Numerical models, like the phenomenological Beam-On-Foundation (BOF) approach, have proven to be an efficient alternative to the analytical European Yield Model (EYM) for the design of dowelled timber joints according to EN 1995-1-1 (2004) (EC 5) (see e.g. Lemaître et al. (2018), Bader et al. (2016)). In contrast to the EYM, BOF-models allow not only for prediction of the load-carrying capacity, but also for prediction of the load-displacement behavior of single-dowel connections, and thus of their stiffness. This makes BOF-models predestined for the design of joints in advanced modern timber structures, which for reason of their complexity rely on a reliable prediction of the jointload-deformation behavior.

BOF-models are used since the early thirties of the last century (Hager, 1930). Models of different complexity were used from simplified (i) rigid-ideal plastic models, which allow only for strength prediction (cf. Johansen (1949)); to (ii) bi-linear elastic approaches, being able to predict stiffness and strength (Sawata and Yasumura (2003), Cachim and Franssen (2009)), and (iii) nonlinear elastic models, which are optimized for numerical simulations (Lemaître et al., 2018). BOF-models might be even used for earthquake design by application of plastic, or even hysteresis models (Izzi et al. (2018), Girhammar et al. (2017)). Developers and users of such phenomenological models face the challenge to find reliable input data on the load-deformation behavior of steel dowels embedded in wood or wood-based products.

Lack of input data for numerical models, like BOF, is one of the main reasons which hinders application of such models in engineering practice and in research. Since the EYM of EC 5 uses only the embedment strength (fh,EC5) but no stiffness, as input, the related European standard for embedment testing, EN 383 (2007), focuses mainly on the embedment strength determination. However, numerical modeling requires information on the entire load-displacement curve from embedment tests.The aim of this contribution is to

1. present methods to analyze and parameterize experimental load-displacement curves for BOF-models, with embedment parameters suitable for this purpose;
2. provide a database of embedment parameters for different wood spieces and wood products, and try to find correlations between parameters;
3. give recommendations for embedment testing, with the aim to exploit data in numerical models.