2.1 Basic phenomena
The study of basic phenomena in combustion is an essential part of the activity of
the team. It feeds its scientific expertise which is a solid basis for the other
activities. During the last three years, a number of different subjects have been
investigated that are either generic or more focused on a specific topic:
-
Flame-wall interaction and wall heat flux in rocket engines
- Stratified combustion modelling
- Autoignition in diffusion layers
- Interaction of a flame with a liquid fuel film on a wall
- Structure and ignition of supersonic coaxial jets
- Propagation of laminar two-phase flames
- Reduced chemistry
These studies all aim at improving our understanding of the mechanisms that controll
turbulent flames from ignition to extinction. The results are
either used to improve the modelling in CFD simulations or directly applied to
industrial applications, as for example the use of the maximal wall heat flux
obtained in the flame-wall interaction study to help the design of the engine.
We develop below four selected topics, that illustrate the present activity.
2.1.1 Interaction of flame with liquid fuel on a wall (G. Deroutter, B. Cuenot, C. Habchi)
The interaction between walls and flames is a critical issue in the
design of many modern combustion systems: in gas turbines, Lean
Premixed Prevaporized (LPP) technologies often lead to the
coexistence of liquid films on combustor walls and of flames (for
example during flashback). In piston engines, Direct Injection (DI)
techniques can create fuel films on the piston or on the cylinder
walls: these films burn at later times, interacting with the flame in a
complex manner which is not understood or modelled at the present time.
The consequences are an increased emission of
pollutant species, lower performance and shorter life time of the engine.
The problem of flame wall interaction with a liquid film
is a dual problem in which the temperature
profile must be solved simultaneously in the gas and in the liquid
film. While a DNS (Direct Numerical Simulation) code (NTMIX) is used for the gas phase problem, the resolution of the
temperature field in the liquid film is done with an
integral method in which the temperature profiles are assumed to be polynomial
functions of the spatial coordinate.
Figure 2.1: Configuration of the flame interaction with a liquid film.
Figure 2.2: Normalized flame / wall distance (triangle), consumption speed
(dashed line) and wall heat flux (solid line) during
the flame interaction with a liquid film.
First computations of a monodimensional premixed flame interacting with a liquid
film have been performed (Fig. 2.1). The results show that the evaporation of the liquid film
changes the equivalence ratio in front
of the flame and modifies its propagation. Typical results are shown on
Fig. 2.2 where the distance between the flame and the liquid surface,
the flame speed and the heat flux to the wall
have been plotted. It appears that the flame does not approach
the film surface as close as a dry wall and that it is quenched by an excess
of fuel and not because of heat loss. As a consequence the heat flux to the wall
stays much lower than in the case of interaction with a dry wall. This project is
continued through a thesis in collaboration with IFP.
2.1.2 Ignition and structure of supersonic coaxial jets (A. Dauptain, B. Cuenot)
Rocket engines technology raises very basic combustion problems. This is the
case for the ignition of the VINCI combustor that operates at very low ambiant
pressures (almost vacuum). The high pressure ratio between the injectors and the
chamber leads to supersonic flow with high compressibility. This raises the
problem of mixing and ignition of the two initially separated reactants (H2 and O2).
Indeed, the underexpanded jets that come out of the injectors have a particular structure with slip lines, weak and strong shocks. The very strong shear at the envelope
of the jet prevents mixing (see Fig. 2.3). Mixing is mostly
due to the turbulence that develops behind the first strong shock.
Simulations of underexpanded jets at high pressure ratios have been performed with
the AVBP code, using a particular numerical treatment of the thin interfaces (slip
lines and shocks). The thermodynamic
conditions correspond to the igniter jet. An example is shown
Fig. 2.3 where the density field of a jet at pressure ratio 7 is plotted.
For comparison a visualisation of the same jet obtained in the laboratory (B. Yip et al, 2003, private communication)
is inserted,
demonstrating the performance of the code on this configuration.
The next step is to study H2-O2 autoignition in this configuration and to perform
simulations with two jets: the igniter and an H2/O2 injector of the first row of
injectors. This project is going on within a thesis supported by SNECMA
Moteurs Fusée.
Figure 2.3: Density field of an underexpanded jet (pressure ratio 7). The inserted
picture is the result from Rayleigh Scattering on the same jet obtained by B. Yip
et al.
2.1.3 Propagation of laminar two-phase flames (B. Cuenot, M. Boileau, S. Pascaud)
The development of the two-phase flow solver AVBP-TPF for combustion applications
requires to master the physics and behaviors of two-phase laminar flames.
Depending on the thermodynamic conditions, droplet size, fuel characteristics and
chemistry, different classes of two-phase flames may be encountered: either all
droplets evaporate before burning and combustion occurs in a pure gaseous phase,
or evaporation is not complete in front of the flame and droplet burning occurs.
The simplest case where
all the liquid fuel is first vaporized before burning through a premixed flame
in a purely gaseous zone, has the great advantage to present
an analytical solution, that can be used as a reference solution to
validate the computation.
As an example, the flame presented in this section burns ethanol C2H5OH,
with a laminar flame speed of 0.42 m.s-1. This value is used to initialize
the velocity profile in order to keep the flame front steady.
Fig. 2.4 shows the comparison between the computed solution and the analytical
solution of the mass fraction profiles. The agreement is very good.
This shows that the mechanisms of two-phase flames are understood and that the
AVBP-TPF code is able to compute such flames. The relative facility to calculate
two-phase burning flows is one of the advantage of the
Eulerian formulation for the dispersed phase used in AVBP-TPF, compared to a
Lagrangian formulation.
Figure 2.4: Comparison between initial and converged profiles of species mass fractions.
2.1.4 Reduced chemistry (K. Truffin, T. Poinsot)
Chemical kinetics are one of the controlling factors in all
combustion applications. Although the chemistry of most fuels is known
the accurate description of the
chemical kinetics in practical computations still remains a problem. This is due
to the complexity of the available chemical kinetic schemes which add many
variables to the system and drastically increase its stiffness.
To address this problem, the team has developed a strategy for chemistry reduction
based on the following statements:
-
the reduced chemistry must guarantee a certain number of flame
characteristics (flame speed and thickness at least)
- the reduced chemistry must decrease the stiffness of the full scheme
- the reduced chemistry must minimize the number of added variables
Such reduced schemes are built either with an empirical method, in which the
chemical species and reactions are chosen and the reaction constants fitted to
match flame characteristics or in a more systematic way using an optimisation
code (EPORCK) based on a genetic algorithm. The obtained reduced schemes are
usually valid on a restricted range of parameters but have the advantage of a
very easy implementation.
|
|
|
