2.1 Basic phenomena
To sustain the development of models for RANS or LES, and to understand the details of turbulent combustion and its interaction with other phenomena, a number of fundamental works are conducted by the team. Basic studies are performed on the role of liquid films in piston engines (in collaboration with IFP), the flow and flame structure in rocket engines, the modelling of perforated plates that can be found in gas turbines, and in parallel on the development of a two-phase flow version of AVBP, the structure and ignition of two-phase flames.
2.1.1 Interaction of a flame with liquid fuel on a wall (G.
Desoutter, B. Cuenot, C. Habchi)
The Direct Injection Engine (IDE)
is an interesting solution to reach a compromise between
low fuel consumption and reduction of pollutant emissions.
However direct injection of fuel in the combustion chamber causes important deposition on
the walls, leading to an increase of unburnt hydrocarbons (HC) release.
In IDE engines, especially during the starting
regime, the fuel jet impacts directly on the piston and creates a liquid film that
is a source of HC production and smoke during the combustion phase. If many studies can be found on the
interaction of a flame with a dry wall, little is known about the flame
behavior when it approaches a liquid film. In particular the quenching of the
flame and the distance of quenching are still open questions.
Here, Direct Numerical Simulations (DNS) of the flame-wall interaction with
a liquid fuel film were performed with the code NTMIX3D and published in the last Symp. on Comb.
[Desoutter, 2004].
DNS allowed to
understand the mechanisms involved in the interaction, like for example the film
evaporation due to the presence of the flame and its impact on the flame structure.
The role of the liquid film was also studied by comparing the interaction
of the flame with a wet and a dry wall.
Simulation results show that the flame quenches much further away from the walls when a
liquid film is present, and that this effect is increased by high wall
temperatures. The liquid film is mainly altered by the wall temperature and not by the flame.
Based on these results, a new model for the
evaporation of a liquid film in the presence of a flame has been established for RANS
calculations and tested in the C3D code of IFP for piston engines.
Figure 2.1: Deposition of a liquid fuel film in Direct Injection Engines (IDE).
2.1.2 H2 - O2 flame ignition and structure (A. Dauptain, B. Cuenot)
The ignition process of rocket engines must control ignition but also minimize the risk of
explosion and destruction of the engine. In such systems, chemical phenomena are
coupled to highly compressible flows, characterized by structures like shocks and slip lines.
The numerical simulation of rocket ignition then requires robust and accurate methods, able to
handle stiff chemistry and flow structures with strong pressure gradients.
In a first step, the autoignition of a mixing layer between hydrogen and hot oxidizer
was simulated with Direct Numerical Simulations, using realistic transport
laws, showing a behaviour which differs strongly from methane autoignition because of the
high diffusivity of hydrogen. Results were
reported in Comb. Sci. Tech. [1].
Figure 2.2: Supersonic H2 - O2 combustion: instantaneous temperature field.
In a second step, LES of underexpanded sonic jets were performed with AVBP
after some adaptation to handle supersonic flows with shocks. This allowed a complete
spectral study of the unsteady features that are the initial source of the acoustic noise, feeding the screech instability.
The compression zone has been particularily studied, evidencing
three different types of excitations
[Dauptain, 2005].
Finally LES was applied to supersonic combustion cases, such as the
flame of Cheng [2]. The agreement between
numerical and experimental results is fair: computations capture the lifted
flame (Fig. 2.2), the mean values of speed, temperature and species concentrations,
while the levels of RMS values are also correct. LES results confirm the
influence of supersonic compressible pattern on combustion.
[1] R. Knikker, A. Dauptain, B. Cuenot and T. Poinsot, (2001), Comparison of computational methodologies for ignition in diffusion layers,
Combustion Science and Technology, 175(10), 1783-1806.
[2] T.S. Cheng, J.A. Wehrmeyer, R.W. Pitz, O. Jarrett and G.B. Northam, (1994), "Raman measurement of mixing and finite rate chemistry in a supersonic hydrogen-air diffusion flame", Combustion and Flame, 99:157-173
2.1.3 Direct and Large Eddy Simulations of Effusion Cooling (S. Mendez,
F. Nicoud, T. Poinsot)
In almost all combustion systems, solid boundaries must be cooled. One possibility
often chosen in gas turbines is to use multiperforated walls. In this technique, fresh air coming from
the casing goes through the perforations and enters the combustion chamber. The associated micro-jets
coalesce to give a film that protects the internal wall face from the hot gases.
The number of submillimetric holes on a perforated plate is far too large to resolve each hole and allow a complete
description of the generation/coalescence of individual jets. Effusion is however known to have drastic effects on the whole
flow structure, and to modify noticeably the flame position.
Figure 2.3: Visualization of the structure of the flow around a perforated plate obtained by Direct
Numerical Simulation (velocity iso-surface).
As a consequence, new wall models for turbulent flows with effusion are required to perform predictive
full scale computations. One major difficulty in developing new wall functions is that the boundary
fluxes depend on the details of the turbulent flow structure between the solid boundary and the fully
turbulent zone.
Measurements in realistic configurations are
difficult to perform and generally do not provide enough detailed information to allow a complete
understanding of the phenomena involved. This can be overcome by using numerical simulation.
Although limited to very small domains, Direct Numerical Simulation
is a first and crucial step to describe the details of the flow in the near zone of a perforation plate.
Figure 2.3 shows an example of the velocity field obtained
around one perforation (the domain has been duplicated four times for
clarity). In this
simulation, the main global characteristics of the flow have been validated by comparison with
experimental data. A data base can then be generated on this configuration in order to build an
appropriate wall model for perforated walls. In a second step, RANS computations and Large Eddy Simulations of a full
burner with the new perforated wall law is performed to validate the model
[Mendez, 2005].
2.1.4 Ignition criteria of two-phase flames (N. Lamarque, G. Boudier,
B. Cuenot
A critical issue for flames in liquid fuel combustors is the ignition process, that combines evaporation
and combustion, and is significantly different from the ignition of gaseous flames. In this
process and in addition to chemical kinetics data, parameters like the droplet size and its distribution
have important impacts.
The ignition of a reactive mixture in simulations is often obtained by unphysical methods. Here a
realistic spark plug model has been implemented in AVBP, where the energy deposition due to spark
discharge is modeled by a source term added to the total energy equation.
Figure 2.4: Typical temporal variation of gas temperature at the center of the flame kernel. td is the deposition time and tsat is the time at which the liquid droplets reach the saturation temperature.
A comprehensive theoretical and numerical study of ignition in a one-dimensional configuration
has been performed, starting from the spark energy deposition within the spray to the flame
stabilisation. Figure 2.4 shows a typical time evolution of temperature at the center of the flame kernel. Three stages can be identified: in a first stage, the energy deposition results in a fast temperature increase; then energy deposition is stopped and the main phenomenon of the following stage is the preheating of the droplets; finally in the last stage, starting when the droplets have reached their staturation temperature, ignition occurs, leading to a sharp increase of temperature and a subsequent stabilisation on a value close to the flame temperature.
Two ignition criteria were determined, based
on global parameters of the two-phase flow and the spark, that can therefore be calculated a priori. The relevance of these criteria has finally been evaluated against a series of one-dimensional ignition simulations [1].
[1] N. Lamarque, G. Boudier, B. Cuenot and T. Poinsot, "Numerical Study of Two-phase Flame Ignition
by a Spark Plug", submitted to Proc. of the Combustion Institute, 31 (2006).
|
|
|
