1. Swirl intensity influence on interaction between non-swirling and swirling co-axial jets in a combustor configuration: LES and modelling study
S. Šarić, B. Kniesner, P. Altenhöfer, S. Jakirlić and Cam Tropea
Fachgebiet Strömungslehre und Aerodynamik, Technische Universität Darmstadt
Darmstadt, Germany
Dalibor Čavar
Fluid Mechanics Sec., Dept. of Mech. Eng., Technical University of Denmark
Lyngby, Denmark
Branislav Basara
Advanced Simulation Technology, AVL List GmbH
Graz, Austria
5th Int. Symposium on “Turbulence and Shear Flow Phenomena” Munich, Germany, August 27-29, 2007

2. Outline Motivation/Objectives Flow Configuration Computational details
Results
Conclusions

3. Single tuboannular model combustor
“Experimental and numerical investigations of flow and mixing in a swirl combustor” (FG SLA)
To systematically investigate the effects of
Swirl intensity
Mass flow rate ratio (Reynolds number)
geometry confinement (in terms of ER)
Flue (PIV)
Inlet Section (LDA)
U/Uf
Streamlines coloured by Ma-number

4. SLA Combustor (flow features)
Complex mean strain tensor: mean and secondary shearing, streamline curvature, adverse p.g. effects
S=1.0

5. Flow Configurations Parameter Range Reynolds number (mean flow)
Mass flow rate (mean flow)
Reynolds number (annular flow)
Mass flow rate (annular flow)
Swirl intensity
The objective:
investigate the swirl intensity influence on interaction between the central non-swirling stream and the swirling co-axial jet

6. Computational Details (inflow generation)
Three sets of computations Rec=49530, Rem=23500, S=0.0, 0.6 and 1.0 :
Swirling inflow generated computationally by assuming a fictitious pressure gradient in azimuthal direction (LES Dynamic Smagorinsky).
Additional forcing of the target U-velocity profile (Pierce 2001): LES-f
(Simplified) Swirl generation system computed separately (swirler): LES (DSM)
Entire combustor computed (swirler + combustor): LES (SM, Cs=0.1)
Separate precursor LES of a fully developed, non-swirling (central) pipe flow corresponding to the experiment (Rem = 23500)

7. Computational Details (discretization, solution domain)
FASTEST 3D
FV method; collocated variable arrangement; SIMPLE algorithm, second order accuracy in both space (CDS) and time (Crank-Nicolson); block-structured, body-fitted, non-orthogonal meshes.
The solution domain consisting of the flue (Lx=4Df), swirler with annular pipe and central pipe is discretized by a Cartesian grid (NxxNrxNq) comprising about 4.6 Mio. cells in total (192x161x128 cells in the flue, 96x49x128 cells in the swirler and 19x(5x32x32) cells in the end part of the central pipe).
The time step chosen ( 6x10-5 or 7.7x10−4 Df/Uf) corresponds to
CFL < 0.5 in the largest part of solution domain.

8. Resolution assessment

9. RANS computations 800 000 grid cells Lx=6Df
AVL SWIFT code (AVL GmbH, Graz, Austria):
finite volume discretization technique,
arbitrary unstructured, collocated grid arrangement,
pressure correction is based on SIMPLE algorithm,
the so-called “universal” wall treatment that
blends the integration up to the wall with the standard wall functions
Popovac and Hanjalic (2007), Basara et al. (2007)
grid cells
Lx=6Df

10. k-z-f turbulence model
k-- f :
k-v2- f :
A robust ER-EV model
Hanjalić, Popovac & Hadziabdić, IJHFF 2004

11. Annular inlet section: mean axial velocity field (LES)
Rec=49530; Rem=23500
S=0.0 and 1.0
S=0.0
S=1.0

12. Results: Inlet Profiles ( annular section, at x= -0.04m)

13. Results: mean u and u’v’ profiles in the combustor flue

14. Results: mean u’u’ and v’v’ profiles in the combustor flue

15. Swirl intensity influence on the flow:
LES predictions of the radial (v) and circumferential (w) velocity

16. Available experimental and LES data
Exp.: Roback & Johnson (1983)
LES: Pierce & Moin (1998)
ER=2.1
S=0.41
Re=47.500
Flow conditions close to the SLA experiment
Measurements of the tangential (circumferential)
velocity available

17. Comparison of the available exp. and LES data

18. Comparison of the available exp. and LES data

19. Comparison of the available exp. and LES data

20. Conclusions
The effects of the increasing swirl intensity on the interaction between the outer, swirling stream and the inner, non-swirling flow in the near field of a model combustor was computationally investigated applying both LES and k-ζ−f RANS model.
The increasingly swirled annular jet promotes an intensive mixing in the near field of combustor. The overall agreement between simulations and measurements is good. This is particularly the case in the shear layer and the outer, wall-affected flow region.
The present predictions exhibit a higher level of agreement with the Roback and Johnson’s experimental results.
The simulations return a ring-shaped recirculation zone with positive centerline velocities along entire combustor (uncertainties in the experimental outlet boundary condition? An issue of LES resolution / grid design?)