A direct procedure for the evaluation of imperfection-sensitivity in bifurc
ation problems is presented. The problems arise in the context of the gener
al theory of elastic stability (Koiter's theory) for discrete structural sy
stems, in which the total potential energy is employed together with a stab
ility criterion based on energy derivatives. The imperfection sensitivity o
f critical states, such as bifurcations and trifurcations, is usually repre
sented as a plot of the critical load versus the amplitude epsilon of the i
mperfection considered. However, such plots have a singularity at the point
with. epsilon = 0, so that a regular perturbation expansion of the solutio
n is not possible. in this work, we describe a direct procedure to obtain t
he sensitivity of the critical load (eigenvalue of the bifurcation problem)
and the sensitivity of the critical direction (eigenvector of the bifurcat
ion problem) using singular perturbation analysis. The perturbation expansi
ons are constructed as a power series in terms of the imperfection amplitud
e, in which the exponents and the coefficients are the unknowns of the prob
lem. The solution of the exponents is obtained by means of trial and error
using a least degenerate criterion, or by geometrical considerations. To co
mpute the coefficients a detailed formulation is presented, which employs t
he conditions of equilibrium and stability at the critical state and their
contracted forms. The formulation is applied to symmetric bifurcations, and
the coefficients are solved up to third-order terms in the expansion. The
algorithm is illustrated by means of a simple example (a beam on an elastic
foundation under axial load) for which the coefficients are computed and t
he imperfection-sensitivity is plotted. Copyright (C) 2001 John Wiley & Son
s, Ltd.