Abstract. Engineering structures, like bridges, dams and towers are designed by considering temperature changes, earthquakes, wind, traffic and pedestrian loads. However, generally, it can not be estimated that these structures may be affected by special, complex and different loads. So it could not be known whether these loads are dangerous for the structure and what the response of the structures would be to these loads. Such a situation occurred on the Bosporus Bridge, which is one of the suspension bridges connecting the Asia and Europe continents, during the Eurasia Marathon on 2 October 2005, in which 75 000 pedestrians participated. Responses of the bridge to loads such as rhythmic running, pedestrian walking, vehicle passing during the marathon were observed by a real-time kinematic (RTK) Global Positioning System (GPS), with a 2.2-centimeter vertical accuracy. Observed responses were discussed in both time domain and frequency domain by using a time series analysis. High (0.1–1 Hz) and low frequencies (0.00036–0.01172 Hz) of observed bridge responses under 12 different loads which occur in different quantities, different types and different time intervals were calculated in the frequency domain. It was seen that the calculated high frequencies are similar, except for the frequencies of rhythmic running, which causes a continuously increasing vibration. Any negative response was not determined, because this rhythmic effect continued only for a short time. Also when the traffic load was effective, explicit changes in the bridge movements were determined. Finally, it was seen that bridge frequencies which were calculated from the observations and the finite element model were harmonious. But the 9th natural frequency value of the bridge under all loads, except rhythmic running could not be determined with observations.
A Bootstrap Method to Test Granger-Causality in the Frequency Domain
