Numerical simulation of
hydro-structural coupling for
marine applications
Guest lecture at the TU Deift,
March21
Paul Groenenboom
E
Conterrfls
° Introduction
D Finite Element Method ° Particle Method (SPH) D Fluid-structure interaction D Benchmarks
° Marine applications
ESI Group 2'
Deift University of Technology
Ship Hydromechanics Laboratory
Library
Mekelweg 2, 2628 CD Deift
The Netherlands
Equations to be solved
Structure:
Kinematics: Newton's laws. Dynamics: Stress-strain relation.
Non-linear behaviour: large displacements, plasticity and failure of material, contact.
Fluid:
Conservation of mass, momentum (energy). Euler or Navier-Stokes.
ESI Group 3
Frame of reference
LAGRANGE:
Nodes have the same motion as the material. Natural system for structures.
No need to interpolate transported properties. Usually not suitable for extreme deformation.
EULER:
Nodes are fixed ¡n space. Natural system for flow of fluids.
Requ ¡red interpolation of transported properties.
Not suitable for structures at small deformations.
Finite Element Method (1)
Equations of mechanics for a (complex) structure. Divide space into a number of small volumes of a simple shape.
Apply variational principle for the energy w.r.t. virtual displacements.
Assume the unknowns may be approximated by simple functions on these volumes.
Solve the resulting (large) set of discrete equations on a computer.
ESI Group
Finite Element Method (2)
Fixed connectivity.
How to account for motion?
Choice of element shape and basic functions. Choice of number, size and distribution of elements.
Check and analyse the results.
ESI Group
IESI GROUP
Solid Elements
ESI Group To discretize bulk materials
Standard: Solid 8 node "brick" elements ' First-order, trilinear isoparametric elements
Single-point integration requires hourglass control SRI (Selective Reduced Integration - i ini. pnt. for
hydrostatic part and 8 i. pnt's for deviatoric part)
also available.
Shell Elements (1)
to discretize structures made of plates and shells
a C° continuity
Number of integration points through the thickness: NINT
ESI Group
Solution ¡n time
Divide the relevant range in time into
anumber of (small) steps.
Update the kinematics and dynamics of
all nodes/particles or particles from one
step to the next.
Investigate the best way to perform this
step.
ESI Group
Explicit integration scheme
I
,
t,_I/2 tn ESI Groupf(t),x,,ï
m-i-kx=f(t) mÏ,+1cc, =f(t) ï,,=m'(j-kx,,) 1,H.I/2 = i lOi'5
X,.ÏX,,_112,,._..+I/2
t X,. k in = + &l/2Xn...1F2IESIGROUP
Explicit integration scheme
'Stab'e time step
L E. p, A
bar spr.flg-flmSS
m = pALJ2
k = EAIL
ESI Group
C[j
isthe speed of sound is the Traversal timeL _L
E/pc
mpllcit integration scheme
f(t),x,k,ï EStGroup ,n+kx=f(t) + kx4.1= f1(t) (2) L = (3) =(m/A72+k)_! +m/Lt2(2X0 -C 12 -C
'6
Comparison of explicit and
imp ici
urne n egration
Number of time steps
Although unconditionally stable, implicit methods needs many time steps in order to trace the physical
phenomena studied
Explicit analysis requires a small time step. This leads to many, but CPU cheap solutions
Equations to be solved
Implicit analysis requires matrix inversion. The solution of NL sets requires iterative solution strategies.
Explicit analysis requires no iteration and no matrix inversion.
ES Group 13
Example: Impact on Beam
Drop of rigid cylinder on supported beam
Impact loading
Elasto-plastic material Spotweld connections
ESI Group 14
PAM-C RASH Applications
Crashworthiness DesignINBAT Project
nnovative Barge Trains for Effective Transporton
Shallow Waters.
5th Framework Programme Project
ESI (roup 15
Mesh plots of a cut-view model at various
moments.
Particle Method SPH
Smoothed Particle Hydrodynamics
Meshless method for continuum mechanics Origins ¡n modelling of cosmic physics Significantly enhanced for fluid-flow
Handles free-surface and other interlaces
SPH - fundamental concept
SPH is an alternative numerical method to FEM for simulation of continuum mechanics. Properties (density) can be interpolated within a certain range.
The interpolation points are identified with the
particles with a specified mass.
With SPH there is no need to define a grid The FE connectivity is replaced by a dynamic nearest neighbor search.
There is no need for numerical evaluation of
partial derivatives as with other methods.
ESI Group (20
10
SPH is a new technology for simulation of
large deformation mechanics
Hypervelocity lmpact.
Ballistic Impact Material Forming Liquid sloshing
Impact on water/free surface flow Heart Valve opening
Flow in porous media Industrial Hydraulics Wave interaction
ESI Group 21
Space Debris Im 'act Simulation
References: ESNESTEC, Alenia Spazio
ESI Group
I Kernel approximation
f(x)= ff(x')w(x-x',h)dx'
Vf(x) = ff(x')VW(x-x', h)dx' Particle approximation N :_L fW(x-x,h)
i=iPi)
=-x,h)
Ji Pj
ESt Group> Smoothing function
V' Normalization;JW(x x',h)dx'=1
Compactly supportedW(x-x')=O
Delta function propert
limW(x- x',h)=ö(x-x')
h - Smoothing length W-Smoothingfuncdon j illustration of a smoothing function a12 [JÈsIGROUPsp
sc Cocep
IFESIGROUPCoustrìincthg smoothing functh»D
Constructing smoothing function
The default smoothing function is the cubic spline W4: W(r,h) = C*(1. - 1.5*x**2 + .75*x**3) if x < i
C*O.25*(2 - x)**3 if I <x <2
O.
ifx>2.,
where x=r/h, with the smoothing length h.
ESI Group
Eoup
Fluid Flow Equations
Cäniervation of Mass
dpt.
Conservation of Momentum
dvt-..
(pj
p--=m+-Conservation of Energy
Equation of State
1f
1
L\ PiI
ESI Group- y)
viwii
13
1.2 0.8 0.6 0.4 0.2 0 0.5 O.Oaum, Kw,wt 1.5 2 2.5 .01 0.2 03 .0.4 .05 06 .5 5ESI Giou
° Artificial Viscosity
2
/
Ci+Cj+ 2
+
M3.)Dissipation term within Governing Eq.
= -
( +;
+H)
=
mi(++iiii)(vi_v).ViWii
ESI Group 27
dv dt
Speca for Shocks
+
h.\(vi_vj).(ri_')) of
v Conservation equationDp_
&w° Dt Dvcr_1aoe Dtpa
De O a DtI
0.aß
afi .aß
ESI Group
(vv).(rr)<O
(vv).(rr) O
rS eqmtti
2(V.v)5' &xa a 3 í2814
sp foruao or the N-S equat
Density Momentum Energy Strain rate p, f___., pi+pJ Dz ,m1.LfltJ(I7(.vf)
Meí +J'j Wi Pipi af =±m
.(JJ-+!L)!i+m (& Dr ' p p2 f í? Pj2 r Pi+Pj (fifi)aWJ/IÇ9 Dt 2 PiPj' ai
2p,2f!m (!+!L)(vfl .vL+..&_efl
Dr 2 '? p 'a(
2p, £: = .V,W,J)Ö" V' Constitutive modeling (Jaumann rate)V' Strain rate and Rotation
rate
V' von Mises flow stress and
von Mises rule
ESI Group j.aß
-
-=Gë i (a 21è ax° R«8 21,ax ar J = jafi..c1fi Tt;e jJ/3J2cj5
ESI GROUPRiI1ae& wth
e1I
ESIGroup 29I'ESI GROUP
SPH for material with strength
y' Discretized equations
ESI Group
SPH approximations in a two-dimensional space with a circular support domain. ,.
sigoup
PAMCRASHI SPH
- Features
Standard Monaghan formulation mcl.
deviatoric strength.
Cartesian coordinates in i D, 2D, 3D. Choice of momentum formulations. Choice of smoothing kernels.
Two types of NN search. Variable smoothing length.
Artificial viscosity: standard FE or M-G Anti-crossing force option.
Symmetry planes
ESI Group (32
IESIGRoup PAM-CRASHISPH - Features (2)
SPH - Features (3)
Moving Least Squares option (order O or 1) Special version for axial symmetry
Special version for porous media flow and two-phase flow (coupled to finite element solution)
ESI Group
17
Material models: JWL(detonation),Hydrodynamic-EP, Johnson-Cook, SESAME Lib.,Mumaghan
Free material surface correction. Material interface treatment.
Coupling to PAM-SOLID FE through contact algorithms.
Particle generation from solids ¡n generis Post-processing by standard PAM-VIEW.
Advantages
a Lagrangian method suitable for large deformations of solids, liquids and gas.
Pre&post-processing same as for PAM-SOLID FE
Straightforward coupling to finite elements. Proven capability for Hyper-velocity impact and ballistic applications.
Suitable for liquids with free surlaces or
interfaces (sloshing and mixing) a Fluid-structure interaction (coupling)
ESI Group
Disadvantages
V' Comparatively lower accuracy; V Boundary deficiency;
V' Particle inconsistency; V Tensile instability.
ESI Group 36
Notable modifications
ESI Group y' Monaghan XSPH (anti-crossing) d v+et m(
Liu W. K.'s modification (RKPM, 1996)V Kuhnert's FPM (with ESI-Group...) I Chen J. K.'s correction (CSPM, 1999)
I Monaghan's tensile stability modification (2001) y' Liu M. B.'s generalization (FPM, 2003)
I
Others (MLSSPH, ASPH, etc.)SPH Apphcatìon Example
Using Smoothed Particle Hydrodynamics for the water
ESI Group 38
The example of a bursting dam has been given by Monaghan, Bonet and Hyncik.
A rectangular column of water is confined between a
vertical wall and the floor. At t=O the gate is removed instantaneously and the water is allowed to collapse
under gravity,
The SPH simulation is performed in 2D with 3249 particles, using the Murnaghan model for the liquid. The results are compared against the experiments by
Martin and Noice, and against the SPH simulation by Monaghan (1994) ESI Group i 39 Contours of the horizontal velocity at various times
Bursting dam
ESI Group 1 4020
!Eao
Bursting dam
cied dip1.et a
Bursting dam
.35 crtst'8pfl eBU1tB daa.fctaap. dLa daa.fctIaaghan 0H Z I-423 : : 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.0 ca1ed t1 ESt GroupFluid Mixin
Mixing of water at different densities
under gravity
ESI Group (42)
PAM-CRASH Contact treatment: Adding dynamic
connectivity between a node and an element
Penalty method
Fluid-structure interaction
ESI Group TN (,JA _jB).jA Sfl = f (5T o}dV -sri + a frc ö(7N)2d1' = O22
PAM-CRASH FE contact interface: Nodes to Surface: (Master-Slave, node to segment) Surface to Surface: (Master-Slave, segment to segment)
It is also possible to tie a node to a surface.
Fluid-structure interaction
Water entry of cylinders
Aircraft ditching Helicopter drop Bird strike Cardiovascular flow í oup
23
IESIGROUPFluid-structure interaction
SPH Application Example
Car entering water pool24
SPH (Smooth Particle Hydrodynamics)
Benchmark test of car vs. water-filled barrier
courtesy H-Y Clzoi - Hong-1k University, Seoul, Korea
ÈsI GROUP
'-l
--:Fuel sloshing
9Fluid
SPH and FEM for ship hydrodynamics