Overstrength in timber engineering: General discussion and proposal for a more reliable and broader application

This review paper explores the concepts of overstrength and ductility as fundamental components of capacity-based design with a focus on timber engineering. It addresses the necessity of implementing overstrength and capacity-based design principles for effectively managing both accidental and characteristic load scenarios. Current approaches define overstrength either within the Load and Resistance Factor Design (LRFD) framework or by means of structural reliability analyses at component level, both with clear limitations in controlling the overall structural reliability and the target failure hierarchy, which is exemplified and demonstrated by a parameter study. In response, a new thematic approach for overstrength factors is proposed, formulated within the structural reliability framework, which enhances the understanding and application of ductility in design. This new methodology is benchmarked against existing approaches, therefore demonstrating its effectiveness. A clear and concise definition of ductility is emphasised as essential for a meaningful application of capacity-based design in clear conjunction with the concept of overstrength. The aim is to create a clear classification of component failure modes either as brittle (non-dissipative, non-ductile) or ductile (and dissipative, i.e. connections which provide also ductility in reverse loadings without severe pinching), which is crucial for reliably achieving a failure hierarchy where ductile failures precede brittle failures. The findings underline the importance of integrating overstrength and ductility into timber engineering practice to improve structural safety and at the same time to achieve resilient designs. • Review of overstrength determination. • Compared LRFD and structural reliability-based approaches to determine overstrength. • A new thematic approach for overstrength factors is proposed. • Analysis of the effects of different statistical distribution models.