x formation in a spatially developing turbulent premixed bunsen flame · spatially developing...
Post on 18-Mar-2020
3 Views
Preview:
TRANSCRIPT
XXXIX Meeting of the Italian Section of the Combustion Institute
II2.1
NOx formation in a spatially developing turbulent
premixed Bunsen flame
S. Luca, A. Attili, F. Bisetti stefano.luca@kaust.edu.sa
King Abdullah University of Science and Technology (KAUST), Clean Combustion
Research Center (CCRC), Thuwal, Saudi Arabia
Abstract
A Direct Numerical Simulation of a three-dimensional lean methane/air flame in a
spatially developing turbulent slot Bunsen burner is performed. This configuration
is of interest as it retains selected characteristics of real devices, such as turbulent
production by mean shear.
The jet consist of a methane/air mixture with equivalence ratio ϕ = 0.7 and
temperature of 800 K. The simulation is performed at 4 atm. The coflow is
composed of Argon at the temperature of the combustion products.
The flame is in the thin-reaction zone regimes and the Reynolds number based on
the jet width and velocity is 5600. The grid has a resolution of 20 µm resulting in a
total of 350 million points. A supporting simulation is performed to generate the
inflow conditions for the jet.
Chemistry is treated with a new skeletal chemical mechanism developed
specifically for the DNS with 33 species.
The macroscopic and microscopic characteristics of the flame are analyzed. Due to
the inert coflow, the flame develops from a location few millimeters above the
nozzle. The flame structure is found to be similar to the one of one-dimensional
premixed flame. Heat release rate and NO rate of formation are analyzed taking
into account four paths of decomposition on N2 as initiation steps for NO
formation: NNH, Thermal, Prompt and N2O.
Configuration & Numerical Methods
The DNS configuration consists of a central premixed jet surrounded on both sides
by a heated coflow. This arrangement is similar to the piloted flame used in
experiments [1].
The jet is a methane/air mixture with equivalence ratio equal to ϕ = 0.7 and
temperature of 800 K. The coflow is composed of Argon at the temperature of the
combustion products. This temperature value is obtained from a fully burnt
equilibrium solution with initial conditions equal to those imposed in the central
jet. The choice of the coflow is made in order to stabilize the flame with the hot
gas, without affecting NOx production with a coflow containing Nitrogen.
A summary of relevant parameters of a freely propagating one-dimensional laminar
flames is given in Table 1.
XXXIX Meeting of the Italian Section of the Combustion Institute
II2.2
Table 1. Premixed flame parameters at T = 800 K for ϕ = 0.7 and p = 4 atm, computed
using PREMIX [2]. SL: laminar flame speed, δHRR: laminar flame thickness based on
half peak width of heat release rate, δL: laminar flame thickness based on the
maximum temperature gradient.
Density (kg/m3) 1.7045 δHRR (mm) 0.0418
Viscosity (kg/s m) 3.62 x 10-5
δL (mm) 0.1124
SL (m/s) 1.0123 τf = δL/SL (ms) 0.1109
The central jet has a width of H = 1.2 mm and a bulk velocity Ub = 100 m/s. The
Reynolds number based on the jet slot width and bulk velocity is 5600.
A supporting simulation is performed to generate the inflow conditions of the
central jet. The velocity field is extracted from a fully developed turbulent channel
simulation. The 2D field is sampled at a selected streamwise location in the
channel and used as the inflow conditions for the main slot burner simulations.
The simulation is performed using the NGA code. The gas phase hydrodynamics
are modeled with the reactive, unsteady Navier-Stokes equations in the low Mach
number limit [3]. The species obey the ideal gas equation of state and all transport
properties are computed with a mixture-average approach [4].
Combustion is modeled using a new skeletal mechanism developed targeting lean
premixed methane-air flame. The skeletal mechanism has 33 species and 181
reactions. It is obtained from the application of the directed relation graph (DRG
[5]) method and sensitivity analysis on the GRI-3.0 [6] detailed mechanism and is
tailored to the lean conditions of the DNS. The mechanism has been validated for
flame speed and flame structure for the target unburnt mixture conditions and good
agreement was obtained.
The DNS is performed with a grid resolution of 20 µm and a domain size of
24Hx16Hx4H. The domain is discretized with 1440x960x256 cells resulting in 350
Million grid points and the mesh size results in a spatial resolution below twice the
minimum average Kolmogorov scale and the thin reaction fronts are adequately
resolved with δL/dx~5.
The simulation necessitates 2 million CPU hours on 16384 cores of the CRAY
XC40 supercomputer ``Shaheen'' available at King Abdullah University of Science
and Technology.
Results
The flame is initially planar near the jet nozzle and shows significant development
and wrinkling with downstream distance. It is strongly wrinkled and the scales of
wrinkling are characterized by wide range of sizes. Instantaneous contour of
temperature is shown in Figure 1.
Due to the presence of mean shear, the turbulent scales and statistics evolve in the
axial direction. Since there is significant axial development of the flame, the results
XXXIX Meeting of the Italian Section of the Combustion Institute
II2.3
are presented at a selected axial location where the flame had time to completely
develop.
Due to the inert coflow the flame starts few millimeters above the nozzle exit as
seen from the contours of the mass fractions of H2O and O in Figure 1. Further
downstream the flame is completely developed and presents the same features of a
flame with a coflow composed of burnt product [7].
Figure 1 shows also the NO mass fraction. NO production is characterized by large
time scales resulting in high values downstream of the flame.
Figure 1. Two-dimensional contour plots of Temperature, H2O, O and NO mass
fractions. Only a small part of the domain is shown in the y direction.
In the region where the flame is fully developed, it presents the characteristics of a
flame in the thin reaction zone. Figure 2 shows selected quantities in a small region
of the domain. The flame is a sharp interface as seen from the temperature and heat
release contours. However it is observed that the rate of production of NO is
broader and extend itself in the burnt gases. This observation justifies the choice of
Argon as coflow to avoid that mixing with a stream containing N2 could influences
the rate of production of NO.
XXXIX Meeting of the Italian Section of the Combustion Institute
II2.4
Figure 2. Two-dimensional contour plots of Temperature, Heat Release Rate, NO
mass fraction, and NO net rate of production in the region of the domain
highlighted with a rectangle in Figure 1.
NO preferred paths of formation are investigated considering four well known
main paths of N2 decomposition: the NNH path, Thermal NO, Prompt NO and N2O
path. The rates are computed considering selected reactions that are the initiation
step of NO formation.
Figure 3. Two-dimensional contour plots of decomposition rates of N2 for four
paths: NNH, Thermal, Prompt and N2O.
Contour plots of the rate of production for the four paths taken into account are
presented in Figure 3. As expected Prompt NO is confined at the flame location,
NO formation from NNH and N2O paths is broader, while Thermal NO formation
is concentrated after the flame.
The domain is sampled at one axial location and scatter plots of the four rates are
also presented in Figure 4 confirming that NNH and N2O paths are broader in
temperature space, while prompt and thermal NO peak at the flame location and in
the burnt gases respectively.
XXXIX Meeting of the Italian Section of the Combustion Institute
II2.5
Figure 4. Scatter plot of decomposition rates of N2 for four paths: NNH, Thermal,
Prompt and N2O.
References
[1] Filatyev, S. A., Driscoll, J. F., Carter, C. D., and Donbar, J. M., “Measured
properties of turbulent premixed flames for model assessment, including
burning velocities, stretch rates, and surface densities”, Combustion and
Flame 141:1-21 (2005).
[2] Kee, R. J., Grcar, J. F., Smooke, M., Miller, J., Meeks, E., “PREMIX: a
Fortran program for modeling steady laminar one-dimensional premixed
flames”, Sandia National Laboratories Report, 1985.
[3] Desjardins, O., Blanquart, G., Balarac, G., Pitsch, H., “High order
conservative finite difference scheme for variable density low Mach
number turbulent flows”, Journal of Computational Physics, 227:7125-
7159 (2008).
[4] Attili, A., Bisetti, F., Mueller, M. E., Pitsch, H., “Effects of non-unity
Lewis number of gas-phase species in turbulent nonpremixed sooting
flames”,Combustion and Flame, 166:192-202 (2016).
[5] Lu, T., Law, C. K., “A directed relation graph method for mechanism
reduction”, Proc. Comb. Inst. 30:1333-1341 (2005).
[6] Smith, G. P., Golden, D. M., Frenklach, M., Moriarty, N. W., Eiteneer, B.,
Goldenberg, M., Bowman, C. T., Hanson, R. K., Song, S., Gardiner Jr, W.
C., et al., “GRI-Mech 3.0” (1999).
[7] S. Luca, A. Attili, and F. Bisetti. “Direct Numerical Simulation of
Turbulent Lean Methane-Air Bunsen Flames with Mixture
Inhomogeneities”, 54th AIAA Aerospace Sciences Meeting, AIAA SciTech,
(AIAA 2016-0189). http://dx.doi.org/10.2514/6.2016-0189
doi: 10.4405/39proci2016.II2
top related