studies of the ercoftac centrifugal pump with openfoam
TRANSCRIPT
Studies of the ERCOFTAC Centrifugal Pumpwith OpenFOAM
Shasha Xie
June 7, 2010
Title page - - Shasha Xie June 7, 2010 1/40
Outline
Outline
I Purpose and goal
I Method and approach
I Geometry
I Boundary condition
I Results
I Conclusions
Outline - - Shasha Xie June 7, 2010 2/40
Purpose and goal
I To investigate the rotor-stator interactions in the ERCOFTACCentrifugal Pump using OpenFOAM 1.5-dev.
I Carry out the unsteady simulation for both 2D and 3D modeling.
I Compare the results of the numerical solution with the experimentaldata.
Purpose and goal - - Shasha Xie June 7, 2010 3/40
Method and Approach
I MethodI 2D and 3D mesh were generated using ICEM-Hexa.I Incompressible Reynolds-Averaged Navier-Stokes equations are solved.I The standard k-ε turbulence model is used.I Generalized Grid Interface (GGI) is used:
I for the steady-state simulations.I for the unsteady simulations.
I ApproachI 2D steady-state simulationI 2D unsteady simulationI 3D steady-state simulationI 3D unsteady simulation
Method and Approach - - Shasha Xie June 7, 2010 4/40
Geometry
I Geometry of the ERCOFTAC Centrifugal Pump.
I The positions where the simulated results are plotted.
Figure: Geometry and positions for plotting the simulated results.
Geometry - - Shasha Xie June 7, 2010 5/40
Geometry
I Geometry of 2D (left) and 3D (right) model.
Geometry - - Shasha Xie June 7, 2010 6/40
Boundary conditions
I Boundary conditions.
Calculated data for the 2D cases for the 3D casesInlet Diameter D0=184 mm D0=200 mm
Z thickness Z=1 mm Z=40 mm
Flow rate Q=ϕU2πD2
24 =0.292 m3/s Q=0.292 m3/s
Inlet speed U0= QA0
= Q2πr0∗0.04=11.4 m/s U0=10.98 m/s
Rotating speed ω = 2000 rpm ω = 2000 rpm
Boundary conditions for the 2D cases for the 3D casesAt the inlet Vradial=U0 Vaxial=U0
µTµ =10 µT
µ =10
k=32U2
0 I 2=0.48735 m2/s2 (I=5%) k=0.4521 m2/s2
ε=Cµρk2
µT=
Cµρk2
µ(µT /µ)=Cµk2
ν(µT /µ)
At the outlet Average static pressure 0
Boundary conditions - - Shasha Xie June 7, 2010 7/40
2D steady-state simulation
2D steady-state simulation.
Results - 2D steady-state simulation - Shasha Xie June 7, 2010 8/40
Set-up for the 2D steady-state simulation
I Set-up for the case 2DSteady.
Schemes Convection schemes of U linearUpwind Gauss
Solvers p GAMGsmoother GaussSeideltolerance 1.0e-08relTol 0.05
U,k ,ε smoothSolversmoother GaussSeideltolerance 1.0e-07relTol 0.1
Results - 2D steady-state simulation - Shasha Xie June 7, 2010 9/40
Results for the case 2DSteady
I Contours of the relative velocity magnitude (left) and static pressure(right) for the case 2DSteady.
Results - 2D steady-state simulation - Shasha Xie June 7, 2010 10/40
Results for the case 2DSteady
I Distribution of the radial (top left), tangential (top right) velocitiesand static pressure coefficient (down) for the case 2DSteady.
0
0.05
0.1
0.15
0.2
0.25
0.3
0 0.5 1 1.5 2
Wr/U
2
yi/Gi
experimental2DSteady
-0.9
-0.8
-0.7
-0.6
-0.5
-0.4
-0.3
0 0.5 1 1.5 2
Wu/
U2
yi/Gi
experimental2DSteady
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 0.5 1 1.5 2
C~p
yi/Gi
experimental2DSteady
Results - 2D steady-state simulation - Shasha Xie June 7, 2010 11/40
2D unsteady simulation
2D unsteady simulation.
I Compare the convection scheme:2DEulerU0.5T, 2DEulerL0.5T
I Compare the time discretization scheme:2DBackL0.5T, 2DEulerL0.5T, 2DCN0.5L0.5T
I Compare the Crank-Nicholson off-centering coefficient:2DCN0.2L0.5T, 2DCN0.5L0.5T, 2DCN0.8L0.5T, 2DCN1.0L0.5T
I Compare the maximum Courant Number:2DCN0.5L0.5T, 2DCN0.5L1.0T, 2DCN0.5L2.0T, 2DCN0.5L4.0T
I Compare the transient solver:2DCN0.5L0.5T, 2DCN0.5L0.5S
Results - 2D unsteady simulation - Shasha Xie June 7, 2010 12/40
Stop time is set 0.3 s
I Pressure development of 2D unsteady simulation at Probe 1 (top left),Probe 2 (top right) and Probe 3 (down left) until fully developed.
-960
-940
-920
-900
-880
-860
-840
0 0.05 0.1 0.15 0.2 0.25 0.3
p/rh
o
Time [s]
Probe 1
-200
-150
-100
-50
0
50
0 0.05 0.1 0.15 0.2 0.25 0.3
p/rh
o
Time [s]
Probe 2
-40
-30
-20
-10
0
10
20
30
40
0 0.05 0.1 0.15 0.2 0.25 0.3
p/rh
o
Time [s]
Probe 3
Results - 2D unsteady simulation - Shasha Xie June 7, 2010 13/40
Compare the convection scheme
I Set-up for the case 2DEulerU0.5T and 2DEulerL0.5T.Case Convection scheme
2DEulerU0.5T upwind2DEulerL0.5T linear upwind
Time discretization scheme Euler
maxCo 0.5
Solver turbDyMFoam
Correctors nCorrectors 2nOuterCorrectors 1nNonOrthogonalCorrectors 1
endTime 0.3 s
Results - 2D unsteady simulation - Convection scheme Shasha Xie June 7, 2010 14/40
Compare the convection scheme
I Contours of the relative velocity magnitude for the case2DEulerU0.5T (left) and 2DEulerL0.5T (right).
Results - 2D unsteady simulation - Convection scheme Shasha Xie June 7, 2010 15/40
Compare the convection scheme
I Distribution of the radial velocity for the case 2DEulerU0.5T and2DEulerL0.5T.
0
0.05
0.1
0.15
0.2
0.25
0.3
0 0.5 1 1.5 2
wr/U
2
yi/Gi
t/Ti=0.126
experimental2DEulerU0.5T2DEulerL0.5T
Results - 2D unsteady simulation - Convection scheme Shasha Xie June 7, 2010 16/40
Compare the time discretization scheme
I Set-up for the case 2DBackL0.5T, 2DEulerL0.5T and 2DCN0.5L0.5T.Case Time discretization scheme Computing time
2DBackL0.5T backward 22.0 hours
2DEulerL0.5T Euler 22.7 hours
2DCN0.5L0.5T Crank-Nicholson 0.5 23.9 hours
Convection scheme linearUpwind
maxCo 0.5
Solver turbDyMFoam
Correctors nCorrectors 2nOuterCorrectors 1nNonOrthogonalCorrectors 1
endTime 0.3 s
Results - 2D unsteady simulation - Time discretization scheme Shasha Xie June 7, 2010 17/40
Compare the time discretization scheme
I Distribution of radial velocity for the case 2DBackL0.5T,2DEulerL0.5T and 2DCN0.5L0.5T.
0
0.05
0.1
0.15
0.2
0.25
0.3
0 0.5 1 1.5 2
wr/U
2
yi/Gi
t/Ti=0.126
experimental2DBackL0.5T2DEulerL0.5T
2DCN0.5L0.5T
Results - 2D unsteady simulation - Time discretization scheme Shasha Xie June 7, 2010 18/40
Compare the time discretization scheme
I Contours of the relative velocity magnitude (left) and static pressure(right) for the case 2DBackL0.5T.
Results - 2D unsteady simulation - Time discretization scheme Shasha Xie June 7, 2010 19/40
Compare the time discretization scheme
I Contours of the static pressure coefficient for the case 2DBackL0.5T(left) and experimental results (right).
Results - 2D unsteady simulation - Time discretization scheme Shasha Xie June 7, 2010 20/40
Compare the Crank-Nicholson off-centering coefficient
I Set-up for the case 2DCN0.2L0.5T, 2DCN0.5L0.5T, 2DCN0.8L0.5Tand 2DCN1.0L0.5T.
Case Time discretization scheme
2DCN0.2L0.5T Crank-Nicholson 0.22DCN0.5L0.5T Crank-Nicholson 0.52DCN0.8L0.5T Crank-Nicholson 0.82DCN1.0L0.5T Crank-Nicholson 1.0
Convection scheme linearUpwind
maxCo 0.5
Solver turbDyMFoam
Correctors nCorrectors 2nOuterCorrectors 1nNonOrthogonalCorrectors 1
endTime 0.3 s
Results - 2D unsteady simulation - Crank-Nicholson coefficient Shasha Xie June 7, 2010 21/40
Compare the Crank-Nicholson off-centering coefficient
I Pressure fluctuation at Probe 1 for the case 2DCN0.2L0.5T (top left),2DCN0.5L0.5T (top right) and 2DCN0.8L0.5T (down left).
-940
-930
-920
-910
-900
-890
0.29 0.291 0.292 0.293 0.294 0.295
p/rh
o
Time [s]
2DCN0.2L0.5T
-940
-930
-920
-910
-900
-890
0.29 0.291 0.292 0.293 0.294 0.295
p/rh
o
Time [s]
2DCN0.5L0.5T
-940
-930
-920
-910
-900
-890
0.29 0.291 0.292 0.293 0.294 0.295
p/rh
o
Time [s]
2DCN0.8L0.5T
Results - 2D unsteady simulation - Crank-Nicholson coefficient Shasha Xie June 7, 2010 22/40
Compare the maximum Courant Number
I Set-up for the case 2DCN0.5L0.5T, 2DCN0.5L1.0T, 2DCN0.5L2.0Tand 2DCN0.5L4.0T.
Case maxCo Time step Computing time
2DCN0.5L0.5T 0.5 0.80 ∗ 10−5s 23.9 hours
2DCN0.5L1.0T 1.0 1.58 ∗ 10−5s 11.7 hours
2DCN0.5L2.0T 2.0 3.13 ∗ 10−5s 6.4 hours
2DCN0.5L4.0T 4.0 6.20 ∗ 10−5s 3.5 hours
Time discretization scheme Crank-Nicholson 0.5
Convection scheme linearUpwind
Solver turbDyMFoam
Correctors nCorrectors 2nOuterCorrectors 1nNonOrthogonalCorrectors 1
endTime 0.3 s
Results - 2D unsteady simulation - Maximum Courant Number Shasha Xie June 7, 2010 23/40
Compare the maximum Courant Number
I Distribution of the radial velocity for the case 2DCN0.5L0.5T,2DCN0.5L1.0T, 2DCN0.5L2.0T and 2DCN0.5L4.0T.
0
0.05
0.1
0.15
0.2
0.25
0.3
0 0.5 1 1.5 2
wr/U
2
yi/Gi
t/Ti=0.126
experimental2DCN0.5L0.5T2DCN0.5L1.0T2DCN0.5L2.0T2DCN0.5L4.0T
Results - 2D unsteady simulation - Maximum Courant Number Shasha Xie June 7, 2010 24/40
Compare the transient solver
I Set-up for the case 2DCN0.5L0.5T and 2DCN0.5L0.5S.
Case Solver nCorrectors nOuter-Correctors
nNon-Orthogonal-Correctors
2DCN0.5L0.5T turbDyMFoam 2 1 1
2DCN0.5L0.5S transientSim-pleDyMFoam
0 1 0
Time discretization scheme Crank-Nicholson 0.5
Convection scheme linearUpwind
maxCo 0.5
endTime 0.3 s
Results - 2D unsteady simulation - Transient Solver Shasha Xie June 7, 2010 25/40
Compare the transient solver
I Contours of the relative velocity magnitude for the case2DCN0.5L0.5T (left) and 2DCN0.5L0.5S (right).
Results - 2D unsteady simulation - Transient Solver Shasha Xie June 7, 2010 26/40
Compare the transient solver
I Distribution of the radial velocity for the case 2DCN0.5L0.5T and2DCN0.5L0.5S.
0
0.05
0.1
0.15
0.2
0.25
0.3
0 0.5 1 1.5 2
Wr/U
2
yi/Gi
t/Ti=0.126
experimental2DCN0.5L0.5T2DCN0.5L0.5S
Results - 2D unsteady simulation - Transient Solver Shasha Xie June 7, 2010 27/40
3D steady-state simulation
3D steady-state simulation.
Results - 3D steady-state simulation - Shasha Xie June 7, 2010 28/40
Set-up for the 3D steady-state simulation
I Set-up for the case 3DSteady.
Schemes Convection schemes of U linearUpwind Gauss
Solvers p,U,k ,ε GAMGsmoother GaussSeideltolerance 1.0e-08relTol 0.05
Results - 3D steady-state simulation - Shasha Xie June 7, 2010 29/40
Results for the case 3DSteady
I Contours of the relative velocity magnitude (left) and static pressure(right) at the midspan position for the case 3DSteady.
Results - 3D steady-state simulation - Shasha Xie June 7, 2010 30/40
Results for the case 3DSteady
I Distribution of the radial (top left), tangential (top right) velocitiesand static pressure coefficient (down) at the midspan position for thecase 3DSteady.
0
0.05
0.1
0.15
0.2
0.25
0.3
0 0.5 1 1.5 2
Wr/U
2
yi/Gi
experimental3DSteady
-0.9
-0.8
-0.7
-0.6
-0.5
-0.4
-0.3
0 0.5 1 1.5 2
Wu/
U2
yi/Gi
experimental3DSteady
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 0.5 1 1.5 2
C~p
yi/Gi
experimental3DSteady
Results - 3D steady-state simulation - Shasha Xie June 7, 2010 31/40
3D unsteady simulation
3D unsteady simulation.
Results - 3D unsteady simulation - Shasha Xie June 7, 2010 32/40
Set-up for the 3D unsteady simulation
I Set-up for the case 3DBackL0.5S.
Case 3DBackL0.5S
Time discretization scheme backward
Convection scheme linearUpwind
maxCo 0.5
Stop time 0.3 s
Solver transientSimpleDyMFoam
Correctors nCorrectors 0nOuterCorrectors 1nNonOrthogonalCorrectors 0
Computing time 113.5 hours < 5 days
Results - 3D unsteady simulation - Shasha Xie June 7, 2010 33/40
Stop time is set 0.3 s
I Pressure development at Probe 1 (top left), Probe 2 (top right) andProbe 3 (down) for the 3D unsteady simulation until fully developed.
-1000
-950
-900
-850
-800
-750
-700
-650
0 0.05 0.1 0.15 0.2 0.25 0.3
p/rh
o
Time [s]
Probe 1
-200
-150
-100
-50
0
50
100
0 0.05 0.1 0.15 0.2 0.25 0.3
p/rh
o
Time [s]
Probe 2
-40
-30
-20
-10
0
10
20
30
40
0 0.05 0.1 0.15 0.2 0.25 0.3
p/rh
o
Time [s]
Probe 3
Results - 3D unsteady simulation - Shasha Xie June 7, 2010 34/40
Results for the case 3DBackL0.5S
I Contours of the relative velocity magnitude (left) and static pressure(right) at the midspan position for the case 3DBackL0.5S.
Results - 3D unsteady simulation - Shasha Xie June 7, 2010 35/40
Results for the case 3DBackL0.5S
I Distribution of the radial (top left), tangential (top right) velocitiesand static pressure coefficient (down left) at the midspan position forthe case 3DBackL0.5S.
0
0.05
0.1
0.15
0.2
0.25
0.3
0 0.5 1 1.5 2
Wr/U
2
yi/Gi
t/Ti=0.126
experimental3DBackL0.5S
-0.9
-0.8
-0.7
-0.6
-0.5
-0.4
-0.3
0 0.5 1 1.5 2
Wu/
U2
yi/Gi
t/Ti=0.126
experimental3DBackL0.5S
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 0.5 1 1.5 2
C~p
yi/Gi
t/Ti=0.0
experimental3DBackL0.5S
Results - 3D unsteady simulation - Shasha Xie June 7, 2010 36/40
Results for the case 3DBackL0.5S
I Contours of the radial (left) and tangential (right) velocities fordifferent span distances for the case 3DBackL0.5S (top) andexperimental (down).
Results - 3D unsteady simulation - Shasha Xie June 7, 2010 37/40
Conclusion
I All the computational cases show some similarities with theexperimental results.
I The unsteady simulations show better prediction of the wakes thanthe steady-state simulation.
I The first-order upwind convection scheme failed in capturing the wakeeffect of the flow unsteadiness.
I Balance between short computing time and damping on the choice ofmaximum Courant Number.
I The transientSimpleDyMFoam solver shows the possibility for the 3Dunsteady simulation but still needs more validations and more testings.
Conclusion - - Shasha Xie June 7, 2010 38/40
Acknowledgements
I would like to say my thanks to
I Division of Fluid Dynamics, Department of Applied Mechanics
I Supervisors Hakan Nilsson and Olivier Petit
Acknowledgements - - Shasha Xie June 7, 2010 39/40
Thanks!
Thank you for listening.Questions?
End - - Shasha Xie June 7, 2010 40/40