Improving density estimation in large cross-section timber from existing structures optimizing the number of non-destructive measurements

Abstract

Abstract The density of 45 structural Salzmann pine timber pieces with nominal dimensions of 150 × 200 × 5000 mm3 from an existing 18th century structure was estimated using the following non-destructive methods: needle penetration resistance (NPR), screw withdrawal resistance (SWR) and drilling chips extraction (DCE). Non-destructive testing (NDT) measurements were correlated with actual and apparent density (considering a rectangular cross-section instead of the actual profile). Ten non-destructive measurements were taken in each specimen with each device. The best results were obtained with DCE and the poorest with NPR. The increase of the determination coefficient obtained by increasing the number of measurements did not show a great improvement from the fifth to the tenth measurement: r2 = 0.12–0.65 for one reading, 0.35–0.77 for five readings and 0.42–0.76 for ten readings, depending on the device. Because density varies along pieces, three accumulated measurement sequences along them were analyzed to correlate better with the minimum number of readings: Linear (LM), End to Centre (ECM) and Centre to End (CEM) methods. CEM was the most effective sequence, giving higher r2 than the others. In practical work involving the in-situ estimation of density it is recommended that a minimum of 3 or 4 measurements using CEM be performed.