Comparison of natural and artificial forcing to study the dynamic behaviour of bell towers in low wind context by means of ground-based radar interferometry: the case of the Leaning Tower in Pisa

The study of Cultural Heritage assets needs the application of non-destructive and non-invasive monitoring techniques. In particular, monuments and historical buildings which are open to the visitors and/or subject to important stress must be studied for their dynamic response. In the last 10 years the new ground-based radar interferometry technology has been developed allowing to monitor displacements from a point of sight far from the studied targets. It virtually provides a continuous mapping of displacements of the observed structures up to 10 mu m with a range resolution of 0.75 m. In this paper, the application of ground-based interferometry on one very important historical building, the Leaning Tower of Pisa in Italy, is reported. The analysis of these kind of structures is important to catch their dynamic response to natural actions in general, and also to assess the effects due to pedestrian and users, and consequently to define functional capabilities and levels of acceptable dynamic stress. The studied structure was subject to artificial loading by synchronous movement of about 20 people. Artificial forcing led the structure to a resonance condition with the same frequency of the one due to the natural noise excitation, which was separately measured, and with an oscillation amplitude more than thirty times greater than the natural one (in conditions of weak wind). During the passive stages of the survey the recorded structural vibrations were very closed to the instrumental sensitivity, making difficult to distinguish vibration amplitudes amplifications of various segments at various heights. Through the spectral analysis of the acquired data it was possible to estimate the vibration frequencies of the first modal shapes of the structure along two orthogonal directions. The power spectra of the passive survey data have the same maximum frequency of the active but contain more noise at low frequency.