Acceleration response spectrum for predicting floor vibration due to occupant walking

Annoying vibrations caused by occupant walking is an important serviceability problem for long-span floors. At the design stage the floor׳s structural arrangement may frequently change to cater for the owner׳s varying requirements. An efficient and accurate approach for predicting a floor׳s acceleration response is thus of great significance. This paper presents a design-oriented acceleration response spectrum for calculating a floor׳s response given the floor׳s modal characteristics and a specified confidence level. 2204 measured footfall traces from 61 test subjects were used to generate 10 s peak root-mean-square acceleration response spectra, on which a piecewise mathematical representation is based. The proposed response spectrum consists of three main parts: the first harmonic plateau ranging from 1.5 to 2.5 Hz, the second harmonic plateau ranging from 3.0 to 5.0 Hz and the descending part going with frequencies from 5.0 to 10.0 Hz. The representative value of each plateau and the mathematical representation for the descending curve were determined statistically for different confidence levels. Furthermore, the effects of factors, such as floor span, occupant stride length, higher modes of vibration, boundary conditions and peak acceleration response, on the proposed spectrum have been investigated and a modification measure for each factor is suggested. A detailed application procedure for the proposed spectrum approach is presented and has been applied to four existing floors to predict their acceleration responses. Comparison between predicted and field measured responses shows that the measured accelerations of the four floors are generally close to or slightly higher than the predicted values for the 75 percent confidence level, but are all lower than the predicted values for the 95 percent confidence level. Therefore the suggested spectrum-based approach can be used for predicting a floor׳s response subject to a single person walking.