152 TIME-DEPENDENT AND NON-LINEAR PROBLEMS
coefficients it is possible to remove the time dependence in the
equations by taking a Laplace (L) or Fourier transform (&). The
resulting equations may then be solved in the transform space with
the usual boundary element method. Equations involving the time
variable are usually hyperbolic (or parabolic) in type and as such are
unsuitable for solution by the boundary element technique, the
transformed equations, however, are in many cases elliptic.
Once the equations are solved in the transformed space the original
variable involving the time dependence may be recovered by inverting
the transformation numerically. The type of transform used, either
Laplace
(L)
or Fourier (&) will depend on the type of solution sought.
We shall illustrate this point by considering
a
simple 'forced vibration'
problem.
Consider the second-order differential equation,
a
2
u du ,
m
—-=-
+
c —-
+
ku
= Fe^
at
2
at
(6.1)
This equation represents the forced vibration problem shown in
Figure 6.1. Here m is the mass of the system, c the damping, k the
stiffness and Fe
lwi
is some forcing function with angular frequency ω;
u is the displacement of the mass.
ζ
Figure 6.1 Forced vibration of a
damped system
Y.
Fe
1
Equation (6.1) has a general solution consisting of the following
two parts.
(i) The complementary function obtained by considering the
homogeneous form of the equation,
d
2
w dw ,
m—-y + c—- +
ku
= 0
at
2
at
(6.2)