Session: 02-10-02 Collision and Crashworthiness 2

Paper Number: 79655

79655 - Discretization Challenges in Crash Simulations: Mesh, Geometry and Failure Criterion Effects From the Energy Perspective 

The simulation of collisions and crashworthiness in general of ships using FEA has a long history
of rules and guidelines to which such simulations should conform. Nevertheless, there are still
subjects that lack thorough understanding in literature and standards. One of these subjects is
the concept of scaling of the failure criterion for different mesh sizes for the use of shell
elements. Strains at which failure occurs differs by definition by the length over which the
strain is calculated. A characteristic of steel in tensile tests is that it exhibits uniform elongation
until the increasing resistance from work hardening can’t keep up with the decrease in cross-
sectional area due to conservation of volume. This necking phenomenon marks the point from
where the strain is no longer uniform and the gauge length starts to affect the measured strain.
Historically, this has been dealt with in FEA by adjusting the strain to failure by the ratio of
element length to the thickness. This data is easily obtained by using multiple gauge lengths on
tensile tests to measure the failure strain. However, since the localization is driven by the
conservation of volume, the ratio of in-plane strains in plate structures, or stress state,
influences the onset of localization. As long as the deformation is caused by a load away from
the localization, one could perform material tests with different stress states and find the gauge
length effect for different stress states. This data can then readily be applied in FEA packages
with GISSMO like failure models. In crash simulations, ship structures are often meshed with
element sizes ranging between 10 times the plate thickness to 1 element per stiffener spacing.
Given that the size of the necked region is in the order of the plate thickness, and reported
failure strains decrease inversely proportional to the element size, failure strains not much
higher than the necking strain can logically be used in these cases. Failure criteria adopted in
standards such as DNV-GL and ADN reflect this, but also include failure strains much smaller
than this necking strain. This is presumably done to both be conservative and in an attempt to
mitigate the issues caused by the inability of the larger shell elements to deform in the way
demanded by the loading. The inability of the mesh to deform to this shape introduces the
problem that one may observe different failure mechanisms depending on the mesh size or
load placement with respect to the element boundaries. In energy based criteria, often found in
the form of an amount of energy being dissipated before X m penetration or tank failure, large
differences may arise because of a different deformation mechanism. Furthermore, with the
knowledge that in equi-biaxial tension necking should not occur and in other stress states the
localization is different from the uniaxial tension case, stress state dependent failure criteria
have seen a rise in popularity. These often focus on the range between uniaxial tension and
equi-biaxial tension (triaxialities between 0.33 and 0.67). Failure in the stress state range
between pure shear and uni-axial tension (triaxialities between 0 and 0.33) is often ignored.
This paper discusses some issues raised in the paragraph above using the scenario of a V-bow
impact on another ship. It is shown that the placement of the incoming V-bow with respect to
both element boundaries and the location of stiffeners significantly influences the energy
absorption. It is also shown that the use of different size meshes and different failure criteria
result in different failure mechanisms and deformation patterns. Lastly, the stress states in
which most of the plastic work is performed (and therefore energy is dissipated) are shown to
be in the triaxiality range of 0.27-0.35. The energy contributions of the regions above that,
towards equi-biaxial tension, and below that, towards pure shear, are very significant on first
contact and shape the initial deformation mechanism.

Presenting Author: Okko Coppejans TNO

Authors:

Okko Coppejans TNO
Noud Werter TNO