Comparative pushover and limit analyses on seven masonry churches damaged by the 2012 Emilia-Romagna (Italy) seismic events: Possibilities of non-linear finite elements compared with pre-assigned failure mechanisms

• Analysis of seven masonry churches damaged by the Emilia Romagna (2012) seismic event.
• Pushover analyses conducted on detailed plate and shell FE models.
• Collapse mechanisms and failure accelerations evaluated with FE upper bound limit analysis.
• Comparison with 28 pre-assigned failure mechanisms provided by Italian Code.
• Analysis of active failure mechanisms in light of the observed state of damage.

The most suitable way to perform a fast but reliable failure analysis of existing masonry churches in earthquake prone areas is investigated in order to predict the state of damage and eventually failure modes. Different failure analyses are systematically applied on a wide variety of different churches, that suffered damage during the recent 2012 Emilia-Romagna (Italy) earthquake sequence. The weaknesses of some specific parts, as for instance the tympanum, the apse and the lateral long walls of the naves, which are typically responsible for the partial collapse of the structures, are highlighted. The large set of examples considered in this study allows a comparative analysis of pros and cons linked with the practical application of the different procedures. In particular, global FE (finite element) pushover and limit analyses, combined with a plate and shell discretization, are adopted to have an insight into (a) active failure mechanisms and (b) accelerations associated with the formation of partial collapses. Results are compared with what stated by Italian Guidelines on Cultural Heritage for the safety assessment of historical masonry constructions in seismic zones. For masonry churches, it is required to analyze 28 pre-assigned failure mechanisms by means of the application of the upper bound theorem of limit analysis in presence of no-tension materials. It is found that FE limit analysis may provide reliable failure mechanisms – when compared with the other approaches discussed – but requiring a reduced processing time, without the need to adopt questionable a priori choices on the macro-blocks active at collapse.