Cross-laminated timber (CLT) has become one of the most popular mass timber products. Its overall behavior is strongly influenced by the CLT layup and the properties of the layers. Existing calculation approaches reflect this through simple calculations using mean stiffness values. This study evaluates the validity of these simplifications for CLT panels made with a highly heterogeneous layup and partially reclaimed wood. 

Five-layer CLT panels were produced using various wood species, namely Dutch-grown Ash and Douglas, and reclaimed Spruce in the cross layers. Ultrasonic non-destructive testing (NDT) was performed on each individual lamella before panel assembly, yielding a detailed dataset of the longitudinal elastic moduli. Lamellae locations within the CLT were recorded during assembly. 

Parametric finite element models were generated via Python scripting in DIANA FEA, and modal analyses were carried out. Several statistical and probabilistic approaches were investigated for assigning the elastic moduli to lamellae to the respective layers within model: (i) assigning the actual measured modulus to each lamella, (ii) assigning a random modulus from the dataset to each lamella, (iv) using the global mean value per layer, (iii) using the statistical mode, and (v) randomly sampling a modulus value from the measured dataset (Monte Carlo approach). The results from the FEA analyses were compared with results from laboratory modal tests. The impact of each modulus assignment strategy on the dynamic response of the panels was evaluated by comparing natural frequencies, mode shapes, statistical indicators, and relative errors.