fluid-structure interaction by the mixed sph-fe method...

Copyright © ESI Group, 2011 All rights reserved.Copyright © ESI Group, 2011. All rights reserved.

Paul GroenenboomESI Group

Delft, Netherlands

Fluid -structure Interaction by the mixed SPH -FE Method with Application to Aircraft Ditching

Conference on SPH and Particle Methods for Fluids and Fluid Structure Interaction

Lille, France, 21.-22.01.2015

Martin SiemannGerman Aerospace Center (DLR)

Stuttgart, Germany

Copyright © ESI Group, 2011 All rights reserved.

Outline

• Introduction• Computational approach

• SPH• Coupling with structures

• Innovations• Pressure correction• Particle regularization• Damping• Initial particle distributions• Periodic boundaries

• Guided Ditching Tests• Conclusions & Perspectives

Fluid-structure Interaction by the mixed SPH-FE Method with Application to Aircraft Ditching


Computational Approach: SPH

The SPH solver within VPS (ESI-Group) is based on the ‘standard’ weakly-compressible SPH algorithm

There are many innovative extensions to improve accuracy and performance, in particular for fluid-structure interaction simulation

The SPH solver is fully integrated within the explicit Finite Element Method (FEM) of VPS


Coupling with structures

SPH is suitable to model violent flow of water.FEM is best suited to model the (aircraft) structureA hybrid SPH-FE approach allows to use ‘best of both worlds’ to model fluid-structure interaction.

VPS/PAM-CRASH software from ESI-GroupThe penaly-based contact algorithm between FE and SPH allows to model fluid-structure interaction.This approach combines accurary with good CPU performance.For regions with limited fluid displacements it is possible to use finite elements for water.


Coupling of FE and SPH

VPS/PAM-CRASH Contact treatment: Adding dynamic connectivity between a node and a contact segment



Ditching simulation of an Airbus 321 model (Courtesy of DLR) involves sliding interface contact between particles and the aircraft model, and tied contact of particles with the brick elements for the water.



It is possible to include tension by definition of a ‘seperation stress’ once contact has been established – this can model suction effects.For aircraft ditching suction effects are important for the aircracft kinematics.


Innovations: Pressure CorrectionNecessity and Requirements

Standard WC-SPH method � poor pressure distributions (high-frequency oscillations in time and space)

Established pressure correction methods like density re-initialization by Shepard filtering and Rusanov flux were recently implemented in the SPH solver of VPS/PAM-CRASH


SPH Pressure Correction MethodsDensity Re-Initialization using Shepard

Filtering

Density re-initialization method which was derived from an interpolation technique initially published by Shepard in 1968

SPH notation for the modified density

Since for WC-SPH the pressure is derived from an equation-of-state, the Shepard filter directly influences the pressure distribution. The density field is periodically re-initialized at user-defined cycle frequency f with recommended values of 20 cycles


SPH Pressure CorrectionRusanov Flux

The Rusanov correction is efficient and robust, but also somewhat diffusive, numerical approximation to solve Riemann problems

1st order accuracy compared to 2nd order accuracy of Riemann flux

Modified continuity equation


Pressure field at t = 30ms in 2D NACA flat plate test case

SPH Pressure Correction MethodsExemplary effects on pressure field

Shepard filteringf = 20 Hz

Rusanov fluxε = 0.5

Standard WC-SPH(no correction)


Pressure correction: Test CaseTwo-Dimensional Rigid Wedge Vertical Impact

Experimental results from Battley et al.Vertical impact (3 m/s , const.)Pressure results available for three positions along center line (keel-chine)

Numerical modelSymmetryRusanov flux with ε = 0.5Smoothing length h = 2 mm (810 000 particles)

Good correlation of peak pressure values


Innovations: Particle Regularization

During flow or deformation the particle distribution may display some irregularities with as consequence:

Pressure oscillations

Clumping of particles

Numerical (tension) instability

Counteract by particle regularization methods which aim to yield a more regular distribution

Effect should be local

Conservation of mass, momentum, and energy

Numerical stability

No significant increase of computational costs


Particle Regularization

Test case for floating boxes with the VJA algorithmParticle distribution with contours of the vertical displacement at the final state for VJA2 (partial view)


Innovations: Damping zone

Absorbing boundary conditions for pressure are available.Absorbing boundary conditions for surface waves have to be different as they involve gross motion of the material.The proposed algorithm is nodal damping in specific regionsTested for wave propagation and impact

Damping volume


Damping zone

Wedge impact with damping zone near the impact region

Contour of the horizontal displacements at the final state for the damping case.


Uniform initial distribution (2D)

volume ratio 1:9

Non-Uniform Initial Particle Distributions

Define suitable initial particle distributions for SPH fluid impact and flow simulations.Two topics of special interest:

1. Particles of non-uniform size2. Filling of arbitrary volumes

Proposed Solution �� Weighted Voronoi Tessellation (WVT)


Filling simulation (2D)Distribution of smoothing length (2D)


� Fast and easy to use



Gravity load test case (2D) – instable; Color denotes velocity magnitude


g

Distribution of smoothing length (2D)

0.05 m/s ca. 1800 ms

� Fast and easy to use



Gravity load test case (2D) – stable; Color denotes velocity magnitude

ca. 10 000 ms0.02 m/s


g


� Stable under gravity load� Fast and easy to use



Cutting plane

y xz

Initial results for guided ditching simulation (3D)



� Significant CPU time savings (>10x through WVT)



Innovations: Periodic boundaries

• Allows to re-enter particles that have reached a boundary on the opposite side.• Provides a significant reduction of the number of particles required for a moving

object in contact with fluid.• Extended to incorporate translating domains• Extended to allow for inflow with undisturbed flow conditions


Ditching

aircraft emergency situation that ends with planned impact of the aircraft on waterInvolves complex phenomena (water impact, suction, spray, cavitation, aerodynamics, structural deformation)Difficult to test (motion control, scale effects, aerodynamics)Challenge for simulation (FSI, free surface…)


Guided Ditching Tests

Test setup (at INSEAN in Rome) :Simple geometries in aerospace design:panels with skin panel dimensions and thickness Represent high inertia of aircraft : guided motion Full-scale ditching conditions (representative impact velocities): sling shot system


High-Speed Camera @ 5400 fps


60 m/s

Challenges for SPH Model

100 m

10 m

5 m

Objectives

- Reduce amount of particles (run time)

- Allow for finer particles in proximity of impacting structure

Large fluid domain due to relative high forward velocity

Fine particle distribution needed to allow for accurate results

Drivers for run time

Guided Ditching Tests


GDT selected Results

Selected Results – Force Z10o 30 – 40 – 46 m/s ALU 15 mm


x

z

Volume cut of trolley & SPH domain


Selected Results – Force Z6o 40 m/s ALU 0.8 mm


S4

xy

S4x

S4y

Selected Results – Strain6o 40 m/s ALU 3 mm

ESIZE ~10 mm


STRAIN YY3x amplified

S4


S1 xy

S1x

S1y

Selected Results – Strain6o 40 m/s ALU 3 mm


STRAIN YY3x amplified

STRAIN XX3x amplified

S1

S1


Suction ForceCN235 sub-scale model

Good correlation of pressure results between test and simulation (overpressure and suction regions)

Time [s]

Pre

ssur

e [P

a]P

ress

ure

[Pa]


Suction ForceCN235 sub-scale model

Comparison suction contact model ON (top) and OFF (bottom)

Good corellation between test and simulation with suction model ONHigh influence on aircraft pitch kinematic

Suction model ON

Suction model OFF

Test (No. 25)Suction model ON

Suction model OFF

Time [s]

Pitc

h A

ttitu

de [d

eg]


Conclusions

SPH-FE approach suitable to simulate deformable aircraft ditchingSignificant reduction in CPU cost through enhanced modeling featuresKinematic and structural behavior in good agreement (velocity, force, strain)Hydrodynamic behavior (pressure results) still very noisy

The relevance to include suction has been demonstrated.A simulation study has been presented to demonstrate the capabilities of the approach.


Perspectives

• Further CPU reduction by optimised particle distribution and combination of SPH with elements for water

• Transfer knowhow to flexible full aircraft ditching simulation

Bottom view on rear fuselage zone

Copyright © ESI Group, 2011 All rights reserved.Copyright © ESI Group, 2011. All rights reserved.

Paul [email protected]

Fluid-structure Interaction by the mixed SPH-FE Method with Application to Aircraft Ditching

ACKNOWLEDGEMENT

Most of the work presented in this paper has received funding from the European Commission’s 7th Framework Programme under grant agreements no FP7-266172 (project SMAES –Smart Aircraft in Emergency Situations).

Thanks for your attention

fluid-structure interaction by the mixed sph-fe method...

Documents