On Linear/nonlinear Seismic Analysis Methods Of Bridges

Seismic analysis methods for long-span bridges have received much attention during the past several decades. The construction of middle- or long-span bridges is also developing rapidly with the high speed economic development in China. The current (JTJ004-89) applies only to bridges with 150m or shorter spans, that has not fitted the current requirements. Therefore its revision work has begun. One of its' purposes is giving conductive proposals or principles for the construction of longer-span bridges. To achieve it, a great deal of theoretical research and numerical computations/comparisons must be carried out. For instance, it is well- recognized that the seismic spatial effects are important for long/middle-span bridges. So far, however, the response spectrum method (RSM) adopted as the most important tool by the code is unable to deal with such problems. The time history method is not popular because of its very low efficiency. The random vibration approach is theoretically superior in dealing with such multiple excitation problems, however it has hardly been practically applied due to its computational complexity, in addition to the problems relating to its input and output. The pseudo excitation method (PEM) has brocken its efficiency-bottleneck, and appears to be quite effective in coping with such multiple excitation random analyses.Therefore introducing the random vibration approach into the bridge engineering as a tool of seismic designs not only meets the objective requirement and the modem technological development, a good methodology background is also available. From another point of view, facing the drastic international bidding competition, a code that involves technical means with our own superiority will certainly benefit our international competition. There are too many examples that some contries restrict others in this manner. The research of this doctoral thesis is aiming at the revision of Code TJT004-89. It not only provides sound theoretical basis and innovative analysis methods for the code under revision, but also makes deep-reaching explorations for some issues that are beyond the current version of the code. The main research work covers the following aspects:(1) The random vibration approach, response spectrum method and time history scheme are compared systematically. A great deal of numerical analyses are conducted to investigate the dynamic responses of bridges with different spans subjected to homogeneous/non-homogeneous seismic excitations. It has been concluded that the adoption of random vibration approach not only leads to safer and more reliable design, but is also very simple and effective. In particular, for the revised version of the code which will cover bridges with span beyond 150 m, this thesis provided necessary theoretical arguments as well as convenient analysis means.(2) Seismic excitations are in essence random, meanwhile the structural parameters are also random. The influences of these two random factors on the bridges are sometimes equally important. However, a single problem that involves these two random factors ( "double-random problem") would be very difficult to deal with. This thesis first turns the double random problem exactly into a single random problem by using the pseudo excitation method which turns the stationary random excitations into deterministic sinusoidal excitations. Furthermore, a scheme is developed by which a few single-parameter basedrandom perturbation analyses are combined to produce the results for a multi- parameter random purturbation analysis. The proposed method is very efficient for complex structures, and its accuracy has been justified in terms of Monte-Carlo simulations. The method has been applied to complex bridge systems to analyse the influence of randomness of structural parameters on their earthquake-resistant capabilities. Such practical numerical-based research has not be seen in the literature, that creates a new approach to the double random...