Evaluation of load carrying capacity of bridge based on ambient acceleration measurements

The load carrying capacities of exciting bridges need to be properly assessed for the safe operation and efficient maintenance. The conventional methods require load tests with controlled vehicles to measure the static and dynamic responses, such as deflections or strains under known loading conditions. However, full or partial block of traffic during the tests may cause not only inconvenience to the traffic but also increase of logistic cost and time. To overcome these difficulties, an alternative method is proposed using ambient acceleration data measured without traffic control in this study. The load carrying capacity of a bridge (P) is commonly evaluated by combining the design live load (P_r), the rating factor (RF), the deflection (or stress) correction factor (K_δ (or K_ε)), the impact correction factor (K_i), and the correction factors for traffic volume and pavement roughness (K_t, K_r) as P = P _r x RF x K_δ (or K_ε) x K_i x K_t x K_r where P_r is a given design value; RF is determined by structural analysis using the initial FE model of a bridge; and K_t and K_r are empirically estimated by the structural engineer. On the other hand, two correction factors, K_δ (or K_ε) and K_i, are generally evaluated by load tests on the bridge. Static load tests are traditionally carried out to obtain the deflection correction factor (K_δ) using loaded trucks, and vehicle running tests are carried out to estimate the impact correction factor (K_i). The RF, which is the ratio of the live load resistance and the design live load, is evaluated by the allowable stress design (ASD) method for steel members and the load and resistance factor design (LRFD) method for concrete members. The proposed method for the deflection correction factor consists of the following procedures: (1) ambient acceleration measurements on the deck due to wind or the traffics on the adjacent bridge, (2) estimation of the modal properties using the stochastic subspace identification method, (3) updating of the initial FE model based on the modal properties, (4) calculating of the static deflection using the updated FE model, and (5) evaluation of the deflection correction factor by comparing the deflection using the initial and the updated FE models. The present method replaces the static deflection measurements of the bridge deck under known loading conditions, which requires expensive deflection measurement devices such as advanced laser devices and traffic control during the tests. The present method for the impact correction factor utilized the pseudo-deflections obtained by double integration of the measured acceleration records, so that dynamic loading tests using a moving loaded truck may be avoided. For stable integration, only the part of the acceleration record corresponding to the duration with the vehicle on the bridge was considered. The shift components were eliminated using a baseline correction procedure during the double integration. For validation of the proposed method a series of ambient vibration tests and load tests with various truck weights were carried out on a steel plate-girder bridge on an expressway in Korea. The tests were performed in 3 different seasons: August 2004, December 2004, and July 2005. Conventional load tests were also performed, which were composed of quasi-static load tests with a speed of 3 km/h and vehicle running tests with a speed of 50 km/h. At first, the deflection correction factors were estimated by the proposed method. The results by the proposed method showed big discrepancy compared with those obtained by the conventional method with quasi-static loads and contact type displacement transducers with connecting wires (OU displacement transducers). However, the results by the proposed method are found to be very similar to those by the conventional method using the data from a laser vibrometer, which are proven to be more accurate than those by the OU gages during the field validation tests. The estimated deflection correction factor by the proposed method shows good consistency regardless of fhe test season. The impact factors were also estimated using the pseudo-deflections obtained from the acceleration data by double integration. The results by the proposed method are very close to those by the conventional method using the measured dynamic deflections, so as the impact correction factors estimated using the impact factors. Using the above correction factors, the load carrying capacities of the example bridge were evaluated. The results of a series of field tests on a bridge may be summarized as (1) The proposed method gives very consistent results for the load carrying capacity regardless of the test season, and the results are reasonably close to those by the conventional method. (2) The accuracy of the deflection sensor is very critical to the conventional method. The conventional OU gage did not provide accurate deflections of the bridge girder, while the laser vibrometer gave good results. (3) The deflection correction factors by the proposed method using the updated FE model are very close to those obtained by the conventional method and the deflection using the laser vibrometer. (4) The impact correction factors by the proposed method using the pseudo-deflection are close to those by the conventional method. (5) Further tests are needed for validation under traffic conditions.