Past research and projects

In the past, major research activity considering the seismic response of RC buildings dealt with monolithic structures rather than with RC precast buildings. As mentioned in the In troduction, the lack of knowledge about the behaviour of precast structures has resulted in strict code requirements and a too-conservative approach for the design of this structural type. In the first draft of Eurocode 8, very low behaviour factors were defined for the design of precast structures, which put them at a disadvantage compared to the cast-in-situ RC structures.

On the initiative of the industrial sector, comprehensive systematic studies of RC precast buildings were performed within several EU research projects combining the efforts of industry and different

academic institutions. As a result, knowledge about the dynamic behaviour of prefabricated structures has improved, and the competitiveness of precast building stock was enhanced.

An excellent overview of the European research of seismic behaviour of precast structures done from the mid-nineties until 2015 was made by Toniolo (2012) and Fischinger et al. (2014). This section summarises the most relevant information and complements it with recent findings.

The first draft edition of the Eurocode 8 in 1994 with a considerably conservative approach to seismic design of precast structures was the main trigger. Soon after that, the Italian association of the prefabrication industry ASSOBETON has supported a series of cyclic and pseudo-dynamic tests on precast columns in pocket foundations (Saisi & Toniolo, 1998). The test campaign confirmed the good behaviour of the precast columns, but there was still no experimental evidence about the response of the complete precast structural system.

The European project ECOLEADER - European Consortium of Laboratories of Earthquake Dynamic Experimental Research has provided the experimental comparison between the precast and cast-in-place one-storey frame structure (Biondini & Toniolo, 2004; Ferrara et al., 2004;

Biondini & Toniolo, 2009). Both prototypes were designed for the same base shear resistance and had the same fundamental period. Experimental results have demonstrated the same seismic capacity and quite similar behaviour of the two tested systems.

The next project, PRECAST - Seismic Behaviour of Precast Concrete Structures with respect to Eurocode 8, was supported by ten partners from Slovenia, Italy, Portugal, Greece and China. Within the project, pseudo-dynamic and cyclic tests of a full-scale one-storey precast structure were performed at ELSA, the European Laboratory for Structural Assessment in Ispra, Italy (Biondini et al., 2008). The tested structure had a realistic roof that proved to behave like a rigid diaphragm.

Research also pointed out the significant effect that the connections between precast elements and cladding panels might have on the behaviour of the complete structure. In the case of the tested structure, the cladding panels changed the response significantly. This response, however, depends on the type of cladding connections used because different types of connections may provide different levels of interaction between the panels and the main structure (see Section 2.3).

As explained by Fischinger et al. (2014), the tested precast structure had large overstrength.

Yielding of the columns was not observed until the last pseudo-dynamic test with a maximum ground acceleration of 0.525 g. Substantial top displacements of 40 cm or 8% drift were achieved, where the yield drift was over 2%. These large drifts match the response of precast RC buildings during the Emilia-Romagna earthquake. Ercolino et al. (2016) reported the large yielding rotations of precast columns at around 2% drift, which was also confirmed with subsequent nonlinear dynamic analyses.

Experimental research performed within the PRECAST project provided valuable data about the seismic response of RC precast structures, which was subsequently used in extensive numerical and analytical studies (Biondini & Toniolo, 2009; Kramar et al., 2010). In the past, numerical modelling of precast structures has been extensively studied at the University of Ljubljana (Kramar, 2008;

Zoubek, 2015; Babič, 2017). An inelastic numerical model for columns was modified to accurately estimate the seismic response of slender columns typical of prefabricated industrial buildings (Fischinger et al., 2008). Such columns have high shear span ratios (height to width of column) of more than 10, low axial compressive load ratios (less than 0.16), high deformability and large deformation capacity (over 2% yield drift and around 8% ultimate drift).

The recommended model for columns (Fischinger et al., 2008) assumed a yield chord rotation (i.e.

yield drift) to be the sum of theoretically determined flexural deformations as proposed by Fardis

& Biskins (2003) and the empirically calibrated contributions of shear and bond-slip. The numerical model was validated by full-scale cyclic and pseudo-dynamic tests of a one-storey precast structure (Toniolo, 2007). The whole building was modelled as an equivalent cantilever column using the lumped plasticity beam-column element model with a zero-length plastic hinge at its base. The hysteretic moment–rotation response was described with two different models: the modified Takeda hysteretic model (Takeda et al., 1970) and the modified Ibarra hysteretic model (Ibarra et al., 2005) calibrated by Haselton (2006). Both were able to adequately describe the response observed within the tests. Verified models were further used in systematic seismic risk studies of realistic one-storey industrial buildings used in practice (Kramar et al., 2010).

Many experimental and numerical studies of the different types of connections, most commonly used in the European design practice of precast buildings, have been done within the European project SAFECAST - Performance of Innovative Mechanical Connections in Precast Building Structures under Seismic Conditions. Four classes of the connections were investigated: floor to floor, floor to beam, beam to column, and column to foundation (Toniolo, 2012b).

The most extensive and essential series of pseudo-dynamic tests of a full-scale, three-storey precast building were performed at the ELSA Laboratory of Ispra (Negro et al., 2013). Also, many tests on sub-assemblies of structural elements connected with joints have been done in other European laboratories. The main interest of the research team at the University of Ljubljana was the study of the beam-to-column dowel connections (Zoubek et al., 2013; Zoubek et al., 2014; Zoubek et al., 2015). The dowel-type connections are the most frequently used beam-to-column connections in Central European precast industrial buildings. Complete information about the failure mechanisms of dowel connections was obtained using numerical models calibrated by the set of experiments reported in Zoubek et al. (2013). The proposed procedure for calculating the resistance of a dowel

connection for different reinforcement layouts (Zoubek et al., 2015) is included in the recently composed draft of the Eurocode 8 standards.

The last joint EU project, SAFECLADDING - Improved Fastening Systems of Cladding Wall Panels of Precast Buildings in Seismic Zones, was devoted to the connections of façade cladding panels to the main structural system of industrial buildings. The goals of the project were identification of basic response mechanisms of different cladding connection types, improvement and definition of the design procedures and a proposal for the improvements of the connections (SAFECLADDING, 2015). Comprehensive experimental (Negro & Lamperti Tornaghi, 2017;

Toniolo & Dal Lago, 2017) and analytical studies performed within the aforementioned European project considerably improved the knowledge about the seismic response of the cladding panel fastening systems.

Within the SAFECLADDING project, cyclic and pseudo-dynamic tests on full-scale structures were performed (Negro & Lamperti Tornaghi, 2017; Toniolo & Dal Lago, 2017), as well as many monotonic and cyclic tests of single connections (Zoubek et al., 2016a; Psycharis et al., 2018;

Yüksel et al., 2018). The behaviour mechanisms of different connection types were defined, and proposed numerical models were used in many subsequent analytical and numerical studies. Three different basic concepts (presented in the following Section 2.3) were assessed and considered within the studies.

Part of the research, performed at the UL (Isaković et al., 2013; Fischinger et al., 2014; Isaković et al., 2014b), was devoted to the fastenings systems of vertical and horizontal cladding panels that are widely used in the existing practice in Central Europe. The behaviour of hammer -head strap connections that are typical cladding connections for vertical panels was explained and studied in detail by Zoubek et al. (2016a). A design procedure was also recommended, and reliable macro models for simulation of hysteretic response were proposed.

Recent projects significantly raised the awareness of many problems with the existing design and construction practice. One of the SAFECAST project outcomes was a manual, Design Guidelines for Connections of Precast Structures under Seismic Actions (Negro & Toniolo, 2012), which became an ISO standard ISO 20987:2019. The SAFECLADDING project resulted in the new design guidelines for precast structures with cladding panels (Colombo et al., 2016b) and wall panel connections (Colombo et al., 2016a).

Similar research campaigns were performed on national levels, mostly in Slovenia and Italy, in parallel with the mentioned European projects. The in-plane and the out-of-plane seismic response of the connections used to fasten the horizontal cladding panels was experimentally and analytically studied by Belleri et al. (2016; 2018). The isostatic types (described in Section 2.3) of connections

for vertical and horizontal panels were extensively investigated by Del Monte et al. (2019). They successfully modified the cladding connections to improve their displacement capacity.

Despite the extensiveness of the projects presented above, past research mainly focused on investigating the response of single components based on the monotonic and cyclic experiments.

Many important observations about the seismic response of cladding panels typical for Central Europe have been obtained at the UL within the SAFECLADDING project. Many experiments have been done, and valuable data were obtained (Isaković et al., 2013; Isaković et al., 2014b; Zoubek et al., 2016a). Results of these studies have also been used within the STREST project (Esposito et al., 2020) to derive fragility functions of industrial precast building classes and perform seismic risk analyses (Babič & Dolšek, 2016). However, this research could not completely reveal and explain all the aspects of this complex response. The behaviour of the cladding systems under the dynamic loading was insufficiently studied. It was not possible to fully determine the role of panel fastenings and their realistic boundary conditions without a more complex study of the whole-system response. Research continued within the UL research project Seismic resilience and strengthening of precast industrial buildings with concrete claddings, funded by the Slovenian Research Agency (CLADDINGS, 2016) to find answers to these questions.

One of the main phases of this project was devoted to the full-scale shake table experiments of an RC building with cladding panels. Different parameters within these experiments were varied, including the orientation of cladding panels, the type of cladding-to-structure connections and the configuration of the specimen (symmetric and asymmetric).

To be able to set up these tests, additional cyclic and dynamic tests of the single components were performed to obtain as much data as possible about their basic seismic response mechanisms and their capacity. Experimental observations and results of the shake table tests with vertical panels are presented in Isaković et al. (2018), while the research on horizontal panels is the topic of this doctoral dissertation. The experimental studies were proceeded by the related analytical studies and numerical analysis.