WHAT ONE NEEDS TO KNOW FOR THE ASSESSMENT OF TIMBER STRUCTURES Helena Cruz, José Saporiti Machado, Pedro Palma. Laboratório Nacional de Engenharia Civil (LNEC), Lisboa, Portugal ABSTRACT The assessment of old timber structures is normally a great source of problems for building interventors and a frequent justification for integral replacement of structures that would otherwise be kept in service. It requires understanding the original structural system behaviour, the estimation of strength and stiffness properties of the timber in use, the evaluation of individual timber members’ quality and effective cross section dimensions. Due to the natural variability of timber, unknown load and environment history, and common biological damage, assumptions have to be made, frequently with a high degree of uncertainty. This paper discusses the common approach in the assessment of timber structures in service and tries to identify advances, problems, knowledge gaps and research needs related to this activity. 1. INTRODUCTION Repair, strengthening and upgrading of old buildings are and will represent in the future a large share of building contractors activity, as the urge for new construction is diminishing due to population stabilization in most developed countries, and the economical benefits, historical or environmental concerns justify the maintenance, as opposed to replacement, of existing structures. Nevertheless, civil engineering and architectural teaching has been mainly directed towards new construction, thus ignoring the old materials and construction techniques. This is now changing, especially at the level of post-graduate studies or optional courses, but even so the time dedicated to timber structures is in most cases negligible. At the same time, professional carpenters, in the traditional sense of long- gained practical experience and father-to-son transmitted knowledge, no longer exist. Often, so-called carpenters do not really understand the material, and frequently base their activity on erroneous principles applied with a frightening sense of self-confidence. This scenario compromises the conservation of old timber structures where specific expertise and knowledge are essential for obtaining data on timber members and joints mechanical behaviour, which is directly linked with wood species used and structural system under evaluation. Assessment of timber structures may be carried out under various circumstances. The easiest case includes the retrofitting of well preserved structure to its original condition, which can be handled with minor replacements or strengthening. In this case it is assumed that the number of years in service is a prove of its safety. However, in most cases, either due to conservation problems or change of use, judgement about material strength and damage quantification is required. In the last situation, the lack of expertise on wood as a structural material often leads to replacement of suitable timber structures, by steel or concrete structures more adjusted to current engineer knowledge. One should say however that, despite the difficulties, there have also been numerous interventions where timber structures receive careful inspection and suitable consideration. Several steps should be tackled before a final answer can be given about the safety of an existing timber structure. Inspection is a basic and crucial step upon which other steps and decisions rely. The purpose of this paper is to identify common methods, problems, knowledge gaps and research needs related to inspection of timber structures. 2. OVERALL STRUCTURAL SYSTEM EVALUATION The original structural system behaviour, although sometimes complex and difficult to understand, can in most cases be handled by structural engineers. One should have present that ancient timber structures are not always structurally sound, even disregarding possible degradation. Some exhibit an enormous degree of improvisation, some have basic conceptual/structural errors, some were poorly made and others have been altered disrespecting safety considerations. Common problems in roof structures are erroneous geometry, eccentric loading at the truss supports and due to rafters placed away from the truss nodes, lack of bracing between trusses and missing elements due to previous interventions. In the case of floor systems, insufficient support length at the beam ends, lack or sloppy bracing between beams, removal of support walls and introduction of intermediate loading partitions are quite common. Joints frequently have some kind of damage (metal corrosion, sloppiness, timber splitting or crushing) and original defects like missing plates or fasteners, minute edge and end distances of fasteners, too small washers, gaps between elements that should be in contact. The engineer should carefully inspect the joints, be able to understand them, identify failure or other dangerous situations like the consequences in the internal forces distribution of a failure in a joint, and propose remedial measures. Structural defects must receive due consideration and the interventors must be aware of the need to guarantee suitable safety levels (sometimes by restraining the use of the structure to less demanding activities) despite the historical or architectural interest of the building. 3. ESTIMATION OF TIMBER BASIC PROPERTIES Estimation of timber properties begins with the identification of wood species. This task although highly specialized, can be easily performed, especially when historical information is available (date of construction, region and wood species normally used in that period and region) thus the wood anatomist only has to confirm the species from a limited range of suspected species. After wood species identification, the tricky task of allocating characteristic values (strength, stiffness) to the different timber structural members should be performed. In general three different types of approach are adopted by contractors. These are discussed below. Sampling Building contractors are often tempted with sampling and testing timber from the structure, copying the approach followed for other construction materials. Their sampling may involve one or two structural members that for some reason are to be replaced; or involve a small amount of wood taken from different locations within the structure: either to check clear wood properties, or state of conservation, as an attempt to evaluate its effects in strength. Due to the natural variability of timber, such sampling can’t be considered representative. To account for variability, the determination of characteristic values for the mechanical properties of timber is generally based in several samples of 40 timber pieces each, of structural dimensions from a certain timber grade (exhibiting representative defects and features of that grade). This is obviously not feasible within the assessment of existing structures, but explains why strength values obtained by sampling from the structure can hardly be used with confidence. Besides, sampling of clear wood disregards the major influence of defects like knots, slope of grain or fissures, and the effect of local biological damage, which may vary considerably within the structure. Sampling is however very useful to provide information on species, moisture content and density. It can also be used to check if strength values of clear wood fall within the expected range or to restrain the quality of wood in a certain visual strength grade; but in order to do that the collected information shall be used in conjunction with methods specially developed to derive strength values appropriate for design. Load tests Load tests may be performed on the whole structure or parts of it (individual floor beams, for instance). Loading is applied to the structure by using hydraulic jacks or, more commonly, dead loads like water tanks or cement bags. The measured deflection of the structure under load is compared to the predictions of a structural model (generally a finite element modelling), and the estimated mechanical properties of the elements and joints (especially stiffness) necessary for the model are adjusted/calibrated to match the measured deflections. These provide an estimation of elastic modulus (MoE) of timber, which alongside with density serves to derive strength values. Bending strength (MoR) and compressive strength are derived from existing correlations of these with MoE+density for that specific species. The other properties are generally derived from more general correlations between different properties. Load tests may also consist of vibration tests. In this case, an instantaneous load (hit) is applied to the structure and the resulting motion is measured by accelerometers. The vibration frequencies are a function of geometry, support conditions and material properties, namely the elastic modulus. This enables an estimation of the MoE, which is subsequently used to derive the other mechanical properties. Both techniques are expensive and time consuming and the interpretation of results depends on how accurately the support conditions and the influence of load-sharing elements can be understood and modelled. Results are known to be affected by moisture content, therefore local variations of moisture content may be another source of errors. Moreover, results depend on how good the correlations between different properties are. It may happen that for the specific sample of timber used in a particular structure, the actual correlation deviates from the average correlation established for that species. Furthermore, for some species such correlations were never determined. It should also be stressed that the whole approach assumes that the correlations between different strength properties obtained for new timber of a given species also apply for old timber, even if it has suffered unknown load and environment history. Studies conducted so far on mechanical performance of structural members removed from old constructions do not seem to support the idea that age on its own is a factor that should be considered. The grading approach This can be considered the most currently followed approach. It involves the identification of the wood species and a general evaluation of timber members’ quality (density, defects). In this process ideally the evaluation of timber quality should be reported to an existing stress grading standard (used to industrially grade timber for structural uses), for which strength values can be allocated. Stress grading procedures check if the defects (specially knots, slope of grain, fissures and density) of individual members are within the limits established for a certain grade. This evaluation allows the allocation of an average grade to all members. After a first structural analysis, a more refined grading can be carried out on those members which importance in the structure and/or high stress level justify extra attention. The grading approach naturally requires an extensive survey of the structure, both costly and time consuming. It also assumes that the strength values allocated to new timber of a certain species and grade also apply to old used timber. The influence of moisture content cyclic variations in the strength and stiffness of timber has been extensively studied and most results indicate that these have a negative effect. Also load history previously applied to timber is known to reduce strength and stiffness, highly depending on the stress levels attained and the environmental conditions. Other studies suggest that time itself enhanced some of the timber properties, although we lack precise information about the properties of that same timber when it was new. Survey of timber structures have to deal with the large range of wood species used for construction, natural variability of timber (combination of wood species, growth region, physical properties and defects) and unknown load and environment history which brings a high degree of uncertainty on the assumptions taken from inspection. 5. BIOLOGICAL DAMAGE AND ESTIMATION OF CROSS SECTIONS Cross section dimensions are in principle easy to assess, although in the case of irregular round elements this may require some simplification. If biological degradation of timber is identified its overall and local effects on strength and stiffness of members and joints must be judged. This is normally a great source of problems for building interventors and a frequent justification for integral replacement of structures that would otherwise be kept in service. In the case of fungal attack, the actual strength loss varies with wood species, specific fungus, time and environmental conditions. The affected volume may be assessed by non-destructive methods. Models to predict strength and stiffness reduction have been tried, but the allocation of equivalent values for the affected section in a practical situation involves a high degree of uncertainty. Consequently, it is common to assume in practice a negligible contribution of those cross sections seriously affected by decay and to completely or partially replace those members or strengthen them as if they were already lost. In those cases where decay is incipient and it seems to affect just the surface of a member, that member may be kept in place if it is clearly overdesigned by report to calculated stresses. Timber affected by insects is a totally different matter, as the decrease of timber members’ properties result from a reduction of material in the cross section (rather than from a chemical modification of the wood cells). When attacked by beetles, it is common that only a surface layer of sapwood is destroyed. The depth of the attacked layer may be assessed with the help of a simple knife or screw-driver or by other non-destructive techniques. A reduced cross section may then be estimated. Alike the procedure followed for timber quality, safety checking may be carried out by assuming the same cross section reduction for all members in the structure, followed by a more detailed assessment and refined verification for specific members of higher importance or higher calculated stress levels. This procedure may however be too conservative, since even the damaged layer is likely to contribute to the load carrying capacity of that cross section. Although less frequent, insect attack may instead take the form of a diffuse damage throughout most of the cross section. This has been observed in softwood members with a small percentage of heartwood. Considering a reduced cross section is meaningless in this case. Alternatively, a reduced quality due to a reduced apparent density may be assumed for such cases. However, both approaches regarding insect damage involve a high degree of uncertainty since the real influence of insect attack in the strength and stiffness of timber members has yet to be clarified. 6. CONCLUSIONS The lack of proper methods to evaluate the strength of timber members on-site is a strong drawback shown by timber structures, as compared to concrete or steel structures. The assessment of old timber structures requires specific expertise and knowledge, given the need to quantify individual timber members and joints mechanical properties and effective strength. Regarding the timber members assessment, possible approaches basically include load tests and grading approach Load tests rely on existing correlations between the timber strength and stiffness in structural dimensions. Such correlations only exist for a few species. Even then, one has to assume that the same correlation found for new timber also applies for timber that grew in different conditions and that was subjected to unknown loading and environment through its service life. In the above “direct evaluation” of individual timber elements strength and stiffness, one assumes that the correlation between species + origin + (quality) grade and strength properties established for new timber also applies to old aged timber. However, this is yet to be confirmed. Besides, grading operations may be difficult in practice when there is no visual access to the ends and some faces of the timber, or when it is dusty or stained. Grading randomly chosen members may be an alternative in these situations but it introduces even another source of uncertainty. The estimation of the effective cross section dimensions represents a huge problem. Options like assuming a nil contribution of decayed cross sections, or a reduced cross section in the case of beetles surface layer attack have been widely used in practice but may be too conservative. Although a number of approaches may nowadays be used to assess old timber structures, they involve steps that still require scientific validation. Wood Science should help to reduce the present degree of uncertainty associated to some necessary assumptions, thus improving safety while avoiding too conservative evaluations that may obstacle the maintenance of still safe structures.