Fethi Bouras^
e-mail: f.bouras@holmail.fr
Azeddine Soudani
Mohamed Si-Ameur
Universify of iHL-Bafna,
05000 Batna, Aigeria
Numerical Study of the Turhulent
Flow Inside an ORACLES
Configuration
This numerical investigation deals with the validation of the experimental results in the
inert cases of Nguyen et ai, obtained in the framework of the European Union-funded
research program MOLECULES (Modelling of Low Emissions Combustors Using Large
Eddy Simulations). This study is based on the benchmark of testing one rig for accurate
comparisons with large eddy simulations configuration (ORACLES), aimed at helping the
design of reliable lean premixed prevaporized) combustion chambers and supplied with
two identical flows of air channels. Therefore, this study is based on the 3D numerical
simulation using large eddy simulation-wall adapting local eddy viscosity (LES-WALE)
model that aims to determine the longitudinal velocity, the longitudinal velocity fluctua-
tion and the length of recirculation zone for the three cases of flow in different inlet Reyn-
olds (Re = 25,000, 50,000, 75,000). Calculations are carried out by the FLUENT CFD.
Tiie results obtained are compared with experimental measurements of Nguyen et al. The
LES WALE eddy viscosity computation presents a good agreement with the experimental
data where we could observe the asymmetrical flow and also detect the recirculation
zones and the differences between the cases of the flow. [DOI: 10.1115/1.4006455]
Keywords: turbulence, large eddy simuiation, WALE eddy viscosity, ORACLES configu-
ration, CFD
1 Introduction
Very common observations teaeh us that the majority of the
flows met in praetiee do not represent the laminar ideal properties.
In 1980, R. Legendre observed the distinction between various
regimes of flows and stated that it is wrong to say that turbulence
is not a property of the fluid; instead it is a eharacteristie of the
flow. Simulating turbulent flows in realistic conditions is a formi-
dable computational task. Many approaches have been proposed
to understand and model the turbulent flow. They are summarized
in the recent reviews [1,2]. Among these techniques, we can men-
tion the Reynolds averaged Navier-Stokes (RANS), the direct nu-
merical simulation (DNS) and the large eddy simulation (LES)
which are very rapidly growing tools, mainly because of the pro-
gress of turbulent flow modeling, numerics, and computer scienee.
These techniques also provide the numerical solutions of the
coupled, nonlinear set of (spatially) three-dimensional Navier-
Stokes equations describing time evolution of the flow over a
mesh of nodes and volumes or elements covering the flow
domain.
In 2000, Besson et al. were testing a facility named one rig for
accurate comparisons with large eddy simulations (ORACLES)
which has been designed and developed in the framework of a Eu-
ropean research program large-eddy simulations for lean pre-
mixed prevaporised combustion, where a zone of combustion of
the type LPP [3], supplied with two flows of channel constituted
of a premixture of air and commercial propane, with the primary
and specific long-term objective of progressively building up an
experimental database adapted and useful for testing the modeling
of LPP flows configurations by a LES based approach [3,4].
Moreover, Reynolds averaged Navier-Stokes based simulations
can also be tested. In particular, it allows a first characterization of
the unsteady structure of the inert and reactive flows that are
'Corresponding author.
Contributed by the Applied Mechanics Division of ASME for publication in the
JOURNAL OF APPLIED MECHANICS. Manuscript received October 27, 2010; final manu-
script received March U, 2012; accepted manuscript posted March 29, 2012; pub-
lished online June 28, 2012. Assoc. Editor; Arif Masud.
developed in the combustion chamber to emphasize the following:
an asymmetrical flow without combustion, a symmetrical flow
with combustion and the presence of a strong component of the
velocity fluctuations in certain configurations with combustion
[3,5]. Following, Nguyen et al. have pursued this objective in the
framework of the European program modeling Of low emission
combustors using large eddy simulations (MOLECULES) aimed
at promoting the extensive use of LES to model low emissions
combustors. The motivations of this benchmark are for testing the
main properties of the veloeity field for different cases of flows
(inert/ reactive). The investigation is based on the inlet Reynolds
to determine each case. However, the reactive flow is character-
ized by the equivalence ratio in order to determine the characteris-
tics of the lean extinction according to the different parameters in
ORACLES configuration [4,5]. In addition, Nguyen et al. meas-
urements confirmed the asymmetrical and symmetrical flow in
inert and reactive case and maitily the important amplitudes of ve-
locity fluctuation in the inert case [3,4]. Kurenkov and Oberlack
were focusing on numerical modeling of the turbulent premixed
combustion using the level set approach for Reynolds averaged
models in ORACLE configuration. They based their simulation
on 2D and all results were computed for half of the chamber and
then mirrored with respect to the symmetry plan. However, this
did not allow visualizing asymmetrical flow and the difference
between the two zones of recirculation [6]. In addition Reynolds
Averaged Models is only applicable enough far from the wall and
is based on unicity of the scale [7,8]. This prevents modeling other
scales which causes impaets on the results [9,10]. Domingo et al.
was interesting in the DNS of a premixed turbulent V flame in the
first part of his work but he did not succeed in finding a fully
appropriate velocity in the Flame/wall interaction. In the seeond
part, he gave LES for validating the case of higher Reynolds num-
ber of flows in ORACLE configuration. For LES numerical treat-
ment, this is preferred for solving the scale equations for the
variance itself [11]. Duwig and Fureby were interested in LES
based on the flamelet models of unsteady lean stratified premixed
combustion in ORACLE geometries in different cases of the grid.
The subgrid viscosity is modeled by a one-equation eddy viscosity
Journal of Applied Mechanics
Copyright © 2012 by ASME
SEPTEMBER 2012, Vol.79 / 051014-1