Abstract
A numerical method is presented for the solution of the damped transient response of tall, slender structures such as towers and telecommunications masts. Mainframe computer packages for structural dynamics are complex to use and often prohibitively expensive. An analysis developed for 'desktop' computers, with an animated display of the response, is shown to be efficient and cost-effective. A time integration procedure and spatially discrete structural idealisation are used, and the performance of the resultant algorithms is investigated and optimised. Pressure loading and base excitation are considered in detail, both as idealised functions and as digitized 'real' data. It is shown that the interaction of different excitation functions can significantly affect the response. A novel form of idealised base restraint facilitates the simulation of various support and excitation conditions. Structural damping is shown to be important, although the conventional viscous damping approach is unsatisfactory for the transient analysis of built-up structures. Several material damping algorithms are examined, a modified 'stiffness proportional' approach being identified as most suitable, and an expression is derived for support damping. The complex nature of energy dissipation mechanisms in structural connections leads to a reliance upon an empirically derived 'interface damping' term. A bolted flange is considered, and a series of static hysteresis and free vibration tests yield useful structural damping data. A power law is shown to adequately describe the relationship between the interface energy loss and the local stress amplitude. A transient test rig capable of decelerations up to 30 g was designed and used to study the response of jointed and unjointed models. Transient recording and digital filtering techniques yielded high quality data, enabling meaningful comparisons to be made with corresponding numerical simulations - an aspect neglected by other workers. As a consequence, the semi-empirical damping algorithms are shown to give adequate practical response predictions.