Ignition delay times of low alkylfurans at high pressures using a rapid compression machine

Transcription

1 Available online at Proceedings of the Combustion Institute 36 (2017) Ignition delay times of low alkylfurans at high pressures using a rapid compression machine Nan Xu a, Yingtao Wu a, Chenglong Tang a,, Peng Zhang b, Xin He c, Zhi Wang b, Zuohua Huang a, a State Key Laboratory of Multiphase Flow in Power Engineering, Xi an Jiaotong University, Xi an , China b State Key Laboratory of Automotive Safety and Energy, Tsinghua University, Beijing , China c Aramco Services Company: Aramco Research Center-Detroit, Novi, MI 48377, USA Received 30 November 2015; accepted 16 July 2016 Available online 10 October 2016 Abstract Auto-ignition behavior of 2-ethylfuran (EF) was investigated in the low to intermediate temperature range ( K) over equivalence ratios of 0.5, 1.0, and 2.0 at 16 and 30 bar using a rapid compression machine. Equivalence ratio was varied by changing the mole fractions of O 2 while fixing the fuel fraction. The fuel did not show negative temperature coefficient (NTC) or two stage ignition behavior in the present measuring range. Simulations were conducted on the basis of the alkylfuran mechanism of Somers et al. (2013) [1], in which the EF sub-mechanism is not fully developed. Optimization of the EF sub-mechanism was attempted. Results show that the modified mechanism shows better agreement with our measurements and other data in the literature. Comparison of the reactivity of EF with that of 2-methyl furan (MF) and 2, 5-dimethyl furan (DMF) under various equivalence ratios was conducted and results show that the relative reactivity of the three furanic fuels depends complicatedly on both equivalence ratio and temperature, which was further interpreted by examining the ignition kinetics. As DMF is more branched, H-abstractions contribute most to the fuel consumption. For EF and MF, OH-additions at C2 and C5 positions dominate the reaction pathways. From reaction flux analyses, the reactions involving alkyl side chains of alkylfurans are analogous with those of alkylbenzenes at low to intermediate temperature oxidation conditions. Moreover, large amount of alkylfurans go through OH-addition reactions on the ring, which is not observed in the reaction pathways of alkylbenzenes by The Combustion Institute. Published by Elsevier Inc. Keywords: Ignition delay times; Ethylfuran; Low temperature; Alkylfurans 1. Introduction Corresponding author. Fax: addresses: chenglongtang@mail.xjtu.edu.cn (C. Tang), zhhuang@mail.xjtu.edu.cn (Z. Huang). Alkylfurans have recently received increasing attention from the combustion field, since they can be produced in large quantities from lignocellulosic biomass [2,3]. DMF (2, 5-dimethylfuran) and by The Combustion Institute. Published by Elsevier Inc.

2 324 N. Xu et al. / Proceedings of the Combustion Institute 36 (2017) MF (2-methylfuran) possess similar energy densities and research octane numbers with commercial gasoline [4]. The application potential of alkylfurans in internal combustion engines has been evaluated. Sudholt et al. [5] reported the DCNs (derived cetane numbers) of alkylfurans: DMF (10.9) > EF (10.2) > MF (8.9). Daniel et al. [6] investigated the influence of operating parameters on the combustion and emissions of DMF. When applied in spark-ignition engines, DMF [7] and MF [8] possess similar indicated specific fuel consumption and indicated efficiency with gasoline, and the emissions are lower except for NOx. For compression ignition engines, the addition of DMF could effectively prolong ignition delay times, resulting in higher NOx emissions [9]. Homogeneous charge compression ignition (HCCI) combustion is a combustion concept realizing compression-engine like high thermal efficiency without abundant NOx or soot emissions [10]. As there is no direct control strategy for the beginning of auto-ignition, fuel modification altering the auto-ignition chemistry of the fuel mixture thus effectively influences the combustion phase [11]. Research toward fundamental combustion characteristics of low alkylfurans facilitates their HCCI application. Experimentally, species profiles of DMF, MF, and furan have been measured to reflect their laminar premixed flame structures and interpret the influence of possessing different numbers of methyl side-chains [12 14]. Comparisons among laminar flame speeds of DMF [15], MF [16], iso-octane, and ethanol has been conducted. Tian et al. [17] concluded that the deviation between the laminar burning velocities of DMF and iso-octane is within 10% at the equivalence ratios of High temperature ignition delay times of DMF have been measured by Somers et al. [1] and Sirjean et al. [18]. Two kinetic models have been developed respectively, yielding reasonable agreement with their shock tube data. With respect to MF, Somers et al. [19] measured the ignition delay times and constructed a kinetic model. Thereafter, Wei et al. [20] extended the ignition delay times for MF to a wider pressure range and modified the mechanism of Somers et al. [19]. In addition to DMF and MF, EF is another alkylfuran of combustion research interest, possessing a side chain C C bond. Moreover, the weaker allylic C H bond and side chain C C bond would results in easier H-abstractions at allylic sites or bond scission at the side chain C C bond, contributing to its reactivity. Eldeeb and Akih-Kumgeh [21,22] conducted systematic investigation toward high temperature combustion kinetics of low alkylfurans in air conditions. From their comparison, DMF is the least reactive, while MF is the most reactive. Furthermore, EF ignites easier than DMF, and they contribute the higher reactivity to the ethyl radicals and more favorable sites for abstraction and decomposition reactions. Xu et al. [23] has compared the ignition delay times of DMF, MF, and furan under a constant fuel load, indicating that substituted furans possess higher reactivity. However, research toward low to intermediate temperature ignition of low alkylfurans has not been sufficiently explored. This study firstly aims to provide the ignition delay times of EF at low to intermediate temperatures and high pressures over various equivalence ratios. Moreover, previous stoichiometric MF ignition data [24] has been complemented with different equivalence ratios. In addition, investigation on the effects of different methyl side chain numbers, alkyl side chain lengths, and isomers was conducted by comparing the ignition behaviors of EF, MF, and DMF. Finally, the submechanism of EF in Somers et al. [1] was updated and validated against the measured ignition delay times and other literature data. Kinetic analyses based on the updated mechanism were applied to interpret the relative reactivity of the three furanic fuels. 2. Methodology The ignition delay times of low alkylfurans were measured using the RCM at Tsinghua University, see details in Di et al. [25]. The RCM comprises fiv e components: the high pressure air tank, the drive section, the driven section, the test section, and the hydraulic section. A creviced piston is employed to ensure the homogeneous temperature distribution in the test section. The length of the test section is variable to study different compression ratios. The gas mixture is prepared in a stainless steel tank according to their partial pressures. The partial pressures of MF and EF were kept below bar to avoid fuel condensation [26,27]. Composition of gas mixtures are tabulated in Table S1. The effective pressure and temperature were adopted to define experimental conditions. The effective pressure is defined as the integral average of the end-of-compression pressure and the lowest pressure due to heat loss before ignition. 1 P eff = ( t P min t P max ) t P min t P max P dt (1) The effective temperature is calculated thusly using isentropic compression integral: T eff ( ) γ P eff dln T = ln (2) γ 1 T 0 P 0 where T 0 is the initial temperature, P 0 is the initial pressure, γ is the specific heat ratio of the gas mixture. The experimental ignition delay times were deduced from pressure traces and defined as the time interval between end-of-compression and the steepest pressure rise, as shown in Fig. 1. Simulation was conducted using the SENKIN code of the Chemkin Ⅱ software, the maximum

3 N. Xu et al. / Proceedings of the Combustion Institute 36 (2017) Fig. 1. The definition of ignition delay times. Solid line: experimental pressure trace; dash line: pressure trace of nonreactive mixture. d P /d t is employed to define ignition delays. Volume time profiles of nonreactive mixtures were introduced into the simulation to account for heat loss. 3. Results and discussion 3.1. Model improvement The subsequent modification and kinetic analysis are based on the mechanism of Somers et al. [1]. Present experimental work and model optimization efforts are the continuation of our previous work [24] which has conducted some refinement to improve the agreement with DMF low to intermediate temperature ignition data. We conduct flux analyses to find the possible absent reactions that would influence the simulation. In analogy with ethylbenzene [28], HO 2 addition and H abstraction by O 2 reactions are included for E2F2J-A and E2F2J-P radical, respectively. The E2F2J-A radical would react with HO 2 to give OH radical, CH 3 radical, and F2CHO. The isomer, E2F2J-P radical would undergo addition reactions with HO 2 radical, generating OH, HCHO, and F2J radicals. In the original mechanism, V2F go through isomerization reactions or ipso-additions by H radicals. H-abstractions of V2F by O 2, H, OH, and CH 3 radicals are implemented by the present work. Addition reactions of V2F and V2FJ-P have been added, see details in Table 1. The thermochemical properties of the species V2FJ-P is absent in the mechanism of Somers et al. [1] and is calculated using THERM [29], see details in Supplemental Information. The rate constant of MF2 + OH = MF22J + H 2 O has been set to half of the value of DMF25 + OH = DMF252J + H 2 O. Moreover, the A-factor of reaction: MF22J + HO 2 = MF22OJ + OH is decreased by a factor of 1.5, in accordance with the corresponding reaction of DMF. For MF, OH-addition reactions are favored at two positions: C2 and C5. The molecular structures of EF and MF are analogous, and the only difference is the length of alkyl side chain. We assume that OH additions can happen both at C2 and C5 position of EF. As only C5 position is possible in the original mechanism of Somers et al. [1], E2F + OH = HCO + EVK is added to the mechanism with a rate constant of Ignition delay times of EF The low to intermediate temperature ignition delay times of EF were measured. The uncertainty of present τ ing is no more than 12%, mainly comes from the uncertainty of the pressure and temperature measurement, see detailed analysis in Supplemental Information. Figure 2 depicts the normalized newly-measured ignition delay times of EF with high temperature data of Eldeeb and Akih- Kumgeh [22]. The predicted ignition delay times with two kinetic mechanisms from low to high temperature range were also presented, it is seen that our modified mechanism show good agreement with the measured ignition delay times at both high and low temperatures.

4 326 N. Xu et al. / Proceedings of the Combustion Institute 36 (2017) Table 1 Details of modification. Reaction A-factor n E a /cal/mol E2F2J-A + O 2 = HO 2 + V2F a 7.0E E3 E2F2J-A + HO 2 = OH + CH 3 + F2CHO a 5.0E E2F2J-P = C 2 H 4 + F2J 3.55E E3 E2F2J-P + O 2 = HO 2 + V2F a 3.0E E3 E2F2J-P + HO 2 = OH + HCHO + F2J a 5.0E V2F + O 2 = HO 2 + V2FJ-P a 2.0E E3 V2F + O = F2J + CH 2 CHO a 3.0E E3 V2F + OH = MF22J + HCHO a 1.4E E3 V2F + OH = MF25CHO + CH a 3 2.8E E3 V2F + H = H 2 + V2FJ-P a 6.6E E3 V2F + OH = H 2 O + V2FJ-P a 1.5E E3 V2F + CH 3 = CH 4 + V2FJ-P a 3.5E E3 V2FJ-P + O 2 = HCHO + MF25CJO a 4.5E E3 a In analogy with ethylbenzene from Husson et al. [28]. Fig. 2. Ignition delay times of stoichiometric EF/O 2 /Ar mixture at 12 bar and dilution ratio of Circles: present data; squares: data from Eldeeb and Akih-Kumgeh [22] ; solid line: the modified mechanism of Somers et al. [1] ; dash line: the mechanism of Somers et al. [1]. Figure 3 a depicts the influence of pressure on ignition delay times of EF. The ignition delay times decrease with the increase of pressure. The effect of equivalence ratio on ignition behaviors of EF is shown in Fig. 3 b. Equivalence ratio was changed by varying the mole fractions of oxygen while keeping the fuel concentration constant. The ignition delay times increase with increasing equivalence ratio, since oxygen fraction is decreased at higher equivalence ratios while fuel concentration is kept constant. Comparison was conducted between experimental data and simulations. As EF and corresponding reactions are absent from the mechanism of Liu et al. [12], the simulation is not favored. As could be seen from Fig. 3 a and b, the ignition delay times of EF are largely over-predicted by the mechanism of Somers et al. [1]. Moreover, as depicted in Fig. 2, the reactivity of EF is underestimated in the high temperature range too. This is unsurprising since EF is regarded as a derivative of DMF or MF in their mechanism, without fully developed secondary reaction schemes. After modifying the mechanism, the mechanism now agree well with low to intermediate temperature ignition data at all equivalence ratios, see the solid lines in Fig. 3 a and b. As there is no other EF data in literature, further validation could not be conducted. However, the OH-addition and subsequent isomerization and decomposition reactions are lumped in the mechanism of Somers et al. [1]. Efforts should be made to refine the mechanism more elaborately, including pressure-dependent rate constants.

5 N. Xu et al. / Proceedings of the Combustion Institute 36 (2017) Fig. 3. Low to intermediate temperature ignition delay times for 1% EF fuel mixtures: (a) at pressures of 16 and 30 bar; (b) at equivalence ratios of 0.5, 1.0, and 2.0. Symbols: experimental data; solid line: the modified mechanism of Somers et al. [1] ; dash line: the mechanism of Somers et al. [1] Comparison among EF, MF, and DMF Effect of equivalence ratio The effect of equivalence ratio on ignition delay times of MF was investigated. Figure 4 depicts model predictions and experimental data. The mechanism of Somers et al. [1] shows reasonable agreement with MF data, while exaggerated curvature exists in the simulations. After modification, the mechanism could better capture the dependence of MF ignition delay times on temperature. The measured ignition delay time for the three fuels is fitted as functions of the experimental condition. The correlation of DMF ignition data is based on our previous work [24]. For EF, τ ign = P eff φ D exp( / R T eff ) (3) For MF, τ ign = P eff φ D exp ( / R T eff (4) For DMF, τ ign = P eff φ D exp ( / R T eff ) (5) where τ ign is ignition delay times in ms, P eff is the effective pressure in bar, φ is equivalence

6 328 N. Xu et al. / Proceedings of the Combustion Institute 36 (2017) Fig. 4. Ignition delay times of 1% MF fuel mixtures at equivalence ratios of 0.5, 1.0, and 2.0. Solid line: the modified mechanism of Somers et al. [1] ; dash line: the mechanism of Somers et al. [1]. ratio, D is dilution ratio (defined as (Ar + N 2 ) / O 2 ), T eff is the effective temperature in Kelvin, R = cal/(mol K), the unit of E a is kcal/mol. The correlations are applicable to low to intermediate temperature range and lower fuel concentrations. As denoted by these correlations, the pressure dependence of EF and MF is similar, both are weaker than DMF Comparison of low to intermediate temperature reactivity of alkylfurans To interpret the influence of alkyl side chain length and number on the ignition behavior of low alkylfurans, the comparison among ignition delay times of EF, MF, and DMF was conducted for three mixtures: φ = 0.5, 1.0, and 2.0 at 1% fuel concentration and 16 bar. It can be seen from Fig. 5 that DMF has a slightly weaker temperature dependence compared to EF and MF. The logarithmic ignition data of DMF as a function of inverse temperature shows a slight curvature. For the three equivalence ratios investigated, EF continuously gives shorter ignition delay times than MF, denoting that longer side chain length induces shorter ignition delay times. The relative reactivity of the two isomers, EF and DMF, is temperature dependent. At the equivalence ratio of 0.5, as shown in Fig. 5 a, a crossover temperature was observed. Above this temperature, EF has a higher reactivity, in accordance with the observation of Eldeeb and Akih-Kumgeh [22] at high temperatures. They concluded that the difference is caused by the weaker allylic C H bond in the ethyl side chain, combined with CH 3 radicals and C 2 H 6 generated from the decomposition of DMF and EF, respectively. Below the crossover temperature, the reactivity of DMF is higher. The crossover temperature moves to lower temperature with decreasing O 2 mole fractions. This discipline is also true for MF and DMF, and DMF possesses higher reactivity below the crossover temperature. The above mentioned features indicate that for EF, MF, and DMF, the more branched alkylfuran ignite faster at temperatures below the crossovers. To further understand the kinetics that govern the relative reactivity of the three furanic fuels, reaction flux analyses were conducted for the three fuels (see Fig. 6 ) at 880 K, 16 bar, and φ = 1.0 at 20% fuel consumption. The primary consumption routes of EF vary little after modification. The fuel is mainly consumed (59.2%) by OHaddition followed by decomposition reactions into C 2 H 5 CO + C 2 H 3 CHO or HCO + EVK. For EF, the furanic C H bond is weaker than that of methylated furans, which is 350 kj/mol. However, the terminal C H bond of the ethyl side chain is kj/mole [5]. Most of H-abstraction reactions (31%) happen at the allylic site of the ethyl side chain, and 1.9% fuel is consumed by H-abstraction at its terminal site. The E2F2J-A radical would either go through H-abstractions with O 2 or add with HO 2 radical yielding OH, CH 3, and MF25CHO radicals. The E2F2J-P radical would go through H- abstraction reactions by O 2 rather than release H radicals directly or experience β-scissions at the relatively low temperature. More consumption routes are provided for V2F in the modified mechanism, and most of the radicals react via additions by O or OH radicals, while only a small fraction is through H-abstraction. Compared with DMF, MF possesses less available furanic H atoms for abstraction. EF contains even less furanic H atoms, while the BDE of the allylic C H bond of EF is slightly lower

7 N. Xu et al. / Proceedings of the Combustion Institute 36 (2017) Fig. 5. Ignition delay times of stoichiometric DMF (squares), EF (circles), and MF (triangles): (a) φ = 0.5; (b) φ = 1.0; (c) φ = 2.0. Solid line: the modified mechanism of Somers et al. [1]. than that of furanic C H bonds. These features result in an order of H-abstraction fractions: DMF (36.4%) > EF (31%) > MF (22.1%). In addition, Fig. 6 shows that OH-addition reactions contribute differently to the three fuels: EF (59.2%) MF (54.9%) > DMF (27.8%). For DMF, H-addition reactions followed by isomerization and decompositions yielding CH 3 and MF consumes 25.7% of the fuel. However, for MF, the H-additions at C5 are favored, but the energy barrier of the subsequent decomposition reaction yielding CH 3 and CH 2 CHCHCO is about 70 kj/mole higher than the corresponding reaction of DMF [30], thus the H-additions only accounts for 4.5% MF Fig. 6. Reaction flux analyses at 880 K, 16 bar, and φ = 1.0 under 20% fuel consumption: (a) EF; (b) MF; (c) DMF. Using the modified mechanism of Somers et al. [1].

8 330 N. Xu et al. / Proceedings of the Combustion Institute 36 (2017) consumption. As a consequence, OH-addition reactions contribute more to fuel consumption for MF than DMF. These differences in the oxidation routes between DMF and EF or MF may account for their relative reactivity. In the present temperature range, the autoignition behavior is not controlled by the typical low temperature oxidation chemistry (O 2 addition to a fuel radical and subsequent isomerization from RO 2 to QOOH ), which explains that none of the three fuels exhibits NTC (negative temperature coefficient) phenomena nor twostage ignitions. Alkylbenzenes have similar molecular structures with alkylfurans. Previously eleven low alkylbenzenes have been investigated and only n-butylbenzene was found to show a distinct NTC behavior, suggesting that alkylbenzenes with longer side alkyl chains would experience the typical low temperature oxidation chemistry and exhibit NTC behaviors [31]. It is assumed that alkylfurans with longer side chains may also display NTC behaviors. Moreover, DMF252J and MF22J radicals undergo exactly the reaction pathways of benzyl radicals [32] in the low to intermediate temperature range. They would go through disproportionation with CH 3 O 2 or HO 2 radicals, followed by H radical release, H-abstractions, CO-elimination, and subsequent ring opening, leading to smaller radicals. With regard to EF, see Fig. 6 a, E2F2J-A and E2F2J-P radicals go through H-abstractions to yield V2F radicals, while E2F2J-A radical can also add with HO 2 radical to generate OH, CH 3, and F2CHO radicals which is also the reaction possessing the largest negative sensitivity coefficient, as shown in Fig. 7 a. This is quite similar with the C 8 H 9 radical forming through H-abstraction reactions at the benzylic site of ethylbenzene [28], which further yields either benzaldehyde or styrene. Consider styrene (alkenebenzene) and V2F (alkenefuran), ipso-additions by O radicals, OH-additions, and H-abstractions constitute the reaction pathways. All the above mentioned features suggest that the reaction routes involving the alkyl side chains of alkylfurans or alkylbenzenes are analogous, although their ring structures are different. Under low temperature conditions, low alkylbenzenes (toluene [32] and ethylbenzene [28] ) are consumed mostly by H-abstraction reactions ( > 70%) from the side chain, and H-abstractions are important to the consumption of low alkylfurans (EF, MF, and DMF) as well ( 30%). Different from alkylbenzenes, OH-addition reactions contribute a lot to the consumption of alkylfurans. For alkylbenzenes, comparison between isomers, namely dimethylbenzenes (m-xylene, o- xylene, and p-xylene) and ethylbenzene [31], indicate that only o-xylene (dimethylbenzene with adjacent methyl side chains) possesses higher reactivity than ethylbenzene in the low temperature range. Silva and Bozzelli [33] pointed out that the Fig. 7. Sensitivity analysis conducted for stoichiometric 1% fuel mixture at 16 bar and 880 K: (a) EF; (b) MF; (c) DMF. Using the modified mechanism. primary fuel radical of o-xylene (resulted from H-abstraction at the methyl side chain) would add with O 2, and this reaction scheme would yield two OH radicals and one H radical, enhancing the low temperature reactivity. By increasing and decreasing the pre-exponent factor by a factor of 2 respectively, the sensitivity

9 N. Xu et al. / Proceedings of the Combustion Institute 36 (2017) coefficient is defined as: sensitivity coefficient = ( τ 2k τ 0. 5k ) / τ k (6) Negative sensitivity coefficients represent promoting effect on auto-ignition and vice versa. Figure 7 shows the sensitivity analysis conducted at 16 bar, 880 K, 1% fuel, and φ = 1.0 for EF, MF, and DMF, respectively. The reactions of primary fuel radicals with HO 2 radicals generating OH radicals show the highest negative sensitivity for the three fuels. For EF and MF, OH-additions taken place at the C2 position are sensitive for ignition as well. At this relatively low temperature, ignition is sensitive to HO 2 and H 2 O 2 related reactions, while the important high temperature chain branching reaction H + O 2 = OH + O only gives moderate sensitivity. side chains but not OH-additions at ring sites. Although the structures of the two families are somehow different, their reactions revolving alkyl side chains are highly analogous, indicating that reaction schemes of alkylbenzenes can be referred for further mechanism construction of higher alkylfurans. Acknowledgment This work is supported by the National Natural Science Foundation of China ( , , and ), the National Basic Research Program ( 2013CB ), and the State Key Laboratory of Automotive Safety and Energy ( KF14102 ). 4. Conclusions Low to intermediate ignition delay times of EF have been measured in this work. Results show no NTC or two-stage ignition behaviors. The ignition delay times of MF at different equivalence ratios were measured. Coupling with the DMF data obtained previously [24], the comparison among ignition behaviors of EF, MF, and DMF was presented. Results show that EF is consistently more reactive than MF, while the relative reactivity of EF and DMF depends on temperature. There exists an intersection temperature below which DMF is more reactive. The phenomenon is also true for MF and DMF. The mechanism of Somers et al. [1] was adopted for simulation. As EF is only treated as a derivative of DMF in the mechanism, the reactivity of EF was under-estimated at both low and high temperatures. By modifying the EF sub-mechanism, agreement between calculated and measured ignition delay times of EF in this work as well as those of Eldeeb and Akih-Kumgeh [22] was improved. Kinetic analyses using the modified mechanism show that in the low to intermediate temperature range, OH-additions and H-abstractions are important consumption pathways for EF, MF, and DMF. Compared with DMF, the available allylic H atoms of EF and MF for abstraction are 2 and 3, while that of DMF is 6. However, the allylic C H bond dissociation energy of EF (350 kj/mol) is 10 kj/mol lower than its counterpart of MF and DMF. All of this results in H-abstractions contributing 32.5% and 36.6% to the fuel consumption of EF and DMF, while only 23.1% MF is consumed this way. H-additions are important for DMF consumption, and OH-additions contribute 59.2%, 54.9%, and 27.8% to EF, MF, and DMF, respectively. Compared with low alkylfurans, low alkylbenzenes (toluene and ethylbenzene) are mainly ( > 70%) consumed by H-abstractions at the alkyl Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi: /j.proci References [1] K.P. Somers, J.M. Simmie, F. Gillespie, et al., Combust. Flame 160 (2013) [2] Y. Roman-Leshkov, C.J. Barrett, Z.Y. Liu, J.A. Dumesic, Nature 447 (2007) [3] H. Zhao, J.E. Holladay, H. Brown, Z.C. Zhang, Science 316 (2007) [4] N. Xu, J. Gong, Z. Huang, Renew. Sust. En. Rev. 54 (2016) [5] A. Sudholt, L. Cai, J. Heyne, F.M. Haas, H. Pitsch, F.L. Dryer, Proc. Combust. Inst. 35 (2015) [6] R. Daniel, G. Tian, H. Xu, M.L. Wyszynski, X. Wu, Z. Huang, Fuel 90 (2011) [7] S. Zhong, R. Daniel, H. Xu, et al., En. Fuels 24 (2010) [8] C. Wang, H. Xu, R. Daniel, et al., Fuel 103 (2013) [9] Q. Zhang, G. Chen, Z. Zheng, H. Liu, J. Xu, M. Yao, Fuel 103 (2013) [10] M.B. Luong, G.H. Yu, T. Lu, S.H. Chung, C.S. Yoo, Combust. Flame 162 (2015) [11] H. Bendu, S. Murugan, Renew. Sust. En. Rev. 38 (2014) [12] D. Liu, C. Togbé, L.S. Tran, et al., Combust. Flame 161 (2014) [13] C. Togbé, L.S. Tran, D. Liu, et al., Combust. Flame 161 (2014) [14] L.S. Tran, C. Togbé, D. Liu, et al., Combust. Flame 161 (2014) [15] X. Wu, Q. Li, J. Fu, et al., Fuel 95 (2012) [16] X. Ma, C. Jiang, H. Xu, H. Ding, S. Shuai, Fuel 116 (2014) [17] G. Tian, R. Daniel, H. Li, H. Xu, S. Shuai, P. Richards, En. Fuels 24 (2010) [18] B. Sirjean, R. Fournet, P.A. Glaude, F. Battin-Leclerc, W. Wang, M.A. Oehlschlaeger, J. Phys. Chem. A 117 (2013)

10 332 N. Xu et al. / Proceedings of the Combustion Institute 36 (2017) [19] K.P. Somers, J.M. Simmie, F. Gillespie, et al., Proc. Combust. Inst. 34 (2013) [20] L. Wei, C. Tang, X. Man, Z. Huang, En. Fuels 27 (2013) [21] M.A. Eldeeb, B. Akih-Kumgeh, En. Fuels 28 (2014) [22] M.A. Eldeeb, B. Akih-Kumgeh, Combust. Flame 162 (2015) [23] N. Xu, C. Tang, X. Meng, X. Fan, Z. Tian, Z. Huang, En. Fuels 29 (2015) [24] N. Xu, Y. Wu, C. Tang, et al., Combust. Flame 168 (2016) [25] H. Di, X. He, P. Zhang, et al., Combust. Flame 161 (2014) [26] National Institute of Health, 2014, available at 2- methylfuran#section=vapor- Pressure. [27] National Institute of Health, 2015, available at 2- ethylfuran#section=vapor- Pressure. [28] B. Husson, M. Ferrari, O. Herbinet, S.S. Ahmed, P.A. Glaude, F. Battin-Leclerc, Proc. Combust. Inst. 34 (2013) [29] E.R. Ritter, J.W. Bozzelli, Int. J. Chem. Kinet. 23 (1991) [30] K.P. Somers, On the pyrolysis and combustion of furans: quantum chemical, statistical rate theory, and chemical kinetic modelling studies, Ph.D Thesis, National University of Ireland, Galway (2014). [31] A. Roubaud, R. Minetti, L.R. Sochet, Combust. Flame 121 (2000) [32] G. Mittal, C.J. Sung, Combust. Flame 150 (2007) [33] G. Da Silva, J.W. Bozzelli, Combust. Flame 157 (2010)