Veer N. Vatsa

1
Case 1 Overall Summary(Synthetic Jet in
Quiescent Air)

• Veer N. Vatsa
• NASA Langley Research Center
• Hampton, VA

CFD Validation of Synthetic Jets and Turbulent
Separation Control Workshop March 29-31,
2004 Williamsburg, Virginia
2
Outline

• Contributors/codes
• Governing equations/turbulence models
• Grid types/sizes, temporal resolutions
• Cavity modeling
• Initial conditions at slot exit
• Sample of time-averaged results
• Sample of phase-averaged results
• Sample of turbulence results
• Concluding remarks

3
Contributors/codes

• ONERA-flu3m structured, compressible 2D laminar
  and URANS (SA) and 3D LES
• NASA LaRC-tlns3d structured, compressible 2D
  URANS, SA SST
• POITIERS-saturne unstructured, incompressible 2D
  URANS, wall functions, k- e RSM
• WARWICK-neat structured, incompressible 2D
  URANS, k- e (standard and non-linear) EASM

4
Contributors cont.

• UKY-ghost/uncle unstructured, incompressible 2D
  and 3D URANS, SST
• WASHU-wind structured, compressible 2D URANS,
  SA SST, hybrid SST/LES
• GWU-ns structured, incompressible 2D 3D,
  laminar
• NCAT-quasi1d structured, compressible quasi- 1D
  2D, laminar

5
Summary of grid sizes and time-steps
Grid Size 5K-199K (2D) 528K-1.4M cells
(3D) Time-step 72-17496 steps/period
6
Cross-section of Actuator cavity
7
Cavity modeling

• Internal cavity modeled diaphragm motion
  simulated by transpiration condition
• ONERA, NASA LaRC, UKY, WASHU, GWU
• Internal cavity not modeled
• WARWICK top hat sinusoidal profile (power law
  b.l.) at slot exit
• POITIERS top hat profile at slot exit from LDV
  data
• Specialized modeling of cavity
• UKY diaphragm velocity form PIV data curve fit,
  nonzero negative (suction) mean velocity
• WASHU diaphragm pressure iterated for zero net
  mass
• GWU modified cavity, transpiration condition at
  bottom surface
• NCAT quasi1D, includes diaphragm motion (unique)

8
Schematic of cuts along which results are shown
9
Time histories of phase-averaged v-velocityat
orifice (x0, y0.1 mm)
from 1.4 in bound volume
10
Time histories of phase-averaged v-velocityat
orifice (x0, y0.1 mm)
from 1.4 in bound volume
11
Time histories of phase-averaged v-velocityat
orifice (x0, y0.1 mm)
from 1.4 in bound volume
12
Some Observations on v-velocity time history at
slot exit

• Significant differences in PIV and hotwire data
• PIV data deviates from sinusoidal profile in
  phase and amplitude
• Max. and min. velocities are not 1800 apart
• Most CFD results are approximately sinusoidal at
  slot exit and deviate significantly from
  experimental data
• Simulating flow inside cavity produced larger
  differences at slot exit

13
Time-averaged centerline (x0) v-velocity
from 1.7 in bound volume
14
Time-averaged centerline (x0) v-velocity
from 1.7 in bound volume
15
Time-averaged centerline (x0) v-velocity
from 1.7 in bound volume
16
Some Observations on time-averaged
v-velocityalong centerline (x0)

• Significant differences in two sets of
  experimental data (PIV Hotwire), especially in
  the vicinity of slot exit
• Significant variation in CFD results
• Laminar computations indicate largest deviation
  from the norm
• For URANS computations RSM and EASM models
  produced large variation compared with SA, SST or
  k-e
• 3-D modeling of the problem did not necessarily
  improve correlation with experimental data

17
Time-averaged v-velocity at y0.1 mm
from 1.8 in bound volume
18
Time-averaged v-velocity at y0.1 mm
from 1.8 in bound volume
19
Time-averaged v-velocity at y0.1 mm
from 1.8 in bound volume
20
Time-averaged v-velocity at y1 mm
from 1.9 in bound volume
21
Time-averaged v-velocity at y1 mm
from 1.9 in bound volume
22
Time-averaged v-velocity at y1 mm
from 1.9 in bound volume
23
Time-averaged v-velocity at y4 mm
from 1.11 in bound volume
24
Time-averaged v-velocity at y4 mm
from 1.11 in bound volume
25
Time-averaged v-velocity at y4 mm
from 1.11 in bound volume
26
Jet-width comparisons
from 1.18 in bound volume
27
Jet-width comparisons
from 1.18 in bound volume
28
Jet-width comparisons
from 1.18 in bound volume
29
Some Observations on time-averaged v-velocity and
jet-width

• Significant variation in CFD results and PIV data
  near slot exit
• Net suction velocity in UKY simulations at slot
  exit possibly due to input velocity profile
  specification at diaphragm
• Agreement of CFD results with PIV data generally
  improved away from slot exit, until y4 mm
• Large spread in jet-width computed from
  v-velocity
• Disagreement with PIV data and among CFD
  solutions grows with distance from slot exit

30
Time-averaged u-velocity at y1 mm
from 1.14 in bound volume
31
Time-averaged u-velocity at y1 mm
from 1.14 in bound volume
32
Time-averaged u-velocity at y1 mm
from 1.14 in bound volume
33
Time-averaged u-velocity at y4 mm
from 1.16 in bound volume
34
Time-averaged u-velocity at y4 mm
from 1.16 in bound volume
35
Time-averaged u-velocity at y4 mm
from 1.16 in bound volume
36
Time-averaged u-velocity at y8 mm
from 1.17 in bound volume
37
Time-averaged u-velocity at y8 mm
from 1.17 in bound volume
38
Time-averaged u-velocity at y8 mm
from 1.17 in bound volume
39
Some Observations on time-averaged
u-velocityprofiles along fixed y-cuts

• Wider spread among CFD results compared with
  v-velocities. Agreement with PIV data
  deteriorates with distance from slot exit
• Incompressible simulations slightly better in
  lower speed regions
• Greater disparity between laminar and turbulent
  simulations
• Loss of symmetry along centerline more evident

40
Time-histories of v-velocity at x0, y2 mm
from 1.5 in bound volume
41
Time-histories of v-velocity at x0, y2 mm
from 1.5 in bound volume
42
Time-histories of v-velocity at x0, y2 mm
from 1.5 in bound volume
43
Some Observations on time-history of
v-velocityat x0, y2 mm

• Maximum expulsion near 900 phase angle
• Most CFD calculations predict max. expulsion at a
  later phase than PIV data (except POITIERS)
• Maximum suction near 2250 phase angle
• Good agreement between CFD results in phase
• Spread among CFD results and PIV data smaller for
  suction magnitude

44
V-velocity contours, phase90
ONERA-flu3m-2d-sa
WASHU-wind-sa
GWU-ns-lam
UKY-ghost-t003(fine)
NASA-tlns3d-sa(baseline)
from 1.96-1.102 in bound volume
45
V-velocity contours, phase90
WARWICK-neat-ke
NCAT-quasi1d
POITIERS-saturne-ke0.25f
POITIERS-saturne-rsm0.125c
from 1.96-1.102 in bound volume
46
V-velocity contours, phase90
ONERA-flu3m-les
UKY-uncle-3d-sst
GWU-ns-lam (3d)
WASHU-wind-sst/les
from 1.124-1.130 in bound volume
47
V-velocity contours, phase225
ONERA-flu3m-2d-sa
WASHU-wind-sa
GWU-ns-lam
from 1.124-1.130 in bound volume
UKY-ghost-t003(fine)
NASA-tlns3d-sa(baseline)
48
V-velocity contours, phase225
WARWICK-neat-ke
NCAT-quasi1d
POITIERS-saturne-ke0.25f
POITIERS-saturne-rsm0.125c
from 1.124-1.130 in bound volume
49
V-velocity contours, phase225
UKY-uncle-3d-sst
ONERA-flu3m-les
GWU-ns-lam (3d)
WASHU-wind-sst/les
from 1.124-1.130 in bound volume
50
Phase-averaged centerline v-velocity, phase90
from 1.31 in bound volume
51
Phase-averaged centerline v-velocity, phase90
from 1.31 in bound volume
52
Phase-averaged centerline v-velocity, phase90
from 1.31 in bound volume
53
Phase-averaged v-velocity profiles at y2 mm,
phase90
from 1.34 in bound volume
54
Phase-averaged v-velocity profiles at y2 mm,
phase90
from 1.34 in bound volume
55
Phase-averaged v-velocity profiles at y2 mm,
phase90
from 1.34 in bound volume
56
Phase-averaged v-velocity profiles at y4 mm,
phase90
from 1.35 in bound volume
57
Phase-averaged v-velocity profiles at y4 mm,
phase90
from 1.35 in bound volume
58
Phase-averaged v-velocity profiles at y4 mm,
phase90
from 1.35 in bound volume
59
Phase-averaged centerline v-velocity, phase225
from 1.49 in bound volume
60
Phase-averaged centerline v-velocity, phase225
from 1.49 in bound volume
61
Phase-averaged centerline v-velocity, phase225
from 1.49 in bound volume
62
Phase-averaged v-velocity profiles at y2 mm,
phase225
from 1.52 in bound volume
63
Phase-averaged v-velocity profiles at y2 mm,
phase225
from 1.52 in bound volume
64
Phase-averaged v-velocity profiles at y2 mm,
phase225
from 1.52 in bound volume
65
Phase-averaged v-velocity profiles at y4 mm,
phase225
from 1.53 in bound volume
66
Phase-averaged v-velocity profiles at y4 mm,
phase225
from 1.53 in bound volume
67
Phase-averaged v-velocity profiles at y4 mm,
phase225
from 1.53 in bound volume
68
Some observations on phase-averaged velocities

• Agreement among CFD and exp. data is qualitative
• Expulsion phase (900)
• Large differences among CFD solutions and PIV
  data, disagreement increases with distance from
  slot
• CFD simulations that modeled the cavity indicate
  phase discrepancy during expulsion cycle
• Biggest differences seen in laminar simulations
  (except ONERA)
• Suction phase (2250)
• Better agreement between CFD and PIV data near
  slot exit, agreement deteriorates further away
  from slot
• Spread among CFD results smaller compared to
  expulsion cycle (except for laminar computations)

69
Phase-averaged uv profiles along centerline,
phase0
from 1.67 in bound volume
70
Phase-averaged uv profiles along centerline,
phase180
from 1.67 in bound volume
71
Phase-averaged uu profiles along centerline,
phase0
from 1.68 in bound volume
72
Phase-averaged uu profiles along centerline,
phase180
from 1.68 in bound volume
73
Phase-averaged vv profiles along centerline,
phase0
from 1.69 in bound volume
74
Phase-averaged vv profiles along centerline,
phase180
from 1.69 in bound volume
75
Phase-averaged uv profiles at y1 mm, phase0
from 1.70 in bound volume
76
Phase-averaged uv profiles at y1 mm, phase180
from 1.70 in bound volume
77
Some observations on turbulence quantities

• Overall agreement among computations and
  measurements poor
• Cause for such large differences not clear
• Modeling vs. interpretation ?
• Non-dimensionalization ?
• Among available turbulence models, no clear
  trends
• Turbulence quantities were not available for RSM
  and EASM models

78
Concluding Remarks

• Significant differences in PIV, Hotwire and LDV
  data, especially near slot exit
• Significant variation in computational results
  from different groups
• Effect of grid density/time-step refinement
  minimal for a given code
• Laminar results most different
• No particular turbulence model distinctly
  superior
• 3-D simulations did not appear to improve
  correlation with exp. data
• Cavity modeling/initial conditions
• No clear advantage seen by modeling of diaphragm
  motion via transpiration condition vs. initial
  condition specification at slot exit (unexpected)
• Phase averaged comparisons suffer from difficulty
  in matching of phase at slot exit with
  experimental data
• Mismatch of velocity profiles at slot exit
  (phase/amplitude) introduce large errors in the
  field
• Comparison of turbulence quantities very poor
• Requires re-examination of the process