09.10.2015 Views

OS-C501

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Offshore Standard DNV-<strong>OS</strong>-<strong>C501</strong>, November 2013<br />

Sec.9 Structural analysis – Page 157<br />

11.5 Buckling analysis for sandwich structures<br />

11.5.1 Sandwich structures may be exposed to highly localised buckling modes such as wrinkling and<br />

dimpling, in addition to more global modes. For simple stress states these local modes may often be checked<br />

using standard formulae.<br />

11.5.2 The wave-lengths for wrinkling are normally very short (often of the order of the sandwich thickness).<br />

If a direct FE analysis of wrinkling is carried out it is essential that a sufficiently fine mesh be used in the skin<br />

laminates, such that the mode shape is well represented. If each skin laminate is modelled using shell elements,<br />

the element size should not normally be greater than λ/12, where λ is the buckling wavelength. The core shall<br />

be modelled with solid elements of similar size. The required element size shall be established using iterative<br />

calculations.<br />

11.5.3 In performing FE analysis of wrinkling it is not normally necessary to model a large area of the<br />

structure, provided the in-plane stress state in the skin is well represented. A portion of the panel extending over<br />

a few wavelengths is normally sufficient. The result is not normally sensitive to the size of the panel selected<br />

for modelling.<br />

11.5.4 In the absence of detailed information about geometrical imperfections and their consequences, these<br />

may be allowed for by reducing the critical wrinkling stress by 40%. The face wrinkling stress in some text<br />

book formulas may already includes such allowance.<br />

11.5.5 Wrinkling of skin laminates may be accompanied by yielding of the core if the core is made of a ductile<br />

material. This may in turn lead to a reduction in the tangent stiffness of the core and a lowering of the critical<br />

stress for wrinkling. This is mainly a problem at points of load application and at joints, where the core<br />

experiences local loading, and may be avoided by adequate thickening of the skin laminate, insertion of higher<br />

strength core material locally or by other local design features. The adequacy shall be proved by testing or<br />

analysis unless previous experience shows the solution is adequate.<br />

12 Partial load-model factor<br />

12.1 General<br />

12.1.1 A deterministic factor shall be assigned to each structural analysis method. It is designated in this<br />

standard as the partial load-model factor γ Sd (see Sec.3, Sec.2 [3.6] and Sec.8 [2.2]).<br />

12.1.2 The load-model factor accounts for uncertainties of the structural analysis method being used to<br />

accurately describe and quantify the response of the structure.<br />

12.1.3 Model factors for the main structural analysis methods are given in the following sub-sections.<br />

12.1.4 In some cases a structure is only evaluated by testing, and such an approach evaluates only the particular<br />

conditions tested. A procedure for this approach is given in Sec.10.<br />

12.2 Connection between partial load-model factor and analytical analysis<br />

12.2.1 When analytical methods are used within their assumptions and limitations a model factor of 1.0 should<br />

be used.<br />

12.2.2 If analytical methods are used outside their assumptions and limitations, it shall be documented that the<br />

magnitude of the model factor ensures that all predicted stresses and strains are higher than in reality. If the<br />

choice of model factor cannot be documented, the analytical method shall not be used.<br />

12.3 Connection between partial load-model factor and finite element analysis<br />

12.3.1 The accuracy of FE methods is generally very good when the structure is properly modelled. The use<br />

of these methods with unsatisfactory models is much more uncertain.<br />

12.3.2 When FE methods are used within their assumptions and limitations (and according to [5]) a model<br />

factor of 1.0 may be used.<br />

12.3.3 If FE methods are used outside their assumptions and limitations, it shall be documented that the<br />

magnitude of the model factor ensures that all predicted stresses and strains are higher than in reality. If the<br />

model factor cannot be documented, the analysis method shall not be used.<br />

12.3.4 If the boundary conditions do not exactly represent the real conditions the effect on the load model<br />

factor shall be evaluated. As a minimum a factor of 1.1 shall be used.<br />

12.3.5 If the load-model factor cannot be determined for calculations in a critical region, e.g. a critical joint or<br />

region of stress concentrations, experimental qualification should be done (see Sec.10).<br />

DET NORSKE VERITAS AS

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!