Recently it has become more and more important to ascertain residual safety in accidental case such as collision, grounding etc. of various kind of sophisticated ships or mobile offshore units. In order to localize an accidental damage of those structures, or to confine within a minor damage, it is very essential to quantify a structural resistance, especially relation between absorbed energy and penetration depth. And for this purpose, it is fundamentally important to point out when a outer shell get teared and inner frames or a deck plate get buckled. To observe and quantify those behaviors of steel structeres, many experimental crush tests have been tried. In this report, to improve predicting method and get more precise estimation of damages, some considerations and basic test results are shown about fracture characteristics of materials used in similitude test and its effect on a law of similitude. It is since a plate thickness can affect not only geometrical dimension, but also material characteristics such as ductility and toughness. Namely, conducting I-beam bend series test of geometrically similar specimen, it's shown that critical strain corresponding to plastic deformation instability of thin plate (2.3 mm thickness) is considerably larger than that of 20 mm thick plate. And the absorbed energy in each specimen at fracture is not consistent to the law of similitude, but deeply dependent on fracture characteristics of the material. That is, the absorbed work including fracture energy can't be estimated by a simple formula, third power to the scale factor of the model. It is emphasized that model test results should be modified by calculating the critical strain to the onset of ductile fracture of used material and that its fracture mode, ductile appearance or brittle one, should be checked by a material strength test to be in a same type of that in original structures.