Here the focuss is not necessarily only on capsiz- ing or the occurrence of extreme angles of roll, but (depending on the transport task of the vessel) also on accelerations, the efficiency of roll stabilizing sys- tems, etc。 Fig。 4 shows examples for such diagrams with respect to the danger of sliding of unlashed cars。

Deterministic wave sequences for model testing

While numerical investigations are very useful to evaluate a design’s performance, further develop- ments especially with respect to the validation of the prediction of combined failures (e。g。 combinations of stability loss, surf-riding and broaching to) and some basic assumptions (e。g。 with respect to roll damping) are necessary。 The most efficient way to investigate these gaps are model tests。 Additionally in discus- sions with authorities today, usually only model tests are accepted。

Thus it is necessary to adequately model the envi- ronmental conditions in the tank。 For achieving this a computer controlled test procedure in combination with the deterministic generation of arbitrary (dan- gerous) wave sequences has been developed。

As a first step, a target wave sequence is chosen as time series at a target position in time and space

— i。 e。 the position where the ship encounters the wave train at a given time。 This can  either  be a wave sequence from the numerical simulation of a rolling/ capsizing scenario or an arbitrary rogue wave sequence。

This wave train is transformed upstream to the po- sition of the wave maker wherefore different ap- proaches for modelling non-linear wave propagation are applied [1]。 The corresponding control signals for driving the wave maker are calculated using adequate transfer functions。 The resulting wave train is gen- erated and measured at the selected positions in the tank。 The ship model arrives at the target position by the corresponding target time (measured from the beginning of wave generation)。 This is achieved by the fully automated test procedure at the Hamburg Ship Model Basin [4]:

The ship moves in parallel with the tank side wall at a required minimum distance。 Registration starts by setting the desired course。 The ship’s course is con- trolled by the master computer by telemetry which commands a Z-manoeuvre at given constant course angle and model velocity。 Ship motions in six de- grees of freedom are registered precisely by computer controlled guidance of both, the towing and the hor- izontal carriage: During the entire test run, the ship model stays in the field of vision of the optical sys- tem line cameras。 Additionally, the wave train is measured at several fixed positions of the wave tank。 When the model reaches the critical safety limit at the wave maker or the absorbers at the opposite side

of the tank, the ship and the carriage stops automat- ically。

Thus, the test is realized by a deterministic course of test events which allows a reproducible correlation of wave excitation and ship motion - both as time se- ries in the moving reference frame of the ship。 As an example, Fig。 2 shows a test with a multipurpose ves- sel。 A wave packet within a regular wave is measured at a stationary wave probe close to the wave board (x = 297。8 m, model scale 1:34)。 It is transformed to the ship position shown in the second picture i。 e。 in the moving reference frame wave train。 The result- ing wave sequence is quite regular and contains the target rogue wave at the location of interaction with the cruising ship。 As a result, the ship behaves in- conspicuously until it encounters the high transient wave。 For obtaining such an apparently simple wave train in the moving frame (compare wave registra- tion close to the main board), the described wave generation technique and test procedure are applied。 Thus, the ship behaviour can be clearly related to the wave sequence。

Figure 2: Roll motion of a multipurpose  vessel (GM = 0。44 m, v = 14。8 kn und µ = ±20◦) in a reg- ular wave from astern (λ = 159。5 m, ζcrest = 5。8 m) with proceeding high transient wave packet。