Bridge live load models with special reference to Hong Kong

Abstract

This thesis presents the development of a methodology for deriving statistical highway bridge live load models for short span bridges. The methodology is applied to using Hong Kong weigh-in-motion (WIM) data to establish live load models for bridge designs in Hong Kong. The methodology can also be applied to the development of bridge live loading models in other parts of the world. Different WIM systems and their brief history are reviewed. Ten year's WIM data collected from five WIM sites in Hong Kong are statistically analyzed as a prerequisite for the development of bridge live load models. The bridge live load models adopted overseas are reviewed and compared. The comparison result shows a wide variety of solutions to representing the traffic loads for bridge designs. It also shows that the live load models supplied in BS 5400, the current bridge design codes in Hong Kong, are not the best models that suit the traffic condition in Hong Kong. Statistical approach is adopted to analyze the collected WIM data. Parameters that affect the formulation of the loading model are identified and considered one by one. The statistical distributions of these parameters are investigated and described using several stochastic processes. The results obtained illustrate that the distributions of Hong Kong traffic vehicles are of three basic distributions, namely Inverse Gaussian, Lognormal and Gamma distributions. It is also interesting to note that the maximum value of gross weight and axle weight during bridge design life of 120 years, calculated according to the proposed statistical method are very close to the legal weight limits of Hong Kong. Two methodologies, statistical analysis and extrapolation, are used and compared in this thesis to obtain the maximum value of gross vehicle weight within bridge design life. The statistical method can provide the mathematical distributions and parameters of studied objectives. The results obtained by extrapolation method used overseas are adopted to verify the statistical method. A methodology based on the equivalent base length concept is also used to derive the lane loading model for Hong Kong in this thesis. The standard truck loading model is also proposed using the statistical approach. The multiple presence effects in two-way bridges of short span are investigated through a case study. The traffic streams are drawn from the WIM records and the effects induced by vehicles in one or two lanes are calculated by convolution integral. The arbitrary-point-in-time distributions for two-lane bridges, derived using a convolution procedure, demonstrate good correlation with that calculated by the simulation of assumed traffic streams. A procedure for the calibration of the proposed Hong Kong bridge design loading models (PHKBDL) is presented in this thesis. The reliability index, load and resistance factors are calculated so that the reliability of bridges designed using the PHKBDL will be at the predefined target level.