X hits on this document





31 / 48

PREDICTION. Comparison of normalized CAR values for full loading with an additional ballast of 10 t for the three hull shapes, determined by experiment (blue) and simulation (red) in the forward (top) and aft (bottom).

COMPARISON. Launch in still water: time curves of the vertical accelerations for experiment and simulation in the forward (top) and aft (bottom).

an inclination of 35 ° (bow down) into a wave with a period of 13 s and a height of 15.8 m in water 33.5 m deep and without any current.

A total of 18 simulations were performed: nine with waves approaching from the front, nine with waves following from behind – the in- itial position of the wave crest being varied by 20 m. The boat was given three degrees of free- dom – as in the previous case – and the mesh had approximately 600,000 cells.

The figure left (SIMULATION I) shows the position of the lifeboat and the deformation of the water surface during entry into the water; the figure right (SIMULATION II) depicts the lifeboat’s position and free surface shape dur- ing resurfacing.

These examinations demonstrated that the highest loads arise for following waves, i.e. when the boat hits the wave before the wave crest (at a distance of approximately 15 per


cent of the wavelength). The peak average pressures measured by a sensor in the keel area exceed 7 bar in the critical case, while the maximum load attains approximately 4 bar for approaching waves. The peak ac- celerations in the critical case amount to 13 g (vertical) or 27 rad/s (rotation); for the reverse wave propagation direction, the maximum values lie at 10 g and 20 rad/s respectively.

For waves approaching obliquely, other drop heights, launching from a ramp, differ- ent inclinations, other wind directions and speeds, etc., the loads will vary accordingly. It will only be possible to identify which sit- uation is the most unfavourable overall af- ter comprehensive analysis with variation of all the relevant parameters. For the final evaluation, the probability of this situation occurring at a certain geographical location must also be considered.

The calculations performed thus far in- dicate that simulation of the lifeboat move- ment during the critical phase of approxi- mately two to three seconds with three de- grees of freedom in full scale with a mesh of approximately half a million cells can be conducted in a day on a single proces- sor (i.e. one core of a multi-core proces- sor). Simulations with a much finer mesh (approximately 1.5 million cells) predicted only a negligibly lower loading, so that the former mesh size may be regarded as ad- equate for the purposes of optimization.

The use of CFD for studying the loads on the lifeboat hull and the accelerations acting on persons seated at various posi- tions inside it during entry into the water hence represents an ideal supplement to experimental investigations. The experi- ments can be limited to the final phase, for validating the simulation of an opti- mized geometry. With a cluster compris- ing a hundred processor units, it is pos- sible to execute thousands of simulations and to evaluate the results within only a few weeks.

SIMULATION II. Lifeboat position and free surface shape during resurfacing (250 ms between frames, from top to bottom).

nonstop 02/2009


Document info
Document views172
Page views172
Page last viewedFri Jan 20 08:39:51 UTC 2017