Annual Report

2.3 LES of unsteady combustion

2.3.1 Unsteady combustion studies using LES and acoustic tools (Y. Sommerer, T. Poinsot)

Since 2000, Large Eddy Simulations (LES) have changed the way scientists investigate turbulent combustion and especially combustion instability phenomena. This evolution has been strong at CERFACS in particular in 2002 and 2003 where fifteen different configurations have been simultaneously under study using the LES code AVBP. Table 2.1 presents some examples of the various test cases which were studied: usually, these configurations were tested experimentally at the same time elsewhere in Europe, providing CERFACS with a unique data base for validation of the AVBP code.

Name of laboratory Type of geometry Type of flame/flow

Ecole Centrale Paris (EU project ICLEAC) Dump combustor (backward facing steps) Lifted turbulent diffusion burner

University of Poitiers (ORACLES) Dump combustor (rectangular section) Partially premixed turbulent burner

University of Twente (EU project DESIRE) Circular swirling with central hub Diffusion turbulent burner

Ecole Centrale Paris Backward facing step Laminar acoustically forced flame

Karlsruhe University (ITS) Double swirler (rectangular chamber) Partially premixed turbulent burner

Karlsruhe University (EBI) Double swirler (circular chamber) Acoustically forced premixed turbulent burner

Ecole Centrale Paris (COS) Circular swirling with premixing tube Lean premixed turbulent burner (with flashback)

Ecole Centrale Paris (EU project FUELCHIEF) Double stage circular Staged partially premixed turbulent burner

Ecole Centrale Paris and DLR Split conic shape Staged partially premixed turbulent burner

Alstom Daetwill Backward facing steps Reheat turbulent burner

SNECMA Double swirler Lean premixed gas turbine (LPP) injector

DLR (EU project PRECCINSTA) Circular swirler with hole injection Partially premixed turbulent burner

ONERA (EU project MOLECULES) Jets in cross flow in a tube Dilution jet mixing

Bochum University Jets in cross flow in a planar channel Mixing enhancement

Table 2.1: Examples of configurations investigated with LES in 2002 / 2003 (gaseous reactants).

As a result, several joint papers have been submitted in 2003: LES of jets in cross flow with Bochum University [1], LES of partially premixed flames in gas turbine burners with Karlsruhe (ITS) and Siemens [2], acoustic forcing of gas turbine burners with Karlsruhe EBI and Siemens [3], cold and hot flow validation of LES in swirled burners with DLR [4], system identification for turbulent diffusion flames with Ecole Centrale Paris [5].

LES alone is not sufficient to understand turbulent combustion in confined chambers: theory and experiments show that acoustics play a very important role in many cases, especially in triggering combustion instabilities. Therefore writing acoustic codes allows to describe the structure of acoustic modes in combustion chambers and their coupling with combustion is required: this is done at CERFACS and discussed in Section 2.3.3.3 and 2.3.3.4.

Many instabilities are caused by insufficient mixing between fuel and combustion air. More generally, mixing in combustors is essential for efficiency and pollution reduction: LES is used at CERFACS to study mixing and methods to enhance mixing for fuel injection or for dilution jets (Section 2.3.3.10).
[1] C. Prière and L.Y.M. Gicquel and A. Kaufmann and W. Krebs and T. Poinsot. LES of mixing enhancement : LES predictions of mixing enhancement for jets in cross-flows. Submitted to Journal of Turbulence, 2004.

[2] L. Selle and G. Lartigue and T. Poinsot and R. Koch and K.-U. Schildmacher and W. Krebs and P. Kaufmann and D. Veynante. Compressible Large-Eddy Simulation of turbulent combustion in complex geometry on unstructured meshes. In press, Combustion ands Flame, 2004.

[3] A. Giauque and L. Selle and T. Poinsot and H. Buechner and A. Kaufmann and W. Krebs. System identification of a large scale swirled premixed combustor using LES and comparison to measurements. Submitted in Journal of Turbulence, 2004.

[4] G. Lartigue and S. Roux and L. Benoit and T. Poinsot and U. Meier and C. Bérat. Studies of unsteady cold and reacting flow in a swirled combustor using experiments, acoustics analysis and Large Eddy Simulations. Submitted in Combustion and Flame, 2004.

[5] Truffin K., Varoquié, Veynante D. and Poinsot T. Large Eddy Simulations of the unsteady response of partially premixed flames. Submitted to Combustion and Flame, 2004.

2.3.2 Submodels used for LES at CERFACS (T. Poinsot, L. Gicquel, D. Veynante)

The efficiency of the LES tool developed by CERFACS jointly with Institut Francais du Pétrole is a compromise between precision and complexity: all submodels for turbulence, boundary conditions, turbulence / flame interaction, ignition, flame wall interaction, etc, are chosen to offer homogeneous performance and implementation costs which are compatible with the use of a compressible explicit solver on very large grids and large number of processors. Most models have been already published by CERFACS and are described in the literature.

The DTF (Dynamically Thickened Flame) model initially written by [1] continues to perform very well in the new configurations and has been extended to partially premixed and diffusion flames (see validations in Section 2.3.3). Many test cases of Table 2.1 could be categorized as diffusion flames in which the fuel (methane or propane) is injected through holes inside the chamber. However results show that these flames are usually lifted with significant mixing upstream of the flame, so that very few diffusion flamelets are found.

An important submodel is the chemical scheme: initially many computations with the DTF model were performed with one step schemes; two-step and four-step schemes are now being developed and the DTF model has been modified to accomodate these multistep schemes. The construction of these schemes is performed by comparing their results on simple laminar premixed flames and using genetic algorithm techniques developed to fit the results of the reduced schemes with the data obtained with full schemes. This was done in 2003 for methane, propane and heptane. These schemes are also being extended to include NOx and CO.

In 2002 and 2003 CERFACS has continued working on boundary conditions and acoustic wave management of boundaries for flame computations. In [Kaufmann, 2002a] it is shown that the behavior of a forced flame depended strongly on the method used to pulsate the inlet of the computational domain: methods based on characteristics are well suited to such computations but need some modifications. In 2003 the 'non reflecting' character of these methods has been analyzed and correct scalings for relaxation coefficients on non reflecting boundaries have been proposed [2]. During the 2002 CTR Summer Program at Stanford another specific point was analyzed: how to initialize a reacting LES or in other words, how to ignite combustion in a LES computation ? This is an important point because LES codes are much more sensitive to 'perturbations' than RANS codes, especially those created by ignition. It is also a field of future studies in which the ignition phase of a combustor will be investigated. In [Selle, 2002] is derived a procedure to ignite a LES which is now used as a standard tool.
[1] O. Colin, F. Ducros, D. Veynante, and T. Poinsot. A thickened flame model for Large Eddy Simulations of turbulent premixed combustion. Physics of Fluids, 12(7):1843-1863, 2000.

[2] L. Selle, F. Nicoud, and T. Poinsot. The actual impedance of non-reflecting boundary conditions: implications for the computation of resonators. In press AIAA Journal, 1-21, 2004.

2.3.3 Typical examples

The following examples do not cover all unsteady combustion studies of CERFACS in 2002 and 2003 but present typical results and recent advances. The first example (Section 2.3.3.1) is used to show the accuracy of LES when it is compared to experimental data. Section 2.3.3.2 shows examples of pulsated flames: the first one is a turbulent diffusion flame, the second example is a premixed swirled flame in a large gas turbine burner and the last example is a new burner dedicated to fluid structure interaction studies during combustion oscillations. The third section shows the importance of acoustics and demonstrates the power of using LES and acoustic analysis together (Section 2.3.3.3). New theoretical methods required to study the budget of acoustic energy in an oscillating burner are simultaneously being developed (Section 2.3.3.4). For the first time in 2003, LES was used to compute flashback in a burner built at Ecole Centrale Paris: results are summarized in Section 2.3.3.5. Section 2.3.3.6 presents a very large LES performed at the French Super Computing Center CINES for a combustion chamber equipped with three burners. The implementation of schemes which can predict pollutant formation is another important topic discussed in Section 2.3.3.7. The specific issue of coupling RANS and LES for reacting flows has been studied for EDF since January 2003 and is discussed in Section 2.3.3.8. One example of LES in a very high-speed reacting flow is presented in Section 2.3.3.9. Finally, examples of mixing studies (between fuel and air or between dilution jets and burnt gases) are described in Section 2.3.3.10.

A small gas turbine burner (S. Roux, T. Poinsot, G. Lartigue)

The first example shows the accuracy of LES by comparing LES velocity fields with and without combustion with measurements performed at DLR. The burner is a swirled premixed injector (Fig. 2.9)where swirl is produced by tangential injection downstream of a plenum.A central hub is used to stabilize the flame.Experiments include velocity measurements for the cold flow as wellas a study of various combustion regimes.The dimensions of the combustion chamber are 86 mm × 86mm × 110 mm.

Figure 2.9: Configuration (left). Location of cuts for velocity profiles (right).

For this chamber the critical question of boundary conditions is avoided byextending the computational domain upstream and downstream of thechamber :the swirlers and the plenum are fully meshed and computed. The mesh includes even a part of the outside atmosphere (not shown onFig. 2.9 forvisibility)to avoid having to specify a boundary conditionat the chamber outlet.The expected precision in terms of acoustic waves interacting withtheoutlet of the chamber is much improved since this section is not aboundary condition but a part of the computational domain.

Swirled flows are very sensitive to an hydrodynamic instability called PVC (Precessing Vortex Core). This mode which induces a rotation of the axis of the recirculation zone is captured by LES (Fig. 2.10) particularly well in the non reacting case.

Figure 2.10: Vector and pressure field in central plane (left) andisosurface of low pressure (right).

For nonreacting flow the LES and experimental profiles are comparedat various sections of the combustion chamber(Fig. 2.9) for average axial (Fig. 2.11)and RMS axial (Fig. 2.12) velocities. Azimuthal velocities are predicted with the same accuracy. All mean and RMS velocity profiles are correctly predicted.Considering that this computation has no boundarycondition which can be tuned to fit the velocity profiles thisconfirms the predictive capacity of LES in such swirling flows which are wellknown to be difficult for RANS.A large central recirculation zone (evidencedthrough negative values of the mean axial velocity) is formed on the chamber axis. This recirculation zonebegins at x=2 mm downstream of the central hub and is not yetclosed at x=35mm. The large values of RMS velocities on the axis in Fig. 2.12 are due to the PVC.

With combustion LES also performs very well. For an equivalence ratioof 0.75, an air flow rate of 12 g/s and a thermal power of27 kW, the velocity fields are presented inFig. 2.13 (mean axial velocity), 2.14 (RMS axial velocity). The overall agreement between mean LES results and experimental datais outstanding. LES also shows that the PVC is damped when combustion starts.

Figure 2.11: Reacting flow; mean axial velocity in the central plane. Circles: LDV; solid line: LES.

Figure 2.12: Reacting flow; RMS axial velocity in the central plane. Circles: LDV; solid line: LES.

System identification of turbulent burners (K. Truffin, A. Sengissen, T. Poinsot, B. Varoquié, L. Selle)

This section presents examples of forced turbulent flames. Forcing is a normal procedure in the investigation of combustors stability. Burners are forced in order to examine their response to acoustic waves entering the air inlet. This type of analysis, called system identification can be performed experimentally or numerically using LES. But the cost of such experiments is very high and LES are presently being developed to try to replace experiments for this task.

At CERFACS, three studies of this type were performed in 2003:

Chamber A: forced response of a turbulent diffusion flame in a rectangular propane / air burner. This burner is installed in Ecole Centrale Paris (ICLEAC EC project).
Chamber B: forced response of a large scaleindustrial gas turbine mounted in a square cross section chamber in Karlsruhe University.
Chamber C: a new experiment built atTwente Univ. in the DESIRE EC project. For this last burner, multiple innovations will take place: the computation will include both mixing from the methane jets and the combustion in the chamber. All LES and measurements will be performed to address the issue of fluid structure interaction during combustion oscillations.

Figure 2.13: Chamber A: burner (left) and set up for system identification (right).

The geometry of burner A is displayed in Fig. 2.15. Loudspeakers produce the acoustic excitation u' at the chamber inlet. A photomultiplier measures the resulting unsteady heat release w'. The transfer function between u' and w' is used in acoustic codes to predict the burner stability (Section 2.3.3.3). A typical view of the flame for Chamber A is given in Fig. 2.16. Note that the flame is very long: it does not swirl and is lifted from the injectors. The comparison with the experimental results of Ecole Centrale is promising [1].

Figure 2.14: Chamber A: isosurface of fuel mass fraction and reaction rate in three planes.

For chamber B, a much larger burner is used. This burner is also swirled and reaches a power up to 800 kW. It is installed in two different institutes in Karlsruhe, either in a square (ITS) or a circular (EBI) combustion chamber depending on the type of experimentation.

Figure 2.15: Chamber B: burner (left) and combustion chamber (right).

Fig. 2.17 shows the main features of the burner in which this identification by acoustic forcing is performed:a central axial swirler (colored in dark) is used to inject and swirla mixture of natural gas and preheated air. The main part of the combustion air as well as fuelis injected by the diagonal swirler through holes located on both sides of the vanes in order to achieve swirling. Perfectly mixed gases enter the diagonal swirler while pureair enters the axial swirler: the flame inside the chamber is in apartially premixed regime.

Figure 2.16: Combustor B: flame shapes (isocontours of temperature T=1000 K) at two instants of the forcing cycle separated by a half period.

This burner has been analyzed in detail in [2], but only the pulsated results [3] are presented here: for this study, the flame is pulsated at various frequencies by modulating the inlet velocity. Results are compared to experiments performed at EBI Karlsruhe. For a pulsation at 120 Hz Fig. 2.18 shows two extreme positions of the flame front during one cycle. The inlet pulsation leads to very large excursions of the flame shape and area so that the total heat release is also oscillating. This response is the building block of acoustic solvers that predict the stability of combustors (see Section 2.3.3.3).

The geometry of burner C is displayed in Fig. 2.19: for this flame the computation must include both mixing from the four methane jets with air and combustion in the chamber. This burner is not yet built but LES has already been used for its design: one of the conclusions is that the geometry of Fig. 2.19 would have led to flashback (see Section 2.3.3.5) and had to be modified before final construction. The walls of the combustion chamber are equipped with accelerometers to measure the structure vibrations and to correlate them with the pressure and heat release oscillations inside the combustor.

Figure 2.17: Combustor C. Left: geometry: the whole shaded area is meshed and computed. Right: isosurfaces of Y_CH4=0.1 and isosurface of temperature T=1000 K.

[1] K. Truffin, B. Varoquié, D. Veynante, T. Poinsot, and L. Lacas. Large eddy simulations and experimental characterization of the unsteady response of partially premixed flames. Submitted for publication to Combustion and flame, 2004.

[2] L. Selle. Simulation aux grandes échelles des interactions flamme/acoustique dans un écoulement vrillé. PhD thesis, Institut National Polytechnique de Toulouse, 2004.

[3] A. Giauque and L. Selle and T. Poinsot and H. Buechner and A. Kaufmann and W. Krebs. System identification of a large scale swirled premixed combustor using LES and comparison to measurements. Submitted in Journal of Turbulence, 2004.

Acoustic tools development. Application to high-frequency modes (L. Benoit, T. Poinsot, A. Kaufmann, F. Nicoud)

The importance of acoustics in combustion has been the subject of a long controversy in which certain authors argued that acoustics could be neglected both experimentally and numerically. Most recent experimental and numerical results actually confirm that acoustics are essential phenomena for turbulent confined flames. To understand confined flames, developing acoustic codes which can solve the wave equation in complex geometries for reacting flows is therefore a necessary step. CERFACS is developing such tools [1].

The first tool called Soundtube is able to provide the low-frequency longitudinal resonant modes in a network of interconnected ducts with variable sections and temperatures. This tool has been installed at SNECMA and EDF. The second tool is a full three-dimensional Helmholtz solver. This Helmholtz solver (called AVSP) provides the solution of the Helmholtz equation in the frequency domain and is coupled to the LES code AVBP: they use the same grids; the mean temperature and mass fraction fields required by AVSP (to know the sound speed) and the flame transfer function (to know the acoustic / flame coupling, see Section 2.3.3.2) are provided by AVBP.

Figure 2.18: Acoustic pressure for the 1200 Hz mode. Left: Helmholtz result.Right: LES result.

Low frequency modes are not the only evident product of flame /acoustics interactions. Higher order acoustic modes of the chambercan also interact with the flame front. These modes are more difficult to understand than low-frequency oscillations and the joint usage of LES andHelmholtz solvers offers a new and powerful approach for such phenomena asshown in this section for a turning mode in the combustor of Fig. 2.17. In theLES, thiscombustor exhibits a natural unstable mode at 1200 Hz which isvisible in the wall pressure traces. The structure of this mode can be visualized by plotting the p' amplitude on the walls of the chamber (Fig. 2.20 right). Even though the data is slightly noisy because of insufficient sampling, a clear mode structure appears.

When the Helmholtz solver is applied to this geometry, it also exhibits two transverse eigenmodes (the (1,1,0) and the (1,0,1) modes) at the same frequency 1220 Hz which can be combined into a turning mode.This turning mode rotates around the axis of the combustion chamber.The resulting average structure is displayed in Fig. 2.20 (left):obviously it perfectly matches the structure measured in the LES: the 1200 Hz mode seen in the LES is a turning mode which is a linear combination of the (1,0,1) and (1,1,0) modes. This turning mode has a direct effect on the flame topology. Fig. 2.21 shows the flame shape at two instants during one 1200 Hz cycle (phase pi/2 and 3pi/2): the acoustic velocity induced by the turning mode at the lips of the diagonal swirler creates ahelicoidal perturbation which is convected downstream and slices flame elements when it reaches the flame extremities.
[1] A. Kaufmann. Towards Eulerian-Eulerian large eddy simulation of reactive two phase flow. PhD thesis, Institut National Polytechnique de Toulouse, 2004.

Figure 2.19: Isosurfaces of temperature T=1000 K at two instants of the 1200 Hz cycle separated by a half period. The turning mode shapes the flame along a spiral motion.

Acoustic / combustion coupling tools: the acoustic energy equation (C. Martin, T. Poinsot, F. Nicoud)

A key issue to control combustion oscillations is to understand them. At the moment, no one can predict at the design stage whether a given combustor will oscillate. If and when it oscillates, measurements and computations sometimes give indications of the reasons of the problem but it is usually by then too late. Being able to predict these phenomena requires the development of a new approach in which LES and acoustics are coupled. Section 2.3.3.3 has shown how acoustic codes were developed at CERFACS. In parallel, new theoretical tools must be built: one of them is a method to examine all terms in the acoustic energy equation e₁ [1] which controls the evolution of acoustic perturbations:

where the ⁰ subscript refers to mean quantities and the ¹ to acoustic values.

If integrated over the whole volume V of the combustor bounded by the surface A, it yields:

where

is the surface normal vector. This surface consists of walls or of inlet / outlet sections. The RHS source term

corresponds to the well-known Rayleigh criterion and is the source of the oscillations. But there are other terms like acoustic fluxes on outlets and inlets

which also have a very strong effect on the instabilities: these terms can not be measured experimentally and most studies consider only the term they can quantify (the Rayleigh term) which is clearly just a small part of the problem. LES offers a new approach by giving access to all terms of Eq. (2.2). A full closure of the acoustic energy equation using LES equation is tested for the first time in [1]. Preliminary results are very promising: the budget equation for acoustic energy seems to be closed reasonably well so that individual terms can now be examined. This has been done for the burner installed in Ecole Centrale Paris.
[1] T. Poinsot and D. Veynante, (2001), Theoretical and numerical combustion, R.T. Edwards Ed., Chapter 8, 473 pp.

LES of flashback in swirled burners (Y. Sommerer, T. Poinsot, J.-P. Légier, D. Galley, D. Veynante)

Flashback is one of the phenomena of high concern for the new generation of lean premixed gas turbine burners: by improving mixing upstream of the combustion chamber (see Section 2.3.3.10), many devices used to reduce pollution by increasing mixing can also create the possibility for the flames to propagate upstream of their normal stabilization zone, thereby risking the destruction of the injector or a part of it. Certain gas turbines manufacturers already imagine future robust designs which would resist a temporary flashback and recover without failure. Being able to predict flashback is therefore a key issue for modern numerical combustion. It is also a challenge for modelling because, during flashback, the flame regime changes considerably from partially premixed to almost purely non premixed flames. The duration of the flashback event can also be long (of the order of a second) requiring significant computing power.

Figure 2.20: Left: geometry of the burner (Ecole Centrale Paris). Center: experimental view for stable regime. Right: flashback.

Figure 2.21: Left: stable lifted flame. Right: flame after flashback. The flame is visualized by an isosurface of reaction rate.

CERFACS has worked on flashback in Lean Premixed Prevaporized burners since 1999 [1] and these studies have been continued since then. The experimental validation is performed in Ecole Centrale (PhD of D. Galley) in a special chamber in which premixing tube and combustion chamber are transparent in order to observe the flame flashback. Fig. 2.22 shows the geometry (left) and the two extreme regimes which can be observed: either the flame starts in the combustion chamber (where it should: Fig. 2.22 center) or it starts in the premixing tube (Fig. 2.22 right). Here flashback is obtained by simply reducing the air flow rate. The whole flashback evolution can be studied using LES (Fig. 2.23). The limits where flashback is obtained experimentally are qualitatively recovered by LES but a major problem arises: hysteresis. For the same regime, the flame can be either lifted or flash backed depending on the history of the ignition and stabilization procedure: increasing or decreasing the equivalence ratio to go from point A to point B for example can lead to two different flows at point B. This point is being investigated with LES and experiments. This configuration will also be run for two-phase flow combustion in 2004.
[1] J.-P. Légier. Simulations numériques des instabilités de combustion dans les foyers aéronautiques. PhD thesis, Institut National Polytechnique de Toulouse, 2001.

Multiburner computations (G. Staffelbach, T. Poinsot)

Most academic studies of combustion in gas turbines are performed using a gas turbine burner and installing it into a laboratory combustion chamber. However real gas turbines use 16 to 24 burners installed in the same annular chamber. In such situations strong coupling may occur between burners: in regions where the issuing flames from neighbouring burners meet, strong turbulence and heat release can take place and lead to instabilities which cannot be obtained in single burners configurations. CERFACS is studying a chamber equipped with three burners, which is the largest LES computation ever performed with combustion in such a geometry (5 million cells). It requires typically 64 to 128 processors to run efficiently. Fig. 2.24 shows an isosurface of temperature (left) and a snapshot of the reaction rate in a surface passing through the three burners axis (right). The flows issuing from the burners interact and the flames influence each other. The next studies will focus on comparison with experimental data that will be obtained in 2004 at DLR.

Figure 2.22: Three-burner combustion chamber of DLR. Left: isosurface of temperature T=1000 K. Right: field of reaction rate in a surface passing through the axis of the three burners.

Prediction of pollutant formation with LES (P. Schmitt, T. Poinsot, D. Veynante, N. Dioc)

A natural extension of the CERFACS LES tools is to predict the formation of pollutants. This can be done in the framework of the Thickened Flame model even though few developments have been done up to now. As an example, LES to predict CO levels are performed : Fig. 2.25 shows the geometry of the burner and a typical instantaneous field of CO. This swirled burner uses a cone cut in two parts and shifted to create a lateral swirling air injection. Fuel (methane) is injected laterally by small holes. In the computation mixing and combustion are handled simultaneously using the DTF model and a two-step chemical scheme. The experimental results corresponding to this burner will be provided in 2004 by Ecole Centrale Paris (at atmospheric pressure) and DLR (at high pressure).

Figure 2.23: LES of combustion including CO formation. Left: configuration. Right: CO instantaneous fields in various cuts in the burner.

Coupling RANS and LES for reacting flows (M. Saudreau, B. Varoquié, T. Poinsot)

A usual question about LES is its computing cost. CERFACS' experience is that the cost is low especially on modern parallel computers once an initial solution is available. Obtaining such a first solution rapidly is the central problem. One way to do this is to compute the flow with a RANS (Reynolds Averaged Navier Stokes) solver and to use this solution to initialize the LES code. This issue is studied in collaboration with EDF in the case of the Oracles rig developed at Poitiers. Various coupling solutions between Saturne (the EDF RANS code) and AVBP are being investigated. The main difficulty is that there are significant differences between a RANS solution and an initial LES field: how to reconstruct the LES data from the averaged RANS data raises a variety of fundamental issues which are now being examined.

LES for high speed reacting flows (B. Varoquié, T. Poinsot, R. Knikker)

Certain innovative combustion concepts require the study of the combustion of gaseous fuels in high speed hot air (typically a few hundreds of m/s): a first classical lean burner preheats the air and this hot air is then sent to a reheat burner in which methane is injected and autoignites rapidly. Combustion in this reheat burner is very different from low-speed combustion due to the very high injection velocities. Understanding unsteady combustion in reheat burners is therefore a priority for developments. This field of applications is also totally new for LES: in such flows, autoignition is the key process for flame stabilization. CERFACS has worked on such a combustor [1] to investigate chemical schemes adapted to such flames, the feasability of LES and the stability of burners based on this reheat approach.

An essential difference between LES for classical and reheat burners is the importance of chemistry: in reheat burners, autoignition is the main factor controlling the flame position while flame propagation plays the same role in low-speed burners. To understand these mechanisms and to construct a chemical scheme suited to LES of reheat burners, the first part of the study was devoted to the computation of the ignition between a hot air layer and a cold fuel jet. This basic problem contains all the phenomena present in the real combustor except turbulence: the fuel and the air must mix before ignition can start but this mixing also lowers the local temperature, slows the chemical reactions down and delays autoignition. The results are described in [Knikker, 2003].
[1] B. Varoquié, C. Martin and T. Poinsot. Large eddy simulation of auto-ignition in gas turbines with sequential combustion. Contract report CR/CFD/04/19, 2004.

Mixing studies (C. Prière, L. Gicquel, T. Poinsot)

Being able to mix fuel and air efficiently is a key issue in multiple combustion problems: insufficient mixing leads to high pollution levels and stability problems while too fast mixing can cause flashback and endanger the safety of the burner. CERFACS has investigated the mixing of fuel jets with air, the effects of mixing enhancement devices and the mixing of dilution jets with burnt gases using LES. These numerical studies were accompanied by experiments performed at Univ. Bochum and ONERA Toulouse.

Fig. 2.26 shows a result where 8 dilution jets are injected into a central duct. The full configuration is computed and no assumption is used on the symmetry of the flow. The comparison with the PIV results of ONERA Toulouse is excellent.

Figure 2.24: Isosurfaces of axial vorticity tracing the dilution air jets injected in the main duct.

Return to TOP

webmaster@cerfacs.fr Last Updated: Apr 11, 2008		Copyright © 2002-2007 CERFACS All rights reserved.