Engineers use models of physical systems that aim at representing accurately the behavior of the underlying system under future conditions. These models are subject to uncertainties which stem from either lack of sufficient data on a parameter value or the intrinsic randomness of a phenomenon, such as a future wind loading condition. Proper quantification of model uncertainties and their impact on the performance of the model is paramount for obtaining accurate predictions. We apply state-of-the-art methods for quantifying the probabilistic description of model parameters and the uncertainty they introduce in the system response. Our methods are able to combine information from different sources and adapt the models and their predictions when new data become available.