Bayesian model updating of eight-storey CLT building using modal data

Abstract

The serviceability of tall timber buildings due to wind-induced vibrations is often a governing design criterion. However, accurate modelling of such buildings for their serviceability is a challenge, partly because certain non-structural building elements (e.g. plasterboards, façade, screed) act structurally, and no guidelines exist on how to effectively include them in the model. Model updating may be helpful in revealing their as-built characteristics. This paper presents findings of a surrogate-based Bayesian model updating of a lightweight eight-storey cross-laminated-timber building. In particular, the focus was on patterns and correlations between mass distribution and modal characteristics, as well as the effects of joints, non-structural building elements, and the foundation. Model updating utilised the results of ambient vibration testing, which is commonly used in civil engineering for structural identification or health monitoring, however, it generally offers a relatively low number of identified modes. To this end, the study investigates the sensitivity of the updating process to the number of modes involved in the analysis.