Aleatoric and epistemic uncertainty in the overstrength of CLT-to-CLT screwed connections

The most common and practical connection between CLT walls can be realized with inclined screws. This choice avoids the realization of more elaborated half-lap or spline joints. The failure mechanism of CLT-to-CLT screwed connections is highly ductile. However, the epistemic and aleatoric uncertainties associated with the capacity estimation of the connection might lead to an undesired overstrength, compromising the expected hierarchy between failure mechanisms. This paper presents the results of an extended experimental campaign to estimate the overstrength of CLT-to-CLT screwed connections. However, the overstrength directly obtained from the experimental tests could be underestimated. In the experimental campaign, the same wood and screw stock is used, which might not represent the actual scatter of the material properties and construction uncertainties of the as-built connection (e.g., the screw inclination). Therefore, this paper attempts to provide a model-driven assessment of the overstrength factor, assuming more realistic values for the parameter uncertainties. The authors propose a method for removing the contribution of epistemic uncertainty to the model-driven estimation of the overstrength based on experimental tests with two Montecarlo simulations. Following the proposed method, the paper compares the overstrength estimations from the experimental tests to the predictions of analytical and nonlinear finite element models. This study has proven that an overstrength factor between 1.8 and 2 can represent the actual uncertainties in as-built CLT-to-CLT screwed connections.