| dc.description.abstract | Full wood characterization tests require expensive and large equipment, found only in research centers and, moreover, require time and skilled labor. Thus, the minimal and simplified characterization are a tool widely used in the design of projects, as they make work easier, faster and cheaper. Among the properties of wood, the bulk density presents itself as a reliable estimator of the other properties used in the design of timber structure projects. As a result, many researchers have analyzed the interaction of the wood density value with the other properties through empirical equations and obtained good results. Among the anatomical factors, porosity is preponderant in influencing the value of wood density. In addition, porosity is also directly related to permeability (a major influence on the preservative treatment process and in the production of wood panels), thermal and dielectric properties, dimensional stability and mechanical strength of timber. Considering the great influence of porosity on important properties of wood and the search for mathematical models that are increasingly accurate, this study aims to generate multivariable regression models for the estimation of the physical-mechanical properties of Brazilian tropical woods as a function of porosity, obtained through the analysis of porosimetry by mercury intrusion, sometimes of the bulk density and sometimes for both properties, with the purpose of generating equations with a highs degrees of precision. For this purpose, eight species of tropical wood will be investigated, namely: Cedro-Amazonense (Cedrela spp.), Cedroarana (Cedrelinga catenaeformis Ducke), Cambará (Erisma uncinatum Warm), Tatajuba (Bagassa guianensis), Cupiúba (Goupia glabra), Jatobá (Hymenaea spp.), Angelim Vermelho (Dinizia excelsa Ducke) and Champagne (Dipteryx odorata). For the eight species, complete characterizations were carried out according to ABNT NBR 7190 (1997), the average values for porosities were obtained through the analysis of porosimetry by mercury intrusion and, finally, the statistical analysis was carried out to obtain of the multivariable regression models. As a result, the estimates of the properties of fc,0, fv,0, fs,0, Ec,0, EM and Et,0, considering the bulk density and porosity as free variables [f(ρ12%; Po)], showed higher coefficients of determination when compared to models considering either only apparent density or only porosity as an estimator of these properties. The ft,90 was possible only considering the porosity as an independent variable, where it presented R2 equal to 51,91%. The difficulty in estimating fs,0 and ft,90 can be partially explained by anatomical issues of the wood, showing how much porosity was crucial in the success of estimating ft,90. Finally, the anatomy factor considered by porosity improved the multivariable regression models in estimating the properties of fH,0 and EM, as they are properties that are strongly influenced by the anatomical factors considered by the average porosity values of wood species. | |
