investigasi numerik pada interaksi struktur keairan menggunakan metoda berbasis partikel untuk...
TRANSCRIPT
-
8/3/2019 Investigasi Numerik Pada Interaksi Struktur Keairan Menggunakan Metoda Berbasis Partikel Untuk Aplikasi Maritim_Marine Transport_Sun_Fanfan
1/1
Numerical Investigations on Fluid-Structure InteractionsUsing Particle Based Methods for Marine Applications
Fanfan Sun [email protected] - School of Engineering SciencesSupervisors Dr. Mingyi Tan and Professor Jing Tang Xing
FSI Away Day 2010
Fluid Structure Interactions
Research Group
Objective
Theory
The formulation of SPH is often divided into two steps: integral representation
and then particle approximation.
Integral representation:
Particle approximation: and
here h is the smoothing length defining the influence area
Navier-Stokes equations:
(1)continuity equation ; (2)momentum equation; (3)energy equation
1). 2). 3).
Applying SPH method to equations 1 , 2 and 3
a). and b).
c).
considering unity property of kernel function showing below other forms of
SPH equation can be funded
Methodology
Investigation of Convergence of SPH
Conclusions
SPH is a pure meshfree, particle based method which is widely used for
fluid simulations especially rough sea motion simulations SPH approximations give accurate and reliable results
Fluid simulation will finally be coupled with structure simulation using,
for example, finite element method.
Introduction & Motivation
Many fluid-structure interaction problems often involve violent fluid
motions in marine engineering field, such as slamming and green water
when a ship travels in rough seas which can produce overall momentum
change and deformation of the hull. Hence, it is important to consider the
fluid-structure interactions, breaking waves and flow separations in order to
avoid damages caused by dynamic loads on the structures.
As it is difficult to obtain analytical solutions for such complicatedproblems, numerical methods and experiments are adopted ininvestigations. Traditional grid-based numerical methods like finite
element method have been developed but they are not efficient forlarge deformation problems. Particle based methods like SmoothedParticle Hydrodynamics (SPH) are an alternative to simulate fluidflows due to their Lagrangian and meshless properties.
To develop a numerical approach combing the smoothed particle
hydrodynamics method for fluid and other methods, FEM for example, for
structures to simulate violent fluid-structure interactions.
Figure 1: Rough sea slamming on offshore structure Figure 2: slamming on ship
(3.bp.blogspot.com/.../s200/freakwave.jpg)
ijj
N
j j
j
i Wxfm
xf )()(1
'),'()'()( dxhxxWxfxf
Hydrodynamics problems are generally expressed in the form of partial
differential equations which is normally impossible to obtain analytical
solutions for. Numerical solutions are necessary and SPH method is one of
the efficient method to solve this type of problems. The key ideas include:
Discreting the problem domain by a set of arbitrarily distributed particles.
No connectivity for these particles is needed (meshfree)
Approximating the field function by the integral representation method
(integral function representation)
Using particles to represent the kernel approximation by replacing the
integration term with the discrete particle volume (compact support)
Performing the particle approximation at every time step (adaptive)
Approximating every term related to field function in the PDEs to
produce a set of ODEs in discrete form with respect to time only
(Lagrangian)
Solving ODEs with explicit integration algorithm to obtain the time
history of all the field variables for all the particles
Convergence of the integral approximation and particle approximation
adopted in SPH is studied using a simple function f(x)=exp(-x) by
comparing the numerical results with analytical data
Figure 3. comparing analytical results and kernel approximation results.
x
v
Dt
D
xDt
Dv 1
x
v
Dt
De
ij
N
j
ji Wm1
i
ij
i
N
j j
j
ii
x
Wv
m
Dt
D
1 i
ij
j
jN
j
j
i
i
x
Wm
Dt
Dv
1
1
ii
i
i
i
ij
ij
j
j
i
iN
j
ji u
x
Wv
ppm
Dt
De
2)(
2
122
1
Figure 4. comparing analytical results and particle approximation results.
For integral
approximation,
Discretise the problem
domain into N points
with smoothing length
h. Results agrees with
analytical solution
'),'('
dxhxxWeexx
For particle
approximation,
Discretise the problem
domain into N particles
Results agrees with
analytical solution well
ij
xN
j j
jxWe
me ji
1
01'),'(1 i
ijN
j j
j
x
WmdxhxxW