Evaluation of mechanical strengths of tropical hardwoods: proposal of probabilistic models

Abstract

The characteristic value of compressive strength parallel to the grain is used to design structural members in bridges, houses and buildings. Such value is obtained based on experimental tests. The Brazilian standard proposes equations to estimate the strengths by means of probabilistic equations that allow obtaining the characteristic values of different mechanical strengths, such as the compressive ($f_{c0,k}$), tensile ($f_{t0,k}$) and shear ($f_{v0,k}$) properties obtained from a set of experimental results. Considering the results for these strengths in parallel direction to grain of 30 hardwoods, with a total of 1080 experimental determinations, the precision of the relations proposed by this standard was duly evaluated from probability distribution models. The Normal, LogNormal, Weibull and Exponential functions were used in order to determine the best adhesion model. Once the values and respective models had been determined for each species, a multivariate linear regression model, based on analysis of variance (ANOVA) and dependent on the mean value (${\bar{x}}$), coefficient of variation ($CV\%$) and lower (LO) and higher (HI) strength values, was adopted to estimate the $f_{c0,k}$, $f_{t0,k}$ and $f_{v0,k}$ adjusted with the most significant terms, in order to infer the quality of the estimator and, consequently, the reliability of such mechanical properties. Finally, the multivariate model proposed here was compared to the empirical proposition of the Brazilian standard, to evaluate the reliability of the model and its adequacy in the estimation of the characteristic strengths values to distinguish the mechanical properties of 30 tropical hardwoods.