Acoustic-structural interaction in an infinite acoustic
medium
This example is intended to illustrate and verify the use of
simple absorbing boundaries and acoustic infinite elements in a coupled
exterior acoustic-structural system.
Problems involving infinite regions of acoustic media, solved
using absorbing boundary conditions and infinite elements, are compared to
converged solutions.
This benchmark problem is a simplified version of a characteristic problem
in exterior acoustics: a thin shell structure immersed in a fluid. Since the
intention here is only to test the effectiveness of the boundary conditions,
the structure's purpose in this model is to introduce nontrivial dynamics in
the acoustic mesh. In particular, the field incident on the radiating boundary
should include significant spatial variation for the test to be meaningful.
The absorbing boundary conditions intended for three-dimensional
applications are tested here in axisymmetric meshes. The two-dimensional
elliptical and circular type radiation conditions are also applicable in
three-dimensional analyses with radiating boundaries in the form of
right-elliptic or right-circular cylinders.
Figure 1
shows the test mesh used for the circular and spherical boundary condition
tests. On the inner radius of the mesh there is a circular structure of 4 units
in radius consisting of shell elements in the axisymmetric case and beam
elements in the two-dimensional case. The shells are connected to the acoustic
domain with a tie constraint. The structure has stiffeners to disrupt the
symmetry of the solution and to produce a nontrivial spatial variation in the
pressure field incident on the absorbing boundary. Surrounding the structure is
a fluid modeled with acoustic elements and terminated at the outer edge by the
absorbing boundary conditions. The mesh shown uses an outer radius of eight
units; it is stretched along the Y-axis by a factor of 1.2
for the elliptical and prolate spheroidal boundary condition tests.
Much larger meshes, 28 units in radius, are also run in both the
axisymmetric and two-dimensional cases. The results from these meshes provide
reference solutions against which to compare the results from the test meshes.
Both steady-state and transient dynamic conditions are considered. The
transient analyses are also performed using
Abaqus/Explicit.
For tests of the acoustic infinite elements, the acoustic infinite elements
are coupled directly to the structural model using a tie constraint. The center
of the shell is used as the reference point for the acoustic infinite elements
(see
Acoustic, Shock, and Coupled Acoustic-Structural Analysis).
While in this case the acoustic infinite elements are coupled directly to the
structural model, an alternative modeling strategy would involve defining an
intermediate region comprising acoustic elements between the structure and the
acoustic infinite elements. In certain situations the second approach may lead
to improved accuracy.
The acoustic medium has a bulk modulus of 2.25E9 Pa and a density of 1000
(the properties of water). For the steady-state analyses the acoustic material
has no volumetric drag. For the transient analyses a volumetric drag parameter
is introduced with a numerical value equal to approximately 10% of the speed of
sound. This value is not negligible in the sense of the approximation made in
the derivation of this boundary condition (see
Coupled acoustic-structural medium analysis);
therefore, this case is a good test of the formulation.
Loading
The structure is excited in all cases with a concentrated load applied to
degree of freedom 2 on the node at the free (innermost) end of the shell
stiffener.
The absorbing boundary tests use nonreflecting boundary conditions to
generate the necessary data automatically at the absorbing boundary. For the
two-dimensional circular case and reference solutions and the spherical
(axisymmetric) case and reference solutions, only the radius of the terminating
circle or sphere need be specified to impose corresponding nonreflecting
boundary conditions. In the two-dimensional elliptic case and the
three-dimensional prolate spheroidal case, additional data need to be included
to specify the size and orientation of the ellipse or spheroid.
For the
Abaqus/Standard
steady-state analyses using infinite elements, the infinite elements model
radiation damping as well as the added mass effect, when coupled to the
structural elements.
Results and discussion
The steady-state analyses are performed using the direct-solution
steady-state dynamic procedure. The analyses are performed for two frequencies:
the frequency corresponding to
to test the effectiveness of the radiation boundary conditions, and the
frequency corresponding to
(in two dimensions) or
(in three spatial dimensions) to test the limits of the acoustic mesh
discretization (less challenging frequencies for the boundary conditions). Two
steps are used. The impedance boundary condition is applied in the first step
(for both frequencies). In the second step these impedances are removed and
surfaces impedances of the same value are applied. The results with both
options are identical.
Each analysis shows good agreement between the reference solution and the
solutions using the absorbing boundary conditions or the acoustic infinite
elements on the smaller test meshes. The lower frequency is a more challenging
test for the boundary conditions, particularly in two dimensions, where the
boundary conditions are only asymptotic.
Figure 2
through
Figure 5
show the pressure amplitudes on the surface of the shell as a function of
angular position. The angle is measured from the Y-axis,
as shown in
Figure 1.
Each figure shows three curves; where differences between them are visible, the
circular and elliptical condition solutions overlay the reference solution, and
the spherical and prolate spheroidal conditions deviate by small amounts. The
results from the acoustic infinite elements are similar.
The transient analyses are performed with an excitation frequency of 100 Hz
using two steps and a fixed time increment of 2.5 × 10–5 units,
approximately one two-hundredth of the wavespeed. The analyses are run for
0.003 time units in the first step, where the impedance boundary condition is
applied. This time period is long enough for the wave to reach the boundary of
the test meshes. In the second step the simulation runs for another 0.003
units, this time with the boundary condition applied using a surface impedance.
The total simulation time, 0.006 units, is not long enough for the wavefronts
emanating from the structure to reach the boundary of the reference meshes, so
the impedance conditions there do not play a role in the simulation.
Figure 6
through
Figure 9
show the pressure amplitudes in the acoustic domain at selected times. In each
case the reference solution is on the left. Very good agreement is achieved in
all cases; the three-dimensional boundary conditions perform slightly better,
for reasons discussed in
Coupled acoustic-structural medium analysis.
The
Abaqus/Explicit
transient analyses performed using acoustic infinite elements give very similar
results to the
Abaqus/Explicit
transient analyses performed using nonreflecting boundary conditions. The test
using two-dimensional acoustic infinite elements gives very similar results to
the test using the circular radiation condition, while the test using
axisymmetric acoustic infinite elements gives very similar results to the test
using the spherical radiation condition.
Figure 10
shows the pressure amplitudes on the surface of the shell at an angle of 90° to
the vertical for the two-dimensional analysis (for a clear comparison the
Abaqus/Standard
transient analysis results are also included). The pressures are seen to match
closely. The tests with acoustic infinite elements do not include any acoustic
elements to model the fluid surrounding the structure. However, the additional
computational cost due to the acoustic infinite elements offsets any savings
obtained from not including the acoustic elements. For this example it was
found that the acoustic infinite element analyses were about 12% more expensive
than the corresponding analyses without acoustic infinite elements; i.e., with
acoustic elements and nonreflecting boundary conditions.