Even if the two shapes differ, the corresponding stress– strain curves are quite similar (Fig。 9)。

0。000     0。050     0。100     0。150     0。200     0。250     0。300     0。350        0。400

Average strain [%]

Fig。 9: Influence of the initial imperfections shape (Standard model with Welds A)

The deflection produced in the panel by lateral pressure is different from the local buckling mode (Standard procedure)。 This deflection is of a thin-horse mode and is composed of many deflection components including the local buckling mode。 If only the buckling mode is given as initial deflection, buckling mode gradually develops when the load approaches the buckling load。 However, when a thin-horse mode is assumed as initial deflection, other components restrain the  development of the buckling mode until the load  exceeds the buckling load。 As a result, very clear  buckling behaviour is observed as indicated in Fig。 10。 This buckling behaviour could be accompanied by a snap- through if the load at which buckling deflection appears is higher than the buckling load in simply supported mode。 Fig。 10 gives the deflection history of the centre point of two adjacent plates。 Due to the initial deflection shape, the deflections at the two points are initially identical。 Latter, for higher compressive load, the two plates buckle but in opposite direction。 As the initial deflection shape differs completely from the shape of the collapse mode, it induces a strengthening of the structure。

Compressive stress (in equilibrium with the tensile stress)

f0。2(HAZ) = f0。2  (of parent material)

-10。00 0。00 10。00 20。00 30。

Deflection along OZ (mm)

Fig。 10: Shape of the plate collapse mode (ISSC model without HAZ)。 Stress–deflection curves at the centre of two consecutive plates。

Amplitude effect (sensitivity assessment): Sensitivity assessment is obtained from Fig。 11。 On average, for the ISSC model, each millimetre of initial  deflection induces about 1。1% of reduction of the  ultimate strength。 Such a small variation shows that the amplitude of the initial deflection is not a key factor (for the considered panel)。

Fig。 12: Residual stress across the HAZ

For the transversal welds (C1 and C2), the residual stresses are acting perpendicular to the residual stresses of the longitudinal welding seams。 The width of the field of compressive stresses in the plate is a problem, as it cannot be the entire length of the  three-spans model。 Therefore, they are considered to act only in the middle span of the model, between the two transversal beams (T2 and T3)。 The magnitude of the transversal residual stresses is determined in the same way as for the longitudinal ones。

Weld positions and weld types (sensitivity assessment): The results are presented in Fig。 12, Fig。 13 and Table 2。 The tendency is the one already expected: the panel has a higher ultimate strength for welds B (extruded element) than for welds A (stiffeners welded on the plate)。

The same trend is obtained with or without residual stresses (compare Fig。 5 to Fig。 13, Fig。 6 to Fig。 14 and Table 1 to Table 2)。 Ultimate strength reduction is a little bit larger with residual stress than without (excepted for Welds B)。

Fig。 11: Effect of the amplitude of the initial deflection (ISSC model and ISSC initial imperfections)

Phase B4––residual stresses

In this phase, the effect of the residual stresses in the HAZ on the ultimate strength of the panel was analysed。 Checking the effect of welding is, of course, a very difficult task。 The simulation of welding itself is beyond the scope of work of the ISSC committee。 However, a simplified distribution of residual stresses was assumed (Fig。 12) which was implemented in the FE model as initial stresses。