Computational Modelling of Free
Surface Flows: wave interaction with
fixed and floating bodies
Derek Causon, Clive Mingham, Ling Qian
Zheng Zheng Hu and Hanbin Gu
Department of Computing & Mathematics
CMMFA Team Photo
B Wang, ZZ Hu, Y Zhang, D Causon, J Armesto, K Bennett, L Qian (Back) S Higgins, N Subramaniam, F Gao, J Shiach and C Mingham (Front)
Outline
• Shallow water code
AMAZON-CC
• Surface capturing Navier Stokes code
AMAZON-SC
• Wave overtopping and wave energy converters
• Further methods development
• High Performance Computing – the GPU
• Future work
Manchester Bobber
The Pelamis
AMAZON two-fluid solver
• Incompressible Navier-Stokes solver — Based on an artificial compressibility solver.
• Surface-capturing method
– Treats the free surface as a contact discontinuity in the density field, allowing the use of modern high resolution “shock capturing” methods.
• Fully two phase approach which solves in both the air
and the water fluid regions.
overlay Cartesian grid
(adaptive) cut cell grid for an island
AMAZON-CC: generation of oblique
waves using cut cells
3D incompressible, Euler equations with variable density: T I T I T I T z I y I x I T S w p w vw uw w v vw p v uv v u uw uv p u u n n n p w v u s t 0 g 0 0 0 , , where . 2 2 2 h g f B h g f F Q B n F Q
is the coefficient of artificial compressibility
• The convective flux (Fk) is evaluated using Roe’s approximate Riemann solver.
• To ensure second order accuracy, MUSCL reconstruction is used
where (x,y,z) is a point inside the cell (i,j,k), r is the vector from
the cell centre to the point (x,y,z), Qi,j,k is the cell centre data
and is the slope limited gradient.
k k k I k I I k
F
Q
F
Q
R
LQ
Q
F
12r
Q
Q
Q
(
x
,
y
,
z
)
i,j,k i,j,kConvective fluxes
k j iQ
, ,Time discretisation
The implicit backward Euler scheme is used together with an
artificial time variable (to ensure a divergence free velocity field) and a linearised RHS. 1 1 1 1 1 1 1 , 1 , , 1 , 1 1 , 1 , 1 where ) ( t t t m m n m n m n ta m n m n m n m diag I R t V I R V I Q Q Q Q Q Q Q
The resulting system is solved using an approximate LU factorisation. A Jameson-type dual time iteration is used to eliminate at each real (outer) iteration.
AMAZON-SC 2D simulation
(air/water)
• Overtopping event occurred 142 seconds into the experiment.
• Seaward boundary located 2m from wall. • Landward boundary is transmissive.
Merged cell
Cut cell
Computational domain and
boundary conditions
Body Air
Reflection boundary (the fictional cell R): Wave maker Water Numerical Tank Non-reflecting Non-reflecting Non-reflecting Wall RO gn p p v v v v z ijk ijk R b ijk ijk R ijk R n n n n) 2( ) ( 2 Free surface
Results
A. Fixed Body
B. Moving Body
C. Wave propagation and Extreme
waves
1. A vertical cylinder in a 1st order wave maker
2
. A horizontal cylinder in a 1st order wave maker3. The Dam break flow past various bodies (a right cube, a skewed cube, a bottom mounted and truncated cylinder) 4. A bottom mounted and truncated cylinder in regular waves
5
. A Pelamis-type Geometry in regular waves• A regular 1st order wave imposed at the inlet:
• Experiments and Theoretical analyses conducted by
A.G. Dixon et al. (1979) and W.J. Easson et al. (1985)
as well as used the STAR CCM by J. Westphalen
2. Horizontal Cylinder
)
cosh(
)
cos(
))
(
cosh(
k h
t
k x
h
z
k
gAk
u
)
cosh(
)
sin(
))
(
sinh(
k h
t
k x
h
z
k
gAk
w
)
cos(
kx
t
A
-0.40 -0.30 -0.20 -0.10 0.00 0.10 0.20 0.30 0.40 0.00 0.50t/T 1.00 R e la ti v e v e rt ic a l fo rc e Theoretical force Experimental force present result -0.60 -0.40 -0.20 0.00 0.20 0.40 0.00 0.50 1.00 t/T R e la ti v e v e rti c a l fo rc e Theoretical force Experimental force Present result Case1: d= 0.0m Case 2: d= -0.075m Case 3: d= -0.15m -0.40 -0.30 -0.20 -0.10 0.00 0.10 0.20 0.30 0.40 0.00 0.50 1.00 t/T R e la ti v e v e rti c a l fo rc e Theoretical force Experimental force present result
2. Horizontal Cylinder
• Pelamis: 0.6 × 0.2 × 0.2m
• Tank: 6.86 × 1 × 1m
• Water depth: 0.45m
• Grid size: 256 × 39 × 39
• Wave gauges: at front of Pelamis
• CPU about 25 days for 8s • Input velocity = Asin t
A=-0.1, =2 , T =1s
5. Fixed Pelamis-type geometry in
regular waves
1 2 3 4 5 6 7
front bac
k
•Tank: 13.0m×1.0m×3.5m and the water depth:h=2.8m
•The front end of the cylinder is placed at 5.0m from the wave makers • Input amplitude: A=0.025,0.05,0.1 & 0.15, wave period: T=1.78s, wave maker number: k=1.277, wave length: L=5.0m
the angular frequency: w=3.534 and frequency: f=0.5625
the radius of cylinder: a=0.095m, the total length of Pelamis=1.40m
Sketch of the Pelamis split to 9
sections
A horizontal position of
Pelamis
5. Pelamis Tests
Sketch of the reference of
B. Moving Body Cases
1. Oscillating cone describing a vertical Gaussian wave packet motion (blind test validation)
2. Water entry of various rigid bodies (a wedge; Bobber-type shape; sphere shape; cone shape)
at a prescribed velocity
• The vertical position of the cone followed the form of a Gaussian wave packets
• Experiments conducted by K. Drake et al.(2008)
• Test case: A=50mm and m=9 • The dead-rise angle is 45º
• Tank: 2.0m×1.6m, the water depth: h=1.02m and the initial
draught of the cone is z=0.148m
• Dimensions: a =0.228m, d=0.2281m, b=0.05m • The time step=0.00005
the CPU for coarse mesh (dx=dy=0.02m) 15 hours the CPU for fine mesh (dx=dy=0.01m) 41 hours
1. Oscillating Cone using
AMAZON-2D (axisymmetric) code
d z
a
Comparison with experimental data (K. Drake et al. 2008) and numerical result for non-dimensional vertical forces
on cone against time
-6 -4 -2 0 2 4 6 t/T -0.8 -0.4 0 0.4 0.8 1.2 n o n -d im e n s io n a lis e d v e rt ic a l fo rc e m=9 AMAZON 001 AMAZON 002 experimental data CV-FE -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -6 -4 -2 0 2 4 6 t/T N o n -d im e n s io n a l v e rt ic a l fo rc e s AMAZON (Δ=0.01) AMAZON (Δ=0.02) Experimental data
• Body moves vertically downwards towards an initially calm water surface at a velocity V=1.19m/s
• Experiments conducted by Tveitnes et al.(2008)
• Tank: 2.0m×0.4m×2.0m and the water depth: h=1.0m
• The dead-rise angle is 45º
• Uniform mesh: 80×16×80=102,400 (dx=dy=dz=0.025m)
• Dimensions: breadth=0.6m, length=0.3m, height=0.3m
• A time step=0.0002s and total CPU about 3 days and 6 hours (t=1s)
Comparison with experimental data (Tveitnes et al. 2008) and numerical result for water entry forces
0 100 200 300 400 500 600 700 0 0.5 1 1.5 z/d F o rc e ( N ) Experimental data Present result d z
Beam sea direction Head sea direction
wave maker number : k=6.63 K=8.36
Wave period: T=0.78s Wave length: L=0.95m
T=0.69s
Wave length: L=0.75m
frequency: f=1.28 f=1.44 amplitude: A=0.013 A=0.013
Head sea direction
Beam sea direction Sx
Radius a
mf=1.2 kg (float mass)
mc=0.4 kg (counterweight mass) Pulley radius=0.0175m
Bobber geometry: a=0.074m, b=0.06m and c=0.07m, Sx = 4a, the water depth =0.46m
Initial free surface
b a c
3. Single and
multiple floating
Bobbers in regular
waves
Manchester Bobber
3a. Free-decay for the Manchester Bobber
flat-bottomed and conical-topped
float
3 6 1 i i i i i
U
f
mg
M
g Mf Mc ds z p a Mf Mc b s ) ( * ) ( 3 n float mass=2.1kg Counterweight mass=1.0kg Mf Mc ds z p b s n m= 1.1kg M = 3.1kg 3a
a. The drop test
-0.10 -0.05 0.00 0.05 0.10 0.00 3.00 6.00 9.00 time (s) V e rt ica l d isp la ce m e n t (m) Experimental data Present result Case 1: mf=2.1kg, mc=1.0kg
Comparison: experimental data (T. Stallard (2008) ) and current results for decay rate
Case 2: mf=2.1kg, mc=1.2kg -0.10 -0.05 0.00 0.05 0.10 0.00 3.00 6.00 9.00 time (s) V e rt ica l d isp la ce m e n t (m ) Experimental data Present result
b. The rise test
Comparison: experimental data (T. Stallard (2008) ) and current results for decay rate
Case 1: mf=2.1kg, mc=1.0kg Case 2: mf=2.1kg, mc=1.2kg -0.15 -0.10 -0.05 0.00 0.05 0.10 0.00 3.00 6.00 9.00 time (s) ve rt ica l d isp la ce m e n t (m ) Experimental data Present result -0.15 -0.10 -0.05 0.00 0.05 0.10 0.00 3.00 6.00 9.00 time (s) V e rt ica l d isp la ce m e n t (m ) Experimental data present result
Beam sea direction Head sea direction
wave maker number : k=6.63 K=8.36
Wave period: T=0.78s Wave length: L=0.95m
T=0.69s
Wave length: L=0.75m
frequency: f=1.28 f=1.44 amplitude: A=0.013 A=0.013
Initial free surface
a
b
Single and multiple hemispherical Bobbers
in regular waves
• mf=1.2kg,mc=0.4kg
• the water depth =0.46m
• The radius of a=0.075m,
b=0.07m
• The floats separated by Sx = 4a
Head sea direction
Beam sea direction
Sx
a. Head seas
direction
Isolation
Displacement vs time -0.04 -0.02 0.00 0.02 0.04 0.06 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 time (s) V e rt ic a l d is p la c e m e n t (m ) Bobber 1 Isolation -0.40 -0.20 0.00 0.20 0.40 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 time (s) V e rt ica l ve lo ci ty (m /s) Bobber 1 Isolation -6.00 -4.00 -2.00 0.00 2.00 4.00 6.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 time (s) V e rt ic a l fo rc e ( N ) Bobber 1 Isolation
Vertical force vs time Vertical velocity vs time
Comparison: The Bobber at same position in isolation and array
-4.00 -2.00 0.00 2.00 4.00 34.00 34.50 35.00 35.50 36.00 36.50 t/T d (t )/ A Exp.data
Present result (float 3)
-4.00 -2.00 0.00 2.00 4.00 34.00 34.50 35.00 35.50 36.00 36.50 t/T d (t )/ A Exp. data
Present result (float 1)
-4.00 -2.00 0.00 2.00 4.00 34.00 34.50 35.00 35.50 36.00 36.50 t/T d (t )/ A Exp.data
Present result (float 5)
-4.00 -2.00 0.00 2.00 4.00 34.00 34.50 35.00 35.50 36.00 36.50 t/T d (t )/ A Exp. data
Present result (float 4)
Displacement
in head seas direction
-4.00 -2.00 0.00 2.00 4.00 34.00 34.50 35.00 35.50 36.00 36.50 t/T d (t )/ A Exp. data
present result (float 2)
Comparison: experimental data (T. Stallard (2008) ) and
b. Beam seas
direction
Isolation
Displacement vs time
-0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 time (s) V e rt ic a l d is p la c e m e n t (m ) Bobber 1 Isolation -4.00 -2.00 0.00 2.00 4.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 time (s) V e rt ic a l fo rc e ( N ) Bobber 1 Isolation
Vertical force vs time
Vertical velocity vs time
-0.25 -0.13 0.00 0.13 0.25 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 time (s) V e rt ica l ve lo ci ty (m /s) Bobber 1 Isolation
Comparison: The Bobber at same position in isolation and array
-4.00 -2.00 0.00 2.00 4.00 30.00 30.50 31.00 31.50 32.00 32.50 t/T d (t )/ A Exp.data
Present result (float 3)
-4.00 -2.00 0.00 2.00 4.00 30.00 30.50 31.00 31.50 32.00 32.50 t/T d (t )/ A Exp.data
Present result (float 1)
Displacement
in beam seas direction
-4.00 -2.00 0.00 2.00 4.00 30.00 30.50 31.00 31.50 32.00 32.50 t/T d (t )/ A Exp. data
Present result (float 5)
Comparison: experimental data (T. Stallard (2008) ) and present results for displacement vs wave period
-4.00 -2.00 0.00 2.00 4.00 30.00 30.50 31.00 31.50 32.00 32.50 t/T d (t )/ A Exp. data
present result (float 4)
-4.00 -2.00 0.00 2.00 4.00 30.00 30.50 31.00 31.50 32.00 32.50 t/T d (t )/ A Exp.data
C. Devices in Extreme waves
1. Wave propagation in empty tank for comparison of the experiments conducted by Gao (2003)
2. Extreme wave in empty tank for comparison of the
experiments conducted by Ning et al. (2008)
Inlet condition for ‘New Wave’ extreme wave:
) ) ( ) ( sin( ) cosh( )) ( sinh( 0 0 1 ) 1 ( i i i i i N i i i i k x x t t h k h z k k gA w N i i i i i k x x t t A 1 0 0 ) 1 ( ) ) ( ) ( cos( N i i i i i N i j i i j i j i j i j j i j i j i j i t t x x k A t t x x k k A A t t x x k k A A 1 0 0 2 1 0 0 0 0 ) 2 ( ) ) ( ) ( ( 2 cos ) ( ) )( ( ) )( ( cos ) ( ) )( ( ) )( ( cos ) ) ( ) ( cos( ) cosh( )) ( cosh( 0 0 1 ) 1 ( i i i i i N i i i i k x x t t h k h z k k gA u ) ) ( ) ( ( 2 cos 2 cosh ) ( 2 cosh ) ( ) ( )] ( ) )( ( ) )( cos[( ) cosh( ) )( cosh( ) ( ) ( ) ( )] ( ) )( ( ) )( cos[( ) cosh( ) )( cosh( ) ( ) ( ) ( 0 0 1 2 1 0 0 ) 2 ( i i i i i N i i j j i i i N i j i j i i j i j i j i j i j i j i j i j i j i i j i j i j i j i j i j i j i j i t t x x k h k h z k D G A k t t x x k k h k k h z k k D G A A k k t t x x k k h k k h z k k D G A A k k u ) ) ( ) ( ( 2 sin 2 cosh ) ( 2 sinh ) ( ) ( 2 )] ( ) )( ( ) )( sin[( ) cosh( ) )( sinh( ) ( ) ( ) ( )] ( ) )( ( ) )( sin[( ) cosh( ) )( sinh( ) ( ) ( ) ( 0 0 1 2 1 0 0 ) 2 ( i i i i i N i i j j i i i N i j i j i i j i j i j i j i j i j i j i j i j i i j i j i j i j i j i j i j i j i t t x x k h k h z k D G A k t t x x k k h k k h z k k D G A A k k t t x x k k h k k h z k k D G A A k k wComparison with experimental data by Ning et al. (2008) and present numerical results for free surface elevation at x=3.0 m
The focus point at x =3m maximum elevation minimum elevation Physical Exp. by Ning et al.
(2008) 0.5653 (t=9.2s)
Present result (1st order) 0.5707 (t =9.147s) 0.4444 (t=8.557s)
Present result (1st + 2nd order) 0.5717 (t =9.147s) 0.4439 (t=8.557s) time (s) S u rf a c e e le v a ti o n (m ) 6 7 8 9 10 11 12 0.44 0.46 0.48 0.5 0.52 0.54 0.56
0.58 present result (1st order)
present result (1st+2nd order) Physical Exp.
Bobber Device in
Extreme Waves
• Input velocity under case 2 (1st +2nd order)
• Tank: 13.0m×0.48m×1m with the water depth: h=0.5m
• The initial position of the apex of Bobber in the tank is
3.0m(=the focus point)×0.24m×0.35m
• The radius of the hemispherical base: a=0.15m, the height of cylindrical section: b=0.15m and non-uniform mesh:
425×22×40=374,000 and the refined regions=0.02m
• The mass of the Bobber = the volume of the hemispherical (m=ρ2 a3/3=7.068kg),which is a little larger than the initial immersed volume
• Wave gauges: 3.48m,3.68m and the front side of the Bobber • A time step=0.0003s and CPU near 25 days up to t=12s
Initial free surface
b
Time history of the heave force on the Bobber
Time history of wave run-up on the front side of the Bobber
Case 2:
Case 3:
Case 2 0.40 0.45 0.50 0.55 0.60 0.00 2.00 4.00 6.00 8.00 10.00 12.00 time (s) S u rf a c e e le v a ti o n ( m ) Case 3 0.40 0.45 0.50 0.55 0.60 0.65 0.00 4.00 8.00 12.00 time (s) S u rf a c e e le v a ti o n ( m ) Case 2 -20.00 -10.00 0.00 10.00 20.00 0.00 4.00 8.00 12.00 time (s) H e a v e f o rc e ( N ) Case 3 -30.00 -20.00 -10.00 0.00 10.00 20.00 30.00 0.00 4.00 time (s) 8.00 12.00 H e a v e f o rc e ( N )Case 3:
Case 2:
Case 2 -0.40 -0.20 0.00 0.20 0.40 0.00 4.00 8.00 12.00 time (s) V e rt ic a l v e lo c it y (m /s ) Case 3 -0.60 -0.30 0.00 0.30 0.60 0.00 4.00 8.00 12.00 time (s) V e rt ic a l v e lo c it y (m /s )Time history of the vertical velocity of the Bobber
Case 2 -0.08 -0.04 0.00 0.04 0.08 0.00 4.00 8.00 12.00 time (s) V e rt ic a l d is p la c e m e n t (m ) Case 3 -0.10 -0.05 0.00 0.05 0.10 0.00 4.00 8.00 12.00 time (s) V e rt ic a l d is p la c e m e n t (m )
Case 2:
Case 3:
The Manchester
Bobber
Improved Resolution of Free Surface
Lin et. al.
Two step projection method
Finite difference
Fast marching
particle level
set
Partial cell + LRS (local relative stationary method)
Non-uniform structured rectangular mesh
Violent Sloshing: Model Requirements
• Finite Volume
• Inviscid
• Compressible Flow
• Change of State Thermodynamics
• Air/Water
Compressible Flow Equations
Finite Volume Euler Equations
1D Shock Tube Problem with Cavitation
Initial Conditions
LEFT STATE RIGHT STATE
ρ = 995.5450 kg/m3 ρ = 995.5450 kg/m3
u = -10 m/s u = 10 m/s
p = 0.9 bar p = 0.9 bar
T = 303.15 K T = 303.15 K
1D Shock Tube with Cavitation: Flow solver with
no
1D Shock Tube with Cavitation: Flow solver with
no
1D Shock Tube with Cavitation: Flow solver with
Change of State Thermodynamics
• Equation of State for
Water (Tait)
• Equation of State for Air
(Ideal Gas)
Change of State Thermodynamics
Properties of Mixtures
• Oldenbourg Polynomials
Modelling Change of State Thermodynamics
• For p: replace EoS by Call Thermo(ρ, e, p, c, α)
• Within
SUBROUTINEThermo:
• Enter with ρ, e
• Solve saturated mixture equation for T using Secant Method • Calculate ρsat_liquid and ρsat_vapour
• If (ρ > ρsat_liquid) then use Tait EoS for water • If(ρ < ρsat_vapour) then use Ideal gas EoS
• Else use saturated mixture relations with p = psat • Calculate mixture speed of sound, c
• Calculate water vapour traction, α • Return with p, c and α
1D Shock Tube with Cavitation: Flow solver with
change of state thermodynamics – 10 Δt
1D Shock Tube with Cavitation: Flow solver with
change of state thermodynamics – 10 Δt
1D Shock Tube with Cavitation: Flow solver with
change of state thermodynamics – 120Δt
1D Shock Tube with Cavitation: Flow solver with
change of state thermodynamics – 120Δt
1D Shock Tube with Cavitation: Flow solver with
change of state thermodynamics – 120Δt
High Performance Computing
• Multi-core high spec desk-top PC
• NEC SX-8 vector supercomputer
• UK National Supercomputing facilities
• Graphical Processor Unit (GPU)
• NVIDIA Tesla C-870 with 2 GPU boards
• NVIDIA CUDA software development kit (SDK)
• Rack of NVIDIA Tesla S1070 systems
• url:
Execution time in seconds for
20,000 time steps
32x32 64x64 128x128 RADIX 36.36 53.97 110.46 TILED 40 142.41 1140.96 UNTILED 41.52 164.92 1895.97 SX8 219.03 1746.27•Three problem sizes with number of real particles used being 32x32, 6x64, 128x128
•Code : SX8: means executed on 1 processor of the NEC SX8, all others executed on
2D Dambreak with Obstacle
using a Smoothed Particle Hydrodynamics
Code
Conclusions
• 2D shallow water and fully 3D cut cell free surface
capturing codes have been developed for the simulation of violent wave loadings on fixed and floating bodies
such as seawalls and wave energy devices.
• A number of test cases have been used to validate the codes.
• The underlying method is generic and can be applied to any application area involving free surfaces and
stationary/moving bodies including compliant bodies (e.g. LPG carrier in steep waves).