Effects of full-height nonstructural partitions on modal properties of two nominally identical building floors

This paper presents a combined experimental and numerical investigation of the modal properties of two fullscale and nominally identical steel-concrete composite floors. The floors were one above the other in the same fully operational multi-storey building. Both floors accommodated open-plan as well as partitioned offices. Multi-input-multi-output (MIMO) modal testing was employed to measure as-built modal properties of both floors. It was found that the two nominally identical floors had different modal characteristics, likely due to the different arrangement of partitions in the floor. It was also found that the measured modes on both floor levels experienced a considerable level of complexity, likely to be caused by nonproportional damping. Finite element (FE) models were developed in ANSYS for both floors using best engineering judgement and their features and properties were then tuned to match the measured counterparts. The tuning was done manually by trial-and-error and then automatically using sensitivity-based FE model updating procedure implemented in the FEMtools software. It was found that the initial and geometrically very detailed FE models, which did not feature any nonstructural components, underestimated the measured natural frequencies by up to a considerable 20%-25%, depending on the floor level. When full-height plasterboard and glass partitions were explicitly modelled as vertical springs connected to the floor and grounded at the other end, the correlation between the experimental and FE results improved considerably