Introduction

Shaking table test on masonry structures are relatively few in literature. Their results are also highly dependent on specific material characteristics and hence of modest generic utility. As it is seen in the brief literature review below somewhat contrasting results can be often obtained.

A series of shaking-table tests on 24 %-scale two-storey masonry models (Benedetti et al., 1998) assessed the importance of the original quality of construction and the significant increase from even simple strengthening on the lateral resistance. In pseudo-dynamic tests conducted by Paquette et al., 2004, a full-scale one-story unreinforced brick masonry specimen with a flexible wood diaphragm and a gap at one corner was excited with selected accelerogrammes. The influence of discontinuous corners is stated to be negligible during high intensity seismic excitation. Stable combination of rocking and sliding mechanisms was found under large deformations without significant strength degradation. The influence of opening ratios on local and global rocking has been studied by Yi et al., 2006 with quasi-static tests on full-scale two-story unreinforced masonry model. It was observed that a large initial stiffness resulted in high initial damage, with stiffness decreasing rapidly with the increase of lateral drift. On the other hand, Adams (Adams, 1996) stated the difficulty in direct comparison of the results from scaled model with those from the full-scale system. Particularly he highlights the gaps between behaviour of models in labs and of real buildings because of differences in loading rates and component characterizations with scaled materials.

Shaking-table tests on masonry models can be basically classified into two groups. The first group uses only a few units of blocks (Casolo, 2000; Lourenco & Ramos, 2004), which is easy to carry out and gives clear behaviour of single rigid block motion. However, as it cannot simulate the interaction among large numbers of units in real masonry walls, the results and derived theoretical models are of limited use for real structures. The second group uses large-scale models(Benedetti et al., 1998; Griffith et al., 2004; Paquette et al., 2004; Lourengo et al., 2005). In this case, as the requirement of reasonable stress state at joints makes very stiff components and large extra loads, these models actually deviate largely from the original structures in the geometrical forms, loading conditions, stiffness and hence the dynamic characteristics. Thus, results obtained require substantial post analysis and interpretation to provide extrapolation to real structure. However, both groups of tests have shown that failure takes place in regular patterns whose arrangements depend on structural and geometrical characteristics of the wall layout both at material and structural element or assembly level (Anderson, 1976; Hendry, 1973; West et al., 1977). This brief review explains the rational for a series of test on 1/10 scale models of brick blocks 3D wall assemblies set in dry work. The aim is to gain better insight into the behaviour of masonry historic structures subjected to seismic action by using the approach of multi rigid body dynamics. The models are relatively simple. Yet a sufficient number of blocks is used to ensure that the fundamental interaction among parts of masonry walls connected in 3D to form buildings is replicated to reasonable accuracy. In the following sections the background to this set of tests and the test set up are first presented, then the results discussed and a simple control approach is proposed to provide a simple and effective modelling.

2 SHAKING-TABLE TESTS 2.1 Background

The series of shaking-table tests presented here is part of an experimental program to identify dynamic behaviour of masonry structures. Previous theoretic and experimental studies conducted by D'Ayala & Speranza; 2003 Restrepo et al., 2004 and Shi & D'Ayala, 2006; Shi et al., 2008, have shown how using limit state analysis and pseudo-static tests is possible to derive a consistent model of behaviour of cracking and damage of historic masonry subjected to lateral action and how the behaviour can be correlated to a relatively small numbers of geometric and structural parameters, without relying on stress analysis. While results of these studies are also confirmed by in situ observation of damage to buildings subjected to earthquakes, their static nature fails to provide insight in the damaging process and hence fails to accurately quantify the strength and "ductility" resources that are available during the hysteretic behaviour. It is argued here that notwithstanding the fact that the constituent materials of masonry, bricks or blocks and mortar, are not ductile, substantial dissipation of energy can take place during the damaging process at the cracks interfaces due to sliding and rocking of portions relative to each other, and hence via friction and impact.

As the processes outlined above are post material capacity, both in tension and compression or shear, it is not essential to replicate the continuum nature of the masonry. This is the main reason for adopting models in dry masonry so that the emphasis is directly on the post-elastic behaviour. It may be argued that the lack of mortared joint would increase the dissipating capacity of the scale model with respect to a real masonry structure, because of the possibility of each unit to randomly move with respect to the adjacent one. This indeed does not occur, as verified by the tests, because the gravity load distribution is proportional to the real case and hence actual relative movement does indeed occur only along main lines of crack pattern. Neglecting the elastic phase has two advantages: the first is that in scaling down the model the considerations on stresses need not to be addressed, the second is that results can be more easily generalised to masonry structures with a variety of materials.

2.2 Experimental setup

Three series of models were built and tested on the Instron Structural Testing Multi-Axis Shaking Table (IST-MAST) system with 6 degrees of freedom, sited at the Mechanical Engineering laboratory of the University of Bath, UK. The control system used is the Labtronic 8800 Control Electronics with a Basic 8800 Console, control cards for each axis position and SCM data acquisition boards-table accelerometer monitoring and customer feedback. Models tested have rectangular plan shape so as to collect information on fa├žades with different height to length ratio. Panels perpendicular to y axis have 7 bricks in length, while the facades perpendicular to the x axis have 5 bricks. They are built without windows and at corners the superposition of bricks is equal to half the length resulting in competent connections. The bricks have dimensions (L x W x H) 100 x 50 x 35 mm, and they are regularly staggered to provide s/h = 50/35 = 1.4, which from previous experimental and analytical studies (D'Ayala, 2005, Shi et al., 2008) has shown to provide good equivalent shear strength to the masonry.

Model dimensions are shown in Figure 1. The first and second series of models were built with 9 courses

Figure 1. Plan (a) and elevation (b) of the three models.

Figure 1. Plan (a) and elevation (b) of the three models.

of bricks, while the third series had 15 courses to investigate the difference in height to length ratio further and to study the effect of different area to mass ratio. The roof structure is simulated by five timber beams set a regular spacing on the top course with weights bolted at each end to simulate the actual roof mass and to ensure proper simulation of the constraining effect of floor structures onto the wall. Of course the issue of rigid or flexible diaphragm action is not addressed at all here. However, to obtain a more even distribution of mass and constraint, especially over the corners, series II models have extra weight applied to the corners for a total 20% more with respect to series I models (Figure 2).

Besides the accelerogrammes registering the six component of motion of the shaking table, each specimen is instrumented with 6 displacement transducers, 3 for each of the x and y alignment, set on the central brick and the corner bricks on each side. As the bricks of the top layer are less stable without mortar, in order to represent the global shaking, displacements of the 2nd top layer are measured. Additionally, there is one more transducer in the middle of the 8th layer in

Figure 2. (a) Series I and II and (b) Series III set up model setup.

each direction for Model III. Details of the transducers layout are shown in Figure 3.

Using a simple sinusoidal input with 50-cycle duration, each of the models is subjected to different group of tests where only the amplitude to the acceleration of the signal was increased, as summarised in Table 1. One of the objectives of the tests is to clarify whether for a same energy input greater levels of damage are triggered by acceleration or increased amplitude of the motion. A second objective is to identify whether the response is frequency sensitive, in other words, notwithstanding the fact that the specimen is not a continuum, whether something akin to natural frequency or range of frequency for which the response is enhanced, can be identified. Hence the input frequency varies following two principles: (1) under constant amplitude. This is used on Model I to identify resonant frequencies. (2) Under constant peak acceleration. To ensure this, as the input wave is defined in terms of frequency and amplitude, the relationship between frequency and amplitude of two successive input series is (4)2/(4)2 = Ab/Aa, where f is the input frequency and A is the corresponding amplitude. For Model II Series, the effect of input

Figure 3. Displacement transducers setup.

Table 1. Group classification of shaking-table tests.

Figure 2. (a) Series I and II and (b) Series III set up model setup.

Table 1. Group classification of shaking-table tests.

Group

No. of tests

Direction

Constant

Series

Was this article helpful?

0 0
Renewable Energy Eco Friendly

Renewable Energy Eco Friendly

Renewable energy is energy that is generated from sunlight, rain, tides, geothermal heat and wind. These sources are naturally and constantly replenished, which is why they are deemed as renewable.

Get My Free Ebook


Post a comment