Analytical framework for predicting stiffness in single wooden dowel timber-to-timber connections

Artikel i vetenskaplig tidskrift, 2026

Engineered wood products (EWPs) connected with wooden dowels, commonly referred to as dowel-laminated timber (DLT), are increasingly adopted due to their recyclability, on site adaptability, absence of synthetic adhesives, rapid assembly, and design flexibility. The structural efficiency of DLT systems is governed by joint stiffness, which serves as a key input for the gamma and the shear analogy method used to evaluate the effective bending stiffness of DLT beams and panels. In practice, this stiffness is derived from tests on single wooden dowel timber-to-timber connections (WDTCs). Despite their importance, a comprehensive evaluation of existing theoretical stiffness models, alternative formulations, and their predictive accuracy for WDTCs remains lacking. Current Eurocode 5 stiffness provisions are not directly applicable to WDTCs because their governing failure mechanisms differ fundamentally from those of metal fastener timber-to-timber connections (MFTCs), involving coupled dowel embedment, bending, shear, and timber deformation. Similarly, the Timber Frame Engineering Council (TFEC) standard addresses the strength of pegged mortise and tenon joints but provides no guidance on stiffness. Unlike metal fasteners, wooden dowels deform significantly; therefore, the influence of key geometric and material parameters on WDTC stiffness differs from that of MFTCs and remains insufficiently quantified. Moreover, empirical equations for predicting WDTC stiffness are scarce in the literature. To address these gaps, this study refines, validates, and proposes several mechanics-based stiffness models for WDTCs, including beam on elastic foundation (BEF), spring in series (S-S), rigid rotation with friction (RRF), embedment stiffness, simply supported dowel (SSB) models, and selected codified formulations. Statistical analyses, including analysis of variance and best subset regression, are conducted to quantify parameter influence and to develop new empirical input model. All prediction models are validated against a comprehensive dataset compiled from the literature, and a new embedment-based stiffness formulation is introduced. The BEF, RRF and embedment stiffness approaches achieve stiffness predictions with maximum deviations below 20 % across the dataset. Finally, a practical and validated design framework for predicting WDTC stiffness is proposed.