One main result is that the maximum first peak doesn’t occur neither at the starting point of the wave trains (located at one quarter length from the free end), but rather around 2/5th length from the free end (see Fig。 8)。 Furthermore, the location where this maximum occurs seems to move closer to the starting point with decreas- ing signal duration。

every curves except for 1 s。 This  comparison shows high levels of initial bending moment peaks  compared to mode 1 in particular。

Even if there are two coexisting kinds of response (high frequency propagating wave and low frequency stand- ing waves), both are linked to the frequency spectrum of the signal。 The fact that the impulse be the same for all the cases means that energy associated to low frequen- cies is similar。 That’s why amplitude of the mode 1 response hardly varies from one load duration to the other。 On the other hand, owing to Fig。 2, energy for high frequency components is not negligible for 10 ms (2/T = 200 Hz) and 1 ms (2/T = 2000 Hz)。

Fig。 7 : Time histories of the bending moment for 1 ms (A) and 10 ms (B) load durations at elements where the maximum initial peaks occur (for A : 110。6 m,

for B : 116。2 m from the free end)

Fig。 9 : Time history of the bending moment for several load durations at constant impulse, at the

elements where the maximum occurs (maximum of the peak for 1 and 10 ms ; maximum of the low frequency modal response for 0。1 and 1 s)

Several filtrations taught that the main part of the fre- quencies participating to the peaks of response (1 and 10 ms) were beyond 20 Hz approximately。 For the 0。1 s load duration, 2/T=20 Hz so that high frequency ampli- tudes are not significative, even less for the 1 s duration of course。 Within this context, the Fig。 10 gives an idea of the wide-ranging of the response spectrums for the two shortest load signals。

Fig。 8 : Spatial distribution of the maximum bending moment peak for 10 ms load duration

The investigation on a wider time range (see Fig。 9), for four different durations from 1 ms to 1 s, at constant impulse, enables to visualize high as well as low fre- quencies。 High frequencies are only obvious for 1 and 10 ms signal durations。 The outward noise mainly cor- responds to the two initial trains of waves previously mentioned but more and more distorted by dispersion, thus superposed, and also reflected at boundaries with time。 These waves are superimposed on much lower frequency response。 In particular, flexural modes 1 (around 0。4 Hz) and 2 (around 2。4 Hz) are visible on

Fig。 10 : FFT of the bending moment time histories for 1

and 10 ms of signal durations

Let’s take advantage of this analysis to make a comment about the so-called “tailing mode”。 The type of de- formed shape of Fig。 3 could correspond to the response of a ship in the very first times of a slamming impact at the aft or fore part。 As shown by Fig。 7 and Fig。 9,  this is rapidly followed by flexural mode 1 vibrations essen- tially。 Thus, the transient deformed shape in question

could eventually be called “transient deformed shape of tailing type” for example, but it seems not appropriate to talk about what would be a local tailing vibration mode, probably not existing, never met in our results at least。

Role of the impulse

Two load triangular signals of same rise time 5 ms and same amplitude were tested, but of different total dura- tions : 10 ms and 15 ms。

The first peak of response is higher for the 15 ms dura- tion (see Fig。 11)。 Nevertheless, the high-pass filtering at 3 Hz showed that the main contribution came from low frequencies。 This was confirmed by the excitation spectrums (see Fig。 12) : at low frequencies (in fact to about 50 Hz), energy is noticeably larger for 15 ms ; beyond differences are less important。 横向载荷冲击下船体梁的瞬态响应英文文献和中文翻译(4):http://www.youerw.com/fanyi/lunwen_94988.html