Sequential Experimental Design of Hybrid Simulations For Bayesian Calibration Of Computational Simulators


METADATA ONLY
Loading...

Date

2021

Publication Type

Conference Paper

ETH Bibliography

yes

Citations

Altmetric
METADATA ONLY

Data

Rights / License

Abstract

Hybrid simulation combines physical and numerical substructures interacting with each other in a real- time control loop to simulate the time history response of a prototype structure subjected to a realistic excitation. Research on hybrid simulation is limited to the coupling of physical experiments and numerical simulations, that is, ensuring compatibility and balance of interface quantities involved by the physics of the prototype structure. No effort has been made yet to link experimental design and hybrid simulation. experimental design can be broadly defined as the process of selecting a particular setup for an experiment, from the parameter space of all the admissible setups, so as to maximize the information obtained from a future experiment prior to data collection. From a hybrid simulation perspective, the experimental design includes the definition of numerical substructures and loading excitations. From this standpoint, this paper presents an experimental design procedure that aims at maximizing the convergence rate of the Bayesian calibration of a computational simulation against hybrid simulation experiments.

Publication status

published

External links

Editor

Book title

Proceedings of the Seventeenth World Conference on Earthquake Engineering Japan 2021

Volume

Pages / Article No.

Publisher

National Information Center of Earthquake Engineering

Event

17th World Conference on Earthquake Engineering (WCEE 2020-21)

Edition / version

Methods

Software

Geographic location

Date collected

Date created

Subject

Hybrid simulation; experimental design; model calibration; Kriging surrogate modeling

Organisational unit

03962 - Sudret, Bruno / Sudret, Bruno check_circle

Notes

Conference postponed due to Corona virus (COVID-19). Rescheduled from September 13–18, 2020 to September 27 – October 2, 2021. Conference lecture on September 29, 2021.

Funding

Related publications and datasets