obtain spatially resolved measurements of soot volume fraction and soot precursors profiles. In the present work, we applied for the first time a recently developed three-excitation wavelength LII/LIF method by coupling excitation wavelengths of 266, 532 and 1064 nm [49]. We thus obtained mappings and profiles of soot (LII at 1064 nm), light soot precursors (LIF at 266 nm) and high-number ring aromatic compounds (LIF at 532 nm) since a shift of the absorption and emission spectra of Polycyclic Aromatic Hydrocarbons (PAHs) towards longer wavelengths is generally observed for aromatic compounds of increasing molecular weight [50–52]. Indeed, Vander Wal et al. already showed that the maximum fluorescence emission around 400 nm corresponds to species composed of 2 or 3 aromatic cycles while the fluorescence between 500 and 600 nm corresponds rather to five rings and more [51]. More recently, Wu et al. [52] attributed fluorescence between 320 and 380 nm to 1- or 2-rings aromatic species while fluorescence around 450 nm has been related to three cycles or more.

150

R. Lemaire et al. / Fuel 107 (2013) 147–161

Consequently, using excitation wavelengths in the UV (266 nm) and visible (532 nm) ranges is of interest to follow the evolution of different classes of PAHs that are closely related to the formation of soot particles. Conventional diesel fuels are composed of several hundred species [53]. Due to the complexity of such commercial blends, the use of surrogates containing a relatively small number of pure hydrocarbons is of interest. Indeed, these simplified mixtures are needed for the development of computational codes. Furthermore, they are very useful to perform well controlled and reproducible experiments. We thus decided to use a Diesel model fuel as done with gasoline in [47] to better analyse the relative influence of dilution and chemical effects on the soot formation process when adding RME. In a previous work devoted to the study of Diesel and Diesel surrogates [39], we used the ‘IDEA’ fuel and we adapted its formulation to match the sooting propensity of a European low-sulphur Diesel. The obtained modified ‘IDEA’ fuel (called m-IDEA in the following) consists of a mixture of 80% n-decane and 20% 1-methylnaphthalene (in volume). Following Ramirez et al. [37] or Feng et al. [54] who used the ‘IDEA’ fuel as a Diesel surrogate, the mIDEA blend has been used here in mixture with RME. The above literature review showed that many authors studied the oxidation behaviour of different alkanes and esters in order to find convenient surrogates for FAME. Nevertheless, the ability of these surrogates to reproduce the sooting propensity of their parent-fuels has only been seldom checked. This is the reason why we also tested different surrogates of RME. This includes n-hexadecane, a simple alkane that has been previously studied by Knothe et al. [17] and considered as a potential RME model fuel by Dagaut et al. [32]. We also tested n-decane since this hydrocarbon is the main constituent of the m-IDEA fuel. Moreover, the sooting propensity of this fuel has already been studied in previous works [39,55] and its oxidation behaviour is well known ([56] and references therein) which is of interest for the modeller. We also studied the combustion of methyl oleate (the major component of RME) while 1-octadecene (an alkene which is representative of the methyl oleate hydrocarbon composition with one double bound but without oxygen) has been used to analyse the impact of the oxygen moiety of the ester on the soot formation. Based on this new set of experimental data, we examined the correlation existing between the threshold soot index of each mixture and the soot volume fraction measured in each flame as in [47]. Finally, LII fluence curves and time decays have been recorded at different heights above the burner in flames of Diesel and B30. The evolution of the LII signals is correlated to some physical properties of soot such as their diameter. This kind of analyses is thus relevant to obtain additional information regarding the influence of the oxygen content of the fuel on the oxidation behaviour of soot particles for example.

bulent diffusion flames have similar dimensions, comparable sooting flame patterns and behave similarly to gaseous diffusion flames beyond the vaporisation stage. This allows focusing analyses on chemical effects only. The European low-sulphur Diesel used in this work (a commercial fuel provided by TOTAL) has been analysed by gas-chromatography/mass-spectrometry. It was found to contain more than 400 compounds including 23% of n-paraffin, 25% of branched paraffin, 7% of cycloparaffin, 17% of aromatics, 8% of esters (mainly methyl oleate and methyl linoleate which are the main components of RME) and 20% of unknowns (in volume). The analysis of the rapeseed methyl ester (a biodiesel provided by Diester Industrie) revealed that it contains 60% of methyl oleate, 25% of methyl linoleate, 8% of methyl linolenate, 5% of methyl palmitate and 2% of methyl stearate. This composition is consistent with that of a standard RME [57]. The physical properties of these two commercial fuels are detailed in [48]. Additional experiments have been carried out with a B30 blend that corresponds to a mixture of 70% Diesel and 30% RME. In a previous work devoted to the study of Diesel surrogates [39], a simplified model fuel (m-IDEA) derived from the ‘IDEA’ fuel was formulated [38]. It was composed of 80% n-decane and 20% 1-methylnaphthalene (in volume). In the present work, we also studied the combustion of mixtures containing 70% of m-IDEA fuel and 30% of additives (RME, n-hexadecane, n-decane, methyl oleate and 1-octadecene). Considering the composition of the m-IDEA blend, the different fuels tested here thus correspond to mixtures of 56% n-decane, 14% 1-methylnaphthalene and 30% additives. A list of the fuels studied in this work together with their TSI (when existing) is summarized in Table 1. These threshold soot index values will be used in Section 3.3. 2.2. LII/LIF procedure and setup The principle of the three-excitation wavelength LII/LIF technique has been presented in detail in [49]. It is based on the fact that when a region of a flame is irradiated with a 1064 nm laser excitation, the measured radiation emission is only assignable to soot. On the other hand, emissions induced by 266 or 532 nm irradiations correspond to the superposition of both LII signal from soot and LIF signals from fluorescent species assigned to light soot precursors (LIF at 266 nm) and high-number-ring aromatic compounds (LIF at 532 nm) [49–52]. It has been demonstrated in [49] that a perfect coincidence of the temporal and spatial LII signals could be achieved at 266, 532 and 1064 nm in a flame region exempt of fluorescent species by using the same laser irradiance

Table 1 List of the fuels studied in this work with their corresponding threshold soot index. Fuels n-Paraffin (23%) Branched paraffin (25%) Cycloparaffin (7%) Aromatics (17%) Esters (8%) Unknowns (20%)

28 [39]

m-IDEA

n-Decane (80%) 1-Methylnaphtalene (20%)

27.6 [39]

RME

Methyl Methyl Methyl Methyl Methyl

/

Additives

n-Decane n-Hexadecane Methyl oleate 1-Octadecene

2. Experimental set-up and procedure 2.1. Burner and fuels A McKenna hybrid burner has been used to standardize the different flames. This burner characterized in detail in [48] is composed of a 60-mm diameter bronze porous plate. A DIHEN (model 170 AA) is inserted into a central 6.35 mm diameter tube for the atomisation of liquid fuels. A lean premixed methane–air flat flame (equivalence ratio = 0.8) stabilized on the porous is used to ignite the fuel jets at the injector outlet. The nebulisation gas is nitrogen (23.8 g/h) and liquid fuels are introduced with a constant mass flow rate of 46 g/h. Thanks to this system, sprays generated by the DIHEN are completely vaporized 15 mm height above the burner (HAB) whatever the fuel used. As a result, the obtained tur-

TSI

European low-sulfur Diesel

oleate (60%) linoleate (25%) linolenate (8%) palmitate (5%) stearate (2%)

4.3 [45] 7.1 [58] / 9.2 [45]

R. Lemaire et al. / Fuel 107 (2013) 147–161

spatial distribution and by adjusting laser energies for the different wavelengths. It is thus possible to deduce the radial LIF profiles at 266 and 532 nm in a zone rich in soot precursors by subtracting the signal obtained at 1064 nm from those obtained at 266 and 532 nm. The absolute calibration of the soot volume fraction is not crucial here as the aim of this work is to study the relative influence of the addition of ester to Diesel on soot propensity. Consequently, the LII measurements have been calibrated as in previous studies [39,47,55] using the correspondence between the intensity of the LII signals obtained in a known laminar diffusion flame of methane and the soot volume fraction measured in the same flame by Cavity Ring-Down Spectroscopy [59]. The use of such a proportionality rule implies that the refractive index function of soot particles (E(m)) issued from the combustion of methane is similar to that of soot generated during the oxidation of the different liquid fuels studied here. It has been demonstrated recently that there was no significant difference between the optical properties of soot formed in flames of Diesel and B30 [49]. This similarity might not be valid in the case of methane soot (additional works being in progress on that issue [60]). Nevertheless, the assumption made above is not critical since the relative comparison between the data obtained in flames of Diesel and Diesel/RME mixtures should stay valid. This point will be discussed in Section 4. Furthermore, the use of such a methodology has already been discussed and validated in the soot formation region of various flames [39,47,55]. A scheme and a complete description of the experimental setup used for the LII/LIF measurements can be found in [49]. It consists of a Brilliant Nd:YAG laser operating at 1064 nm equipped with frequency doubling and quadrupling crystals to generate laser excitations at 532 and 266 nm. A 1-mm diaphragm enables the selection of the central part of the unfocused near-Gaussian laser beams and the diaphragm position is adapted so that the spatial distribution of the laser beams in the flame is the same at 266, 532 and 1064 nm (section at 1/e2 = 0.002 cm2). According to the method used in [49], the energy after the diaphragm (measured with a power-meter) has been fixed to 0.08, 0.22 and 0.44 mJ/pulse at 266, 532 and 1064 nm, respectively. The broadband LII/LIF signals have been collected at right angle on an ICCD camera with a gate width of 20 ns triggered by the peak of the laser pulse (prompt detection). 500 laser shots were accumulated and radiations at 266 and 532 nm have been rejected using a WG280 filter (cut-off below 280 nm) and two 532 nm dielectric mirrors. Finally, a head-on photomultiplier tube (Photonis XP2237) has been used for time-resolved measurements and LII fluence curves recordings. In this case, a thin volume of the flame was imaged on a 300-lm horizontal slit placed in front of the photomultiplier. Radiations below 550 nm were rejected using a Schott OG550 filter and the position of the diaphragm was adapted to obtain a Gaussian laser profile at 1064 nm with a section of 0.015 cm2 at 1/e2.

3. Effect of RME on sooting propensity 3.1. Study of soot formation in flames of Diesel, RME and B30 The mappings of fluorescent soot precursors and soot particles in flames of Diesel, RME and B30 are reported in Fig. 1. The emission patterns of the three flames recorded with a digital camera are also presented (see Fig. 1a). The spray flames stabilised with our hybrid burner are similar in shape and have identical heights and widths. This behaviour is consistent with that observed in previous works related to the study of various liquid fuels [39,47,55]. The Diesel and B30 flames are furthermore very close in terms of sooting pattern while the flame of RME exhibits a different structure relatively close to that of the n-decane spray flame studied

151

in Refs. [39,55]. These observations are clearly illustrated by Fig. 2 that displays the profiles of the peak LIF and LII signals as a function of the height above the burner. Fluorescent soot precursors are thus mainly present at low HAB in flames of Diesel and B30. The intensities of the LIF signals then increase to peak at around 34 mm HAB in the case of a 266 nm laser excitation (see Fig. 2a) against 62 mm HAB when using a 532 nm excitation wavelength (see Fig. 2b). After these critical heights, the intensities of the LIF signals decrease while the formation of soot progresses. A shift of the absorption and emission spectra of PAHs towards longer wavelengths is commonly observed for aromatic species of increasing molecular weight [50–52]. The evolution of the intensity of the LIF signals depicted here is therefore consistent with the soot formation process that implies the transition from light soot precursors (LIF at 266 nm) to high-number ring aromatic compounds (LIF at 532 nm) that subsequently leads to the formation of small soot particles (LII at 1064 nm) [61]. The soot volume fraction then reaches a peak at 92 mm (see Fig. 2c). Soot particles finally enter into an oxidation zone where their concentration decreases to finally become zero at around 160 mm HAB. Despite the abovementioned structural similarities, mappings of Fig. 1 and profiles of Fig. 2 already illustrate the effect of RME on soot propensity. Indeed, reductions of LII and LIF signals intensities of 34–36% are observed when adding 30% of RME to Diesel. The structure of the flame is even changed when using pure ester as said before. Fluorescent soot precursors are formed much later in this case. LIF signals obtained at 266 and 532 nm are much less intense and they pass through a maximum value at 46 mm and 76 mm HAB, respectively, instead of 44 and 62 mm in the case of Diesel and B30. As a consequence, soot particles are formed in very low quantity with a measurable peak at 24 ppb while the soot volume fraction obtained in flames of Diesel and B30 reaches values of 153 and 98 ppb, respectively (see Fig. 1d). One can note that intense signals are measured at low heights above the burner (<15 mm HAB) for all the tested fuels. These signals captured by the camera whatever the wavelength used come from the scattering of the laser beam through the fuels droplets. These signals completely disappear at around 15 mm HAB in each flame indicating a complete vaporization of the fuels as explained in [48]. The lifetime of these scattering signals (less than 10 ns) is rather short especially in comparison to the lifetimes of the LII and LIF signals which are of the order of 1 ls and 40 ns, respectively [39,49]. It is thus possible to avoid the interferences due to the scattering signals by triggering the gate width of the ICCD camera to start 15 ns after the beginning of the laser pulse. We recently used a similar temporal selection process to suppress the LIF contribution from LII signals recorded using laser excitations in the UV and visible ranges [49]. In the present case, measuring the 15 nsdelayed emissions from 266, 532 and 1064 nm excitations enables the detection of the LII and LIF signals free from the contribution of the laser scattering through the fuels droplets. Peak intensities of LII and LIF signals obtained using such a detection configuration are presented in Fig. 3. The intense signals that were present below 15 mm HAB on the profiles of Fig. 2 have completely disappeared whatever the wavelength used. The emissions collected for excitation wavelengths of 532 and 1064 nm are equal to zero at the nebuliser outlet (see the insets of Fig. 3b and c) while LIF signals at 266 nm are present immediately at the output of the fuel injector in flames of Diesel and B30 (see Fig. 3a). We attribute these signals to the fluorescence of 2-rings petrogenic PAHs [62] that are present in our low-sulphur Diesel (mainly methylnaphthalenes, dimethylnaphtalenes and trimethylnaphtalenes). Such aromatic species are indeed able to fluoresce upon UV laser excitations [50]. These LIF signals at 266 nm are absent at the injector outlet in the flame of pure ester since RME does not contain any aromatic compound. By contrast, the LIF signals measured after the fuels

152

R. Lemaire et al. / Fuel 107 (2013) 147–161

(b)

(c)

(d)

LIF at 266 nm (a.u.)

LIF at 532 nm (a.u.)

Soot volume fraction (ppb)

LIF at 266 nm (a.u.)

LIF at 532 nm (a.u.)

Soot volume fraction (ppb)

LIF at 266 nm (a.u.)

LIF at 532 nm (a.u.)

Soot volume fraction (ppb)

RME

B30: Diesel + RME (70/30)

Diesel

(a)

Fig. 1. 2D images of the flames emission (a) and reconstituted 2D-images of soot precursors LIF at 266 nm (b) soot precursors LIF at 532 nm (c) and soot LII at 1064 nm converted into absolute volume fraction (d) in flames of Diesel, B30 and RME.

injection zone are attributed to the presence of pyrogenic PAHs issued from the oxidation of the hydrocarbons contained in the fuels. This is especially the case of the LIF signals obtained with a 532 nm laser excitation since such signals are generally related to aromatic species containing five rings and more [51]. The absence of aromatic species in the ester explains why the formation of pyrogenic PAHs is delayed when using RME instead of Diesel or B30 that already contain a pool of aromatic species. The delayed measurements presented in Fig. 3 thus illustrate one of the main effects associated to the addition of RME on soot propensity which is the dilution of the aromatic part of the base fuel in which the ester

is added. Indeed, the main intensity of the LIF signal at 266 nm decreases of around 34% when adding 30% of ester to Diesel ( 31% at the injector outlet where signals are mainly attributed to the presence of petrogenic species) subsequently leading to equivalent reductions of PAHs LIF at 532 nm ( 36%) and soot volume fraction ( 36%). When using pure RME instead of Diesel, the reductions observed are of around 85% for LIF signals and 84% for LII signals. It is noticeable that reductions of PAHs and soot concentrations are very close in magnitude for a given level of ester content. We already observed a similar positive correlation between the intensity of the LIF signals and the quantity of soot produced in a given flame. This behaviour is depicted in Fig. 4 that summarizes measurements carried out in a set of 23 turbulent spray flames [39,47,55]. The results obtained with mixtures of m-IDEA and RME, methyl oleate, n-decane, n-hexadecane or 1-octadecene are also presented on this figure but will be treated in Section 3.2. The correlation existing between the peak intensity of the LIF signals at 532 nm and the soot volume fraction underlines the precursor role played by the fluorescent species we detect at this wavelength. Furthermore, the evolution of the intensity of LIF signals at 532 nm as a function of the LIF intensity at 266 nm (see Fig. 4b) as well as the evolution of the soot volume fraction as a function of the intensity of LIF signals at 266 nm (see Fig. 4c) also both follow proportionality rules in the three flames investigated in this work. The assignation of the LIF signals at 266 and 532 nm to light and high-number ring soot precursors thus seems coherent since the graphs of Fig. 4 clearly illustrate that the gaseous species which fluoresce in our flames upon UV and visible laser excitations are directly correlated to the quantity of soot produced. Additional works are necessary, however, especially to identify the nature of the aromatic species formed in the turbulent flames generated with our burner. Analyses of the emission spectra of PAHs that fluoresce with laser excitations of 532 nm are therefore in progress [63,64] while additional measurements at 266 nm will be carried out soon. By summing the LII signals in the vertical plane scanned by the laser beam (see Fig. 1d), the obtained signal intensity is directly proportional to the total soot volume fraction contained in this plane. Considering that the investigated flames are axisymetric and have identical soot patterns as explained before, the comparison of the summed LII signals allows a direct comparison between the total soot volume fractions contained in each flame. We thus integrated the quantity of soot produced in Diesel, B30 and RME flames and took into account the difference of calorific value between ester (40 MJ/kg) and Diesel (42 MJ/kg) [2]. We then obtained a 44% reduction of the total quantity of soot produced when adding 30% of RME to Diesel for a same quantity of heat released during the combustion. By the same way, we find that the use of pure RME reduces the total quantity of soot produced of 85%. This evolution of the soot production as a function of the ester content is represented in Fig. 5. It follows an exponential law similar to the correlation proposed by the EPA [13]. Nevertheless, the reductions observed in our case are significantly different from those reported by the EPA since our work focuses only on soot particles formed in flame conditions while the EPA results concern the total particulate matter (PM) emitted at the exhaust of engines fuelled with various biodiesels. The graph of Fig. 5 also includes PM reductions obtained at the exhaust of Diesel engines fuelled with methyl oleate (the main components of RME) [5,17] and pure RME [28,65]. The combustion conditions are in these cases significantly different from those investigated here. It is, however, of interest to note that the main trend observed in our flames is similar to the evolution observed at the exhaust of engines. Such a result might be related to the fact that soot particles are one of the dominant constituent of Diesel engine exhaust PM. Consequently, the study of combustion generated soot is relevant as it offers

153

R. Lemaire et al. / Fuel 107 (2013) 147–161

3,8E+07

9,0E+06 Diesel

B30

RME

3,0E+07 2,5E+07 2,1E+07 1,7E+07 1,3E+07 8,4E+06

(a)

4,2E+06

(a)

RME

6,0E+06

7,0E+06 6,0E+06

4,0E+06

5,0E+06

2,0E+06

4,0E+06 0,0E+00

3,0E+06

0 5 10 15 20 25

2,0E+06

0,0E+00 0

50

100

150

200

0

Height Above the Burner (mm) 2,0E+06 Diesel

B30

RME

1,8E+06

7,8E+06

1,6E+06

LIF intensity (a.u.)

8,9E+06

6,7E+06 5,6E+06 4,4E+06 3,3E+06 2,2E+06

(b)

1,1E+06

50

100

150

200

Height Above the Burner (mm)

1,0E+07

LIF intensity (a.u.)

B30

1,0E+06

0,0E+00

(b)

Diesel

B30

RME

6,0E+05 4,0E+05

1,3E+06 1,1E+06

2,0E+05

8,9E+05 0,0E+00

6,7E+05

0 5 10152025

4,4E+05 2,2E+05 0,0E+00

0,0E+00 0

50

100

150

0

200

7,5E+06

7,2E+06

Diesel

6,7E+06

B30

50

100

(c)

RME 6,4E+06

5,8E+06

Diesel

200

LII intensity (a.u.)

5,0E+06 4,2E+06 3,3E+06 2,5E+06 1,7E+06

(c)

B30

RME

6,0E+05

5,6E+06

8,3E+05

150

Height Above the Burner (mm)

Height Above the Burner (mm)

LII intensity (a.u.)

Diesel

8,0E+06

LIF intensity (a.u.)

LIF intensity (a.u.)

3,4E+07

4,0E+05

4,8E+06 4,0E+06

2,0E+05

3,2E+06 0,0E+00

2,4E+06

0 5 10152025

1,6E+06 8,0E+05

0,0E+00

0,0E+00

0

50

100

150

200

Height Above the Burner (mm) Fig. 2. Profiles of the peak intensity of soot precursors LIF at 266 nm (a), soot precursors LIF at 532 nm (b) and soot LII at 1064 nm (c) as a function of the height above the burner in flames of Diesel, RME and B30. (Prompt detection: the gate width of the CCD camera is triggered to start with the laser pulse.)

attractive complements to engine studies and it may allow a better understanding of the data derived from this kind of practical configurations. To further analyse the results presented in this first section, a Diesel surrogate (m-IDEA) composed of only two components (n-decane and 1-methylnaphthalene) has been used. Mixtures of this simplified model fuel with RME and four different RME surrogates have also been studied. The use of such blends is useful to test the validity of the different model fuels proposed in the literature for RME as it eases the analysis of the obtained results. Furthermore, the soot propensity of these simplified mixtures can easily be associated to the threshold soot index as shown in [47]. Such correlations will help the identification and quantification of

0

50

100

150

200

Height Above the Burner (mm) Fig. 3. Profiles of the peak intensity of soot precursors LIF at 266 nm (a), soot precursors LIF at 532 nm (b) and soot LII at 1064 nm (c) as a function of the height above the burner in flames of Diesel, RME and B30 flames. (Delayed detection: the gate width of the CCD camera is triggered to start 15 ns after the beginning of the laser pulse.)

the different effects associated to the addition of RME in terms of soot formation. 3.2. Study of the sooting propensity of different RME surrogates The ability of the m-IDEA fuel to reproduce the sooting behaviour of our European low sulphur Diesel has already been demonstrated in [39]. Consequently, we used this simplified model fuel instead of Diesel in the second part of this paper in order to isolate more easily the effect of ester addition on soot propensity. In a first time, the impact of the adjunction of 30% of rapeseed methyl ester to the m-IDEA fuel has been analysed. LIF and LII profiles obtained

154

R. Lemaire et al. / Fuel 107 (2013) 147–161

Soot volume fraction (ppb)

180 160 140 120 100

LIF intensity at 532 nm (a.u.)

200

n-decane RME LS30 E30 LS20 E20 LS10 E10 m-IDEA/RME (70/30) m-IDEA/methyl oleate (70/30) B30 E0 LS0

60 40 20

3,0E+06

8,E+06

RME B30 Diesel

6,E+06 4,E+06 2,E+06

(b)

0,E+00 0,00E+00 1,33E+07 2,67E+07 4,00E+07

LIF intensity at 266 nm (a.u.)

m-IDEA/n-decane (70/30) Hex-11 m-IDEA/n-hexadecane (70/30) Modified Hex-12 Hex-12 m-IDEA/1-octadecene Kerosene Diesel m-IDEA (a) IDEA

80

0 0,0E+00

1,E+07

6,0E+06

9,0E+06

LIF intensity at 532 nm (a.u.)

1,2E+07

1,5E+07

160

Soot volume fraction (ppb)

220

140 120

RME B30 Diesel

100 80 60 40 20

(c)

0 0,00E+00 1,33E+07 2,67E+07 4,00E+07

LIF intensity at 266 nm (a.u.)

Fig. 4. Evolution of the peak soot volume fraction as a function of the peak intensity of LIF signals at 532 nm in a set of turbulent diffusion flames (a). The data presented on this graph are issued from Refs. [39,47,55] and from the present work. The insets represent the evolution of the peak intensity of LIF signals at 532 nm as a function of the peak intensity of LIF signals at 266 nm (b) and the evolution of the peak soot volume fraction as a function of the peak intensity of LIF signals at 266 nm (c) in flames of Diesel, B30 and RME (results issued from the present work).

during the combustion of this mixture are plotted in Fig. 6 and are compared with those issued from the analysis of the B30 flame (see Section 3.1). The results obtained with the two base fuels (i.e. Diesel and m-IDEA [39]) used as references here are also reported especially to illustrate the good ability of the surrogate to reproduce the sooting propensity of its parent fuel. Only the LIF signals measured with a 532 nm laser excitation are presented here in order to focus the analysis on large soot precursors which are closely related to the formation of nascent soot. The evolution of the peak LIF and LII signals depicted in Fig. 6 highlights the very good agreement existing between the two blends containing rapeseed methyl

Evolution of the quantity of particles Δ PM (%)

0 -10 -20

ΔPM = (exp(-0.006384

B)

- 1)

100

ΔPM = (exp(-0.018784

B)

- 1)

100

-30 -40 -50 -60 -70 EPA engine test values [13] Experimental values issued from the present work Knothe value [5,17] Wu et al. value [28] Krahl et al. value [65]

-80 -90

-100 0

20

40

60

80

100

Biodiesel content B (in vol %) Fig. 5. Reduction of the total quantity of particles as a function of the ester content. Data issued from the present work are related to the total soot volume fraction measured in flames of Diesel, B30 and RME while the measurements reported from Refs. [5,13,17,28,65] concern the total particulate matter emitted at the exhaust of Diesel engines.

ester (Diesel/RME and m-IDEA/RME). The profiles obtained with these mixtures are very close and the peak LIF and LII signals are measured for identical HAB with similar intensities. Diesel and m-IDEA fuel thus behave similarly in terms of sooting propensity when they are blended with RME. This indicates that the soot suppressing effect of the ester is very close in both cases. This also suggests that the m-IDEA fuel is a suitable 2-components surrogate to be used in mixture with ester in order to mimic the sooting behaviour of complex blends made up of petrodiesel and FAME. On the basis of these results, we used the m-IDEA fuel to test 4 RME surrogates including 2 alkanes (n-decane and n-hexadecane). Compared to the m-IDEA/RME blend, the use of n-decane or n-hexadecane as additives leads to overestimate the quantities of soot and soot precursors produced. Indeed, the peak intensities of LIF and LII signals are around 24% and 23% higher, respectively, when using n-hexadecane instead of RME. It is worthy to note that these results are in good agreement with those obtained by Knothe et al. in engine conditions by adding n-hexadecane to petrodiesel instead of various methyl esters [17]. In the case of n-decane, the peak intensity of both signals is about 12% higher than in the case of RME. The fact that n-hexadecane generates more soot than n-decane is consistent with its higher sooting propensity [45] (see Table 1). This point will be treated in next section. The reduction of the total quantity of soot produced is thus of 28% and 23% when adding 30% of n-decane or n-hexadecane to the m-IDEA fuel. It is of 44% when using RME as additive as seen before. These results demonstrate that simple n-alkanes such as n-decane or n-hexadecane are not suitable model fuels to simulate the sooting behaviour of the large fatty acid methyl esters contained in RME when this biodiesel is used at intermediate concentrations in mixture. One can also note that the structure of the two selected alkanes is simpler than the hydrocarbon composition of FAME such as methyl oleate (the main constituent of RME) that contains 19 C atoms and one double bound. Consequently, the fact that these simple molecules achieved lower soot reductions than RME suggests that the oxygen content of the ester significantly influences the soot formation process in addition to the dilution effect corresponding to the

155

R. Lemaire et al. / Fuel 107 (2013) 147–161

replacement of the aromatic part of the base fuel by a cleaner molecule. The decarboxylation of the ester group that removes C atoms from the product pool of species potentially involved in soot production [7,18,19] is not sufficient here to explain the higher soot suppressing effect of RME. Indeed, the C atoms content of this FAME is higher than that of the two tested alkanes. The presence of the ester functional group should thus influence the sooting propensity as suggested by Feng et al. [54]. The authors of this study especially demonstrated that the combustion of 12 different fatty acid mono-alkyl esters generates significantly less soot particles than the oxidation of their corresponding alkanes. Based on a kinetic modelling of these results, Feng et al. concluded that the presence of the ester group in the fuel molecular structure leads to different reaction pathways compared to n-alkanes. Lower concentrations of soot precursors are thus formed. This includes C2H2, C3H3 and C2H4 [54]. Such a conclusion is consistent with the trend observed in the present work. On the other hand, the results obtained in the case of n-decane are particularly worthy to note since we previously observed that n-decane generates far less particles than pure RME. The measurable peak soot volume fraction obtained with n-decane was indeed of 0.7 ppb [39,55] while it is of 24 ppb in the case of RME (see Fig. 1d). Consequently, this suggests that the soot suppressing effect of RME is more important when it is used at intermediate concentrations as it is the case here. This point which needs to be clarified will be treated more in detail in the next section especially through the analysis of the correlation existing between the threshold soot index of the different studied mixtures and the peak soot volume fraction measured in the corresponding flames. Following Choi et al. [12], we also tested 1-octadecene as a RME surrogate. This alkene is indeed of interest to isolate the influence played by the ester function in molecules such as methyl oleate which is the main constituent of RME as said before (60% in volume). This characteristic is due to the molecular structure of 1-octadecene which has 18 C atoms, one double bond but no oxygen. This alkene can thus be considered as a good representative of the methyl oleate hydrocarbon composition even if the location of the double bound is different in the two compounds (a parameter which may influence the early CO2 production in the case of fatty acid esters [20]). In parallel to the use of 1-octadecene, we also analysed the impact of adding 30% of methyl oleate to the m-IDEA fuel in order to compare the sooting behaviour of 1-octadecene

1,0E+07

3.3. Correlation between soot volume fraction and Threshold Soot Index (TSI) We used the Threshold Soot Index (TSI) as indicator to represent the sooting tendency of the different fuels investigated in a previous work related to the study of ethanol [47]. To determine the TSI of hydrocarbon mixtures like those met in surrogates, we used the additivity rule proposed by Gill and Olson [66] defined by TSImix = RiXiTSIi, where TSIi and Xi are the soot index and the mole fraction of the ith component, respectively. Following

Diesel m-IDEA B30 m-IDEA/RME (70/30) m-IDEA/n-decane (70/30) m-IDEA/n-hexadecane (70/30) m-IDEA/ 1-octadecene (70/30) m-IDEA/methyl oleate (70/30)

(a)

8,9E+06

7,5E+06

6,7E+06 5,6E+06 4,4E+06 3,3E+06

Diesel m-IDEA B30 m-IDEA/RME (70/30) m-IDEA/ n-decane (70/30) m-IDEA/n-hexadecane (70/30) m-IDEA/1-octadecene (70/30) m-IDEA/methyl oleate (70/30)

(b)

6,7E+06 5,8E+06

LII intensity (a.u.)

7,8E+06

LIF intensity (a.u.)

with that of its corresponding ester. The obtained profiles which are depicted in Fig. 6 first illustrate the very good agreement existing between the LIF and LII signals collected in flames burning RME or methyl oleate as additives. The fact that methyl oleate is a good model fuel to mimic the sooting behaviour of RME is not surprising. Nevertheless, it clearly illustrates the effect of the oxygenated moiety on soot reduction especially when comparing the data obtained for the m-IDEA/methyl oleate blend and those issued from the oxidation of the m-IDEA/1-octadecence mixture. Indeed, the addition of 30% of 1-octadecene to the m-IDEA fuel induces a reduction of around 19% of both LIF and LII signals intensities against 39% when methyl oleate is used as additive. Considering that the main difference between these two compounds is the ester functional group, this leads us to separate the contribution of the dilution effect (estimated at around 40%) from that of the ester function (estimated at around 60%) in the reduction of the total quantity of soot produced (integrated on the whole heights and widths of the flames). Such an analysis shows that when methyl oleate is used at intermediate volume percentage (30%), the oxygen content of the ester as well as the dilution of the aromatic part of the base fuel both play a significant role in the soot suppression. This contrasts somewhat with the analysis made by Pepiot-Desjardins et al. in the case of small FAME like methyl butanoate added at lower concentrations [7]. Indeed, the authors argued that the dilution effect was the most important factor involved in the soot reduction for additive volume fractions of the order of 0.1–0.2. Additional measurements using various quantities of ester are thus necessary. Furthermore, the use of the threshold soot index (TSI) could be of interest to further analyse the obtained results with a view to estimate the relative contribution of the different effects involved in the observed soot reductions.

5,0E+06 4,2E+06 3,3E+06 2,5E+06

2,2E+06

1,7E+06

1,1E+06

8,3E+05 0,0E+00

0,0E+00 0

50

100

150

Height Above the Burner (mm)

200

0

50

100

150

200

Height Above the Burner (mm)

Fig. 6. Profiles of the peak intensity of soot precursors LIF at 532 nm (a) and soot LII at 1064 nm (b) as a function of the height above the burner in flames burning mixtures of 70% Diesel with 30% RME (B30) and 70% m-IDEA with 30% of additives (RME, n-hexadecane, n-decane, methyl oleate and 1-octadecene). Results previously obtained for mIDEA fuel [39] are also reported and compared with those of the Diesel flame (Prompt detection).

156

R. Lemaire et al. / Fuel 107 (2013) 147–161

McEnally and Pfefferle [67], we studied the correlation existing between the TSI of a set of different fuels and the peak soot volume fraction measured in the corresponding turbulent flames [39,47,55]. The obtained data are summarized and plotted in Fig. 7. We also reported on this figure the results issued from the study of the mixtures containing m-IDEA fuel and n-decane, n-hexadecane or 1-octadecene. The TSI of these blends have been determined using the additivity rule presented above and the soot index values issued from [45,58] (see Table 1). By integrating these new experimental data, we obtain a linear correlation between the peak soot volume fraction and the TSI as in [47]. We also reported in Fig. 7 the results presented in Sections 3.1 and 3.2 for pure RME and for the m-IDEA/RME blend (70/30). We finally added a set of additional experimental data obtained at the peak soot location in flames burning mixtures of m-IDEA fuel and RME with 10%, 20%, 50%, 60% and 80% of ester (in vol.%). The composition of these 3-components mixtures is perfectly known by contrast to the Diesel/RME blends that contain hundreds of different species. It is thus possible to estimate the TSI of these simplified fuels using the additivity rule proposed by Gill and Olson. Such a method requires the estimation of the TSI of RME, however. In a first time, we used the same methodology than in [47] and we considered that the sooting propensity of the oxygenated additive was equal to zero. This hypothesis implies that this additive is considered as non-sooting which means that it only plays a dilution role. By taking a value of 0 for the soot index of RME, TSI values of 26, 24.3, 22.4, 17.9, 15.2, 8.7 and 0 are found for the blends containing 10%, 20%, 30%, 50%, 60%, 80% and 100% of RME, respectively. Using the linear correlation obtained before (i.e. peak soot volume fraction = 5.6621  TSI), we thus determine theoretical soot volume fractions of 147, 138, 127, 101, 86, 49 and 0 ppb for the above-mentioned mixtures. These theoretical points (noted as ‘‘calculated points’’) are represented by grey squares in Fig. 7 and correspond to the effect associated to a simple dilution of the mIDEA fuel. We experimentally measured peak soot volume fractions of 142, 122, 105, 84, 75 and 48 ppb in flames of m-IDEA/ RME blends containing 10%, 20%, 30%, 50%, 60% and 80% of ester, respectively. In the case of pure RME, a peak soot volume fraction of 24 ppb was found as seen before (see Section 3.1). These points are represented with dark gray squares in Fig. 7. They follow an exponential law. One can note that the measured peak soot volume fractions are lower than the calculated values for additive concentrations comprised between 0% and 80%. Such a global trend indicates that the effect of RME on soot reduction is not limited to a pure dilution one. The observed additional soot reductions suggest that the oxygen contained in the rapeseed methyl ester influences the soot formation process. Furthermore, results depicted in Fig. 7 illustrate the fact that the sooting tendency of RME is negative when it is added up to 80% in mixtures which is consistent with the analysis made by Pespiot-Desjardins et al. in the case of small FAME [7]. On the basis of the curves plotted in Fig. 7, we can estimate the contribution of the effect that would be associated to a simple dilution of the base fuel and separate this contribution from other additional ones. The obtained trends thus reveal that the role played by the additional effects on soot reduction is more important at low and intermediate additive volume percentages (between 10% and 30%) than at high levels of RME addition (see Fig. 7). This suggests that the soot suppressing effect of RME is more important when it is used in relatively small quantities. Furthermore, at very high concentrations (more than 80%), the ester becomes less effective to reduce soot than a pure diluent. This is highlighted by the fact that the exponential curve that represents the experimental data obtained with the m-IDEA/RME mixtures passes above the theoretical dilution curve which corresponds to the linear correlation between TSI and soot volume fraction (see explanations above). This transition from a soot reduction regime

to another is depicted in Fig. 7 by the means of a two zone scheme that separates the region in which the soot suppressing effect of RME is higher than that of a pure diluent (Zone A) and the region in which the oxygenated additive is less effective to reduce soot (Zone B). The fact that the reduction of particles is more effective with low and intermediate biodiesel concentrations in blends is in agreement with trends observed in the majority of engine studies [11,13,21,68,69]. Lapuerta et al. especially reported during engine based experiments that the relative reductions of particulate emissions were more drastic when the reference fuel they used was compared with a 25% blend while additional reductions were not so important for higher biodiesel concentrations. The authors then argued that the PM reductions were mainly caused by a decrease of soot emissions which is consistent with the results presented in the present work. Lapuerta et al. also argued that at high ester concentrations, the decrease in PM emissions was weaker (with respect to the biodiesel content) due to the increase of the soluble organic fraction (SOF) of the particulate matter which is composed by hydrocarbons derived from lube oil [68]. This explanation is not suitable in our case due to the experimental configuration used and to the fact that we focus the analysis on soot particles and hence on insoluble fraction (ISF) of PM. To us, the trends depicted in Fig. 7 could be explained by a combination of different competing effects. Thus, the low and intermediate concentration ranges of additive would be dominated by an effective soot suppressing effect of the oxygen moiety of RME associated to a dilution effect corresponding to the replacement of the aromatic part of the base fuel by cleaner compounds. At high concentrations, the effect of the ester function would then play a less important part in the soot reduction process. Indeed, the contribution of the long alkyl chains contained in RME increases with the quantity of ester added. Consequently, this induces (in parallel to the above-mentioned effects) an inverse dilution process that corresponds to the replacement of the low sooting alkane part of the base fuel by the long alkyl chains of RME which sooting propensities are much higher. The sooting behaviour of RME thus deviates from that of a pure diluent. It generates more particles due to its complex hydrocarbon composition as observed for high additive concentrations (Zone B of Fig. 7). Of course, the transition from a soot suppressing regime to another is progressive since all these effects coexist. Consequently, the separation between the two regions depicted on the graph of Fig. 7 should not be considered as a physical transition but only as a way to represent and highlight the evolution of the propensity of RME to reduce soot which seems strongly correlated to the quantity of additive used. The analysis conducted above (especially regarding the contribution of the alkyl chains contained in RME) implies that the hypothesis made in the first step concerning the negligible soot index of RME is not suitable. Such an observation was already demonstrated by the fact that pure RME generates soot. Consequently, considering that 1-octadence is a good representative of the alkyl part of methyl oleate (the major component of RME), we can use the TSI of 1-octadecene (9.2 [45]) in order to simulate the contribution of the dilution effect induced by the addition of RME to the m-IDEA fuel. We thus obtain TSI values of 26.5, 25.3, 24, 20.9, 19.1, 14.8 and 9.2 for ester volume percentage of 10%, 20%, 30%, 50%, 60%, 80% and 100%, respectively. The corresponding peak soot volume fractions calculated using the linear correlation presented before are of 150, 143, 136, 118, 108, 84 and 52 ppb for the above-mentioned additive concentrations. This new theoretical points are represented by light grey squares in Fig. 8 while experimental data are represented by dark gray squares. The estimation of the contribution of the different effects involved in soot reductions revealed that the oxygen content of RME would be the main factor governing the decrease of

157

R. Lemaire et al. / Fuel 107 (2013) 147–161

240

n-decane LS0 Hex-11 Modified Hex-12 Hex-12 Kerosene Diesel m-IDEA IDEA m-IDEA + 30% n-decane m-IDEA + 30% n-hexadecane m-IDEA + 30% 1-octadecene

220

200

y = 5,6621x R² = 0,9337

120

100

58%

59% 87%

Soot suppressing effect of RME < Soot suppresing effect of a nonsooting diluent

77%

Zone B

140

69%

Peak soot volume fraction of the base fuel (m -IDEA )

160

31%

42%

10% RME

41% 20% RME

99%

Peak soot volume fraction (ppb)

180

Calculated points (Dilution effect for a non-sooting additive) Experimental points Contribution of the dilution effect Additional contributions

30% RME

23% 13%

80

50% RME 60% RME

60 y = 26.295e 0.0643x R 2 = 0.9938

40

1% 80% RME

20

Zone A 100% RME

Soot suppressing effect of RME > Soot suppresing effect of a nonsooting diluent

0 0

5

10

15

20

25

30

35

40

TSI Fig. 7. Evolution of the peak soot volume fraction measured in different turbulent spray flames as a function of the threshold soot index of the fuels. Estimation of the contribution of the different effects involved in soot reduction when considering the RME as a non-sooting fuel (TSIRME = 0).

the sooting propensity when the ester is used at intermediate volume percentages (20% and 30%). On the contrary, the contribution of the dilution effect would be significantly more important than the effect of the ester functional group at high (50% and more) and low (10%) additive percentages. This result is more in adequacy with the trend observed by Pepiot-Desjardins et al. even if the contribution of the dilution effect was much more important in their experiments based on the use of methyl butanoate [7]. The fact that the dilution effect is less important with large FAME like those contained in RME is probably due to the presence of long and complex aliphatic chains having a higher sooting tendency than the alkane part of light molecules such as methyl butanoate. To conclude, the data presented in Fig. 8 corroborate the fact that the effect of the ester functional group (represented as ‘‘additional contributions’’) seems to be less effective to reduce soot at high additive volume percentages than at low and intermediate proportions. The additional contributions associated to the presence of the ester function are indeed of 20% for pure RME while it is comprised between 39% and 57% for ester volume percentages comprised between 10% and 60% in mixtures. The global trend depicted in Fig. 8 was already noticeable in Fig. 7 even if it was not possible to draw a clearcut conclusion due to the dispersion of some experimental points around the proportionality rule existing between the TSI and the peak soot volume fraction. Of course, a modelling of these results would be necessary to better interpret the trends highlighted during this work. The new set of experimental data presented here with simplified model fuels should be of help to this aim.

4. Effect of ester addition on morphological and optical characteristics of soot particles LII time decays and fluence curves have been recorded at different heights above the burner in flames of Diesel and B30. Such evolutions of the LII signals as a function of time and laser energies are of interest to obtain information regarding the morphological and optical properties of soot particles. For instance, the LII time decays are strongly correlated to the soot diameter, to the particles size distribution and to the aggregate morphology. On the other hand, the evolution of the LII signals as a function of the laser energy is representative of different parameters governing the LII process including some physical properties of soot such as their density (q), their specific heat (Cp) and their absorption function (E(m)) [49]. The measurements presented here have been carried out in both flames at 92 mm HAB (in the soot formation region) and at 125 mm HAB (in the oxidation zone). The comparison of the data obtained at such heights will help the study of the influence of ester additions on the characteristics of soot and on their oxidation behaviour. Additional measurements have also been carried out in the flame of pure RME. The obtained results have not been reported here yet since the LII intensities measured at 125 mm HAB were too weak to obtain workable LII time decays and fluence curves. The variations of the LII signals with the laser fluence plotted in Fig. 9 are typical of the use of a Gaussian laser profile [49,70] with a sharp increase after 0.03 J/cm2 and a plateau region at high fluence corresponding to the competition between wings effect and

R. Lemaire et al. / Fuel 107 (2013) 147–161

Fig. 9b). This shows that the particles size decreases much faster during the oxidation process when RME is added to Diesel. We attribute these changes to a faster oxidation of the B30 soot. The results obtained here thus suggest that the oxygen content of ester promotes the oxidation of soot especially by modifying their structure inducing changes in the physical properties of soot. This analysis is consistent with the observations made by MerchanMerchan et al. [22], Boehman et al. [23,24,26] and Jung et al. [25] even if a recent work demonstrated that soot issued from the combustion of a small ester (methyl 2-butenoate) exhibit only slightly higher oxidative reactivity compared to n-pentane soot [72]. According to the authors, this suggests that the fuel-bound oxygen of the ester moiety does not significantly enhance the soot oxidative reactivity. The formation of hollow cavities in the internal structure of primary particles caused by the internal oxygen of biodiesel molecules could explain the faster oxidation observed in the present work as explained in [24].

sublimation [70,71]. One can note that the fluence curves obtained at the peak soot location (Fig. 9a) after normalization at high fluence are very similar. Such a behaviour can be observed at different HAB up to 100 mm. This suggests that the physical and optical properties of soot are nearly the same at these heights. This is a key point justifying that the LII signals measured in these flames, especially with an energy of 0.44 mJ/pulse corresponding to a fluence of 0.22 J/cm2 (see section 2.2), are directly comparable in terms of soot volume fraction. On the other hand, the inset of Fig. 9a shows that the LII time decay is longer in the case of B30. This indicates that the particles size is more important when RME is added to Diesel. This result is consistent with a recent morphological analysis of soot which have been probed in flames of Diesel and B30 and imaged by transmission electron microscopy [49]. Furthermore, a similar behaviour has also been observed at different heights in both flames up to 100 mm HAB. Consequently, this leads us to conclude that the mean size of soot particles formed during the combustion of fuels containing RME is more important in the soot formation region. A completely different behaviour can be observed in the oxidation region of the flames. Indeed, compared to Fig. 9a, a slight shift of the obtained fluence curve towards higher fluences (x axis) can be noticed at 125 mm HAB when RME is added to Diesel (see Fig. 9b). This suggests that the evolution of the physical and optical properties of soot is slightly different depending on the nature of the fuel which was not observable in the soot formation region. Furthermore, the LII time decay obtained with B30 becomes identical or slightly shorter than that obtained with Diesel at 125 mm HAB while it was the contrary at 92 mm HAB (see the inset of 240

n-decane LS0 Hex-11 Modified Hex-12 Hex-12 Kerosene Diesel m-IDEA IDEA m-IDEA + 30% n-decane m-IDEA + 30% n-hexadecane m-IDEA + 30% 1-octadecene

220

200

The effects associated to the use of rapeseed methyl ester on the soot formation process have been studied through the analysis of turbulent spray flames of Diesel, B30 and pure RME. Mappings and concentration profiles of soot particles, high-number-ring aromatic species and light soot precursors have been obtained using a recently developed three-excitation wavelength LII/LIF technique based on the coupling of laser excitations at 1064, 532 and 266 nm. Soot volume fraction measurements have also been

Calculated points (Dilution effect) Experimental points Contribution of the dilution effect Additional contributions

y = 5,6621x R² = 0,9337

43%

57%

68%

120

100

47%

54%

61%

140

43%

Peak soot volume fraction of the base fuel (m -IDEA )

160

10% RME

57% 20% RME

80%

Peak soot volume fraction (ppb)

180

5. Conclusion

53%

158

46% 30% RME

39%

80 50% RME 32%

60% RME

60

y = 10.954e 0.0965x R 2 = 0.9938

40

20%

80% RME

20 100% RME

0 0

5

10

15

20

25

30

35

40

TSI Fig. 8. Evolution of the peak soot volume fraction measured in different turbulent spray flames as a function of the threshold soot index of the fuels. Estimation of the contribution of the different effects involved in soot reduction when considering the TSI of RME identical to that of 1-octadecene (TSIRME = TSI1-octadecene = 9.2).

159

R. Lemaire et al. / Fuel 107 (2013) 147–161

1,00

Diesel

(a)

1,00

0,80

Diesel

(b)

0,80

Diesel

1,00

B30

Diesel

1,00

B30

0,40

0,20

0,80 0,60 0,40 0,20 0,00 150

300

450

600

750

0,20

0,40

0,30

0,40

Laser fluence

0,50

0,80 0,60 0,40 0,20 0,00 0

900

150

300

450

600

750

900

Time (ns)

Time (ns)

0,10

0,60

0,20 0

0,00 0,00

LII Signal (a.u.)

0,60

LII signal (a.u.)

B30

LII Signal (a.u.)

LII signal (a.u.)

B30

0,60

0,70

(J/cm 2)

0,00 0,00

0,10

0,20

0,30

0,40

0,50

0,60

0,70

Laser fluence (J/cm 2)

Fig. 9. LII fluence curves obtained at 92 mm (a) and 125 mm (b) HAB in flames of Diesel and B30. The insets represent the LII time decays measured at the same heights with a fluence of 0.22 J/cm2.

carried out in mixtures containing a Diesel simplified model fuel (the m-IDEA blend) and 10–80% of RME in order to help the analysis of the main trends highlighted during this work. The obtained results have been correlated to the threshold soot index of the corresponding mixtures to isolate and estimate the contribution of the different effects associated to the addition of RME in terms of soot reduction. Different RME surrogates proposed in the literature have also been analysed in order to test their ability to reproduce the sooting propensity of their parent-fuel. Finally, some physical properties of soot produced during the combustion of Diesel and B30 have been compared by analysing the LII time decays and fluence curves measured at different heights above the burner. The new set of experimental data obtained during this work has been analysed with a view to further examine some aspects of FAME combustion that have been pointed out during engine studies or laboratory scale experiments and which seemed unclear especially regarding the impact of ester additions on the soot particles formation and characteristics. The following results have been found:  Important reductions of soot precursors and soot quantities (up to 85%) have been observed when using increasing quantities of RME (from 10% to 100%). It has also been found that the reduction of the total quantity of soot produced as a function of the concentration of ester in the fuel follows an exponential law of the same type than the correlation proposed by the EPA for PM emissions measured at the exhaust of engines fuelled by biodiesels [13].  The absence of petrogenic PAHs in RME as well as the effect of the ester functional group have been pointed out as being the main factors explaining the observed reductions of soot and soot precursors concentrations. The respective contribution of each of these effects has been found to be strongly correlated to the quantity of ester added in the fuel mixtures.  The m-IDEA fuel (a Diesel surrogate previously formulated in [39]) has been found to reproduce well the sooting propensity of the European low-sulphur Diesel used in this work and especially its behaviour towards ester addition.  Measurements carried out in flames burning mixtures composed of 70% of m-IDEA fuel and 30% of RME or various RME surrogates showed that large n-alkanes (n-decane and n-hexadecane) are not suitable model fuels to reproduce the sooting propensity of RME even though such hydrocarbons were con-

sidered as convenient surrogates for modelling the oxidation behaviour and the overall reactivity of some FAME [32,36]. The analysis made in the particular case of n-decane revealed that the use of such an alkane at low concentrations leads to underestimate the level of soot reduction achieved with RME. Absence of oxygen in the alkane can explain this tendency. On the other hand, for high levels of additive, the soot reductions measured with n-decane significantly overestimate those obtained with ester. This is due to the long and complex alkyl chains contained in FAME that generate much more soot than low sooting molecules such as n-decane or n-hexadecane. Methyl oleate has been found to perfectly mimic the sooting behaviour of RME. 1-octadecene which is a good representative of the hydrocarbon composition of methyl oleate has been found to produce much more soot than RME. Such an analysis demonstrates that both the alkyl part and especially the oxygen moiety of esters influence the soot formation process. This also demonstrates that kinetic details of the oxidation of biodiesels including the mechanisms involved in the soot formation process can be mimicked only using large methyl esters as surrogates.  The analysis of the correlations existing between the threshold soot index and the peak soot volume fraction measured in different flames of m-IDEA/RME blends allowed the estimation of the relative contribution of different effects involved in the reduction of soot formation. Such an analysis revealed that the soot suppressing effect of RME was more important than that of a pure diluent at low and intermediate volume percentages of additive (between 10% and 60%). An opposite behaviour has been observed at high level of ester adjunction (more than 80%). Additional analyses then suggested that the oxygen content of RME was the main factor governing the decrease of the sooting propensity of the fuel mixtures when the ester was used at intermediate volume percentages (20% and 30%). On the contrary, the contribution of the dilution effect has been found to be significantly more important than the effect of the ester functional group at high (50% and more) and low (10%) ester concentrations in the blends.  The higher efficiency of RME to reduce soot at low and intermediate concentrations (with respect to the biodiesel content) is consistent with the main trend depicted in engine studies. It has been explained here by the contribution of the oxygen content of the ester molecules which coexists with two competing

160

R. Lemaire et al. / Fuel 107 (2013) 147–161

dilution effects. These effects correspond to the dilution of the aromatic part of the base fuel and to the replacement of its low sooting alkane portion by the highly sooting alkyl chains contained in esters.  The soot particles mean diameter is more important in the soot formation region when B30 is used instead of Diesel. On the other hand, B30 soot oxidize faster than Diesel ones. This conclusion is consistent with observations made by other authors [23–26] and explains why particles emitted by engines are generally smaller when biodiesels are used [21].

Acknowledgments The laboratories participate in the Institut de Recherche en ENvironnement Industriel (IRENI) which is financed by the Communauté Urbaine de Dunkerque, the Région Nord Pas-de-Calais, the Ministère de l’Enseignement Supérieur et de la Recherche, the CNRS and the European Regional Development Fund (ERDF). The authors thank the Agence Nationale pour la Recherche through contract ANR-06-BLAN-0349-01 and Diester Industrie for providing the RME used in this work. Fabrice Cazier from the center of measurement (MREID) of the ‘‘Littoral Côte d’Opale’’ university (ULCO) is warmly thanked for providing the GC composition of the different fuels used in this work. Finally, the authors also thank Xavier Mercier and Pascale Desgroux from the PC2A laboratory and Sébastien Menanteau from Mines Douai - EI for their help in the experimental procedures and for fruitful discussions.

References [1] Demirbas A. Political, economic and environmental impacts of biofuels: a review. Appl Energy 2009;86:S108–17. [2] Agarwal AK. Biofuels (alcohols and biodiesel) applications as fuels for internal combustion engines. Prog Energy Combust Sci 2007;33:233–71. [3] Demirbas A. Progress and recent trends in biodiesel fuels. Energy Convers Manage 2009;50:14–34. [4] Basha SA, Gopal KR, Jebaraj S. A review on biodiesel production, combustion, emissions and performance. Renew Sustain Energy Rev 2009;13:1628–34. [5] Knothe G. Biodiesel and renewable diesel: a comparison. Prog Energy Combust Sci 2010;36:364–73. [6] Schmidt K, Van Gerpen J. The effect of biodiesel fuel composition on Diesel combustion and emissions, SAE paper 961086. [7] Pepiot-Desjardins P, Pitsch H, Malhorta R, Kirby SR, Boehman AL. Structural group analysis for soot reduction tendency of oxygenated fuels. Combust Flame 2008;154:191–205. [8] Krahl J, Munack A, Bahadir M, Schumacher L, Elser N. Review: utilization of rapeseed oil, rapeseed oil methyl ester or diesel fuel: Exhaust gas emissions and estimation of environmental effects. SAE paper 962096. [9] McCormick RL, Ross JD, Graboski MS. Effect of several oxygenates on regulated emissions from heavy-duty Diesel engines. Environ Sci Technol 1997;31:1144–50. [10] McCormick RL, Graboski MS, Alleman TL, Herring AM. Impact of biodiesel source material and chemical structure on emissions of criteria pollutants from a heavy-duty engine. Environ Sci Technol 2001;35:1742–7. [11] Haas MJ, Scott KM, Alleman TL, McCormick RL. Engine performance of biodiesel fuel prepared from soybean soapstock: a high quality renewable fuel produced from a waste feedstock. Energy & Fuel 2001;15:1207–12. [12] Choi CY, Bower GR, Reitz RD. Effects of biodiesel blended fuels and multiple injections on D.I. Diesel Engine Emissions. SAE paper 970218. [13] Assessment and Standards Division (Office of Transportation and Air Quality of the US Environmental Protection Agency). A comprehensive analysis of biodiesel impacts on exhaust emissions. EPA420-P-02-001; 2002. [14] Sison K, Ladommatos N, Song H, Zhao H. Soot generation of diesel fuels with substantial amounts of oxygen-bearing compounds added. Fuel 2007;86:345–52. [15] Pradeep V, Sharma RP. Use of hot EGR for NOx control in a compression ignition engine fuelled with bio-diesel from Jatropha oil. Renewable Energy 2007;32:1136–54. [16] Tsolakis A, Megaritis A, Wyszynski ML, Theinnoi K. Engine performance and emissions of a diesel engine operating on diesel-RME (rapeseed methyl ester) blends with EGR (exhaust gas recirculation). Energy 2007;32:2072–80. [17] Knothe G, Sharp CA, Ryan TW. Exhaust emissions of biodiesel, petrodiesel, neat methyl esters, and alkanes in an new technology engine. Energy Fuels 2006;20:403–8.

[18] Szybist JP, Boehman AL, Haworth DC, Koga H. Premixed ignition behavior of alternative diesel fuel-relevant compounds in a motored engine experiment. Combust Flame 2007;149:112–28. [19] Westbrook CK, Pitz WJ, Curran HJ. Chemical kinetic modelling study of the effects ofoxygenated hydrocarbons on soot emissions from Diesel engines. J Phys Chem 2006;110:6912–22. [20] Zhang Y, Boehman AL. Premixed ignition behavior of C9 fatty acid esters: a motored engine study. Combust Flame 2009;156:1202–13. [21] Lapuerta M, Armas O, Rodriguez-Fernandez J. Effect of biodiesel on diesel engine emissions. Prog Energy Combust Sci 2008;34:198–223. [22] Merchan-Merchan W, Sanmiguel SG, McCollam S. Analysis of soot particles derived from biodiesels and diesel fuel air-flames. Fuel 2012;102:525–35. [23] Boehman AL, Song J, Alam M. Impact of biodiesel blending on Diesel soot and the regeneration of particle filters. Energy Fuels 2005;19:1857–64. [24] Song J, Alam M, Boehman AL, Kim U. Examination of the oxidation behavior of biodiesel soot. Combust Flame 2006;146:589–604. [25] Jung H, Kittelson D, Zachariah MR. Characteristics of SME biodiesel-fueled Diesel particle emissions and the kinetics of oxidation. Environ Sci Technol 2006;40:4949–55. [26] Song J, Alam M, Boehman AL. Impact of alternative fuels on soot properties and DPF regeneration. Combust Sci Technol 2007;179:1991–2037. [27] Soltic P, Edenhauser D, Thurnheer T, Schreiber D, Sankowski A. Experimental investigation of mineral diesel fuel, GTL fuel, RME and neat soybean and rapeseed oil combustion in a heavy duty on-road engine with exhaust gas aftertreatment. Fuel 2009;88:1–8. [28] Wu F, Wang J, Chen W, Shuai S. A study on emission performance of a diesel engine fueled with five typical methyl ester biodiesels. Atmos Environ 2009;43:1481–5. [29] Kegl B. Influence of biodiesel on engine combustion and emission characteristics. Appl Energy 2011;88:1803–12. [30] Maricq MM. Physical and chemical comparison of soot in hydrocarbon and biodiesel fuel diffusion flames: a study of model and commercial fuels. Combust Flame 2011;158:105–16. [31] Fisher EM, Pitz WJ, Curran HJ, Westbrook CK. Detailed chemical kinetic mechanisms for combustion of oxygenated fuels. Proc Combust Inst 2000;28:1579–86. [32] Dagaut P, Gaïl S, Sahasrabudhe M. Rapeseed oil methyl ester oxidation over extended ranges of pressure, temperature, and equivalence ratio: Experimental and modelling kinetic study. Proc Combust Inst 2007;31:2955–61. [33] Sarathy SM, Gaïl S, Syed SA, Thomson MJ, Dagaut P. A comparison of saturated and unsaturated C4 fatty acid methyl esters in an opposed flow diffusion flame and a jet stirred reactor. Proc Combust Inst 2007;31:1015–22. [34] Gaïl S, Sarathy SM, Thomson MJ, Diévart P, Dagaut P. Experimental and chemical kinetic modelling study of small methyl esters oxidation: methyl (E)-2-butenoate and methyl butanoate. Combust Flame 2008;155:635–50. [35] Salamanca M, Mondragon F, Agudelo JR, Benjumea P, Santamaria A. Variations in the chemical composition and morphology of soot induced by the unsaturation degree of biodiesel and a biodiesel blend. Combust Flame 2012;159:1100–8. [36] Herbinet O, Pitz WJ, Westbrook CK. Detailed chemical kinetic oxidation mechanism for a biodiesel surrogate. Combust Flame 2008;154:507–28. [37] Ramirez HP, Hadj-Ali K, Diévart P, Dayma G, Togbé C, Moréac G, et al. Oxidation of commercial and surrogate bio-Diesel fuels (B30) in a jet-stirred reactor at elevated pressure: experimental and modeling kinetic study. Proc Combust Inst 2011;33:375–82. [38] Hentschel W, Schindler KP, Haahtela O. European diesel research IDEA – experimental results from DI diesel engine investigations. SAE paper 941954. [39] Lemaire R, Faccinetto A, Therssen E, Ziskind M, Focsa C, Desgroux P. Experimental comparison of soot formation in turbulent flames of diesel and surrogate diesel fuels. Proc Combust Inst 2009;32:737–44. [40] Lai JYW, Lin KC, Violi A. Biodiesel combustion: advances in chemical kinetic modeling. Prog Energy Combust Sci 2011;37:1–14. [41] Hakka MH, Glaude PA, Herbinet O, Battin-Leclerc F. Experimental study of the oxidation of large surrogates for diesel and biodiesel fuels. Combust Flame 2009;156:2129–44. [42] Westbrook CK, Naik CV, Herbinet O, Pitz WJ, Mehl M, Sarathy SM, et al. Detailed chemical kinetic reaction mechanisms for soy and rapeseed biodiesel fuels. Combust Flame 2011;158:742–55. [43] Bax S, Hakka MH, Glaude PA, Herbinet O. Experimental study of oxidation of methyl oleate in a jet-stirred reactor. Combust Flame 2010;157:1220–9. [44] Naik CV, Westbrook CK, Herbinet O, Pitz WJ, Mehl M. Detailed chemical kinetic reaction mechanism for biodiesel methyl stearate and methyl oleate. Proc Combust Inst 2011;33:383–9. [45] Calcote HF, Manos DM. Effect of molecular structure on incipient soot formation. Combust Flame 1983;49:289–304. [46] Wang X, Kuti OA, Zhang W, Nishida K, Huang Z. Effect of injection pressure on flame and soot characteristics of the biodiesel fuel spray. Combust Sci Technol 2010;182:1369–90. [47] Lemaire R, Therssen E, Desgroux P. Effect of ethanol addition in gasoline and gasoline-surrogate on soot formation in turbulent spray flames. Fuel 2010;89:3952–9. [48] Lemaire R, Maugendre M, Schuller T, Therssen E, Yon J. Original use of a direct injection high efficiency nebulizer for the standardization of liquid fuels spray flames. Rev Sci Instrum 2009;80:105105.1–5.8.

R. Lemaire et al. / Fuel 107 (2013) 147–161 [49] Yon J, Lemaire R, Therssen E, Desgroux P, Coppalle A, Ren KF. Examination of wavelength dependent soot optical properties of diesel and diesel/rapeseed methyl ester mixture by extinction spectra analysis and LII measurements. Appl Phys B 2011;104:253–71. [50] Zizak G, Cignoli F, Montas G, Benecchi S, Donde R. Detection of aromatic hydrocarbons in the exhaust gases of a gasoline i.c. engine by laser-induced fluorescence technique. Recent Res Develop Appl Spectrosc 1996;1:17–24. [51] Vander Wal RL, Jensen KA, Choi MY. Simultaneous laser-induced emission of soot and polycyclic aromatic hydrocarbons within a gas-jet diffusion flame. Combust Flame 1997;109:399–414. [52] Wu J, Song KH, Litzinger T, Lee SY, Santoro R, Linevsky M. Reduction of PAH and soot in premixed ethylene-air flames by addition of dimethyl ether. Combust Sci Technol 2006;178:837–63. [53] Pitz WJ, Mueller CJ. Recent progress in the development of diesel surrogate fuels. Prog Energy Combust Sci 2011;37:330–50. [54] Feng Q, Jalali A, Fincham AM, Wang YL, Tsotsis TT, Egolfopoulos FN. Soot formation in flames of model biodiesel fuels. Combust Flame 2012;159:1876–93. [55] Lemaire R, Therssen E, Pauwels JF, Desgroux P. Experimental comparison of soot formation in turbulent flames of kerosene and surrogate model fuels. In: Bockhorn H, D’Anna A, Sarofim AF, Wang H, editors. Combustion-generated fine carbonaceous particles. KIT Scientific Publishing; 2009. p. 619–36. [56] Dagaut P, Cathonnet M. The ignition, oxidation, and combustion of kerosene: a review of experimental and kinetic modeling. Prog Energy Combust Sci 2006;32:48–92. [57] Van Gerpen J, Shanks B, Pruszko R, Clements D, Knothe G. Biodiesel Production Technology, National Renewable Energy Laboratory Subcontractor, Report NREL/SR-510-36244; 2004. [58] Olson DB, Pickens JC, Gill RJ. The effects of molecular structure on soot formation – II. Diffusion flames. Combust Flame 1985;62:43–60. [59] Bouvier Y, Mihesan C, Ziskind M, Therssen E, Focsa C, Pauwels JF, et al. Molecular species adsorbed on soot particles issued from low sooting methane and acetylene laminar flames: a laser-based experiment. Proc Combust Inst 2007;31:841–9. [60] Bejaoui S, Lemaire R, Therssen E, Desgroux P. Study of the wavelength dependence of the soot absorption function using the two-excitation wavelength Laser Induced Incandescence. application to fluorescent species

[61]

[62]

[63]

[64] [65]

[66] [67]

[68]

[69] [70]

[71]

[72]

161

detection. Fifth International Workshop on Laser-induced Incandescence, vol. 865; 2012 – paper n°2, Ed. CEUR workshop proceedings. ISSN: 1613–0073. McEnally CS, Pfefferle LD, Atakan B, Kohse-Höinghaus K. Studies of aromatic hydrocarbon formation mechanisms in flames: progress towards closing the fuel gap. Prog Energy Combust Sci 2006;32:247–94. Dobbins RA, Fletcher RA, Benner Jr BA, Hoeft S. Polycyclic aromatic hydrocarbons in flames, in diesel fuels, and in diesel emissions. Combust Flame 2006;144:773–81. Bejaoui S, Lemaire R, Therssen E, Desgroux P. Study of the effect of rapeseed methyl ester addition in diesel fuel on soot formation. Proc European Combust Meet 2011;5:107. Desgroux P, Mercier X, Thomson KA. Proc Combust Inst; 2012. http:// dx.doi.org/10.1016/j.proci.2012.09.004. Krahl J, Knothe G, Munack A, Ruschel Y, Schröder O, Hallier E, et al. Comparison of exhaust emissions and their mutagenicity from the combustion of biodiesel, vegetable oil, gas-to-liquid and petrodiesel fuels. Fuel 2009;88:1064–9. Gill RJ, Olson DB. Estimation of soot thresholds for fuel mixtures. Combust Sci Technol 1984;40:307–15. McEnally CS, Pfefferle LD. Improved sooting tendency measurements for aromatic hydrocarbons and their implications for naphthalene formation pathways. Combust Flame 2007;148:210–22. Lapuerta M, Armas O, Ballesteros R, Carmona M. Fuel formulation effects on passenger car diesel engine particulate emissions and composition. SAE paper 2000–01-1850. Armas O, Hernandez JJ, Cardenas M. Reduction of diesel smoke opacity from vegetable oil methyl esters during transient operation. Fuel 2006;85:2427–38. Schulz C, Kock BF, Hofmann M, Michelsen H, Will S, Bougie B, et al. Laserinduced incandescence: recent trends and current questions. Appl Phys B 2006;83:333–54. Bladh H, Bengtsson PE, Delhay J, Bouvier Y, Therssen E, Desgroux P. Experimental and theoretical comparison of spatially resolved laser-induced incandescence (LII) signals of soot in backward and right-angle configuration. Appl Phys B 2006;83:423–33. Zhang Y, Boehman AL. Oxidation behavior of soot generated from the combustion of methyl 2-butenoate in a co-flow diffusion flame. Combust Flame 2013;160:112–9.