Lire la première partie
88 pages

Lire la première partie


Obtenez un accès à la bibliothèque pour le consulter en ligne
En savoir plus
88 pages
Obtenez un accès à la bibliothèque pour le consulter en ligne
En savoir plus


Niveau: Supérieur, Doctorat, Bac+8
Lire la première partie de la thèse

  • transparent view

  • reacting test bench

  • fuel injection

  • main stage bowl

  • counter-rotating relative

  • swirler

  • pilot bowl

  • swirler can

  • swirler channels



Publié par
Nombre de lectures 26
Langue English
Poids de l'ouvrage 5 Mo


la première partie
de la thèsePart IV
Application to an aeronautical
multipoint injector
181Chapter 9
Description of the TLC configuration
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
9.2 The SNECMA staged premixing swirler . . . . . . . . . . . . . . . . 183
9.2.1 Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
9.2.2 Injection of liquid fuel . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
9.3 The ONERA non-reacting test bench . . . . . . . . . . . . . . . . . . 185
9.3.1 Measurement methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
9.4 The numerical setup. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
9.4.1 Modifications of the original geometry . . . . . . . . . . . . . . . . . . . 187
9.4.2 The computational grid . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
9.1 Introduction
experimental and numerical studies in the TLC (for “Towards Lean Combustion”) project of
the 6th framework programme of the European Union. The configuration is therefore referred
to as “TLC configuration”throughout this manuscript.
9.2 The SNECMA staged premixing swirler
9.2.1 Geometry
An isolated, cut-away view of the premixing swirler is shown in figure 9.1. It represents one of
around 20 injectors that are typically mounted on the upstream wall of an annular combustion
chamber of an aero-engine. It is a staged design with the objective to divide fuel injection and
premixing in two separatly controllable zones, in order to allow the optimization of the system
for different operating points (see chapter 1 for a detailed explanation). The device is thus
composed of two stages, that can be identified by two conical “bowls”, where the central, pilot
bowl is nested inside the main stage bowl.
24 multipoint
injectors! Pilot injector!
Pilot stage bowl!
Main stage bowl!
Figure 9.1: Staged premixing swirler, cut-away view.
of channels, divided by guide vanes, that are inclined relative to the main axis and impose
a swirling motion to the flow. Figure 9.2 presents a transparent view, highlighting the three
swirler stages. Two are of radial type and lead into the pilot bowl. While the innermost swirler
troughacircularslotinthesidewalljustbeforethepilotflowexitsintothechamber. Thethird
swirler can be considered to be of radial type, although it is slightly inclined, leading the flow
into the main stage bowl with a non-zero axial velocity component. All three swirler stages are
where the flows meet.
pilot stage !
swirler channels!
Main stage!
swirler channels!
Figure 9.2: Staged premixing swirler, transparent view with highlighted swirler channels.
The bulk of the airflow (approx. 90 %) passes through the main swirler stage. The remaining
10 % is divided between the innermost pilot swirler (3 %) and the outer pilot swirler (7%).9.3. THE ONERA NON-REACTING TEST BENCH 185
9.2.2 Injection of liquid fuel
Liquid fuel is fed into the injector via two separate circuits, which are well visible in figure
9.1. One is connected to the pilot injector, which creates a hollow-cone spray. The pilot fuel
atomizer is of the so-called piezo type, featuring a ring of very small orifices on a cone-shaped
injector head. The second circuit leads to the multi-point injection system of the main stage,
a series of 24 holes located on the inner wall of the main stage, each placed just downstream
of a swirler channel’s exit. At each point, liquid fuel is injected perpendicularly to the surface
through orifices of 0.5mm in diameter.
The partition of fuel mass fluxes between the pilot stage an the swirler stage can be used as
a tool to optimize local equivalence ratio values for different phases of flight. Throughout the
present work, the pilot stage will be completely deactivated in order to study the phenomena
related to multipoint injection in an isolated way. This, of course, is not a realistic operation
condition and only used in the framework of an academical study.
9.3 The ONERA non-reacting test bench
For measurement purposes, the inector described in the preceding section was mounted on var-
ious test benches. These include a completely open setup (with the inejctor directly exiting
into the atmosphere) that has been studied experimentally at ONERA DMAE in Toulouse
and numerically by Lavedrine [81]. A configuration adapted to reactive experiments has been
studied experimentally at ONERA DMPH in Palaiseau [105]. Numerical simulations have been
Figure 9.3: Photography of the installation at the ONERA Fauga-Mauzac center.
In the present work, a third configuration is considered. It was mounted at the ONERA center
at Fauga-Mauzac and allows a detailed study of the non-reacting, two-phase flow [82]. The
test bench, pictured in figure 9.3 allows to pressurize and pre-heat the chamber, which is of a
simple rectangular shape with a square cross-section, where large observation windows provide
optical access for measurements. Air enters through an admission duct that expands into a186 CHAPTER 9. DESCRIPTION OF THE TLC CONFIGURATION
Staged premixing swirler!
Outlet nozzle!
Figure 9.4: TLC configuration ONERA Fauga-Mauzac
plenum. The injector is mounted on the dividing wall between this plenum and the chamber.
Additionally, this divider is perforated to feed air into cooling films exiting into the chamber at
about half the distance between the injector outer diameter and the lateral chamber walls (see
figure 9.5). It has to be noted that this film serves no real purpose in the present configuration.
It is a remnant of the reacting test bench, where these films are located in direct proximity of
thelateralwallsandserveasacoolinglayertoprotecttheopticalaccesswindows. Thechamber
exit is formed by a nozzle that reaches supersonic flow at the throat, leading to an acoustically
non-reflecting outflow.
Figure 9.5: TLC configuration ONERA Fauga-Mauzac - view from the plenum
9.3.1 Measurement methods
The test bench was equipped for different measurement techniques briefly described in the
following. The goal of these measurements was to obtain data on:
• The gaseous phase velocity, using a LDA technique9.4. THE NUMERICAL SETUP 187
• The droplet velocity and diameter, using a PDA technique
• The local droplet size distribution, using laser diffraction spectroscopy
• The spatial distribution liquid volume flux, using a patternator technique
The LDA (for Laser Doppler Anemometry) measurement method uses a pair of coherent laser
beams that are crossed at the location where velocity data is measured. At this location, the
beams form interference fringes, which illuminate particles that cross the pattern periodically.
The frequency of this light signal can be detected and translated into a velocity. This value
corresponds to the velocity component perpendicular to the fringes and the measurement has
to be repeated to obtain other velocity components. The gaseous flow is seeded with particles
of a very low Stokes number in order to minimize errors due to droplet inertia and to exclude
two-way coupling effects.
by Durst et al. [40], that uses two detectors for the light scattered by the particles, arranged
at different locations in space. The resulting phase shift between both doppler signals can be
translated into a diameter information of the recorded particle.
The laser diffraction spectroscopy uses a laser beam to illuminate the spray. For a single
droplet, in close forward direction, diffraction patterns are observed that can be related to the
size of a spherical particle using Mie theory [95]. For a polydisperse spray, a complex light
intensity distribution is recorded, which can be translated into a droplet size distribution using
methodsdescribedbyHirleman[62]. Theadvantageofthistechniqueisthatthedistributionis
obtained instantaneously, from the post-processing of a single image.
Thepatternator technique is based on the verysimple principle of placingan array of recip-
ients in the direction of spray movement. The spatial distribution of liquid volume flux can be
reconstructed by the amount of liquid that is present in each of the recipients.
For additional information, the reader is referred to the TLC report [82] and for theoretical
background to the book of Frohn and Roth [47].
9.4 The numerical setup
9.4.1 Modifications of the original geometry
Modifications are made relative to the original geometry in order to make it suitable for com-
putations. These modifications comprise:
• The air exits the chamber through a supersonic nozzle. In order to render the supersonic
boundary condition more stable, the narrowest section is followed by a short, gently di-
vergingtubetoreliablyestablishMa>1attheactualdomainexit. Thishasnoinfluence
on the computed flow as perturbations do not travel upstream towards the sonic throat.
• The cooling films connected to the plenum through slots in the upstream wall of the
chamber are modeled with a surrogate geometry, as explained in detail in chapter 7.
• A cooling film located along the circular outer rim of the injector (called “collar perfora-
tion”) is entirely replaced by an equivalent inlet condition. The corresponding mass flux
is substracted from the flux perscribed at the plenum.188 CHAPTER 9. DESCRIPTION OF THE TLC CONFIGURATION
• All very small scale features are removed from CAD data. This applies for example to
very fine gaps, chamfers, etc. that would lead the grid generation algorithm to create
diminutive cells.
9.4.2 The computational grid
The computational grid is a cornerstone of the numerical approach and has proven to be a
critical contributor to the quality of the results in the present study. Elements describing the
importance of particular features of the grid can be found throughout this manuscript. Here,
its most important characteristics are summarized and reference the corresponding chapter.
Figure 9.6: Staged premixing swirler, transparent view with highlighted swirler channels.
Figure 9.6 shows an overview of the mesh. It is of unstructured, hybrid type, composed of
tetraedral, prismatic and pyramidal elements. The (triangular) prisms form a single, closed
layer in all regions where wall functions are used to model the turbulent boundary layer. This
layer is necessary for the application of wall functions in the no-slip formulation (see section
6.2.6 for details) but is is also advantageous in terms of the overall number of cells, because the
near-wall grid refinement can easily be controlled by adapting the prism aspect ratio without
leadingtoanexcessivenumberofnear-walltetraedra. Asinglelayerischosenbecauseitallows
consistent meshing in complex geometries including sharp edges, when multiple layers tend to
bemoreheavilydeformed. Figure9.7showsadetailofsuchaprismaticregionontheseparator
between pilot and main stage. Pyramidal elements are used to connect the edge of a prismatic
boundary area with an adjacent tetraeral zone.
Thegridhasseveralrefinedzonesinsideanddownstreamtheinjector. Theswirlerchannelsare
optimized for the use of wall-functions, with a target for near wall prismatic layer thickness of
+y ≈ 100 (see chapter 6 as well as section 10.3.3). Furthermore, the relative grid resolution
in the volume of the swirler channels is kept constant between the swirler stages in order to
avoid mass flux imbalances (see chapter 7 for details). The second important area of high grid
resolution is located in the pilot and main bowl, stretching outwards in areas where the main9.4. THE NUMERICAL SETUP 189
Figure 9.7: Staged premixing swirler, transparent view with highlighted swirler channels.
Mesh type unstructured, hybrid
Cell types tetraedrae (domain volume)
prisms (domain boundary, regions intended for wall-functions)
pyramids (link between prisms and pure tetra regions)
Number of grid nodes 1619357
Number of grid cells 8540311ber of boundary nodes 186367
Table 9.1: Mesh parameters
recirculationzonesarelocated. Theserefinementswereadaptedsuccessivelytoaccomodatethe
shear layers over separated boundary layer zones (see chapter 10). Finally, the grid is refined
in proximity of the cooling films. Here, the meshing parameters are determined at the creation
of the surrogate geometry used in this area to avoid very small-scale geometrical features (see
section 7.3 for details of this method).
as well as inside the plenum. Although resolution inside these zones is insufficient for a proper
LES, this coarsening is needed to render the simulation feasible. The penomena resolved in
these zones are therefore limited to very large scale motion (like the central recirculation zone)
or acoustics effects in the case of the plenum.
pointinjection. Duetotheartificiallyenlargedinjectionzone(seesections8.2.3and11.2.1),the
resulting cell sizes are close to the surrounding near-wall resolution. The primary spray regions
are already sufficiently refined due to the needs of the gaseous phase. The same mesh is used
for all calculations presented in this study, both gaseous and two-phase, regardless of the liquid
phase approach (EL or EE). Global parameters of this common mesh are summarized in table

  • Accueil Accueil
  • Univers Univers
  • Ebooks Ebooks
  • Livres audio Livres audio
  • Presse Presse
  • BD BD
  • Documents Documents