Prediction of global bending stiffness of timber beams by local sampling data and visual inspection

The results of visual inspection according to UNI 11119:2004 and bending tests made on 20 old chestnut (Castanea sativa Mill.) beams according to EN 408:2010 were statistically analyzed in order to provide a consistent and feasible procedure to predict their modulus of elasticity (MOE) in bending. Local data obtained from smaller size specimens was used for the prediction of the global mechanical properties of full structural size members and compared with the results of mechanical tests. The prediction models took into account the visual strength classes and the influence of defects on the determination of the MOE. Moreover, random sampling was considered in order to demonstrate the possibility of using smaller representative samples, thus avoiding excessive need of removal of on-site samples and allowing for a smaller number of mechanical tests. The models using random sampling predicted the behavior of full size scale elements accurately, with strong correlations to the experimental results (coefficient of determination r ² ranging from 0.70 to 0.79) and a percentage error lower than 20 %, thus allowing a reliable estimation of mechanical characteristics of existing timber members with a combination of visual inspection and local sampling.