The Brazilian standard ABNT 7190 (1997) establishes the strength classes C20, C30, C40 and C60 for the proper framework of the different wood types in the group of hardwoods. Associated with the strength class, which is based on the compressive strength characteristic value parallel to the fibers (fc0,k), the standard stipulates the respective values representing the stiffness (Ec0), with 19500 MPa being the reference value for the class C40, essential variables in structural design. For being the C40 class is the one with the greatest amplitude (20 MPa), it is possible that the value 19500 MPa is not the best representation of stiffness. This work aimed to verify the representativeness the stiffness value established by the Brazilian standard for C40 wood. The result obtained from the average confidence interval indicates the value of 14110 MPa as being the most representative, which may imply structures that are supposedly more rigid than they really are.
For safety reasons, wood strength values are calculated based on their characteristic values. Brazilian national standard (NBR, in Portuguese “Norma Brasileira Regulamentadora”) 7190 (1997) establishes ratios for characteristic strength estimation and three forms of wood characterization, with an emphasis on the simplified procedure for common species, which allows obtaining the strength characteristic values through equations correlating different mechanical properties. The present work evaluates the accuracy of the relation proposed by NBR 7190 (1997) of shear strength along the grain (fv0,k) to compression strength along the grain (fc0,k) (fv0,k=0.12 fc0,k). 960 experimental measurements of shear and compression strength values were performed for 40 hardwood species, and the precision of the relation proposed by the Brazilian standard was evaluated using the analysis of variance (ANOVA) method. Linear, exponential, logarithmic, and geometric regression models were used as an alternative to the NBR relation for shear strength estimation. The statistical analysis revealed that the geometric regression is the model of best fit.