Front

1
Front
CAME-GT 2nd international conference on
Industrial Gas Turbine Technologies, 29th April,
Bled, Slovenia
Measurements and Large Eddy Simulation of
combustion instabilities in a scaled Gas Turbine
Combustor
DLR-Stuttgart experiment Dr. U. Meyer, Dr. X.
Duan, Dr. P. Weigand, Dr. R.Giezendanner Work
initiated by Turbomeca within the EC project
PRECCINSTA, Dr. C. Bérat
CERFACS -- Sébastien ROUX -- roux_at_cerfacs.fr --
33 (0)5 61 19 30 13
2
Todays presentation
2
Todays presentation

• PRECCINSTA project
• Description of LES solver
• Comparisons for the COLD flow
• Comparisons for the HOT flow
• Hysteresis and influence of BOUNDARY conditions

3
PRECCINSTA
3
PRECCINSTA project

• PREdiction and Control of Combustion
  INStabilities in Tubular and Annular gas turbine
  combustion systems enrich our understanding of
  unsteady combustion phenomena.
• WP2 target Extensive comparison of LES
  calculations and experimental results have been
  performed for the unperturbed flame, including
  the instability modes. The unsteady flame and the
  issue of the stability of the flame is still
  being investigated.
• Need of proper unsteady 3D investigation both
  experimental numeric.
• Partners involved in WP2 DLR, CERFACS, and
  Turbomeca.

4
AVBP
4
The LES solver AVBP

• high-order schemes (3rd order in space)
  (J. Comp. Phys 1998). No extension of existing
  Reynolds averaged code. Compressible.
• hybrid meshes, parallel code written from
  scratch (AIAA J. 1995)
• specific boundary conditions (NSCBC various
  additional capabilities to handle and control
  acoustic waves) (J. Comp Phys 1992)
• Large Eddy Simulation models for the flow Smag
  filtered Smag (JFM 1996) Wale (ICFD 1998)
• reduced chemistry (ICC, Comb Flame 2000)
  thickened flame (TF) model (Phys. Fluids 2000,
  Flow Turb Comb3 2001, CTR Research Briefs 1999,
  2000, 2002)
• validated for non reacting swirling flows
  (Int.J.Num.Meth Fluids 2002), for mixing in jets
  in cross flow (Flow Turb Comb2 2001).
• Used in many EC contracts LESFOIL, STOPP,
  ICLEAC, DESIRE, PRECCINSTA, MOLECULES,
  FUELCHIEF...
• Industrial users ABB, RR, Siemens, Alstom,
  Snecma, TurboMeca...

5
AVBP
5
The LES solver AVBP
- LES is definitely well adapted for dynamic
phenomena investigation Vortex
shedding Precessing vortex core Acoustics
Ignition Quenching - LES can handle
complex geometries likes those found in gas
turbines.
6
Full geometry
6
Geometry plenum burner chamber atmosphere
NB no questionable boundary conditions
7
Cuts position
7
Comparisons position of cuts
X 1,5 mm
W
X 5 mm
U
X 15 mm
X 25 mm
X 35 mm
TO OUTLET
FROM SWIRLER
NB LES performed before LDV
8
Cold mean Ux
8
Results for cold flow
Mean velocity field of
axial component
(12,87g.s-1)
9
Cold mean Ut
9
Results for cold flow
Mean velocity field of
azimuthal component
(12,87g.s-1)
10
Cold mean Ur
10
Results for cold flow
Mean velocity field of
radial component
(12,87g.s-1)
11
Cold RMS Ux
11
Results for cold flow
RMS velocity field of
axial component
(12,87g.s-1)
12
Cold RMS Ut
12
Results for cold flow
RMS velocity field of
azimuthal component
(12,87g.s-1)
13
Cold PVC
13
Results for cold flow
Dynamic features are predicted
Pressure isosurface shows a Precessing Vortex
Core (PVC).
Axial velocity at the outlet of the burner shows
a rotating structure.
14
Cold f520Hz
14
Results for cold flow
Frequency of the PVC in the chamber (520Hz)
LES 520Hz
Expe 510Hz
15
Cold f310Hz
15
Results for cold flow
2nd mode gt frequency in the plenum (310Hz)
LES 310Hz
Expe 300Hz
16
AVSP
16
Results for cold flow
Acoustic solver
A finite element acoustic solver developped at
CERFACS (AVSP) shows that 310Hz is close to an
eigen mode of the configuration.
freq1 171Hz, 1/4 wave of the whole device freq2
360Hz, 1/4 wave of the plenum freq3
1403Hz freq4 1540Hz freq5 1732Hz
Eigen frequencies found by AVSP
17
Mode structure
17
p along burner axis
Results for cold flow
total measured
18
Cold frequencies
18
Results for cold flow
Conclusion concerning frequencies
310Hz Acoustic mode
520Hz Hydrodynamic mode (PVC)
19
Hot conditions
19
Results for hot flow
Operating conditions

• Equivalence ratio 0.75
• Mass flux 12,87g.s-1
• Power 27kW
• quiet flow
• Fully premixed

20
Hot mean Ux
20
Results for hot flow
Mean velocity field of
axial component
(phi0.75, quiet flow, 27kW, 12,87g.s-1)
21
Hot mean Ut
21
Results for hot flow
Mean velocity field of
azimuthal component
(phi0.75, quiet flow, 27kW, 12,87g.s-1)
22
Hot mean Ur
22
Results for hot flow
Mean velocity field of
radial component
(phi0.75, quiet flow, 27kW, 12,87g.s-1)
23
Hot RMS Ux
23
Results for hot flow
RMS velocity field of
axial component
(phi0.75, quiet flow, 27kW, 12,87g.s-1)
24
Hot RMS Ut
24
Results for hot flow
RMS velocity field of
azimuthal component
(phi0.75, quiet flow, 27kW, 12,87g.s-1)
25
Hot RMS Ur
25
Results for hot flow
RMS velocity field of
radial component
(phi0.75, quiet flow, 27kW, 12,87g.s-1)
26
Hot mean T
26
Results for hot flow
Mean Temperature field
(phi0.75, quiet flow, 27kW, 12,87g.s-1)
27
Hot RMS T
27
Results for hot flow
RMS Temperature field
(phi0.75, quiet flow, 27kW, 12,87g.s-1)
28
Hysteresis
28
Experimental hysteresis equivalence ratio
history gt flame stability
UNSTABLE
STABLE
?
0.60
0.70
0.90
0.65
0.75
1.30
29
Hysteresis -gt BC
29
Experimental hysteresis
Influence of boundary conditions in LES
The hysteresis may be explained by LES by
considering the extreme sensitivity of the
solution to small changes in boundary
conditions. gt 2 computations with the same
regime but two different outlet conditions.
30
BC Temperature
30
Experimental hysteresis
Influence of boundary conditions in LES
CASE 1 stable flow
CASE 2 unstable flow
CASE 2 unstable flow
The LES suggests that the acoustic conditions
effect should be carefully investigated.
31
BC flame
31
Experimental hysteresis
Influence of boundary conditions in LES
CASE 1 stable flow
CASE 2 unstable flow
CASE 2 unstable flow
These acoustic conditions depend on the mean
temperature field in the exhaust pipes
32
Conclusion
32
Conclusions

• The COLD flow is well predicted and understood.
• The HOT case is well predicted at equivalence
  ratio 0.75
• Note No inlet boundary condition to tune
  (experiment fully meshed).
• Open points
• LES has shown that the flow was very sensitive
  to boundary conditions influence of outlet
  boundary conditions under way.
• Influence of mixing not investigated yet (LES is
  fully premixed).
• Other cases phi0.7 is planned to be computed.

33
(No Transcript)