A stochastic variational principle (SVP) is presented in this paper. Based
on the SVP, the stochastic finite segment method (SFSM) is developed for an
d applied to the analysis of shear-lag effect on a box-girder. The random p
roperty of physical and/or geometrical parameters is incorporated consisten
tly into the total potential energy functional of the box-girder. Then the
second-order perturbation techniques are employed to develop a set of expan
ded deterministic expressions of the functional. It is shown that the secon
d-order functional is necessary and sufficient to derive all the governing
equations. Then the recursive stiffness equations of the SFSM are obtained
by the stationary conditions of the second-order energy functional. The sto
chastic shear-lag effect on the box-girder is evaluated by means of the SFS
M. An example is given to demonstrate the efficiency and accuracy of the pr
esent method. Results are compared with the analytical variational solution
s and those achieved by the technique of direct Monte Carlo simulation and
the finite strip method, respectively, showing desirable agreement. (C) 200
1 Elsevier Science Ltd. All rights reserved.