The following presentation describes the use of numerical simulation to optimize the shape, elasticity and volume of the compensation chamber in an axial pump while minimizing pressure peaks.

1. Using MpCCI to Model
Fluid-Structure-
Interactions with
Abaqus and 3rd party
CFD Tools
Bettina Landvogt, scapos AG
Burak Ozturk, VIAS
Arindam Chakraborty, VIAS

2. Agenda
▪ Introduction: scapos and MpCCI
▪ scapos AG
▪ MpCCI Products
▪ MpCCI for FSI
▪ Fluid-Structure Interaction Examples
▪ Elastic Flap => MpCCI GUI
▪ Hydraulic Pump
▪ Airfoil Benchmarks
▪ Racing Car Spoilers
▪ Rotating Fans

3. Introduction: scapos and MpCC

4. Introduction – scapos AG
scapos AG supports research institutes and small
enterprises (software development) with
marketing and sales&distribution expertise
Founded in January 2009 (Fraunhofer spin-off)
Employees 10
Location Schloss Birlinghoven
Sankt Augustin, Germany

5. scapos –Selected Software Products
Knowledge Mining
Digital
Manufacturing
CAE Tools
Nesting &Packing
Parallel Software
GPI
Audio
Mining

6. • MpCCI CouplingEnvironment: vendor-neutral interface for coupling of simulation codes:
– Bi-directional data transfer via fast socket connections, handling of non-matching
interface geometries
– Different coupling schemes: explicit or iteratively implicit transient, steady-state or
mixed (steady-state and transient)
– Easy usage via GUI or batch, visualization of results
• MpCCI FSIMapper: file-based mapping of CFD loads to FEA (pressures or thermal quantities
can be used as boundary conditions or loads)
• MpCCI Mapper: Independent Mapping Tool for Integrated Simulation Workflows (e.g.
mapping from manufacturing simulations to crash simulations input decks)
Multiphysics Products

7. Fluid
• ANSYS ICEPAK
• ANSYS Fluent
• Fine/Turbo
• Fine/Open
• OpenFOAM
• STAR-CCM+
• STAR-CD
Radiation
• RadTherm/
TAITherm
Structure
• Abaqus
• ANSYS Mech.
• MSC.Nastran
• MSC.Marc
System Models
• FlowMaster/
FloMASTER
• MATLAB
• MSC.Adams
• SIMPACK
• FMI/EAS-Master
Electro
• ANSYS Emag
• FLUX
• JMAG
Inhouse Codes
• API To MpCCI
MpCCI Coupling Environment

8. Fluid-Structure Interactions
Aerodynamic fields
incorporating current deflection
• OpenFOAM
• ANSYS Fluent
• Star-CCM+
CFD
Deformation of part (e.g. wing,
spoiler)
• Abaqus
• ANSYS Mechanical
• MSC.Nastran
FEA
RelWallForce
OverPressure
• CFD Code has to adapt the mesh to the
new position sent by FEA code. This can
either be handled by the CFD code itself
(Fluent, Star-CCM+) or by the MpCCI
morpher module (particularly for
OpenFOAM)
NPosition
• Different Coupling Algorithms are
available (transient, steady-state, mixed,
iterative, Gauß-Seidel, Jacobi,..)
• Examples for FSI: Valves, Ships, Cars,
Airplanes, Container, Thermal FSI

9. Fluid-Structure Interaction Examples

10. Elastic Flap
• MpCCI Tutorial
• Show Workflow and
GUI

11. Problem Description
CFD FEA
• ANSYS Fluent
• Model of channel
• Dynamic mesh
• Abaqus
• Model of flap
• BC: fixed top
RelWallForce
OverPressure
NPosition

12. Hydraulic Axial Pump
• Complicated CFD Model
• Compressible Liquid

13. Problem Description
During a research project at the University of Gdansk a
new axial (constant and variable displacement) pump
with a cam-driven commutation unit was developed.
Prototypes showed good performance but possibly
harmful pressure peaks at beginning of periods of
disconnection.
Idea: shorten the period of disconnection
with an additional chamber
(compensation chamber) equipped with
an elastic wall.

14. Pressure peaks reduced by 50%,
without affecting efficiency!
TODO:
Use Numerical Simulation to
optimize the shape, elasticity and
volume of chamber while
minimizing pressure peaks.
Compensation Chamber
Elastic Wall =>
FSI Simulation

15. CFD and FEA Models
Symmetric half model with two chambers
and full model with seven chambers in Fluent
Symmetric Abaqus model
with shell elements
CFD FEA

16. CFD FSI
(FLUENT) (Abaqus-MpCCI-Fluent)
Pressure Plots at three discrete points of the pump.
Results of Coupled and CFD only Simulations
FSI results fitted significantly
better to experiments than CFD
stand-alone simulations

17. Airfoil Benchmarks
• Validation of Coupled FSI
• Coupling Algorithms

18. ▪ High Reynolds Number Aero-Structural Dynamics, DFG
funded project at RWTH Aachen
(=> HIRENASD)
▪ Reynolds numbers up to 80 million, high transonic Mach
numbers (strong non-linearities like compression waves
and steady and transient stalls )
=> passenger aircraft cruising
Problem Description
Experiments:
Steady and transient measurements using a 1.3 m wing with a profile typical for a
big passenger aircraft
Measuring equipment:
250 miniature pressure sensors, 11 acceleration sensors, 22 strain sensors

19. Material Properties
Young’s modulus 2.1e11 Pa
Poisson number ~0.3125 (dep. on T)
Density 7860 kg/m3
Therm. Expansion
Coefficient
~1.3e-05 (dep. onT)
Discretisation ~200 000
Tet10 elements
FEA Model

20. Material properties - Air
Density Ideal gas
Viscosity 1.54434 kg/m s
Boundary Conditions
PressureInlet: Pressure 76239 Pa
Temperature 265.957 K
Pressure Outlet 0 Pa
Operating Pressure 137764 Pa
▪ Experimental measurements are available for different configurations.
▪ For example: Ma=0.8, Re=7 million
CFD Model

21. CFD Model
Discretisation:
~15 million tetrahedral and prismatic
cells
14 boundary layers
Solver:
Transient: density-based solver, 2nd order
implicit, steady state starting solution
Steady: implicit, density-based solver
Spalart-Allmaras turbulence model

22. The pressure coefficient is
evaluated at seven “cuts”
on the wing.
Comparison with Experiments

23. FSI solution – coarse
FSI solution – medium
FSI solution – fine
Experiment (lower)
Experiment (upper)
X/CX/C X/C
Cp Cp Cp
Section 1
Section 3
Section 7
Comparison with Experiments

24. Racing Car Spoilers • Selective Morphing
• Cluster/batch integration
• FSIMapper for “Pre-Check“

25. Problem Description
Example: Rear Spoiler of F1 racing car
Made up of three spoiler elements, attached to
the side plate.
Simulation models are symmetric half models.

26. CFD and FEA Simulation Models
CFD FEA
• Star-CCM+, OpenFOAM
• Model of 8 m long car in 30 m
long channel
• Different simulations: steady
state, transient
• Abaqus
• Only spoiler or other
interesting parts
• Interior structure and material
properties can be complicated
RelWallForce
OverPressure
NPosition

27. Mesh motion in CFD
In the CFD code: mesh has to be adapted to deformation
calculated by FEA.
All CFD codes offer multiple methods for mesh motion.
For OpenFOAM (as an alternative to OF mesh motion solvers):
MpCCI Mesh Morpher
• spring-based morphing tool for MpCCI
• Offers the possibility to morph only small regions close to
deforming boundary => save computational time
Cell zone defined
for morphing
Selective Morphing
CFD NPosition

28. Animation of the
displacement for the
two coupling surfaces:
FEA model (Abaqus)
CFD model (Star-CCM+)
Results of Coupled Simulation

29. Results of Coupled Simulation
Overpressure contour on coupling surfaces for
Abaqus and Star-CCM+ as shown in the MpCCI

30. Example Case HPC Formula 1 Case
CFD model size 16 million cells >> 100 million cells
CFD run time 10 min per 50 iterations
(steady state, 32 cpus)
Confidential
FEA model size 14,000 cells 1 million cells
FEA run time << 1 minute per coupling
step, static, 2 cpus
Confidential
Coupling surface size 14,000 cells 1 million cells
multiple regions
Add. Coupling run time
Initialization
Per Coupling Step
Morphing of OF mesh
12.4 s
<< 1 second
~20 seconds
Confidential
Run Times and Size of Models

31. Spoiler CFD FSI
Drag
Coefficient
Top 0.1812 0.2139
Middle 0.061 0.0678
Downforce
Coefficient
Top 0.1778 0.2396
Middle 0.2900 0.3087
Maximum
Deformation
Top - 2 mm
Middle - 6 mm
Drag and Downforce Coefficient for the two coupled spoilers for
CFD stand-alone and coupled FSI simulations.
(Coupling of steady state CFD simulation with static FEA.
Subcycling for CFD: 10 iterations.)
Comparison of FSI and CFD stand-alone Results
Results are different: FSI
simulation captures more
effects and is more realistic!

32. Rotating Fans
• Rotating Reference Frame
Definitions

33. • CFD (Fluent) and FEA (Abaqus) simulation of fan rotating with 1700 rpm
• Both rotation and local deformation considered
• Transient coupling scheme
• Mixed mesh motion model
Problem description

34. CFD and FEA Simulation Model
CFD FEA
• Fluent
• Dynamic Mesh, Sliding
Mesh
• 900,000 tetrahedral elements
• Abaqus
• Moving Reference Frame
• 34,000 quadratic tetrahedral
elements
RelWallForce
OverPressure
NPosition

35. • Nodal distance (mm) of the
FSI meshes
• Overpressure (Pa) on the FSI
meshes
Results of Coupled Simulation

36. Vibrations of
control points
when FSI
considered
Vibrations of
control points
when FSI not
considered
Results of Coupled Simulation
Results are different: FSI
simulation captures more
effects and is more realistic!

37. Conclusion
Coupled Simulations are easy to set up:
Stand-alone models can be used without (almost) any additional work.
Settings can be handled in the MpCCI GUI.
Monitoring the Coupled Simulation
Coupled quantities can be monitored during and after the simulation.
Simulation codes write their usual output files (e.g. Abaqus .odb).
Additional tool: MpCCI FSIMapper to check whether FSI is necessary
=> Efficient use of Simulation Time

38. Conclusion
Results of Coupled Simulations:
FSI simulations include more physical effects than CFD (or FEA) stand-alone simulations.
FSI gives better predictions of the real world.
Coupled Simulations are easy to set up:
Stand-alone models can be used without (almost…) any additional work.
Settings can be handled in the MpCCI GUI.
Monitoring the Coupled Simulation
Coupled quantities can be monitored during and after the simulation.
Simulation codes write their usual output files (e.g. Abaqus .odb).
Additional tool: MpCCI FSIMapper to check whether FSI is necessary
=> Efficient use of Simulation Time

39. Thank you for your
attention.
Any Questions?