1

SINGLE PARTICLE IGNITION F BIOMASS FUELS: EXPERIMENTS AND MODELING

Gonçalo Simões

IDMEC, Mechanical Engineering Department, Instituto Superior Técnico, Universidade de Lisboa, Lisboa, Portugal

Abstract

The current work focused on the study of the ignition of single-particles of pulverized solid fuels. The experimental tests were carried out in an optical flat-flame McKenna burner that produced a confined laminar flow of combustion products with a specific composition and temperature. Particles of pine bark, wheat straw, vine, kiwi and sycamore branches were injected in different atmosphere conditions in the size range of 212 - 224 µm. The ignition of single-particles was analyzed for three different oxygen concentrations in the ignition zone, namely 3.5 %, 5 % and 6.5 % (dry concentration). The ignition mode and the ignition delay time were evaluated through the analysis of images obtained by means of a CMOS high-speed camera. In order to understand better the phenomena involved in the ignition process an additional numeric study was done. A time-dependent Euler-Lagrange approach was taken where the fluid phase is treated as a continuum by solving the Navier-Stokes, energy and the species transport equations, and the discrete solid phase followed an user-defined-function (udf), which considered sequentially the phases of: evaporation, devolatilization, char combustion. Besides the conditions studied in this work, three other conditions (tested in a previous work [42]) with different temperature in the ignition zone, namely 1500 K, 1650 K and 1800 K, were used in order to numerically evaluate the effect of the temperature on the ignition process. A benchmark case was selected to validate the model. The model was then used to identify the dominant phenomena on the ignition delay time of the solid biomass particles.

The experimental results showed that (i) some biomass particles, namely the wheat straw and vine branch particles under certain conditions, despite their high content of volatiles presented surface ignition, (ii) this incidence did not seem to show a relation with particle composition or its shape/morphology, however it increased with increasing oxygen concentrations, (iii) even though the ignition delay times decreased slightly with increasing oxygen concentrations, the effect of this parameter was marginal. For the selected benchmark case, the numeric results showed a good agreement with the experimental data displaying that even a simple model, as the one used, is able to reproduce in a reasonable way the combustion process of the single-particles, particularly the ignition delay time. The numeric ignition times followed the same experimental trends: (i) marginal effect of the oxygen concentration, (ii) decreasing ignition delay times with increasing temperatures. Finally, the numerical results showed that heating and drying time dominate the particle ignition delay time.

Keywords: biomass, single particle, high-speed imaging, ignition delay time, numeric modeling.

1. Introduction

In the context of the current world energy crisis and increasing CO2 emission constraints, alternatives to conventional coal combustion, such as coal-biomass co-firing, are becoming more common. Co-firing arises as an alternative that allows the use of biomass fuels in existing power plants, thus making it an attractive technology. Biomass is a highly reactive fuel and has a high volatile matter content, resulting in lower ignition temperatures for coal-biomass blends than for pure coal [1], helping with flame stability. Additionally, emissions of SO2 and NOx are typically reduced [2]. Despite some successful long-term demonstrations [3], there are still many open questions on the combustion process, particularly in the area of particle ignition behavior.

The ignition process has a critical influence over flame stability, and is, thus, crucial in the design of boilers. Depending on the conditions, ignition can result from the interaction between the volatiles released from the particles and the oxidizing gas in the gas-phase, i.e., homogeneous ignition, or it can result from direct oxidation of the particle surface, i.e., heterogeneous ignition [4]. Most studies on ignition focus on ignition mode and ignition delay time and the parameters that affect them, namely, size and composition of the particle, and combustion atmosphere (temperature and composition). Despite the importance of the ignition process, its definition is ambiguous and depends on the diagnostic methods used. Early studies of Howard and Essenhigh [5] defined the ignition event based on the percentage of loss of carbon and volatiles during the early moments of combustion; however, since then most works have used optical diagnostics to characterize this event. Molina and Shaddix [6] defined the onset of ignition of bituminous coal diluted jets based on the CH* chemiluminescence. In a more recent work, Shaddix and Molina [7] measured the ignition delay time of individual coal particles by capturing the visible light signal emitted by igniting particles, and defined the ignition onset as the point at which 60% of the maximum luminosity intensity was reached. Yuan

2

et al. [8] used a similar criterion to characterize the ignition of streams of coal particles. In their work, the residence time when the intensity reached 10% of the maximum peak intensity value was considered to be the ignition delay time.

Studies on the ignition of biomass fuels are very scarce, with most works focusing on coal [4-10]. Ignition studies on coal have shown that different types of coal have completely distinct ignition and combustion behaviors [4,9], but generally ignition delay times (i) decrease with increasing temperature [8], (ii) decrease with increasing oxygen concentration [4,6,7,9], and (iii) decrease with decreasing particle size [10]. Even though the behavior of coal is relatively well characterized some differences may arise when the transition to biomass fuels is made. In fact, Riaza et al. [11] studied the combustion of four biomass fuels under different atmosphere conditions and found significant differences between biomass fuels and the coals previously studied by [4,9].

In this context, the main objective of this study is to characterize the single particle ignition behavior of biomass fuels, particularly for temperatures higher than those reported in literature, under atmosphere conditions much closer to the ones that occur inside industrial boilers. The tests were performed in an optical flat flame McKenna burner and ignition events were captured with a CMOS high-speed camera. In this study, the ignition behavior is characterized in terms of ignition mode and ignition delay time. A second objective is to implement a numeric model to replicate the experimental results. Once validated this model allows to access what are the dominant phenomena in the ignition delay time.

2. Experiment

2.1. Experimental setup

Figure 1 shows a schematic of the experimental setup used. It consists of a biomass feeding unit, a McKenna flat flame burner, a gas feeding system and an image acquisition system. The biomass feeding unit consists of a rotameter (to measure the transport air flow rate), a 10 ml syringe and a vibrating motor. The biomass particles stored in the syringe were fed (by gravitational force) into the stream of transport air and injected upward through the central hole of the burner (I.D. 1.55 mm) into the ignition zone. The vibrating motor avoided the clogging of particles in the syringe hole and ensured a low feeding rate.

Figure 1. Schematic of the experimental setup.

The Mckenna flat flame burner consists of a stainless-steel cylinder enveloping a water-cooled bronze

porous sintered matrix of 60-mm diameter. Two mass flowmeters allowed the control of the methane and primary air flow rates to the burner. In addition, cooling water was fed to the burner through copper tubes (labelled 1 and 2 in Fig. 1). Above the burner, a high-grade fused quartz of I.D. 70-mm, height of 500-mm, and thickness of 2-mm confined the flow and avoided the entrainment of ambient air, while providing optical access.

The image acquisition system consists of a CMOS high-speed camera Optronis CamRecord CR600x2, and a diffuse led backlight. The camera was equipped with two lenses: a 2x tele-converter and a Nikkor 60 mm f/2.8D. The position of the camera was adjusted so that its axis of the camera was perpendicular to the axis of the burner, and the ignition zone was in frame. A target with millimeter marks was used for reference calibration before and after each testing period. Table 1 shows the camera settings used.

Probe

CH4 N2/O2

High-speedcamera

Particlefeeder

OpticalMcKennaflat flame

burner (1)(2)

(1) - Water in(2) - Water out

Rotameter

Flowmeters

3

Table 1. Imaging settings. Parameter Setting Aperture 2.8-5.6 Exposure time [s] 1/15000 FPS 3500 Resolution [pix] 250×452

The ignition mode was determined by direct visual observation of the recorded images, whilst ignition

delay time was measured by image post-processing. The ignition criterion used herein was based on the visible light signal following the procedures used by Yuan [8]. This technique is simple to use and the visible light signal is strong and easy to capture. Although soot and char emissions are included in the visible light signal captured this has proved to be a good indicator of ignition [7,8]. In this work, ignition was considered to occur when 15% of the maximum pixel luminosity was reached for either volatile homogeneous or heterogeneous ignition. A Matlab routine was used to calculate the ignition delay time. The code reads the data from the recorded frames, eliminates the background noise, and iidentifies the frame in which the pixel intensity value was closest to 15% of the maximum luminosity intensity. The code ‘tracks’ the position of the particle, and uses the frame-rate to compute the frame-to-frame velocity of the particle. Finally, these velocities and the position of the particle at the ignition frame are used to calculate the ignition delay time. For each experimental condition a minimum of 20 single particle ignition events were analyzed.

2.2 Test Conditions

Table 2 shows the operating conditions used. The fuel and oxidizer flow rates and burner thermal input were varied to yield different operating conditions. The fuel and oxidizer flow rates were varied to yield three different equivalence ratios and, hence, different oxygen concentrations for the same temperature, O1-O3. Throughout the tests the transport air flow rate was kept constant and at the lowest flow rate that ensured particle feeding; specifically, 0.14 dm3/min.

Table 2. Operating Conditions.

Parameter O1 O2 O3 Thermal input [kW] 1.0 1.1 1.4 Equivalence ratio ϕ [-] 0.79 0.71 0.67 O2 concentration [dry vol. %]

3.5 5.1 6.5

Gas temperature [K] 1680 Figure 2 shows the measured gas temperature and O2 concentration profiles above the burner along its

axis for the conditions O1-O3. Local mean temperature measurements were obtained using 76 µm diameter fine wire platinum/platinum: 13% rhodium (type-R) thermocouples and local mean O2 concentration measurements were obtained with the aid of a probe and a paramagnetic analyser. Note that Fig. 2 includes the values of the gas temperature corrected for radiation losses following the method described in Ref. [12], and that the values of temperature and O2 concentration included in Table 2 correspond to averages of those measured in the ignition zone.

Figure 2. a) Measured gas temperature, and b) corrected gas temperature for tests O1-O5 following the

method described in [12]. c) Measured oxygen concentration for tests O1-O3. Vertical bars represent 98% confidence statistical error.

Table 3 lists the properties of the solid fuels used in this study, which included five biomass fuels, wheat

straw (WS), kiwi branches (KB), vine branches (VB), sycamore branches (SY), and pine bark (PB). The fuels were ground and sieved to the range of 212-224 µm. All biomass fuels were burned under conditions O1 to

4

O3 (see Table 2) to study the effects of the O2 concentration and biomass composition on the ignition behavior.

Table 3: Properties of the solid fuels.

Parameter WS KB VB SY PB BC Proximate analysis (wt. %, as fired)

Volatile matter 64.9 71.7 70.2 73.5 58.9 37.6 Fixed carbon 11.5 17.6 17.2 16.0 25.9 58.8 Moisture 8.9 8.3 10.5 9.3 13.9 1.6 Ash 14.7 2.4 2.1 1.2 1.3 2.0 Ultimate analysis (wt. %, as fired) Carbon 39.4 48.2 49.3 47.6 47.8 76.9 Hydrogen 5.2 6.4 6.19 5.98 4.3 5.1 Nitrogen 0.5 0.72 0.67 0.54 0.3 1.6 Sulfur < 0.02 < 0.16 < 0.15 < 0.16 < 0.02 0.7 Oxygen 31.3 44.52 43.69 45.72 32.4 12.1 Low heating value (MJ/kg) 18.8 17.9 17.7 18.1 17.1 32.7

A particle morphology analysis was performed using scanning electron microscopy (SEM). A quantitative

analysis of the aspect ratio of each type of fuel was performed using a minimum of 50 particles per sample. The SEM images and the results of this analysis are presented in Fig. 3. The aspect ratio was simply calculated as the ratio between the largest length of the particle and the corresponding perpendicular length. This analysis revealed that the pine bark particles had a quasi-spherical shape (aspect ratio < 2) and presented a homogeneous size distribution. The remaining biomass fuels were typically more elongated with values of aspect ratios generally close to 4, and close to 5 for the sycamore branch particles. The latter presented higher heterogeneity in the size distribution when compared to the samples with more spherical particles.

Figure 3. SEM images of tested fuels, with the respective calculated aspect ratios (AR).

3. Numeric work

3.1. Overview

The commercial CFD software Fluent v.14.5 was used to model the ignition and combustion of the single-particles in the McKenna burner. A time-dependent Euler-Lagrange approach was taken. In this approach, the fluid phase was treated as a continuum by solving the Navier-Stokes, energy and the species transport equations. The dispersed phase was solved by tracking the particles through the calculated flow field, and solving mass and energy balances. The dispersed phase exchanged momentum, mass and energy with the fluid phase.

The simulations were carried in two steps. Firstly, a steady-state simulation was performed for the gas phase to set the environment conditions in which the particle was injected. After that, the particle was introduced in the domain in a transient simulation. A time step of 1e-6 was used with 100 iterations per time step. The coupled algorithm was used for pressure-velocity coupling with second-order spatial and temporal discretization. This solver belongs to the family of pressure-based solvers, which use a pressure equation (derived from the continuity and momentum equations) to ensure the conservation of mass.

The physical domain was represented by a 2D-axy-simmetric computational domain. The dimensions of the domain followed the dimensions of the burner. Figure 4a shows the geometry of the domain, where the x-axis corresponds to the vertical direction. The top 80 mm above the burner were modeled.

5

Figure 4. Geometry, mesh and boundaries of the computational domain.

Figure 4b shows the mesh used with labels for the boundaries of the domain. The computational grid

consisted of approximately 8000 rectangular cells and was refined so that a grid-independent unsteady solution was obtained. The mesh was structured and non-uniform, with a larger cell density close to the axis of the burner, where the reactions occurred. The flat flame sheet was not modeled in this work. Instead, a flow of hot combustion products gas was injected into the domain from the top of the burner (boundary 3 in Fig. 4b). Some simplifications were made in boundaries 2 and 4, respectively corresponding to the transport air tube wall and quartz tube wall, namely by considering that these walls were at a constant temperature (close to the neighbouring gas temperature). These boundaries are far from the region where the reaction will occur and have a marginal influence on the final result. Finally, Table 4 summarizes the boundary conditions (BC) considered.

Table 4: Summary of the boundary conditions. Bound

ary Region Type of BC Thermal BC

1 Transport air inlet Velocity-inlet Fixed temperature

2 Transport air tube wall Wall Fixed temperature

3 Hot combustion product gas inlet Velocity-inlet Fixed temperature

4 Quartz tube wall Wall Fixed temperature

5 Top border of the domain Pressure-outlet Null temperature gradient

6 Axis Axis Null heat flux In section 3.2 the main phases implemented in the user-defined function are briefly described.

3.2. Discrete-phase

Each particle was transported in a Lagrangian frame of reference where its position, velocity, mass, and diameter were evolved. The position of the particle in the continuous phase is computed by integrating the force balance on the particle, which is written in a Lagrangian reference frame:

𝜕𝑣 !𝜕𝑡

= 𝐹!(𝑣 − 𝑣 !) +𝑔 (𝜌! − 𝜌)

𝜌! (3.1)

where 𝑣 !, 𝜌! and 𝑣 , 𝜌 represent the respectively velocity and density of the particle and gas phases. 𝑔 is the gravity acceleration and 𝐹!(𝑣 − 𝑣 !) is the drag force per unit mass. Heat and mass exchange between the particle and the gas phase were governed by a sequential model of evaporation, devolatilization and char combustion. A consequence of this is that the model was not able to predict the occurrence of heterogeneous ignition. This is a strong limitation, however ignition of biomass particles generally occurred in the gas-phase, and thus this can be an interesting first approach. The moisture content evolution is given by:

𝑑𝑚!!!

𝑑𝑡= 𝑘!

𝑝!!!,!"#𝑅𝑇!

−𝑝!!!𝑅𝑇

𝐴!𝑀!,!!! (3.2)

where 𝑘! is the mass transfer coefficient of steam into air, 𝑝!!!,!"# is the saturation pressure of water at particle temperature, and 𝑝!!! is the partial pressure of water in the gas phase. The devolatilization was modeled based on a single kinetic rate law [52]:

6

𝑑𝑚!

𝑑𝑡= 𝐴!𝑒𝑥𝑝

−𝐸!,!𝑅𝑇!

𝑚! (3.3)

where 𝐴! and 𝐸!,! are the pre-exponential factor and activation energy. The char combustion model has no influence over the ignition according to the model followed. For this reason, and since the main focus of this work was to model the ignition process, a simple diffusion-limited model was used. The energy balance across the surface of a particle, assuming that the temperature within a particle is uniform is given by,

𝑑𝑇!𝑑𝑡

=−𝐴!𝑚!𝑐!,!

ℎ! 𝑇! − 𝑇! + 𝜀𝜎(𝑇!! − 𝑇!!) + S! (3.4)

where 𝑇!, 𝑇! and 𝑇! are the particle, wall, and gas temperatures respectively. 𝑐!,!, 𝑚!, 𝐴! and 𝜀 are the particle heat capacity, mass, surface area and emissivity respectively, 𝜎 is the Stefan–Boltzmann constant, ℎ! = 𝑁𝑢 𝑘/𝑑! is the convective heat transfer coefficient with 𝑁𝑢 = 2 + 0.6𝑅𝑒!/!𝑃𝑟!/! , and S! is the temperature source term due to vaporization, S!,!"#$, and heterogeneous reactions, S!,!!",

4. Experimental results

4.1. Ignition process

Figure 5 shows selected signal intensity profiles and images of the ignition process where it is possible to identify the different steps of the ignition process and the early stages of the particle combustion.

Figure 5. Selected signal intensity profiles and images of the ignition process. a) Visible light signal intensity maximum measured in each frame normalized by the maximum recorded signal intensity; b) gas-phase

ignition of a particle of a kiwi branches; c) surface ignition of a particle of vine branches, both under condition O3 in the size range 212-250 µm.

4.2. Ignition mode

Figure 6 presents the relative frequency of the gas-phase ignition for the operating conditions O1-O3. Ignition of the biomass fuels occurred mostly in the gas-phase for all test conditions with a few exceptions as discussed below.

7

Figure 6. Relative frequency of the gas-phase ignition for operating conditions O1-O3.

Ignition of the tested biomass particles occurred dominantly in the gas-phase as seen in Fig. 6. In fact,

particle surface ignition was only observed for particles of wheat straw and vine branches with a slight tendency of increased frequency when increasing the gas oxygen concentration. These findings reinforce that the biomass type has a minor influence on the ignition mode, considering the high temperature condition. In the case of particles of wheat straw and vine branches, the increased frequency of surface ignition with increasing oxygen concentration is consistent with the definition of critical particle size of Howard and Essenhigh [13] in which increasing the oxygen partial pressure results in an increase of the critical particle diameter. In these conditions (constant temperature and varying oxygen concentration), the ignition mode did not show any correlation with the particle shape

4.3. Ignition delay time

Figure 7 shows the ignition delay times for the operating conditions O1-O3 for all biomass fuels in the size range 212-224 µm. It is observed that for a given O2 concentration all biomass fuels present very similar ignition delay times, revealing that at high temperatures the fuel composition (cf. Table 3) has little impact on the ignition delay time. These results are consistent with the observations of Riaza et al. [11], who concluded that, unlike the coal type, the behavior of the biomass seemed to be more unified in terms of ignition behavior.

Figure 7. Ignition delay times for the operating conditions O1-O3 for all biomass fuels in the size range

212-224 µm. Vertical bars represent 98% confidence statistical error. In addition, Fig. 7 shows a slight decrease in the ignition delay time with an increase in the O2

concentration. This is consistent with the observations for different coal types [4,6,7,9]. Following Shaddix and Molina [7], this behavior can be explained by the increasing reactivity of the local fuel-oxidizer mixture with an increase in the O2 concentration. In the case of the particles of wheat straw and vine branches the increasing surface ignition fraction may contribute to the decay in the ignition times, which, however, does not seem to be the dominant factor, since particles of kiwi branches show a similar decay.

5. Numeric results

5.1. Benchmark case setup and comparison with experimental results

Pine bark was selected as the benchmark fuel to validate the model. Besides the three oxygen concentrations tested in this work O1-O3, three additional conditions with different mean temperature in the ignition zone, taken from a previous work, were analyzed, T1-T3. Figure 8 shows the axial profiles of oxygen concentration and temperature along the centerline of the burner, respectively for conditions O1-O3 and T1-T3. Generally, a good agreement is observed.

0

5

10

15

20

25

30

Ignitiondelaytim

e(ms)

WS KB VB SY PBBiomass type

O1 O2 O3

8

Figure 8. Numeric and experimental profiles of a) oxygen concentration and b) temperature along the

centerline of the burner, respectively for conditions O1-O3 and T1-T3. Similarly to what was done by Vascellari et al. [14], the numeric ignition criterion was based on the fraction

of OH, as a OH* representative. Consistently with the experimental analysis, the numeric ignition was considered to occur when 15 % of the maximum OH mass fraction was reached.

Figure 9 shows the contours of temperature and OH mass fraction around the particle between t = 21 ms and t = 30 ms (close to ignition) and images of an ignition event from a pine-bark particle under test condition 3. In Fig. 8 it is possible to see the relatively slow dissociation reactions by looking at the slowly increasing concentration of OH, which then shoots and quickly reaches OH concentrations close to the maximum, with a simultaneous increase in gas temperature around the particle.

Figure 9. Contour plots of temperature and normalized OH mass fraction in the vicinity of the injection hole

and close to the axis of the burner between t = 21 ms and t = 30 ms. Figure 10 shows the experimental and computational ignition delay times, as computed by the criterion

above stated.

Figure 10. Experimental and numeric ignition delay times for conditions a) O1-O3 and b) T1-T3. The

vertical bars represent 98 % confidence interval. At the high temperatures tested the numeric ignition occurred just briefly after the onset of devolatilization,

not varying significantly with the different oxygen concentrations tested. This reduced influence of the oxygen concentration over the ignition delay time can be observed in Fig. 10a. An increase in the gas temperature, on

9

the other hand, results in an anticipation of the onset of the devolatilization, which in turn reflects in a shorter ignition delay time. Thus, in these simulations the surrounding gas temperature shows a much more intense effect on the ignition time (see Fig. 10b). In general, a good agreement between the experimental and numeric results was achieved, especially taking into account the complexity of the studied phenomena.

5.2. Analysis of the dominant phenomena on the ignition delay time

Figure 11 shows the ignition delay times decomposed by their constituent phases: sequentially heat-up, drying, heat-up, and devolatilization up to ignition.

Figure 11. Decomposition of ignition times in heat-up, drying, heat-up and devolatilization up to ignition for

test conditions a) O1-O3 and b) T1-T3. There is a clear contrast between Fig. 11a and b. While the oxygen concentration has practically no effect

in any of the phases, the temperature of the atmosphere seems to affect all of them. This can be seen in the decreasing duration of these phases with increasing gas temperature. In particular it can be seen that the heat-up and drying phases dominate the ignition delay time, consisting of up to 80-90% of the total ignition delay time in the tested cases. Properties of the particle, which have a strong influence during these phases such as the density, specific heat and moisture content of the particle, will thus have a strong impact on the ignition delay time. It then becomes interesting to analyse the effect of the moisture content.

The effect of this parameter was analysed by considering a particle with the same properties of pine bark, namely the specific heat and density and with a variable moisture content. This study was done only for the three different temperature test conditions since, as seen in Fig. 11, the oxygen concentration had a marginal effect either in the drying or ignition delay times. Figure 12 shows the drying times as a function of the particle moisture content for tests T1-T3.

Figure 12. Drying times as a function of the moisture content of the fuels for conditions T1-T3.

In any of the three conditions it can be seen that, as expected, drying times increase with the moisture

content. This growth is very rapid for low moisture fractions, but quickly reaches a slower steadier pace, which is seen in the higher moisture percentages.

The drying phase is governed by the moisture concentration gradient between the surface of the particle and the surrounding atmosphere, which promotes the transport of this water through diffusion (the model is able to ensure that high evaporations do not occur in saturated environments). On the other hand the moisture concentration at the particle surface increases with increasing particle temperature, exhibiting a exponential behavior. In the case of particles with a lower moisture content, given the low quantity of water to evaporate, the drying phase occurs in a short period in which the particle does not have time to heat-up significantly. In this case the moisture content at the particle surface does not reach high values. In particles with higher moisture contents the drying phase occurs in a longer period, due to the additional mass to evaporate. In this case the particles have more time to heat-up and the drying phase will occur at higher temperatures, when compared with lower moisture particles. In this condition there is an increase in the drying rate due to the increasing moisture concentration at the particle surface. Higher moisture particles reach higher drying rates than those with a lower moisture content.

10

5. Conclusion

The current work focused on the experimental and numerical study of the ignition of single-particles of pulverized solid fuels. The experimental study extended previous work, in order to include a broader diversity of biomass fuels and the effect of atmosphere composition, more specifically the oxygen concentration. Furthermore, a numeric strategy using a commercial CFD software to model the combustion behavior, and more specifically ignition, of single biomass particles was validated. Subsequently, the dominant phenomena on the ignition delay time were numerically investigated. The main conclusions of the experimental study are: (i) some biomass particles, namely the wheat straw and vine branch particles under certain conditions, despite their high content of volatiles presented surface ignition; (i) this incidence did not seem to show a relation with particle composition or its shape/morphology, however it increased with increasing oxygen concentrations; (iii) even though the ignition delay times decreased slightly with increasing oxygen concentrations, the effect of this parameter was marginal. The main conclusions of the experimental study are: (i) for the selected benchmark case, the numeric results showed a good agreement with the experimental data showing that the model is able to reproduce in a reasonable way the combustion process of the single-particles, particularly the ignition delay time. (ii) the numeric ignition times followed the same experimental trends: marginal effect of the oxygen concentration, and decreasing ignition delay times with increasing temperatures. (iii) the model was then used to investigate the effect of the fuel moisture content. As expected, drying times increased with increasing moisture and decreasing atmosphere temperature. The drying phase was seen to represent a significant fraction of the ignition delay, showing that an increase in the moisture is likely to be reflected in greater ignition delay times.

References

[1] J. Riaza, M.V. Gil, L. Álvarez, C. Pevida, J.J. Pis, F. Rubiera, Energy 41 (1) (2012) 429-435.

[2] J. Smart, R. Patel, G. Riley, Combust. Flame 157 (12) (2010), 2230-2240.

[3] D.A. Tillman, Biomass Bioenerg. 19 (6) (2000) 365-384.

[4] R. Khatami, C. Stivers, Y. Levendis, Combust. Flame 159 (12) (2012) 3554-3568.

[5] J.B. Howard, R.H. Essenhigh, Combust. Flame 9 (3) (1965) 337-339.

[6] A. Molina, C. Shaddix, Proc Combust. Inst. 31 (2007) 1905-1912.

[7] C.R. Shaddix, A. Molina, Proc. Combust. Inst. 32 (2009) 2091-2098.

[8] Y. Yuan, S. Li, G. Li, N. Wu, Q. Yao, Combust. Flame 161 (9) (2014) 2458-2468.

[9] Y.A. Levendis, K. Joshi, R. Khatami, A.F. Sarofim, Combust. Flame 158 (2011) 452-465.

[10] Y. Liu, M. Geier, A. Molina, C. Shaddix, Int. J. Greenh. Gas Control 5 (1) (2011), S36-S46

[11] J. Riaza, R. Khatami, Y.A. Levendis, L. Álvarez, M.V. Gil, C. Pevida, F. Rubiera, J.J. Pis, Biomass Bioenerg. 64 (2014) 162-174.

[12] C.R. Shaddix, in: 33rd National Heat Transfer Conference NHTC'99, American Society of Mechanical Engineers, Albuquerque, NM, USA, 1999, p. 1150.

[13] J.B. Howard, R.H. Essenhigh, Proc. Combust. Inst. 11 (1967) 399-408.

[14] M. Vascelari, H. Xu, C. Hasse, Proc. Combust. Inst. 34 (2013).