Estimation of the static modulus of elasticity of wood from the dynamic modulus of elasticity

Aim of study: Analyze the behavior and predictive capacity of different non-destructive testing (NDT) methods—ultrasound, longitudinal and transverse vibrations, and static bending— for estimating the MOEGTO in two species of great forestry interest in Europe: Pinus sylvestris and Populus × euramericana. Additionally, this study focuses on developing global prediction models for MOEGTO using the NDT modulus of elasticity from both species, and on improving the reliability of prediction through the cross-validation process. Area of study: Provenance Region 8 (Northern Iberian System, Spain) and Ribera del Carrión (Palencia, Spain). Material and methods: A total 75 sawn lumber samples (45 of Pinus sylvestris and 30 of Populus x euramericana) with dimensions of 60 × 40 × 1200 mm were analyzed using ultrasonic testing, longitudinal and transverse vibration methods, and both destructive and non-destructive static bending tests performed with a universal testing machine. Regression models were developed to predict the static modulus of elasticity (MOEGTO). Main results: The prediction model using the NDT method of longitudinal vibrations with Fast Fourier Transform (FFT) software for Pinus sylvestris showed the highest coefficient of determination with R2= 0.56, while for Populus × euramericana, the ultrasound method with the Sylvatest 4 device yielded R2= 0.67. For both species, the best-fit global prediction model was obtained with the ultrasound method using the Sylvatest Trio device, achieving R2= 0.60. Additionally, the application of cross-validation improved the predictive performance of the global models, particularly for the three ultrasound methods and the longitudinal vibration method, all of which attained R2 values above 0.6. Research highlights: This study confirms the reliability of ultrasound and longitudinal vibration techniques for estimating MOEGTO and contributes new insights by comparing multiple NDT methods across two species and developing robust cross-validated global prediction models.