Comparison of Stochastic Identification Methods applied to the Natural Response of Millau Viaduct

The efficiency of different alternative output-only modal identification methods applied to data collected during the dynamic tests performed at the Millau Viaduct at the commissioning stage is evaluated here. In particular, modal estimates identified on the basis of an ambient vibration test are correlated with data from free vibration tests and from modal properties obtained numerically at the design stage. Figure 1. General view Millau Viaduct. 3 DESCRIPTION OF THE DYNAMIC TESTS 3.1 Instrumentation Owing to the predicted dynamic properties of this multi-span cable-stayed bridge, the type of dynamic tests required, the time available to perform them (3 days) and the number of sensors available, the CSTB team defined a Plan of Tests trying to conjugate the instrumentation of the permanent monitoring system installed by the SITES company with a complementary equipment. This plan, schemmatically represented at Figure 2, comprised 21 accelerometers disposed in 5 out of the 8 spans, measuring vertical (13 measurement sections on the deck) and longitudinal (1 measurement section on the deck and 8 in the pylons) accelerations. 12 of these accelerometers (designated by S in Figure 2) belonged to the system installed by SITES, located in the third and fourth spans supported by the highest piers, whereas the remaining 9 (indicated as C in Figure 2) belonged to CSTB and FEUP, and were installed in two consecutive spans and at the corresponding pylons, where the data acquisition system from CSTB was installed. Figure 2. Location of the instrumentation used in the free vibration test All the sensors were installed inside the box girder, along the longitudinal axis of the viaduct. Only one pair of accelerometers (C2, C4) was located at the lateral edges of the structure, with the purpose of identifying some torsional modes. Despite the different characteristics of the data acquisition systems from SITES and CSTB, it was possible to synchronize them. A sampling frequency of 40Hz was employed. An ambient vibration test was developed complementarily using 4 seismographs, including triaxial accelerometers, three of them located at positions G825, G826 and G827, indicated in Figure 2, during the free vibration test. These seismographs, duly synchronized using GPS sensors, were placed outside the deck, as shown in Figure 3. Figure 3. Accelerometers installed inside the deck and seismograph used outside 3.2 Ambient Vibration Test This test was performed on the 24 November 2004, using the 4 previously mentioned seismographs. With the purpose of identifying as many modes of vibration as possible, essentially of vertical and transversal bending nature, two of those tri-axial sensors were used as references (sections R1 and R2, Figure 4), keeping them in fixed positions, while the other two were successively placed in each one of the remaining 26 sections schematically indicated in Figure 4. The seismographs were programmed using a laptop in order to acquire signals with a sampling rate of 100Hz in periods of 960s every 20 minutes, the last 4 minutes being used to change the position of the moving sensors for the following measurement setup. The test was completed in 7h30, and during this period the wind speed was always very low and the traffic very sporadic, as the bridge had not been open to traffic yet. Figure 4. Instrumented sections of the deck in the ambient vibration test 3.3 Free Vibration Tests . Figure 5. Free vibration test: tensioned cable, cut from inside the box girder 78.52 92.48 92.48 78.52 171m 342m 342m