PSI - Issue 78

Vittoria Borghese et al. / Procedia Structural Integrity 78 (2026) 1229–1236

1235

For the transverse modulus ( E Y ) and shear modulus ( G XY ), no destructive data were available, but a comparison between numerical predictions and NDT results shows significantly higher di ff erences: 15.6% for E Y and 17.3% for G XY . These discrepancies suggest greater sensitivity to (material) modelling assumptions and experimental variability in directions dominated by thinner and potentially more heterogeneous layers, such as reclaimed Spruce. These results suggest that while the proposed FE framework is well suited to capture longitudinal behaviour with high accuracy, additional refinements are necessary to improve the prediction of E Y and G XY , potentially through more advanced shear modelling or improved material characterisation.

Fig. 4. Comparison of the sti ff ness values for eight hybrid CLT panels: (a) longitudinal modulus ( E X ), (b) transverse modulus ( E Y ), and (c) shear modulus ( G XY ). Experimental results (destructive and non-destructive) are compared with numerical predictions using di ff erent modulus assignment strategies. Destructive test data are available only for E X . To account for discrepancies between simplified design assumptions and actual structural behaviour, an adjustment factor λ X is introduced. It relates the elastic modulus from the numerical model E X , num to the one measured experi mentally via destructive testing E X , real using λ X = E X , real / E X , num . This factor can be used to adjust the average sti ff ness value E X , avg typically adopted in design, yielding a corrected modulus E X , e ff = E X , avg · λ X . Based on the average of the eight numerical predictions in the longitudinal direction and the corresponding experimental result, the correction factor is computed as λ X = 1 . 016. This study presents an experimental-numerical approach for the non-destructive assessment of hybrid cross laminated timber (CLT) panels manufactured of Dutch-grown Ash and Douglas-Fir, and reclaimed Spruce wood. The longitudinal moduli of the individual lamellae were determined using an ultrasonic testing before panel manu facturing. Full-scale hybrid CLT panels were formed and tested by modal analysis. A fully parametric finite element modelling framework was developed in DIANAFEA , allowing detailed representation of each lamella and evaluation of di ff erent strategies for assigning sti ff ness properties. The numerical framework accurately predicted the longitudinal modulus ( E X ), with a relative error of only 1.58% compared to destructive testing and 4.1% compared to non-destructive results. This confirms that the model cap tures the dominant sti ff ness direction e ff ectively, largely governed by the outer Ash layers. In contrast, larger devi ations were observed for the transverse modulus ( E Y ) and shear modulus ( G XY ), with relative di ff erences exceeding 15%. These discrepancies highlight the increased sensitivity of transverse and shear behaviour to assumptions on orthotropic ratios and cross-layer heterogeneity. The three investigated sti ff ness-assignment strategies, direct use of measured moduli, global layer averages, and random sampling from lamella datasets, produced similar results in terms of natural frequencies and mode shapes, particularly for the longitudinal direction. This indicates that using global average properties is acceptable for practical engineering purposes when modelling the primary sti ff ness direction. A correction factor λ X = 1 . 016 was introduced to adjust simplified design sti ff ness values based on the discrepancy between destructive and numerical results. This factor o ff ers a useful calibration tool for reconciling simplified as sumptions with observed mechanical behaviour. Notably, accurate sti ff ness predictions are especially relevant within Service Limit State design, where displacement-based criteria govern performance under regular usage and moderate 4. Conclusion