Reliability Assessment And Life Prediction For Existing RC Bridges Under Multi-Source Uncertainties

Bridge structures are a key component of the transportation system. Currently, China has the largest number of bridges all over the world. Over the past decades, more than 730 000 bridges have been built in China, and more than 90% of bridges are concrete bridges. However, for existing reinforced concrete (RC) bridges subjected to environmental attack and load effect, corrosion of reinforcing bar is one of the major factors that affecting bridge service security. Huge uncertainties are associated in the reliability assessment and life prediction of existing bridges, and the uncertainty mentioned above is generally categorized as aleatory uncertainty and epistemic uncertainty. Failure to distinguish the two categories uncertainties may result in a large error for structural assessment. This paper focused on chloride-induced corrosion and studied the structural reliability assessment and remaining life prediction method under multi-source uncertainties. The main research works are as follows.(1) A probabilistic updating prediction model of corrosion-induced strength degradation for flexural beams under model and parameter uncertainties is proposed. Field inspection data is the very important reference of strength degradation prediction. Considering the current structural condition can decrease the uncertainty and improve the prediction accuracy. The relationship between corrosion loss and the yield strength of steel is established based on the experimental investigation of tensile tests of 452 corroded rebars from accelerated corrosion members and real bridges. A critical corrosion loss is proposed to distinguish the ductile and brittle failure modes of reinforcing bars. Static tests on 48 beams are performed and finite element method analysis are conducted to evaluate the effects of corrosion-induced bond degradation on carrying capacity. Following that, considering the likelihood of general corrosion and pitting corrosion, the inspection information is incorporated based on Bayesian theory and a probabilistic strength degradation model including aforementioned factors is proposed. The method is validated by the testing results from a removed bridge.(2) A Bayesian network updating-based method to tackle uncertainties for the remaining strength prediction of existing bridge is proposed. It is difficult to efficiently use the intermediate information for the current bridge assessment menthod. The load testing method is a very important method for bridge’s assessment, however, this method can not directly reflect the structural strength since destructive-based test is impractical. Considering the practical operating conditions, this study proposes a stiffness degradation model for corroded beams based on experimental investigation. Combining a real RC bridge, a Bayesian network that includes stiffness degradation, corrosion damage, load-deflection response and other factors is established to predict structural strength degradation. A Markov Chain Monte Carlo method is used to implement the calculation procedure. The load-deflection data from in situ load testing is integrated into this Bayesian network to update the entire network. The Bayesian network updating method using indirect observations enhance the use of the measured information, which provides a new method for structural estimation.(3) A fuzzy probabilistic reliability estimation method for existing RC bridges is developed. In practical applications, the statistic parameters of some variables may not be precisely obtained. In addition, it is sometimes impossible to obtain the initial information for the earlier stage of time-variant structure without the health management system. These incomplete information mentioned above is classified as fuzziness and randomness. The probability characteristics of the corrosion loss and the strength loss of reinforcing bars are simulated and verified by experimental results. The evolutionary framework of structural strength under fuzziness and randomness is established, and the probabilistic distribution pattern and the distribution parameters are also obtained. Membership functions are used to describe the fuzzy characteristic of small numbers of detection results. Following this, the fuzzy variables are transformed to the equivalent random variables and the classical reliability method can be used to calculate the reliability index. The proposed methodology is compared with the conventional probabilistic approach using goodness-of-fit method. The effect of data scarcity is also discussed in detail. The developed methodology is demonstrated with a RC bridge.(4) A systematic method is proposed to quantify hybrid uncertainties for the probabilistic durability life prediction in aging RC bridges. Corrosion damage happens when the concrete crack from first corrosion cracking to a limit width. The corrosion process is divided into two stages with three characteristics. The probability distributions of the time to corrosion-induced cracking and time for crack to a limit crack width are estimated based on the numerical simulation method. The sensitivity of input parameters on the model is also discussed. This study considers uncertainty in the distribution parameters of variables since hybrid uncertainty is associated in the assessment of corrosion damage. The key idea is to use a likelihood-based approach to calculate the probability distribution function of the variable described by sparse data and an entropy-based transformation method to obtain the probability distribution function of variable described by expert-based information. Then, a hybrid description of uncertainties is proposed using the marginal integration. Following this, a risk probability model of durability damage over time including a developed time-variant corrosion rate model (after cracking) is established. The proposed methodology is demonstrated by a numerical example of corrosion damage prediction of an existing RC bridge.(5) Anew corrosion fatigue life prediction method for aging RC beam considering the uncertainty in crack growth is proposed. The mechanism of material damage under the coupling effect of corrosion and the cyclic load is very complicated. This study is based on the equivalent initial flaw size method. The proposed model couples the corrosion growth kinetics and fatigue crack growth kinetics together. The developed model is integrated with an asymptotic method to calculate the stress intensity factor for the crack at corrosion pit roots. A phenomenological model is proposed to obtain the stress concentration factor model under different corrosion loss conditions. The fatigue life is predicted by the integration of the fatigue crack growth rate curve from the equivalent initial flaw size to the critical length. In addition, a parametric study is performed to investigate the effect of fatigue degradation of concrete on the fatigue life of RC beams. A inverse first-order reliability analysis method is adopted to consider various sources of uncertainties for the fatigue life prediction. Fatigue life prediction results are validated with experimental observations for various corroded steel bars and beams.