
Cold-State Investigation on a Flame Holder
Copyright © 2013 SciRes. JPEE
Figure 1. Definition of the co-ordinate and the geometrical
parameters for the flow past a slitty bluff body.
cal geometries are generally bluff body and require spe-
cial attention when attempting to predict the associated
flow. In most problems, the flow is unsteady and turbu-
lent with vortex shedding. It is theoretically possible to
directly resolve the whole spectrum of turbulent scales
using an approach known as direct numerical simulation
(DNS) to capture the fluctuations. DNS is not feasible for
practical engineering problem, as DNS approaches are
too computationally inten se. An alternative f or numerical
simulation of complex turbulent flows is the large eddy
simulation (LES). It should, however, be stressed that the
application of LES to engineering simulations is in its
infancy. Evidence shows that LES is improper to calcu-
late the 2D bluff body flow.
Traditionally, numerical si mulations of these flows are
performed using the Reynolds-averaged Navier-Stokes
(RANS) equations and using phenomenological model to
fully represent the turbulence. The available turbulence
models vary extensively in complexity, in particular, from
simple algebraic eddy viscosity relationships to complex
formulations involving several additional differential equa-
tions. In each model, values for the empirical constants
are obtained from turbulent flows that are fundamentally
simple.
The RANS equations represent transport equations for
the mean flow quantities only, with all the scales of the
turbulence being modelled. A computational advantage is
seen in transient situations, since the time step will be
determined by the global unsteadiness in the mean flow
rather than by the turbulence. The Reynolds-averaged ap-
proach is generally adopted for practical engineering cal-
culations.
The RNG k-ε model is one of the k-ε variants of
RANS derived using a rigorous statistical technique
called renormalization group theory. It is similar in form
to the standard k-ε model, but includes the following
refinements:
• The RNG model has an additional term in its ε equa-
tion that significantly improves the accuracy for ra-
pidly strained flows.
• The effect of swirl on turbulence is included in the
RNG model, enhancing accuracy for swirling flows.
These features make the RNG more accurate and reli-
able for a wider class of flows than the standard k-ε
model. Thus RNG k-ε model in FLUENT 6.0 is used in
the present investigation using.
The computational meshes employed are non-uniform
grids. The number of grid cells is about 140,000 (slightly
varied with different gap ratio), these quadrilateral cells
are obtained with interval size 1mm using the Quad/
Tri-Pave meshing Scheme in GAMBIT, which creates a
paved mesh that consists primarily of quadrilateral ele-
ments but employs triangular mesh elements in any cor-
ners, the edges of which form a very small angle with
respect to each other.
Velocity Inlet boundary conditions include the flow
velocity, and all relevant scalar properties of the flow.
The total properties of the f low are not fixed, so they will
rise to whatever value necessary to provide the pre-
scribed velocity dis tribution. Figure 1 defined the inf low
velocity Re = 470,000 (nor mal to the boundary).
Pressure Outlet boundary conditions specify the static
pressure at flow outlets (and also other scalar variables,
in case of backflow). The use of a pressure outlet boun-
dary condition instead of an outflow condition often re-
sults in a higher rate of convergence when backflow oc-
curs during iteration. The outlet boundary is far enough
from the bluff body (x = Lx) with constant static pressure
which can be measured experimentally.
In order to model the natural perturbations in any real
flow, many numerical simulation s usually use an explicit
perturbation at the onset of the trans ient calculation. This
numerical disturbance exists in the form of a deranged
initial flow field often formed by applying slip velocity.
This explicit perturbation is said to be necessary in order
to disturb the Navier-Stokes equations and provoke or
“kick-start” the vortex shedding process by Anderson
(1993) [2]. D.G.E. Grigoriadis et al. (2003) investigated
incompressible turbulent flow past a long square cylinder
using LES [3]. They used a uniform stream U superim-
posed with Gaussian random divergence-free perturba-
tions of intensity 2% - 5% w.r.t. the local value. After a
transient time the flow rejects the initiate unrealistic co n-
dition and the shear layers at the cylinders’ faces initiated
vortex shedding.
In our investigation all flow fields were firstly calcu-
lated with a “stable” solver on the assumption that the
flow field can be time-independent. This assumption is of
signality, although it is undoubted that the real flow past
a bluff body is a ti me-dependent problem. To investigate
the intrinsic mechanism in the w ake flow of a slitty bluff
body, the inflow perturbation is unwanted. The perturba-
tion can be reduced to a certain level, but it can never be
removed completely or be diminished to very small level
in a real flow. However the simulation can initiate the
inflow perturbation to zero. All further time-dependent
simulations are on the basis of the stable results.