Transcript Document

Experimental
Investigation of
Impeller-Diffuser
Interaction
Rita Patel, Eric Savory and Robert Martinuzzi
Outline
• Background
• Motivation
• Current Work
– Design of experimental rig
– CFD analysis
• Results and discussion
• Conclusions
• Future Work
Terminology
Cumpsty (1978)
Types of Impellers and
Diffusers
Impellers
• Radially ending
• Backswept
• Pre-swirl
• Above w/splitter
blades
Diffusers
• Vaneless
• Vaned
– Radial
– Wedge
• Discrete-passage
Pictures courtesy of Compressor Branch NASA Glenn Research Center
Radial Impeller
Discharge
• Increasing BL on
shroud-suction side
due to curvature
Jet
Wake
PS
SS
– Separation
– Wake on shroudsuction side
– Jet displaced to hubpressure side
Dean and Senoo (1960), Eckardt (1976)
& Krain (1981)
Diffuser Inlet
• Large inlet distortions due to impeller
wake
– Angle and velocity fluctuations
• Distortions have least effect in passage
diffusers than vaned, and most in
vaneless
• Mixing-out of jet-wake stimulated by
presence of vanes
Impeller-Diffuser Interaction
• Vanes
– Stationary vanes produce unsteady pressure
disturbances to rotating impeller, Gallus et al.
(2003)
– Velocity fluctuations of 17-20% in vaneless space,
Krain (1981)
– Cause of backflow to impeller, Cui (2003)
– Decrease traveled distance of impeller discharge
distortion, Ghiglione et al. (1998)
Impeller-Diffuser
Interaction (cont’d)
• Radial Gap
– Too small = increase backflow, Cumpsty and
Inoue (1984)
– Too large = less mixing-out of jet-wake
Gallus et al. (2003)
Motivation
Why study impeller-diffuser interaction
when numerous studies have been done?
• All configurations are different
– This project will lead into the
study of a tandem-bladed
impeller coupled with a fishtail
diffuser
– Study the magnitude and effect
of pressure disturbances in
vaneless space
– Validate previously obtained
CFD results
Picture courtesy of Douglas Roberts (P&WC)
Current Work
• Design a test facility (SCR*) that
simulates a typical radial impeller exit
flow field in steady state through a nonrotating cascade configuration
– 5 stationary radial impeller blades
– Diffuser with 5 flat plate splitters
– Pipe to provide required inlet flow
• Obtain LDV data of flow field
*Stationary Cascade Rig
Purpose of SCR
• CFD
– Experimental validation of results obtained on SCR
• Seeding flow distribution, flow patterns, etc…
• Better understanding of
how to apply LDV
technique to a full-scale rig
– Test use of very small
optical access ports
– Type of seeding for this
specific flow
Picture courtesy of Douglas Roberts (P&WC)
SCR
Impeller + Diffuser
Close-up of impeller
Optical Access
10mm diameter
15mm diameter
Upstream Impeller Blades
Blade passages ‘hub side’
Seeding Ports
Six Ports
SCR Specifications
Ttip
Thub
hblade
Ablade
A5 blades
Ageometrical
A5 passages
m
alpha
c2
cr2
cx1
Inlet
2.51
2.62
23.37
59.95
299.75
2849.67
2549.92
0.288
0.00
n/a
n/a
94.84
Outlet
1.04
4.27
15.75
41.80
209.00
2998.19
2789.19
0.288
72.00
291.69
90.14
n/a
[mm]
[mm]
[mm]
[mm 2]
[mm 2]
[mm 2]
[mm 2]
[kg/s]
deg
[m/s]
[m/s]
[m/s]
•
•
•
•
Outlet Ma: 0.85
Total length: 2.0 [m]
Total height: 1.4 [m]
Similar physical
dimensions of fullscale rig
CFD Analysis
• ICEM CFD 10.0 with CFX 10.0
Mesh:
1.1 million tetrahedral
+ 0.2 million prism
element mesh
• SST k-ω model
Boundary Conditions:
Inlet:
Ptotal = 172.4 [kPa]
Ttotal = 288.15 [K]
Outlet: Pstatic = 101.3 [kPa]
mexpected= 0.245 kg/s
mCFX= 0.242 kg/s
Mach Number Contour Plot
- Impeller-diffuser
only
- Region of high
velocity in left
most passage
- Obtaining close
to desired Ma of
0.85 at impeller
exit
50% blade height
Flow Behaviour
• Shock wave at trailing
edge of each blade
• Passage width
increasing, while height
decreasing
• Greater shock wave in
left passage as result of
diffuser sidewall
Flow Behaviour
Blade Passages
Flow behaviour similar
in passages up to outlet
Migration of high velocity
region to shroud suctionside and vice versa
Pressure Contour Plot
-impellerdiffuser only
- Typical blade
suction/pressure
behaviour
Suction side
Pressure side
- Corresponding
region of low
pressure in left
most passage
Conclusions
• From CFX results
– Presence of separation in diffuser
– No separation in impeller
– Good flow pattern agreement between blade passages
• Close agreement between theoretically
calculated and CFX values at boundaries
• SCR will provide a good understanding of how
to apply LDV technique in a high-speed, highlyconfined, compressible flow
Future Work
• Experimental
– Measurements in SCRF
– Compare with current CFD results
• Computational
– Track seeding particles
• Apply LDV technique and CFD model on fullscale rig at P&WC
Acknowledgements
• Advanced Fluid Mechanics Research Group
– http://www.eng.uwo.ca/research/afm/default.htm
• Kevin Barker and Doug Phillips
– University Machine Shop
• Rofiqul Islam
– University of Calgary
• Suresh Kacker, Douglas Roberts, Feng Shi and Peter
Townsend
– Pratt and Whitney Canada
Thank You
Questions?