Quantifying mass changes with ambient vibration measurements

The static loading is one of the key unknowns in the structural design process. While weigh in-motion stations and load cells allow load to be measured over small areas or small structures, they are not viable for more complex structures such as long-span bridges or larger buildings, and can be prohibitively expensive. Operational modal analysis can reveal changes in the dynamic behaviour of civil structures measured with accelerometers, potentially offering a low-cost, durable technique for estimating changes of mass in-service. In order to do this, the dynamic parameters of the structure must be estimated with sufficient accuracy, and then processed using a mass change estimation method robust enough to accommodate noise and measurement error. In this paper a new technique for estimating changes in mass, random sampling mass estimation (RSME), is introduced and applied to experimental data for a laboratory-scale beam structure. RSME incorporates the uncertainty in the estimates of the dynamic parameters and allows least-squares estimation of the total change in mass applied to a structure if the locations of the changes in mass are known, or prediction of both the magnitude and location of a single concentrated change in mass through a novel graphical implementation. For individual masses added to the beam, 92.0% of mass location predictions have errors smaller than 2% of the length of the span and 69.4% of added mass magnitude predictions are within ±10 g, approximately 0.05% of the mass of the beam. For concurrent added mass locations, the mean error of the estimates for the total change in mass where 2 or more added masses are present on the beam is -2 g, approximately 0.011% of the mass of the beam.