An evolutionary gain formulation is proposed for minimizing the performance
damage index of steel buildings subjected to earthquake forces. The gain formulation
herein is used to develop the evolutionary control law of a control algorithm applied
to inelastic systems. The optimal evolutionary gain is subsequently used to control
building damages by satisfying desired performance objectives per time step “as
needed”. The performance objectives are defined for various “damage-safe” and elastic
demands. When the structure responds in the post-yield (inelastic) state, the
material is assumed to follow a kinematic rule for strain hardening, which
consequently may redefine the performance objective window at each unload/reload
response state (cyclic control).
A control nonlinear time-history analysis program, dubbed CONON,
was developed to simulate the stress-strain responses of structural members
and to compute the optimal control forces per time step. The minimization
of the cost function is independent of weighing matrices, thus alleviating
cumbersome calculations that also lack physical description. Instead, an
iterative Riccati matrix is computed per time step and is used to generate the
evolutionary gain for the system leading to an appropriate evolution of the state
transition between time steps. The calculated control responses are compared to
uncontrolled responses. The results are also compared using various methods of gain
calculation by examining the force-deflection hysteresis plots, the strain energy
dissipation in the structural members, and the member accelerations of a
steel frame. The proposed optimal system shows an excellent capability
to control the desired target responses and meet acceptable performance
objectives.
California State University,
Fresno
Department of Civil and Geomatics Engineering
M/S EE94
2320 E. San Ramon Avenue
Fresno, CA 93740-8030
United States