Framework for Online Model Updating in Earthquake Hybrid Simulations

Abstract

Hybrid simulation has been effectively utilized to assess structural response subjected to intense dynamic loads. The process comprises dividing the structure into experimental and numerical modules. The experimental modules represent the critical components responses, which cannot be idealized reliably through analytical approaches. The responses of the different modules are combined through a stepwise integration scheme. In conventional hybrid simulations, the number of experimental components is restricted by the capacity of the test facility; usually 1-3 components, and the numerical simulation does not benefit from the information acquired from the tested component during the analysis. In this article, a framework is proposed to identify the material constitutive relationship from the tested component(s) and to update the corresponding numerical parts that share close characteristics with the physical tests. Optimization tools and neural networks are presented as alternatives for the identification procedure; the framework is however extendable and scalable. The communication protocol between the different structural components is also discussed within the proposed framework. Several analytical examples are presented to prove the feasibility of the presented framework, while experiments are used to verify the process in a companion article.

Parameter identification for on‐line model updating in hybrid simulations using a gradient‐based method

“…They can feature the desirable simulation of the observed kinematic and isotropic hardening phenomena. [14][15][16][17][18][19][20] In hybrid simulation, also called hybrid testing, [21][22][23][24][25][26][27][28][29] the target structure can be partitioned into multiple substructures. When constitutive models such as the Bouc-Wen model and the 2-surface model are adopted in nonlinear structural response simulations, the lack of relations between the control parameters and the mechanical properties may result in difficulties in finding the parameter values.…”

“…Conventional model fitting is done using the trial and error method. [14][15][16][17][18][19][20] 2.2 | Analytical model for the BRB In order to improve the efficiency of model fitting, various techniques for parameter identification were developed in the past years.…”

“…To investigate the seismic behavior of a structure with distributed plasticity developed on BRBs with hybrid simulation, several physical substructures corresponding to the BRBs are required. From the literature, [14][15][16][17][18][19][20] it should be noted that there is a key assumption in model updating. Thus, the benefit of the hybrid simulation may diminish in such applications.…”

Development of a ten‐element hybrid simulation platform and an adjustable yielding brace for performance evaluation of multi‐story braced frames subjected to earthquakes

“…Therefore, the capabilities of the hybrid simulation framework need to be extended to test a larger number of physical components simultaneously. Approaches like the efficient selection of the physical substructures or model‐updating methods in which experimental data can be used to update numerical elements in each time step of simulation can also be employed to further increase the efficiency of the hybrid simulation …”

Hybrid testing with model updating on steel panel damper substructures using a multi‐axial testing system

“…However, if only one physical specimen is constructed and tested while all other relevant elements are numerically simulated, the advantages of HS would be significantly diminished due to the inaccurate modeling of the not‐well‐understood substructures with numerical approach. To address this issue, the development of the online model updating (OMU) technique has gained increasing attention in recent years . One effective approach for OMU is adopted in a series of hybrid tests presented in this paper.…”