STATISTICS OF VIBRATION ENERGY-FLOW IN RANDOMLY PARAMETERED TRUSSES

The problem of determining the ensemble averages of vibration energy f low in harmonically driven trusses with random parameter variations is considered. The mass, elasticity, damping and length of the truss mem bers are modelled as random variables. The uncertainty associated with member lengths result in randomness in truss geometry. Four different linear damping models for truss members, namely, strain rate dependen t viscous/hysteretic models and velocity dependent viscous/hysteretic models are considered. The analysis employs random dynamic stiffness m atrices and Monte Carlo simulation procedures. A comparison of the ens emble averages of subsystem energies obtained using this approach and those using statistical energy analysis (SEA) formalism is also made. The energy coefficient matrix arising in SEA studies is determined usi ng the direct dynamic stiffness matrix approach. The paper illustrates the relative importance of alternative damping models and alternative sources of system randomness on the behavior of spectra of subsystem vibration energies in a 13-member truss. The uncertainties in randomne ss in truss geometry and the choice of damping model are shown to sign ificantly affect the energy statistics. (C) 1998 Academic Press.