Submodeling is the technique whereby a portion of a structure is
analyzed with a different (usually finer) mesh by “driving” the nodes on the
boundary of that mesh from the interpolated solution on the original “global”
mesh.
To perform a submodel analysis, nodal quantities such as
displacements, temperatures, pressures, displacement phases, etc. must be saved
on the file output in the global analysis (usually done with a coarse mesh).
The global model file output is attached to the submodel run (via the
globalmodel parameter on the
Abaqus
execution procedure) to drive the boundary nodes on the submodel (usually done
with a finer mesh). The same reference frame must be used in both models. The
global and submodel meshes can have different element types within the same
group of elements: planar solid to planar solid, axisymmetric solid to
axisymmetric solid, three-dimensional solid to three-dimensional solid, general
shell to general shell, etc. For shell-to-solid submodeling the global model
consists of shell elements and the submodel consists of three-dimensional
continuum elements. The procedure types can be different between the global
analysis and the submodel analysis. For example, a linear static analysis in
the global model can drive an elastic-plastic static solution in the submodel
(as long as plasticity will not influence the driven boundary nodes), or a
dynamic analysis in the global model can drive a static solution in a submodel
(this assumes that inertia forces can be neglected at the submodel level). In
addition, the global procedure can be performed in
Abaqus/Standard
to drive a submodeling procedure in
Abaqus/Explicit
and vice versa. For example, an
Abaqus/Standard
static analysis in the global model can drive a quasi-static
Abaqus/Explicit
analysis in the submodel.
The verification tests are divided into sections according to the element
types supported in the submodel capability. Within each section a combination
of elements and procedures is tested on small models with a limited number of
elements. The values (or amplitudes) at the driven nodes, interpolated from the
global analysis, are verified. In most cases the stress and strain fields in
the submodel analysis match the results of the global analysis. However, in
certain problems the meshes are too coarse to produce good agreement in stress
and strain.
Each test consists of two input files: the global analysis and the
submodel analysis. The same global file can drive several submodel analysis
runs, each using a different mesh with elements that may or may not be the same
as in the global analysis.
An example of running a sequentially coupled thermal-stress analysis is
also given.