Abstract
Using wood for mechanical engineering as solid wood or glued solid timber (or glued laminated timber) requires a good knowledge of the properties of each elementary piece. Due to this lack of knowledge, woods are still not used in modern mechanical design.In the Languedoc-Roussillon French region, pines are the dominant forest resource (approx. 30 % of standing wood). The development of this resource is a major interest for the inter-branch organisation. The aim of this PhD work is to examine the possibility to predict basic useful properties of pine clear wood, based on available non-destructive methods. The study has been made on four of the most present species: Laricio black pine, Maritime pine, Mountain pine and Scots pine.Basic relevant indicators are the density (obtained by mass and volume measurements), the specific modulus of elasticity (obtained by vibrational measurements) and indicators from Near Infrared Resonance Spectroscopy (NIRS). Technological properties are the nine components of the elastic matrix (obtained by resonant ultrasonic measurements on a single specimen), hygroscopic dilatations and compressive breaking strength along the fibres (obtained in standard testing conditions).Data analysis allows to build some predictive models, unique for the whole pine selection, of basic properties, from indicators measured on each specimen of clear wood.Therefore it is possible to associate to a given piece of wood, a technical specification sheet of useful properties for mechanical design. Selected indicators are common measurements in industry and knowledge transfer is conceivable. Each interesting piece of wood (i.e. with few defects) could be rapidly characterized. Related to a good traceability this process could lead to identify clearly each piece and its properties, giving the possibility to the designer to select the best materials to his or her needs.