Papers »
Hybrid simulation of uncertainty contaminated structural systems
In the last two decades, hybrid simulation (HS) has been recognized as a very effective testing strategy for simulating the dynamic response of structural systems whose dimensions and complexities exceed the capacity of typical testing facilities. The hybrid model of the overall structure, which combines numerical and physical substructures (NSs and PSs), accurately reproduces the overall dynamic response of the structure with reduced costs and efforts. Nonetheless, the considerable influence of uncertainties on system behavior, which has also led the scientific community to recognize the importance of a stochastic approach to engineering problems, is largely unexplored in hybrid models. In fact, a rational treatment of the uncertainties, for example by means of probability theory and statistics, cannot be addressed rigorously if the traditional HS approach is employed. From this perspective, the present paper explores the possible enhancement of HS for uncertainty-contaminated hybrid models. Owing to obvious limitations, the use of computationally advantageous and non-intrusive solution algorithms is of paramount importance in this endeavor. In particular, the broad class of spectral methods, which appears to be particularly suitable for this purpose, is investigated. First, a benchmark linear case study is introduced with uncertainty in both boundary conditions and input force. Since the order of the polynomial expansions influences the number of HS runs, “optimal” truncations are evaluated by comprehensive numerical simulations. Then, the same study is extended to a slightly nonlinear system. Finally, the proposed approach is validated on a three degree-of-freedom experimental benchmark system PS.Author(s):
Giuseppe Abbiati
ETH Zürich
Switzerland
Catherine Whyte
ETH Zürich
Switzerland
Stefano Marelli
ETH Zürich
Switzerland
Luca Caracoglia
Northeastern University
United States
Bozidar Stojadinovic
ETH Zürich
Switzerland