Experimental and Theoretical Analysis of the Structural Behaviour of Ancient Timber Elements

Abstract

In the present thesis the evaluation of the structural behaviour of ancient timber elements subjected to bending and axial forces has been investigated. Such type of structures represented the bearing structure of floor and trust of buildings of the Italian and Mediterranean historical centres. More particularly, this study is focused on the typical timber floors of the buildings realized until the first years of the twentieth century. Such timber structures are often constituted by natural beams of chestnut, larch, fir or spruce having circular or square cross-sections. In the case of simple floor typologies, the only beam frame was directly leaned on perimeter walls, with a 4-5 m span length at a distance on center of 50-80cm, with a superposed secondary wood elements and filler materials. With the composite floor typologies, thanks to the double frame of principal floor beams, it was possible to cover longer spans-length. Anyway, generally, beams were set up without providing any special treatment. Therefore, they present remarkable shape defects and a considerable cross-section variation between the end-beams. It is worth noting that these structural members present the major calculation uncertainties on theoretical bases, due to the strong influence of the behavioural response exercised by knots and structural defects. A wide experimental campaign based on a series of ancient chestnut beams in actual dimension have been carried out. First of all, a preliminary visual inspection have been executed on the timber elements, in order to have an exhaustive relief of the actual geometry of the beams, the quality of wood and the deterioration state conditions. Later the main physical and mechanical properties of the base material has been determined. In particular, the moisture content and the density have been determined together with the elastic and post-elastic structural behaviour of wood. Compressive (in longitudinal and transversal direction), bending and shear tests on defect-free small specimens extracted from the members have been carried out in order to obtain the compressive strength, the longitudinal elastic modulus, the tensile resistance and the shear strength. These properties have been successively compared with the one obtained by means of full-scale tests on timber elements in actual dimension. Next phase of the research program concerns the evaluation of the structural behaviour of ancient timber elements subjected to full-scale static tests. Such type of elements present many calculation uncertainties due to the presence of not negligible degradation phenomena which can produce a strong reduction of the load bearing capacity. This experimental activity is based on a series of compressive and bending tests. In particular, compressive tests were carried out on three timber elements previously subjected to Non-destructive Resistographic analyses, while bending tests have been performed on four beams of about 4 m span length. They have been tested according to two different loading condition (three and four-point bending tests) as indicated in international regulations. Some interesting considerations have been made on their partially unexpected failure mechanism. Later, an experimental campaign on samples jointed by means of metal fasteners has been discussed. First, embedding tests in longitudinal and transversal direction have been carried out on ancient chestnut samples, in order to determine the embedment strength of the material. In addition, a series of bending tests on connected timber beams in small dimension have been performed. Tests were executed according to the same loading conditions used for bending tests on small and actual whole beams. Several differences in the cross-section dimensions and in the number of the metal connectors have been analysed. Next experimental phase refers to Non-Destructive Tests (NDT) on chestnut members. In particular, Resistographic analyses have been carried out on the samples successively used for full-scale compressive tests. A large number of local drilling measures have been executed on such elements both in longitudinal and transversal direction, in order to have in-depth information. Several considerations have been made on the correlation between longitudinal and transversal measures, which are the ones commonly operable in the in situ investigations. In addition, comparison between longitudinal analyses and the corresponding compressive strength have been investigated. In order to guarantee a more reliable correlation between destructive and nondestructive tests, Resistographic analyses and compressive tests have been carried out also on spruce defect-free small specimens, obtained from classified spruce beams usually available in commerce. By means of these analyses a correlation curve has been determined, which allow to deduce the mechanical properties of wood exclusively by means of Resistographic analyses.