Recent Advances in Simulation for Marine Applications Milovan Peri - - PowerPoint PPT Presentation

Recent Advances in Simulation for Marine Applications

Milovan Perić CD-adapco, Nürnberg Office

www.cd-adapco.com Milovan.Peric@de.cd-adapco.com

Introduction

 Numerical towing tank has slowly become a reality...  The major advances in simulation technology which led to

the acceptance of simulation as a design and optimisation tool are:

− The ability to handle complex geometry with all relevant

details;

− An adequate modelling of turbulence; − An adequate modelling of free surface effects; − An adequate modelling of cavitation; − The ability to handle non-linear waves; − Coupled simulation of flow and flow-induced motion of bodies; − Coupled simulation of flow and structural deformation.

Handling of Complex Geometry, I

 During the past decade, the ability to handle complex

geometry with all relevant details has been greatly improved:

− Tools for automatic and manual repair of CAD-models

(which are often imperfect) have been developed;

− A surface-wrapping tool has been introduced, which creates

a closed surface around assemblies of solid parts;

− Tools for automatic generation of polyhedral, trimmed

hexahedral or extruded meshes have been developed;

− Automatic and manual definition of local mesh refinement

requirements have been created, based on:

 local curvature,  proximity of other walls,  pre-defined volume shapes, etc.

Handling of Complex Geometry, II

Surface-wrapping of an

• il rig

Handling of Complex Geometry, III

Re-meshed surface of an

• il rig

Handling of Complex Geometry, IV

Simulation of air flow around an oil platform

Handling of Complex Geometry, V

Volume shapes used to enforce local mesh refinement in a study of flow around a KVLCC-hull

Turbulence Modelling, I

 For many types of analysis, the standard k-ε or k-ω

turbulence models are adequate...

 In order to predict secondary flows better, more

sophisticated models are needed, e.g. Reynolds-stress model...

 A special turbulence model is needed to predict transition

from laminar to turbulent flow, e.g. when predicting resistance of a racing yacht...

 For predicting noise sources, pressure fluctuation etc.,

large-eddy-simulation (LES or DES) type of analysis with special subgrid-scale turbulence models is needed.

 The analyst needs to select the most appropriate model for

his analysis...

Turbulence Modelling, II

Predicted iso-lines of constant velocity, showing boundary layer growth KVLCC half model, comparison with wind tunnel test...

Turbulence Modelling, III

Comparison of computed and measured data...

Experiment CFD

Reynolds-stress turbulence model used in simulation

Turbulence Modelling, IV

y/L

0.00 0.02 0.04 0.06 0.08 0.10

u/U, v/U, w/U

• 0.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2 u/U (exp.) v/U (exp.) w/U (exp.) u/U (CFD) v/U (CFD) w/U (CFD)

Velocity cut in propeller plane at z/L=-0.05

Comparison of measured and predicted velocity profiles

Turbulence Modelling, V

Comparison of computed and measured resistance...

Modelling of Free-Surface Effects, I

 Interface-capturing methods have been developed, which

allow for the simulation of both gas and liquid flow...

 Where the two fluids are not expected to mix, a sharp

interface (within one control volume) is produced, with minimized numerical mixing.

 Trapped gas bubbles of liquid droplets are adequately

accounted for (gravity and surface tension effects).

 Arbitrary free surface deformation and fragmentation can

be accounted for.

 Phase change models (cavitation, boiling) are integrated

into this method to allow more complex phenomena to be modelled.

Modelling of Free-Surface Effects, II

 Distribution of liquid volume fraction after 101 periods of roll-

• scillation in an LNG-tank...

 No numerical mixing: even unresolved drops and bubbles can be

tracked until they reach free surface...

Roll-motion of a full-size tank... Free-surface resolution within

• ne cell...

Modelling of Cavitation, I

 If the aim is to avoid cavitation, one only needs to predict

its onset (pressure below saturation level)...

 If cavitation cannot be avoided, its effect on performance

needs to be assessed, so one has to model cavitation...

 Models based on bubble dynamics (Rayleigh-Plesset

equation) have proven robust and sufficiently accurate...

 One additional equation for volume fraction of vapour is

solved, with two parameters:

− seed density (number of seeds per m3 of liquid); − initial radius of seed bubble.

 These parameters are related to “liquid quality” and

depend on region (for sea waters) or treatment (like de- gassing or filtering in a laboratory)...

Modelling of Cavitation, II

Cavitation in experiment: flow around NACA0015 foil at 10.3° angle of attack (chord length 0.2 m, water speed 6 m/s, channel 0.57 x 0.57 m, cavitation number 1.7, absolute pressure 32000 Pa); HSVA in 1999.

t0 t0 + 10,4 ms t0 + 20,7 ms t0 + 31 ms t0 + 41,4 ms t0 + 51,7 ms

Modelling of Cavitation, III

Simulation of cavitation: flow around NACA0015 foil at 10.3° angle of attack (chord length 0.2 m, water speed 6 m/s, channel 0.57 x 0.57 m, cavitation number 1.7, absolute pressure 32000 Pa): CD-adapco, 2009

t0

10 ms 20 ms 30 ms 40 ms 50 ms

Modelling of Cavitation, V

The original design of Voith Water Jet led to substantial cavitation in the upper range of speeds... Experiments (performed after simulation) confirmed this... Simulation Experiment

Courtesy of Voith Turbo Schneider Propulsion GmbH & Co. KG

Modelling of Cavitation, V

Courtesy of Voith Turbo Schneider Propulsion GmbH & Co. KG

With the optimized design, cavitation starts at a much higher speed and is less intensive – a substantial increase of efficiency has been achieved...

Waves, I

 Simulation of wave impact and added resistance requires

realistic waves in the solution domain...

 Creation of incoming waves at a boundary requires long

computing times...

 The solution can be initialized using wave theory (e.g.

Stokes 5th-order theory)...

 This can substantially reduce solution domain size and

the duration of simulation...

 Coupling of solution methods based on RANSE and

potential theory is useful when wave propagation over long distance is of interest:

− solving RANSE in regions where potential flow theory is not

applicable (boundary layer, wake, breaking waves etc.);

− using potential theory elsewhere...

Waves, II

 Waves break when they come into shallow water...  The Stokes 5th-order wave with19.9 m wave height in

33.5 m water depth breaks after about 1 period...

Wave shape shortly after initialization (top), after one period (middle) and after 1.5 periods (bottom)

Waves, III

 Water velocity in the crest region increases with time as the

wave tends to break: from 9.9 m/s after initialization to about 22.9 m/s during overturning (wave propagation: 17.6 m/s)...

• max. 9.9 m/s
• max. 22.9 m/s
• max. 17.7 m/s
• max. 21.1 m/s

Waves, IV

Impact of the first wave (nearly Stokes-wave)

Waves, V

Impact of the second wave (steeper, nearly breaking)

Waves, VI

Horizontal force on the platform during two wave encounters

Impact with the first wave (nearly Stokes wave) Impact with the second wave (steeper, not yet breaking) More than twice higher load!

Waves, VII

Courtesy of Germanischer Lloyd AG

Waves, VIII

Courtesy of Germanischer Lloyd AG

Floating Bodies, I

 Implicit simulation of flow and flow-induced motion of

floating (or flying) body:

− has no restriction on time-step size for stability reasons (time

step can be chosen according to accuracy requirements);

− allows extreme motions of bodies, like lifeboat launching; − handles well even very light bodies...

 Many extensions are under way:

− to allow for multiple bodies moving relative to each other; − to allow for body connections (elastic spring, rigid body,

flexible with constraints, etc.);

− to allow for external forces (thrust, mooring, towing etc.); − to allow for relative motion of parts (hull, propeller, blade...).

Floating Bodies, II

Flow chart of coupled solution

• f equations for fluid flow and

body motion...

Floating Bodies, III

 Free fall of a

lifeboat into flat water: Experiment by NORSAFE

Floating Bodies, IV

 Comparison of computed and measured accelerations

(from Mørch et al, 2008):

Prediction of effects of changes in lifeboat geometry has also been validated...

Floating Bodies, V

 Initial wave position varied

by 20 m (drop from 32 m height).

 Following wave (180°)  Wavelength ca. 220 m,

wave height 13.5 m, water depth 33.5 m

 The questions to be

answered:

− When is the load on the

structure the highest?

− When are accelerations

the highest?

Floating Bodies, VI

Vertical acceleration at center of gravity for different hit points, following wave (180° incidence).

Floating Bodies, VII

Pressure at one monitoring point for different hit points, 180°

Fluid-Structure-Interaction, I

 Coupled simulation of flow and flow-induced deformation of

solid structures is of high interest...

 Coupling of FV-codes simulating fluid flow and FE-codes

computing solid deformation is nowadays explicit – thus not applicable or inefficient in many practical situations...

 An implicit coupling – as for the special case of floating

body – is required for robustness and computational efficiency...

 In many cases, structural model can be simplified – e.g.

ship may be represented as a beam which can sustain bending and torsion...

 Such simplified structural models have been included into

CFD-simulation in an implicit way – with success...

Future Trends

 More powerful and affordable computers = higher

demands from simulation:

− More transient simulations (URANS, DES and LES); − Prediction of pressure fluctuation and noise sources

(turbulence, cavitation);

− More fluid-structure-interaction applications (slamming,

sloshing);

− Simulation of manoeuvring tests (circle, zig-zag etc.) and

• ther experiments – to be done in the design phase...

− More automatic optimization studies...

Simulation Experiments

 Voith uses a ship simulator to

train captains...

 Hydrodynamic coefficients

used to come from experi- ments – now they come from CFD analyses...

Simulator Experiment Simulation

Courtesy of Voith Turbo Schneider Propulsion GmbH & Co. KG

Optimization

An automatic calculation strategy is developed that includes the following tasks:

• parametrization of geometry
• automatic generation of geometry
• automatic mesh generation
• automatic computation and post-processing of results
• embedding the procedure in an optimization loop

automatic optimization loop

Azimuth-Thruster Optimisation

Courtesy of Voith Turbo Schneider Propulsion GmbH & Co. KG

Smart Engineers, I

 No matter what the software can do, it remains just a tool...  How quickly and well problems get solved depends on the

craftsman using the tool – the engineer remains indispensable!

 Engineers need to be educated how to best use the tools

at hand:

− theoretical analysis, − numerical simulation (different kinds, different efforts), − experiment (model, full scale).

 The solution should be:

− reliable (one needs to estimate numerical and modelling

errors);

− obtained with minimum cost (time and money)...

Smart Engineers, II

An example: A mega-yacht with 4 propellers suffered from vibrations at a cruising speed of about 16 - 18 kn. German Lloyd and CD-adapco solved the problem with the help of 2 CFD- simulations, one FE-analysis and two field experiments (on real object)... Old design New design Vortex shedding was the cause...