1. Y. Hardalupas Mechanical Engineering Department London SW7 2BX Experiments with Lean Partially Premixed and Premixed Combustion, Atomisation and Droplet dispersion

2. Outline of experiments Lean Partially-premixed model gas turbine combustor Lean Premixed Rijke combustor Lean premixed flames in fractal grid generated turbulence Instabilities along interfaces during atomisation Droplet dispersion in a “Box of turbulence”

3. Lean Partially Premixed Combustion in a model gas turbine combustor Alan E. Bayley, Y. Hardalupas, A.M.K.P. Taylor

4. Combustor Air

5. Operating conditions Two parameters varied: Reynolds number (Re) = 15000, 23000 Global equivalence ratio (Φ) = 0.57, 0.62, 0.70 Equivalence ratios produce three distinctly different flames Lifted (Φ=0.57)Stable (Φ=0.62)Oscillating (Φ=0.70)

6. Analysis of flame-fronts: Final results Lifted (Φ=0.57)Stable (Φ=0.62) Oscillating (Φ=0.70) Image area 40mm 60mm

7. Statistical uncertainty of flame-front curvature Φ=0.62, Re = 15000 Φ=0.70, Re = 15000 Φ=0.70, Re = 23000 Φ=0.62, Re = 23000

8. Re = 23000 case. Curvature distribution increases with global Φ. Asymmetry in distribution present. Results: PDF independent of location

9. Lean Premixed Rijke combustor Thomas Sponfeldner, Y. Hardalupas, Frank Beyrau, A.M.K.P. Taylor

10. Combustor arrangement

11. Flame Characteristics

12. N. Soulopoulos, J. Kerl, F. Beyrau, Y. Hardalupas, A.M.K.P. Taylor, J.C. Vassilicos A turbulent lean premixed flame in fractal-grid generated turbulence

13. Space filling fractal grid 3 iterations of a main pattern The length and the thickness of the bars change with the same ratio at each iteration A standard and a fractal grid are tested Same (effective) mesh size,13 mm Same blockage ratio, 22% Same pressure drop

14. Borghi diagram Slide 14 ○ : square grid ● : fractal grid

15. Premixed flame burner 600 mm 200 mm 50 mm Flame stabilisation rod Flow development Flow conditioning Fuel – air mixture inlet Grid 62 mm

16. Mean flow field – Cold flow using hot wire □ : square grid ■ : fractal grid

17. Constant Taylor length scale Slide 17 □ : square grid ■ : fractal grid

18. Increasing turbulence intensity Slide 18 □ : square grid ■ : fractal grid

19. High-speed OH Planar Laser Induced Fluorescence x y z Dye laser Nd:YAG laser Camera + Intensifier dP/dz z Nd:YAG @ 532 nm, 5 kHz Dye laser, Sirah Allegro LaVision HighSpeedStar 6, HighSpeed IRO Stabilization rod diameter = 1.5 mm Burner size = 62mm x 62mm Mean velocity U = 4.6 m/s CH 4, equivalence ratio = 0.7 Measurement duration is 600 integral time scales – 30 eddy turnover time scales

20. Progress variable - 1 Slide 20

21. Progress variable - 2 Slide 21

22. Progress variable - 3 Slide 22

23. Turbulent flame speed Slide 23 8◦12◦ squarefractal

24. Flame curvature

25. The scalar dissipation rate in an unsteady turbulent jet N. Soulopoulos, Y. Hardalupas, A.M.K.P. Taylor

26. Unsteady turbulent gas jet Experimental setup Optical measurementGas injector

27. Instantaneous Scalar Dissipation (log 10 χ) ~60 s -1 t/T = 2.153.44.655.9 7.159.6510.913.4

28. Mean Scalar Dissipation (log 10 χ) ~60 s -1 t/T = 2.153.44.655.9 7.159.6510.913.4

29. Measurements of instabilities and intact liquid jet core in a coaxial air-blast atomizer G. Charalampous, Y. Hardalupas, A.M.K.P. Taylor

30. Optical connectivity technique Introduction of the laser beam through liquid injection nozzle Propagation of laser beam along intact liquid jet core, which acts as an optical fiber Break-up of liquid jet interrupts light propagation, so light intensity change identifies the intact core length Addition of fluorescing dye in the atomizing liquid, so that all the liquid volume is observed Tracking the intact liquid jet core by fluorescent intensity eliminates scattered light from droplets by adding an optical filter to remove the wavelength of the incident laser light Intact Core Length Laser Beam introduced into nozzle Liquid flow Dye in intact liquid core is fluorescing Dye in detached droplets not fluorescing because initial laser light does not propagate after liquid jet breaks

31. 31 Test Flow and Instrumentation Co-axial air blast atomizer with 4 inlets (no swirl) Modification for laser beam entry through the liquid inlet tube Double cavity Nd:YAG laser (532nm, 5ns pulse) for internal liquid flow illumination Two 12 bit cameras Rhodamine WT Fluorescent dye

32. Measurements of liquid jets in coaxial atomisers PhotographyOptical connectivity Photographic images are influenced by droplets around the liquid jet core. Not easy to identify location of continuous core length. LIF image is clear. No interference by surrounding droplets/ligaments.

33. Detection of Core geometry by Thresholding Thresholding of the image of the fluorescent core results in the unambiguous detection of the continuous liquid jet contour

34. Simultaneous visualisation of Liquid core from two sides Flow 2a Flow 2e At low air flow-rates Surface disturbances evolve chaotically after about half the continuous core length Small deflection of core from nozzle axis At high air flow-rates Core surface disturbed very close to the nozzle exit No axi-symmetry observed at the deflection of the core Camera 1 Camera 2 Camera 1 Camera 2

35. Average location of liquid core Flows type 3 Flows type 2Flows type 1 Averaging of images of core show no preference of the core location in the mean. Intact length can be tracked from average image Intact length correlates well with MR

36. Destabilization modes of confined coaxial liquid jets G. Charalampous, Y. Hardalupas, A.M.K.P. Taylor

37. 37 Experimental setup Transparent axisymmetric chamber Circular oil jet injected by central nozzle of 3.3mm diameter Annular flow water stream accelerated in the contraction compartment of 12mm diameter and Evolution of central jet morphology observed in mixing compartment Mixed flow decelerated in expansion compartment before exhausting from the test chamber NozzleContraction compartment Mixing compartment Expansion compartment Oil jet Water annular flow

38. Sinusoidal jet mode Central jet develops in a sinusoidal fashion Wavelength of the order of the nozzle diameter Diameter of the central jet is smaller than the nozzle internal diameter due to acceleration Note MR>~15 indicates oil jet diameter thinning 12mm

39. Clustering of mono-disperse and poly- disperse particles in a “box of turbulence” G. Charalampous, Y. Hardalupas

40. Experimental set-up Imaging and Particle Image Velocimetry (PIV) Synthetic jet array PIV camera Double pulse Nd:YAG laser with inter-frame camera Imaging at a plane (60mmx60mm) that crosses the centre of the “box” in the horizontal (X-axis) and vertical (Y–axis) directions –Gas phase seeded with micron sizes particles Measurement of gas velocities across the cross section at the centre of the box –Seeding with monosized particles and water droplets from an air assist atomiser Planar imaging of droplets to evaluate clustering X Y Imaged plane Z

41. Air flow characterisation - Isotropy : RMS x /RMS y RMS x /RMS y : 0.95 Range between: 0.90 to 1.01

42. Characteristics of particles in the “box” of turbulence Diameter (  m) St  St  SpSp ρ p (Kg/m 3 ) 11.9e-52.1e-33.4e-51000 207.5e-38.6e-11.4e-21000 606.7e-27.71.2e-11000 801.2e-1140.221000 1001.9e-1210.31000

43. Radial distribution function Seeding particles (1-3  m ) Unity down to 0.7mm (~2.7  ) Small increase below 0.7mm Negligible mass loading Evidence of clustering for large mono-disperse particles Mass loading ~1.1%±0.6% Evidence of clustering for poly- disperse particles Radial distribution function 80  m mono-disperse particles Radial distribution function Poly-disperse particles (20-100mm)