Date post: | 31-Dec-2016 |
Category: |
Documents |
Upload: | hoangtuyen |
View: | 220 times |
Download: | 1 times |
CITEPH Project
Simulation of extreme waves impacts on a FLNG
Pierre-Michel GUILCHER, Julien CANDELIER
(HydrOcean)
Ludovic BEGUIN, Guillaume DUCROZET, David Le TOUZE
(Ecole Centrale Nantes)
Your Partner in MarineComputational Fluid Dynamics
Page 2
Context of the study
� Strong wave impacts knowledge => structural design
� Hydrodynamics loads
� Induced by waves
� Strongly nonlinear
� multiphysics
~ms, ~mm
~s, ~m
Gas compressibility,
hydroelasticity
Your Partner in MarineComputational Fluid Dynamics
Page 3
Experimental setup
8th Spheric Workshop – Trondheim (Norway), June 4-6,
Your Partner in MarineComputational Fluid Dynamics
Page 4
Experimental setup
Experimental Wave Tank ECN� 50x30x5 m
�Multiflap wave generator
Simplified FNLG model� 1.1m width
Instrumentation�Wave probes in tank
�Wave probes on deck
� Pressure probes on breakwater
Water waves� Regular waves
� Wavelength=7.3m
� Amplitude = 0.44m
Your Partner in MarineComputational Fluid Dynamics
Page 5
Wave-Forcing procedure
8th Spheric Workshop – Trondheim (Norway), June 4-6,
Your Partner in MarineComputational Fluid Dynamics
Page 6
Forcing procedure
� Main algorithm
Complete problem through direct simulation
� Wave generation
� Wave propagation from generator to structure
� Impacts
� High cpu time consuming
� Numerical methods not adapted
=Wave generation/propagation
� Spectral methods
� No dissipation
� No structure
� Low cpu time
� Computed once before SPH computation +Impact
� SPH method
� Inlet/outlet wave boundaries
Your Partner in MarineComputational Fluid Dynamics
Page 7
Forcing procedure
� Incident wave models: potential spectral methods
• Rienecker & Fenton
�Monochromatic regular waves
�Bidimensional
�Fully nonlinear
• HOS (Higher Order Spectral)
� Irregular waves
�Multidimensional
�Fully nonlinear
�Applications: focused waves, irregular sea
states, etc.
Your Partner in MarineComputational Fluid Dynamics
Page 8
Forcing procedure
� Incident wave solution
8th Spheric Workshop – Trondheim
(Norway), June 4-6, 2013
� HOS solution computed once
� File storage of a set of cartesian grids covering
the impact area, at various instants
File storage
Cartesian grid at time t
Linear interpolation in timeBilinear interpolation in space
V, p at particle position
Your Partner in MarineComputational Fluid Dynamics
Page 9
Forcing procedure
� SPH-flow solver
• Developed by ECN and HydrOcean
• Improved SPH solvers
�Riemann solvers for stability
�Renormalization for accuracy
• High Parallel efficiency
�domain decomposition (MPI comm.)
�Efficient scalability (linear scalability up to 40000 cores / 1 billion particles)
�Variable-h capability
�3D complex geometries/domains
Your Partner in MarineComputational Fluid Dynamics
Page 10
Forcing procedure
� Imposition of incident field
�No remeshing
�Enough particles at start time in the buffer zone is required
�Vitalization/unvitalization of particles through inlet/outlet boundary
Incident waves
Dummy particles in the inlet/outlet area
� Pressure, velocity from potential solution
� Position updated with incident velocity
Free standard particles
� Standard SPH scheme
� Standard flux interactions with dummy particles
Your Partner in MarineComputational Fluid Dynamics
Page 11
Numerical simulation of Greenwater event
� Free surface elevation
� Reproduction of HOS signal along
the ship in the undisturbed area
� No phase shifting of SPH/reference
HOS
� Small damping
Your Partner in MarineComputational Fluid Dynamics
Page 12
Numerical Simulation of a Greenwater event
Your Partner in MarineComputational Fluid Dynamics
Page 13
Numerical simulation of Greenwater event
� Numerical Set-up
• dx = 0.01 m
• ~ 250 neighbours
• L/dx ≈ 100 (L = deck width)
• λ/dx ≈ 750 (λ = wave length)
• ~ 1.5 millions particles
• h-variable discretisation
• Use of 512 cores
wave probes
colored according to h
Your Partner in MarineComputational Fluid Dynamics
Page 14
Numerical simulation of Greenwater event
� Qualitative description
Impact of the plunging jet, Flooding of lateral flowsIncident wave exceeding freeboard
Your Partner in MarineComputational Fluid Dynamics
Page 15
Numerical simulation of Greenwater event
� Qualitative description
Converging flow impacts the wall Flow is deviated vertically
Your Partner in MarineComputational Fluid Dynamics
Page 16
Numerical simulation of Greenwater event
� Qualitative description
Water escapeCollapse of the water column
Your Partner in MarineComputational Fluid Dynamics
Page 17
Numerical simulation of Greenwater event
� Water elevation on deck
�The flooding of water is captured
� Impact time occurrences well captured
�Good estimation of water elevation near the ship fore
�Progressive damping of the water elevation on deck
� Initial conditions for wall impact not met
reverse flow
Your Partner in MarineComputational Fluid Dynamics
Page 18
Numerical simulation of Greenwater event
� Pressure probes on breakwater
time (s)
pre
ssu
re(P
a)
14 16 18
0
2000
4000
6000
8000
10000
12000
14000experimental M1SPH dx=1cm M1SPH dx=2.5cm M1
14 16 18
0
2000
14 16 18
0
2000-Good synchronisation
-Impact pressures not captured
-Need of higher refinement
M1
M4M3
Dam break on an obstacle
Your Partner in MarineComputational Fluid Dynamics
Page 19
Industrial application
Your Partner in MarineComputational Fluid Dynamics
Page 20
� Selection of greenwater event
Industrial application
� Irregular sea state statistically
described as (Hs, Tp)
� How to determine most severe
conditions?
� Not possible with CFD
� Use of ‘old’ linear potential solvers
Your Partner in MarineComputational Fluid Dynamics
Page 21
� Selection of greenwater event
Industrial application
Linear seakeeping solver
HOS/SPH-flow
Your Partner in MarineComputational Fluid Dynamics
Page 22
� Selection of greenwater event
Industrial application
Your Partner in MarineComputational Fluid Dynamics
Page 23
Conclusions and perspectives
Your Partner in MarineComputational Fluid Dynamics
Page 24
Conclusions
� wave-structure interactions simulation• forcing procedure between non-linear potential flows and SPH is effective
• uses the advantages of each solver, without drawbacks for simulations with no diffracted field at open boundaries
� Numerical simulation of greenwater events:• propagation phase: no phase shifting, small damping
• Qualitative behaviour of deck flooding is captured
• Kinematics OK, dynamics (pressure) not => Need of higher refinement => local refinement
• Still a very demanding problem in terms of CPU