DEFINITION OF MECHANICAL PROPERTIES OF EXISTING MASONRY ACCOUNTING FOR EXPERIMENTAL KNOWLEDGE BY BAYESIAN UPDATING


Abstract eng:
The current approach of the Italian and European building codes to account for knowledge-based uncertainties in the seismic assessment of masonry buildings consists in selecting a knowledge level and reducing material strengths by means of the associated value of confidence factor. Previous works showed that, in the case of masonry structures, the codified approach leads in many cases to inconsistent and unrealistic results and does not properly consider the experimental tests performed. This paper proposes a methodology for taking into account the knowledge on material properties acquired by experimental tests, using Bayesian updating techniques. The use of the Bayesian approach allows to update the values of the material properties assumed a priori as knowledge on the building increases, by taking into account all the information (experimental or judgement-based) gained during the assessment process. The material parameters resulting from this method could be used as input in numerical models, with the aim of calibrating confidence factors on material properties. Finally, analytical expressions are provided, which approximate the results obtained by Bayesian updating, allowing the analyst to obtain the values of material properties to be used in the analysis, as a function of the experimental information gained, without the need of performing a case-by-case Bayesian updating.

Contributors:
Publisher:
National Technical University of Athens, 2015
Conference Title:
Conference Title:
COMPDYN 2015 - 5th International Thematic Conference
Conference Venue:
Crete (GR)
Conference Dates:
2015-05-25 / 2015-05-27
Rights:
Text je chráněný podle autorského zákona č. 121/2000 Sb.



Record appears in:



 Record created 2017-06-22, last modified 2017-06-22


Original version of the author's contribution as presented on CD, section: .:
Download fulltext
PDF

Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)