University of Birmingham Laminar burning characteristics of 2-methylfuran and isooctane blend fuels

(cid:2) We studied laminar burning characteristics of MF–isooctane blended fuels. (cid:2) The highest un-stretched ﬂame speeds occur in a range of U 1.1–1.2. (cid:2) Markstein number and burning velocities of MF20 and MF50 are presented. (cid:2) MF fraction has larger effects on burning velocities at higher temperatures. investigate the characteristics of gas-eous MF–isooctane at varying and equivalence with of in a The spherical ﬂame to determine the stretched ﬂame speeds. The un-stretched ﬂame speeds, Markstein lengths, Markstein number, laminar burning velocities and laminar burning ﬂux of MF20 (20% MF and 80% isooctane) and MF50 (50% MF and 50% isooctane) under different equivalence ratios and temperatures are then deduced and compared to MF and isooctane. The results show that the un-stretched ﬂame speeds and laminar burning velocities of MF20 and MF50 are between those of MF and isooctane under all conditions. The peak un-stretched ﬂame speeds of the blends occur in an equivalence ratio range of 1.1–1.2 at all temperatures, closer to the case of MF at higher temperatures. Both blended fuels have Markstein lengths closer to isooctane at an equiv- alence ratio lower than 1.2 at all temperatures. The burning velocities of MF50 are very close to the average values for MF and isooctane, particularly at 393 K. MF in the blended fuel presents larger effects on burning velocities at higher temperatures.


Introduction
Alternative bio-fuels present a potential pathway to provide sustainable-renewable sources and address the challenging issues of fossil fuel depletion and global warming. Previous research on alternative fuels has been wide ranging [1][2][3] and fuels that can be produced from celluloses have been considered as the most promising candidates. Ethanol has been thought of as the market-leading gasoline alternative [4][5][6] due to its mature mass production methods [7,8]. However, a new production method for furan-type fuels has been developed in recent years [9] since high efficiency was achieved in producing 2,5-dimethylfuran (DMF) and 2-methylfuran (MF) from celluloses, indicating the prospect of industrial mass production [9][10][11]. MF and DMF have become quite attractive in alternate fuel studies because of their similarities to gasoline. Table 1 shows the properties of MF and gasoline [12,13]. The properties of isooctane are also presented here because isooctane has been used as a representative component for gasoline in many studies [14].
The authors' group has studied previously the combustion and emissions of DMF and MF in a direct-injection spark-ignition (DISI) Table 1 Properties of the test fuels compared to gasoline and DMF [12,13]   single-cylinder engine [15][16][17]. The results show that DMF can produce combustion and emission qualities competitive with gasoline and MF has more significant advantages when it is applied in a DISI engine. It has been revealed that the combustion of MF is faster than that of DMF and gasoline, which is good for knock suppression, and the PM emissions of MF are lower than those of gasoline in DI mode because of the molecular oxygen content. Similar results can also be found in the work of Thewes et al. [18].
Mixing an alternative fuel with gasoline is a realistic choice for the application of an alternative fuel, because current engine designs and configurations need fewer changes than would be needed to re-optimize their fuel supply systems for a pure alternative fuel. Meanwhile, low-percentage blending may also reduce the costs for modifying fuel production, transportation and storage/ sale systems for the new fuels, which are critical issues to the commercial market. Much work has been done in alternative-blended fuel studies [19][20][21]. Rothamer and Jennings [22] compared the knocking propensity of DMF-gasoline blends with that of ethanol-gasoline blends. The results indicated that a blend with 10% DMF provided the best performance in fuel consumption. Wu et al. [23][24][25] studied the laminar flame propagation characteristics of DMF-isooctane blends at elevated pressures and temperatures and their further application in a DISI engine. Laminar flame propagation characteristics are important fundamental physicochemical properties of a fuel-air mixture for validating the chemical reaction mechanisms and gaining a better understanding of the combustion process in engines [26,27]. Little is known about the laminar flame propagation characteristics of MF and MF blended fuels. In the present paper, the schlieren photography method was used to investigate the laminar flame speed, Markstein length, Markstein number, laminar burning velocity and burning flux. This has followed our work on the similar measurement of DMF-air mixtures at elevated temperatures in a wide range of equivalence ratios compared with ethanol and gasoline [28]. As gasoline, a compound of many species, is too complex for detailed chemical reaction mechanism analyses, isooctane was used in this study as a representative component of gasoline, similarly to previous works [12,[29][30][31]]. Fig. 1 shows the schlieren system used in this study. A constantvolume vessel with two circular quartz windows (100 mm in diameter) on opposite sides, and eight heating units, one in each corner, was used for the experiment. The air-fuel mixture was heated by closed-loop controlled heating units. A gasoline direct injection (GDI) nozzle was mounted in the top cover of the vessel for fuel injection, which was driven by an ECU-computer system. A pair of electrodes was installed on two sides of the vessel. A high-voltage unit was used to supply the power for generating sparks between the electrodes.

Experimental setup
A 500-W xenon lamp was coupled with a lens group, and a pinhole was used to generate a point light source for the schlieren test. A concave mirror produced parallel light which passed through the test field in the vessel. After being reflected by a second concave mirror, the light was then cut by a knife edge at the focus for the schlieren effect. A Phantom V710 high-speed camera was synchronized with the spark timing to record the image sequences at a sampling rate of 10,000 fps and a resolution of 800 Â 800 pixels.
Compressed air was used to scavenge the burned gases in the exhaust. After flushing and before each test, the vessel chamber was opened to the ambient air until the air temperature inside the vessel stabilized at the test point. Fuels were injected into the chamber after the vessel was sealed. Then the chamber was left undisturbed for 5 min to ensure the homogeneousness and quiescence of the mixtures. Finally, the mixtures were ignited by a triggered electrode discharge, and the camera was triggered at the same time.
The tests were performed at initial temperatures of 333 K, 363 K and 393 K, with an initial pressure of 0.1 MPa. The equivalence ratios varied from 0.8 or 0.9 to 1.4. Each point of the tests was repeated at least three times.

Image processing
The schlieren images were processed using an in-house-developed MATLAB code to determine the laminar flame characteristics as in the previous work [28]. The flame radii were measured in four directions at an included angle of 45°with the electrodes to reduce the effect of quenching near the surface. The flame fronts were identified by detecting the gradients in the grayscale images (see Fig. 2). All the results in the analysis were averaged from the three tests.
In order to avoid the effect of the spark ignition disturbance [32,33], the pressure increase resulting from the burning [34] and the space confinement [35], only images with flame radii of 6-25 mm were used. This range was validated as an optimized range with the current setup by previous study [28]. The stretched laminar flame speedS n was determined by using the following equation: where r u is the flame radius and t is the time after ignition. By knowing the stretched laminar flame speed, the stretch rate, a, is calculated by [36,37]: The linear correlations between the stretch rate and flame speed are expressed by [36,37]: where S L is laminar flame speed and L b is Markstein length. S L is determined by extrapolating S n to a zero stretch rate. L b is the negative value of the gradient of the flame propagation speed against the stretch rate curve.
The laminar burning velocity (l l ) can be obtained from the equation [36,37]: where q b and q u are the burned and unburned mixture densities, respectively. Assuming the pressure is constant, the burned (q b ) and unburned gas densities (q u ) can be found from the conservation of mass equation: where n u and n b are the number of moles of reactants and products, and T u and T b are the initial and adiabatic flame temperatures.
The density ratio is defined as: The adiabatic flame temperatures were calculated using HPFLAME [38], which incorporates the Olikara and Borman equilibrium routines [39].
The flame thickness is calculated by the ratio of kinematic viscosity to laminar flame velocity via [24,25]: The Markstein number is calculated from Markstein length and the flame thickness [36]: The laminar burning flux, which reveals the eigenvalue of the flame propagation, is calculated by [36]: 4. Results and discussion

System validation
In order to validate the current measurement data, the obtained isooctane burning velocities at different equivalence ratios are compared with the data from previous studies in the literature. Fig. 3 shows that the results at 363 K in the current work agree well with the data from Bradley et al. [40] which were obtained under similar conditions. The differences at high equivalence ratios may be attributed to the different methods [25,41]. are between those of MF and isooctane. When the flame approaches the vessel wall, the shape of the flame becomes distorted with a flatter surface on the upper side due to the influence of the internal geometry [35]. Due to the quenching effect of the electrodes, all flame propagation speeds are slower along the direction of the electrodes than in the vertical direction; thus the flame is not perfectly spherical. The wrinkling near the electrodes is also attributed to the quenching effect. All the images for calculation were chosen such that significant large wrinkling on the flame front surface, which may affect the results, was avoided. The stretched flame propagation speed versus stretch rate (marked as a, in the figures) for the three fuels at different equivalence ratios under 393 K are shown in Fig. 5. As the flame expands in the vessel, the flame stretch rate reduces due to the inverse proportionality between the flame stretch rate and flame radius. The linear correlation between the flame stretch rate and the flame radius at a large stretch rate is considered as representative of the laminar flame characteristics [28]. However, in some cases, the non-linear trends also appear at a large stretch rate. For instance, in Fig. 5a and b, most of the results show a bending trend at maximum stretch rates. This is because of the unstable combustion at the stage shortly after ignition. In some lean-burn cases, e.g., the U = 0.9 case of isooctane, the problem appears more serious. In order to extract the un-stretched flame speed and Markstein length correctly, those points which are too far from the linear trend will be removed as bad data. The used points should give a deviation less than 5% for the fitting result, and the offset to the fitting line of the single points are also limited within 5%, while keeping as many data points as possible. Fig. 5ashows an example of dropping the bad data points in the fitting process and the circled points were removed for the fitting results.

Un-stretched flame propagation speed
Along with the stretched flame propagation speeds, the unstretched flame propagation speeds and the Markstein length were obtained by extrapolating the un-stretched flame propagation speeds to a zero stretch rate (a = 0) and calculating the gradient of the stretched flame propagation speed using the stretch rate slope in the linear range, respectively. Fig. 6 reveals the unstretched flame speeds of the four fuels at different temperatures and equivalence ratios. The scattering points indicate the experimental results, and the solid lines are quadratic fit curves. For all temperatures (333 K, 363 K and 393 K), MF has the highest unstretched flame propagation speeds at all the equivalence ratios, while isooctane has the lowest. The results of the two blend fuels are between those of the pure fuels as expected. In all the cases, the peak un-stretched flame speeds appear around an equivalence ratio of 1.1-1.2. Higher temperature leads to a larger difference between MF and isooctane. It is notable that the differences between MF20 and isooctane also increase when the temperature increases. For instance, at 333 K and under most equivalence ratios, the unstretched flame propagation speeds of MF20 are about 0.1 m/s faster than those of isooctane, while the differences increase to about 0.5 m/s at 393 K. The results for MF50 are generally in the middle of MF and isooctane.

Markstein length, flame thickness and Markstein number
The Markstein length indicates the influence of stretch rate on flame propagation speed, which characterizes the diffusion-thermal instability [26,42]. The asymptotic theory [43] points out that the Markstein length depends on the Lewis number of the fuel for a lean mixture, or that of oxidizer for a rich mixture. Normally, the Markstein length decreases with an increase of equivalence ratio for heavy hydrocarbon-air mixtures, while the trend is opposite for light hydrocarbon-air mixtures [44]. Fig. 7 shows the Markstein lengths of the four fuels at different temperatures and equivalence ratios. The Markstein lengths decrease with an increase of equivalence ratio as a trend, which is consistent with the theory as in [44] since both MF and isooctane have more than five carbon or/and oxygen atoms per molecule.
The Markstein lengths of MF are significantly smaller than those of isooctane at equivalence ratios lower than 1. differences between them are much smaller at equivalence ratios higher than 1.0. The data for the two blended fuels are between those for the pure fuels at equivalence ratios below 1.2; both are however closer to the data of isooctane indicating its dominating role here. Positive Markstein lengths indicate that the flame speed decreases with an increase in the stretch rate, while a negative Markstein length indicates that the flame speed increases with an increase in the stretch rate. Thus in the range of equivalence ratios below 1.2, all the fuels have lower flame speeds when the stretch rate increases. Bradley et al. [40] pointed out that if the Markstein length is larger than 1.5, the flame will be initially stable until a critical flame radius is reached. This means isooctane and the two blended fuels have better initial flame stabilities than MF in a lean burning condition. With respect to temperature, the differences between the Markstein lengths of MF at 363 K and 393 K with the same equivalence ratios are quite small; while the Markstein lengths of MF at 333 K are larger than those at the other two temperatures with low equivalence ratios (0.7-1.0). Negative values of the Markstein length under 393 K appear at rich equivalence ratios and this indicates that the MF flame is more unstable at these temperatures under rich conditions. Correspondingly, MF20 and MF50 also have similar trends. There is no significant difference between the two blended fuels under most conditions, and the temperature has no significant effect on the trends of the blended fuels.
Previous theory [26,45] indicates that there are mainly two kinds of flame surface instabilities acting on the flame front in conditions similar to those in the current work: the diffusion-thermal instability and the hydrodynamic instability. The diffusion-thermal instability is characterized by the Markstein length. The hydrodynamic instability, which is induced by the density transition across the flame front, is characterized by the flame thickness and the density ratio. A decrease in the flame thickness indicates the promotion of this kind of instability. Fig. 8 shows flame thickness versus equivalence ratio for all of the test fuels at different initial temperatures. The trends revealed by the data are generally the same, showing a range of lower flame thickness near an equivalence ratio of 1.0, which indicates higher instability. Isooctane and MF have the highest and lowest values, respectively. The results of MF20 and MF50 are between those of the two pure fuels. For all four fuels, the flame thickness data is not sensitive to the variation of initial temperature except for some points at the lean and rich ends. This means the initial temperature is not the most important parameter affecting flame thickness in the test range, and thus the Markstein number is mainly determined by the Markstein length for the same fuel.
The Markstein number, which characterizes the effect of local heat release on the flame morphology and the flame front curvature, quantifies the response of a laminar flame to stretch and can be used to indicate the stability of laminar and turbulent flame fronts. Fig. 9 gives the values of the Markstein number at different initial temperatures. The Markstein numbers are generally decreasing with an increase in equivalence ratios, similar to the Markstein lengths. It is shown that the results of MF20 and MF50 are between those of the two pure fuels at most equivalence ratios and are closer to isooctane. Generally, the ranking of the fuels in terms of Markstein number for a given equivalence ratio does not change with temperature, except that some points at high equivalence ratios are very close. A cross-over point appears at equivalence ratio 1.1 under the initial temperature of 363 K, as shown in Fig. 9b, which also proves that the Markstein numbers of blended fuels depends on those of the pure fuels. The effect of temperature on the Markstein length for different fuels cannot be represented by a simple tendency [28], nor can the results for the Markstein number. The instability at high equivalence ratios may also result in larger errors. Therefore, it is quite difficult to discuss the ranking of fuels under different initial temperatures at a given equivalence ratio. More investigations are needed to provide a more detailed explanation in future studies.

Laminar burning velocities and burning flux
Laminar burning velocity is the speed at which the flame advances into the unburned mixture. Fig. 10 shows the laminar burning velocities at the equivalence ratios with different initial temperatures. The laminar burning velocities of MF, MF20 and MF50 under varying initial temperatures have peaks near the equivalence ratio of 1.1, which is correlated to the state of the un-stretched flame speeds. Among all the fuels, MF has the highest laminar burning velocity under all conditions. The burning velocities of the two blended fuels are between those of the pure fuels, which is also similar to the trend shown by the un-stretched flame speeds. The burning velocities of MF50 near the equivalence ratio of 1.1 are quite close to the average values of MF and isooctane. Particularly for the cases at 393 K, the values for MF50 are close to the average values of MF and isooctane at all tested equivalence ratios.
Previous studies have pointed out that the laminar burning velocity is strongly related to the equivalence ratio and initial temperature of the reactants [46][47][48]. In Fig. 10 the laminar burning velocities for all the fuels increase when the initial temperature increases. For MF20 and MF50, the laminar burning velocities near the peaks at 393 K are about 0.05-0.07 m/s faster than the results at 363 K, and about 0.15-0.18 m/s faster than the results at 333 K. At higher equivalence ratios, the differences between blended fuels are smaller at 333 K, and the laminar burning velocities of the blended fuel are closer to those of isooctane, whereas at 393 K, no such phenomena are observed. This means MF has a greater effect in determining the laminar burning velocity in blended fuels at higher initial temperatures. As mentioned in a previous section, the work of Wang et al. [17] found the combustion duration of MF is significantly shorter than that of gasoline in a DISI engine. The burning velocity data here clearly proves the conclusion that MF has faster combustion; which results in a higher indicated thermal efficiency than gasoline. At high temperatures, MF in the blended fuels significantly promotes the burning velocity because of its high chemical activity. This offers fundamental support to improving performance by adding MF to SI engine fuel. Fig. 11 shows the burning flux versus equivalence ratio for MF at different initial temperatures. At each temperature and equivalence ratio, laminar burning flux is in the order of MF, MF50, MF20 and isooctane from the highest to the lowest value, corresponding to the order of the burning velocities. The peak values of the burning flux of all the fuels at three temperatures appear between equivalence ratios of 1.0 and 1.2. With respect to temperature, the burning flux of all the fuels increases with an increase in the initial temperature. The burning flux of the two blended fuels varies in a range of 0.45-0.62 near the stoichiometric state at all the initial temperatures.

Conclusions
In this study, the laminar combustion characteristics of 2-methylfuran (MF) and MF-isooctane blended fuels was investigated using high-speed schlieren photography at elevated temperatures (333 K, 363 K and 393 K) and varying equivalence ratios (U = 0.7-1.4) under 0.1 MPa initial pressure in a constant volume vessel. The characteristics of the blended fuels were compared to the cases of MF and isooctane. The following conclusions are drawn from the results: 1. The un-stretched flame speeds of MF20 and MF50 are between those of MF and isooctane under all the tested conditions. The highest un-stretched flame speeds of all four fuels occur in an equivalence ratio range of 1.1-1.2 at all tested temperatures. The peak un-stretched flame speeds of the blended fuels are closer to the data of MF at higher temperatures. 2. The blended fuel flames are less stable than isooctane's flame but more stable than MF's flame at equivalence ratios lower than 1.0 at all the tested temperatures. The difference between the flames of MF20 and MF50 is not significant. The Markstein numbers show similar trends as the Markstein lengths. Both blended fuels have the Markstein number closer to that of isooctane at equivalence ratios lower than 1.