Aravind B.a,
Ratna Kishore Velamatib,
Aditya P. Singha,
Y. Yoonc,
S. Minaevd and
Sudarshan Kumar*a
aDepartment of Aerospace Engineering, Indian Institute of Technology Bombay, Powai, Mumbai, India. E-mail: sudar@aero.iitb.ac.in; Tel: +91-22-2576-7124
bDepartment of Mechanical Engineering, Amrita School of Engineering, Amrita Vishwa Vidyapeetham, Amrita University, Coimbatore, India
cSchool of Mechanical and Aerospace Engineering, Seoul National University, Seoul, South Korea
dFar Eastern Federal University, Vladivostok, Russia
First published on 17th May 2016
Detailed numerical studies through unsteady simulations with detailed hydrogen chemistry have been reported for premixed H2–air flames in straight microtubes to understand the role of flame-wall coupling and its effect on flame dynamics for a range of wall heat transfer conditions. Depending on the wall heat transfer conditions, and tube diameters, varying flame shapes were observed. These flame modes are represented with flame shape angles and corresponding flame shape is correlated to wall heat transfer conditions. It has been observed that an increase in wall heat transfer coefficient, h, though it increases the heat loss from a propagating flame, does not necessarily lead to a monotonic decrease in flame propagation speed. A transition regime, where the propagating flame changes its shape, has been identified. The variation of mass flux in the vicinity of the propagating flame has been used to gain a better understanding of flow-redirection and its impact on flame shape and flame propagation speed for premixed H2–air mixtures.
Maruta et al.5 reported studies on flame stability in externally heated straight micro tubes. They observed that flame instabilities occur for moderately low velocities resulting in a FREI (flames with repetitive extinction and ignition) propagation mode. Subsequently, Fan et al.10,11 reported the formation of various rotating and oscillating flame patterns in radial micro channels for a range of operating conditions with premixed methane–air mixtures. Khandelwal et al.7 and Akram and Kumar8 reported detailed studies on flame propagation in externally preheated diverging channels. Various stable, unstable and partially stable flame propagation modes were observed for a range of mixture flow rates and mixture equivalence ratios.
A number of studies aimed at understanding flame propagation in such micro channels with H2–air mixtures are reported in literature.12–26 Li et al.13 reported experimental studies to understand the effect of wall temperature and radiative heat flux for a series of cylindrical backward-facing step micro combustors. Cao and Xu14 suggested a stable operating regime and different methods to reduce the heat loss from micro-combustor for micro-gas turbine applications. Jianlong et al.20 performed parametric studies to understand the flame blow-off limit and combustion efficiency of H2–air mixtures in a micro channel with bluff body. Flame instabilities and effect of design parameters on extinction limits were investigated by Yuasa et al.21 for a possible application in MIT micro gas turbine with H2–air and CH4–air mixtures in a new micro combustor concept. Stabilization of lean premixed H2–air flame in a radial micro-channel was experimentally and numerically investigated by Zamashchikov et al.25 Wang et al.24 carried out experimental investigations on catalyst assisted micro combustors for premixed H2–air mixtures to understand the effect of catalyst on flame stability. Catalyst helped improving the flame stability in the micro combustors significantly.26
Although various researchers have reported interesting insights on flame stability limits, flame propagation modes, effect of equivalence ratio and inlet velocities through various experimental studies, many difficulties arise while investigating flame propagation at such small scales due to smaller dimensions of these channels. Numerical studies allow better understanding through detailed analysis of flame propagation characteristics. Chen et al.16 carried out numerical investigations on combustion characteristics of hydrogen fuel in the micro combustor with multi catalyst. They also extended their studies to various flow conditions, reactor dimensions and wall materials. Effect of diameter on temperature, velocity and species profiles of micro combustors were analyzed numerically and analytically by Li et al.17 They observed that volumetric heat loss and wall shear stresses increase with a decrease in physical dimensions. Hua et al.18,19 performed three-dimensional numerical studies to understand the effect of dynamic flow behavior on a micro gas turbine engine. They suggested an optimized design for the micro combustor through a systematic parametric analysis. Two-dimensional direct numerical simulations of Pizza et al.22,23 with detailed chemistry for lean H2–air mixtures in planar mesoscale channels showed the existence of various unstable flame propagation modes. This work primarily focused on understanding the flame stabilization dynamics of lean H2–air mixtures in mesoscale channels with varying inflow velocities (0.3 < Uin < 1100 cm s−1). They considered channel heights of 2, 4 and 7 mm for their study. In order to provide the physical attributes to the mesoscale combustor, a temperature profile was assigned as a wall boundary condition. Numerical studies of Baigmohammadi et al.27 in a stepped micro tube combustor with inserted wire method report improvement in flame stabilization limits, temperature distribution, and combustion process by activating certain chemical reactions affecting combustion initiation reaction. Jiang et al.28 conducted detailed H2–air premixed flame simulations to analyze effects of flow velocity, mixture ratio and heat recirculation. Although these studies provide interesting insights on various issues related to flame propagation in such micro channels, they cannot be extrapolated to accurately model realistic wall thermal boundary conditions of the micro channels.
Present work is aimed at understanding flame propagation characteristics, effect of flow redirection on flame shapes and flame–flow interactions in micro-channels (1.0, 1.4 and 2.0 mm) for different wall thermal boundary conditions (0–1000 W m−2 K−1) with stoichiometric H2–air mixtures. Straight micro-channel geometry is considered to investigate these aspects related to flame propagation with H2–air mixtures in straight tubes.
The paper is organized as follows. Computational details such as details of computational domain, assumptions, chemical kinetics modelling details, solution methodology and boundary conditions are summarized in Section 2. Section 3 presents the results obtained in the present study. The major contributions of this work are summarized at the end.
There is insignificant difference in heat release rate for both the grids considered in the present work, hence a grid of 15 μm size was considered for all numerical studies reported in this paper and a procedure similar to that of Nair et al.30 and Singh et al.31 was considered to validate the present results. A tube diameter slightly larger than flame quenching distance (dq) and smaller than twice the quenching distance (1dq ≤ D ≤ 2dq) is considered for detailed studies. Therefore, a micro tube with diameter D = 1.4 mm was chosen for the detailed investigations (0.74 mm as quenching distance obtained through detailed numerical analysis).31 To understand the effect of tube diameter on flame dynamics, various tubes of diameters, 1.0 mm, 1.4 mm and 2.0 mm are also studied. To validate the present model, laminar burning velocity and flame structure of H2–air mixture is calculated by treating the upper wall as a symmetric boundary. This eliminates the wall effects and gives a flat flame in the micro-channel. Hence, the laminar burning velocity of the propagating flame can be obtained as the difference between the flame propagation speed and corresponding unburnt mixture velocity. A one-dimensional freely propagating flame model using PREMIX code is used to validate the present computational results. Laminar burning velocity is calculated for three different uniform grid sizes, 50 μm, 30 μm and 15 μm. It is found that, laminar burning velocity (∼210 cm s−1) for 15 μm and 7.5 μm grid matches exactly with the PREMIX result as shown in Fig. 2b. Furthermore, the simulated flame structure for 15 μm grid depicting the variation of major species for stoichiometric hydrogen combustion provides a good agreement with the PREMIX code result as indicated in Fig. 2c.
Olm et al.32 have tested all the major hydrogen combustion mechanism that were published since 1999. Out of these mechanisms, NUIGNGM-2010, Konnov-2008 and Conaire-2004 have been shown to give an overall good performance. These mechanisms were found to predict the flame speed and ignition delay of hydrogen combustion exactly, when compared with the various experimental results. Therefore, the detailed mechanism proposed by Conaire-2004 (ref. 33) with 9 species and 21 reactions has been employed in the present study. Numerical simulations were carried out by integrating the governing equations in a physical domain using a finite-volume based general-purpose CFD code Fluent.34 Transient simulations were carried out using pressure–velocity coupling algorithm, PISO.35 Few initial steady state iterations with finite-volume method of SIMPLE algorithm36 are carried out to prevent solution divergence at the beginning due to small domain and grid size. The flow is laminar with flow Reynolds number being equal to 218. Therefore, Arrhenius reaction based finite rate combustion model is adopted to perform H2–air combustion studies.
The boundary conditions adopted for the present study are specified as described in Fig. 1. Stoichiometric H2–air mixture with a flow velocity much smaller than the flame propagation velocity, 0.3 m s−1 is applied at the domain inlet. A small inlet velocity of 0.3 m s−1 does not affect the flame dynamics, as the flame propagation velocity is relatively much higher. This further ensures smooth convergence of numerical simulation with a small flow velocity through the channel inlet. Channel walls are considered of negligible thickness and chemically inert.37,38 Various thermal boundary conditions are applied at the walls, through the variation of the convective heat transfer coefficients, h. The fuel–air mixture is initially ignited by providing a thin ignition patch of 2400 K near the exit plane of the channel. Many researchers29,39,40 have reported that final solution depends on the initial flame shape.
![]() | (1) |
![]() | (2) |
![]() | (3) |
![]() | (4) |
Fig. 3a. Shows the HRR contour at different flame positions along the axial location of the tube for h = 600 W m−2 K−1. It is observed from Fig. 3a, that the shape of the propagating flame remains similar for all the positions in the channel as it propagates upstream. Fig. 3b shows the transition of the flame shape from a concave shape to convex shaped flame and its sensitivity around h ∼ 600 W m−2 K−1. It can be observed that, for values 580 ≤ h ≥ 620 W m−2 K−1; ∂h/∂θ ≈ 0, which further implies that ∂θ/∂h → ∞ or θ is most sensitive in this regime. Therefore, the occurrence of transition in flame shape can be observed as the flame propagates upstream. The flame changes its shape from a positively stretched flame, θ > 0 to a negatively stretched flame, θ < 0 when the wall heat transfer coefficient is in 580 ≤ h ≥ 620 W m−2 K−1 range.
Fig. 4a, shows the variation of flame shape angle (θ) with wall heat transfer coefficient, h as the flame propagates upstream in the axial direction for premixed H2–air mixtures. It can be observed from Fig. 4 that an increase in heat transfer coefficient, h results in a flame shape angle variation from θ > 0 to θ < 0, indicating a strong effect of wall heat transfer coefficient on flame shape. The flame shape changes significantly for 580 < h < 620 W m−2 K−1 wall heat transfer conditions. Fig. 4b shows the HRR contours of the propagating flame for h = 300, 600 and 700 W m−2 K−1 conditions. The flame shape angle, θ changes from 43.95° to 41.26° as h value is changed from 300 to 700 W m−2 K−1 with θ = 3.46° at h = 600 W m−2 K−1.
![]() | ||
Fig. 5 Variation of flame propagation speed with (a) flame shape angle (b) wall heat transfer coefficient for d = 1.4 mm. Here dq = 0.74 mm, quenching diameter. |
Fig. 6a shows the variation of axial mass flux, ṁx (ρu), where ρ is density and u is axial velocity component at the axis. It is non-dimensionalized with mixture properties, ρinSoL (ρin – density of incoming mixture and SoL = laminar burning velocity, 2.04 m s−1). Fig. 6a clearly shows that the peak value of ṁx occurs near the flame front and it decreases significantly with an increase in wall heat transfer coefficient, h. The change in mass-flux can also be related to a change in flame shape angle, θ due to change in wall heat transfer condition as discussed in Fig. 4a. This is due to the fact that for lower h values, higher mass flux, ṁx at the axis pushes the flame downwards, creating a concave shaped flame (θ > 0). For higher values of h, beyond a certain critical value, the axial mass-flux, ṁx values at axis decrease, resulting in a change in flame shape. This clearly explains the significant change in the mass flux distribution at the axis and transition in flame shape from θ > 0 to θ < 0 with an increase in the values of h.
Fig. 6b presents the variation of total mass-flux (ṁt) variation in radial direction, defined as ṁt = ṁx + ṁy, where ṁy = ρv (product of density and radial velocity component). It can be seen from Fig. 6b that the distribution of ṁt becomes nearly uniform for h > 620 W m−2 K−1 values. A relatively lower value of ṁt is observed near the axis and the distribution of mass-flux increases towards the walls (for y/ymax > 0.9) for higher h values which can help explain the transition in flame shape from concave shape (θ > 0) to convex shape (θ < 0) as can be seen from the curve corresponding to h = 0 and 1000 W m−2 K−1.
![]() | ||
Fig. 7 Variation of flame shape angle with wall heat transfer coefficient for 1.0 mm, 1.4 mm and 2.0 mm tube diameters. (Data plotted at 6dq from inlet). |
![]() | ||
Fig. 8 Normalized HRR contours of H2–air flames in different tube diameters. Contours normalized to 1.6 × 1010. |
Fig. 9 shows the effect of tube diameter on flame propagating speed with respect to various heat transfer conditions. For 1.4 mm and 2.0 mm diameter cases, flame speed is observed to be maximum for adiabatic conditions and minimum for wall heat transfer conditions, where flame shape changes from θ > 0 to θ < 0. It is to be noted that for 1.0 mm diameter case, the flame propagation speed shows a random behavior perhaps due to the reason that flame shape change is insignificant with wall heat transfer coefficient. Flame thermal wall coupling plays an important role in determining the flame dynamics for channels with dimensions close the flame quenching distance (i.e. 1.0 mm diameter case in present work). A two-dimensional analytical model considering conjugate heat transfer developed by Veeraragavan41,42 endorses the effect of thermal wall coupling on the flame speed. Due to this, perhaps a peculiar behavior on the variation of flame speed with heat transfer coefficient has been observed in the present work for 1.0 mm case. Fig. 9b shows the variation of normalized flame propagation speed with normalized wall heat transfer coefficient. Both the variables are normalized using following relations.
![]() | (5) |
![]() | (6) |
![]() | ||
Fig. 9 Effect of tube diameter on (a) flame propagation speed with wall heat transfer coefficient. (b) normalized speed with heat transfer coefficient. |
It is interesting to note that for hnorm < 1, the values of flame propagation speed almost fall into a single line indicating that flame propagation behavior is independent of tube diameter. For hnorm > 1, the value of flame propagation speed for 1.4 mm and 2.0 mm diameter fall into single line indicating a similarity in flame propagation behavior. However, it is interesting to note that for 1.0 mm diameter tube, the flame propagation speeds remain relatively much smaller for higher values of wall heat transfer coefficient, h. This phenomena of lower flame propagation speeds for 1.0 mm diameter tube can also be observed from Fig. 9a and could be attributed to very small change in flame shape angle with wall heat transfer coefficient, h.
Fig. 10 and 11 show the variation of the velocity angle (tan−1(v/u)) to understand the extent of flow redirection near the flame zone. The velocity angle is plotted for tube diameters of 1.0 mm and 2.0 mm cases with adiabatic wall conditions and h = 600 and 655 W m−2 K−1 conditions which correspond to lowest flame propagation velocity in 1.0 and 2.0 mm cases. A significant difference in flow redirection for adiabatic wall conditions can be observed from Fig. 10a and 11a due to the effect of tube diameter (1.0 and 2.0 mm). For smaller tube diameters, the effect of flow redirection becomes pronounced near the flame front location (1dq) as compared to large tube diameters (2dq). For instance, the point of maximum flow redirection moves from 2dq to 1dq as the tube diameter decreases from 2.0 mm to 1.0 mm. The effect of flow redirection continues to be there in larger tube diameters even at the flame stabilization point (x = 0) as clear for 2.0 mm diameter case shown in Fig. 10a. A similar comparison of flow redirection for transition conditions (hmin) shows that the flow redirection continues to be more prominent for larger tube diameters. The comparison of Fig. 10b and 11b shows that maximum flow redirection occurs for 1.0 mm diameter case at the flame location. However, the point of maximum flow redirection gets shifted upstream to 0.5dq as tube diameter is increased to 2.0 mm. An overall comparison shows that an increase in wall heat transfer coefficient pushes the point of flow redirection towards the location of the flame front and increase in tube diameter also increases the extent of flow redirection and pushes it in the upstream direction.
![]() | ||
Fig. 10 Variation of velocity angles for H2–air flame in a 2.0 mm channel (a) h = 0 (b) h = 655 W m−2 K−1. |
D | Tube diameter (mm) |
L | Axial distance (mm) |
dq | Quenching distance or its numerical value (mm) |
θ | Flame shape angle (degrees) |
h | Wall heat transfer coefficient (W m−2 K−1) |
hmin | h corresponding to minimum flame propagating speed (W m−2 K−1) |
HRR | Heat release rate |
SL | Flame propagating speed (cm s−1) |
u | Axial velocity component (m s−1) |
v | Radial velocity component (m s−1) |
ρ | Density (kg m−3) |
ṁt | Total mass flux (kg m−2 s−1) |
ṁx | Axial mass flux (kg m−2 s−1) |
Yi | Species mass fraction |
SL | Laminar flame velocity |
SL,norm | Normalized flame propagating speed |
hnorm | Normalized wall heat transfer coefficient |
This journal is © The Royal Society of Chemistry 2016 |