Seismic behaviour of masonry buildings with timber diaphragms

Abstract

Existing masonry buildings are very frequently part of the architectural context for several countries all over the world. These constructions often feature masonry walls as vertical structural elements, and timber floors and roofs as horizontal components. The often poor characteristics of masonry, along with the in-plane flexibility of the floors and their frequently weak connections to the walls, make such buildings very vulnerable against seismic actions, as proved by the destructive consequences of several earthquakes in the last decades.
The improvement of seismic capacity of existing masonry buildings is still an open research topic, with several retrofitting techniques for masonry walls and timber floors being proposed and tested. An acknowledged way of increasing the structural performance of a masonry building under an earthquake consists of the development of the so-called box behaviour, enabling the construction to react as a whole to the ground motion. To pursue the box behaviour, the main adopted retrofitting methods are linked to in-plane stiffening of the timber floors, and seismic strengthening of the connections. In this context, several (nonlinear) analysis method for masonry buildings
(e.g. the pushover analysis) assume that rigid floors are present, and out-of-plane failure mechanisms of masonry are prevented. Therefore, also in the context of numerical modelling, the in-plane response of the diaphragms is generally not taken into account in detail, since they are only considered as linear elastic orthotropic membranes or stiff elements.
However, past seismic events demonstrated that an excessive stiffening of the diaphragms can also be detrimental. This triggered the study of several lighter, moderately stiff strengthening methods for timber floors, referred to various architectural frameworks. Since existing floors proved to be excessively flexible, but also too stiff diaphragms could not be recommendable, the overarching research question of this dissertation arises:

“How is it possible to predict the global seismic behaviour of existing and retrofitted masonry buildings, and to optimize it by quantifying the influence of strengthening interventions on timber floors and timber-masonry connections?”

The research question is answered starting from the specific situation of the Groningen area, in the northern part of the Netherlands, where human-induced earthquakes caused by gas extraction take place. The local building stock is composed for more than 50% of low-rise masonry constructions with timber floors and roofs, none of which was designed or realized with seismic events in mind, since earthquakes were absent until recently.
Up to now, these events have not caused extensive structural damage, because their intensity was from low to moderate, but according to probabilistic studies, more intense earthquakes might also occur. For this reason, a seismic characterization of local timber and masonry structural components was firstly necessary: this dissertation focused in particular on the testing and modelling of as-built and retrofitted timber diaphragms and timber-masonry connections.
As a first step, because it was not possible to test a large number of whole structural components on site, a replication method based on material properties was defined. This ensured that the specimens constructed in laboratory could be representative for the actual structural components in practice. With regard to timber diaphragms, in-plane quasi-static reversed-cyclic tests were performed on five as-built samples, which showed an approximately linear, and very flexible response. For these diaphragms, a retrofitting technique enhancing not only strength and stiffness, but also energy dissipation of the diaphragms, was designed. This dissipative contribution of the floors can be relevant, because it can potentially (greatly) dampen the seismic shear forces on masonry walls, and this characteristic can be even more important for the Groningen area, where low-quality masonry and very slender piers are often present. The strengthening technique for the floors consisted of an overlay of plywood panels screwed along their perimeter to the existing sheathing: the retrofitted diaphragms exhibited a great enhancement in seismic properties, with relevant increase in strength, stiffness, and energy dissipation.
The great potential of the developed retrofitting technique could not be limited to an experimental characterization: an analytical model was also necessary to enable the design of the strengthening for other contexts or floor configurations. Hence, starting from the analytical formulation of the load-slip behaviour of the single screws connecting planks and plywood panels, the global in-plane response of the floors was derived, including their characteristic pinching behaviour. This analytical model had also another important function, because it was the basis for an advanced numerical implementation of the seismic response of timber diaphragms in finite element software. This enabled to account in detail for the (dissipative) in-plane behaviour of the diaphragms, so that the seismic response of existing masonry buildings could be optimized with a well-designed retrofitting of the floors.
Yet, this energy dissipation can only be activated by means of the in-plane deflection of the diaphragms: to avoid out-of-plane collapses of masonry walls, this deflection has not to be excessive, but at the same time also effectively strengthened timber-masonry connections are needed. Therefore, an experimental characterization of two as-built and five strengthened timber-masonry connections was conducted. The joints were tested under monotonic, cyclic, and also high-frequency dynamic loading, by subjecting them to an induced Groningen seismic signal. Seven replicates per joint type were built and tested, and analytical models for evaluating strength and stiffness of the joints were derived, useful for design purposes or as input for numerical models.
Finally, in order to study the possible optimization of seismic capacity of existing masonry buildings, it is also necessary to define proper criteria for an optimal retrofitting. The current seismic design framework is extensively based on peak ground acceleration, which cannot, however, take into account factors such as load duration and quantification of structural damage. These parameters can play a crucial role for the Groningen region, because of the transient nature of the local earthquakes, featuring short, high-frequency and sudden signals, if compared to the longer and more damaging tectonic earthquakes.
Therefore, an energy-based approach was adopted, which allows to predict and quantify the hysteretic energy provided by a building as a function of its period and the load duration of the earthquake.
This approach opened up the opportunity to quantify structural damage in terms of number of cycles on the system: the role of timber diaphragms becomes, then, even more relevant, because with an optimized retrofitting the floors are only moderately stiff, thus the period of a building would be higher than that of the same structure featuring stiff diaphragms. Furthermore, the possibility to include load duration enables a characterization of the seismic capacity independent of the context or the earthquake type.
Thus, to prove the beneficial, dissipative effect of the optimized retrofitting of floors, as well as the effectiveness of the adopted modelling strategies, numerical time-history analyses were performed on three case-study buildings. The first two were typical Dutch constructions, subjected to induced earthquakes, while the third was a country house from the Italian context, to which tectonic earthquakes were imparted. This additional building was included because it enabled to demonstrate that the developed retrofitting and modelling principles, along with the energy-based characterization of seismic capacity, can be generalized to other context besides the reference Dutch one.
The results from the analyses show that excessively flexible floors cause, as expected, out-of-plane collapses in masonry walls, while excessively stiff floors limit the energy dissipation to masonry piers only, thus reducing the seismic capacity of the building. On the contrary, an optimized retrofitting is able to retrieve the global base shear of the building, and at the same time its maximum displacement capacity within masonry drift limits. The optimal strengthening also corresponds to the maximum hysteretic energy that can be provided by the structure. Furthermore, the period of the building is also increased compared to stiff floor configurations, meaning that the structure is subjected to a lower number of cycles, besides benefitting from the additional damping effect activated by dissipative diaphragms. This dissipative contribution can be brought into play provided that an effective strengthening of timber-masonry joints is realized. The beneficial, dissipative effect of well-retrofitted, optimized timber floors was quantified in terms of an equivalent hysteretic damping ratio of 15% (additional to the dissipation already provided by masonry walls), and of an increased behaviour factor (q) range for masonry structures: from the usual values of q = 1.5 ÷ 2.5 to q = 2.5 ÷ 3.5 in presence of dissipative diaphragms.
This research study can contribute to a more efficient seismic retrofitting of existing buildings, enabling preservation of the architectural heritage and more dissipative, earthquake-safe masonry structures.