ABSTRACT
Remaining fatigue life of road bridge deck which is subjected to heavy traffic is estimated by data assimilation procedure. Crack pattern which is visually inspected on site is fed as internal damage data in multi-scale analysis. Because crack data is not precise, compatibility of deformation is not satisfied in the initial step. Then, most possible damage field is generated based on laws of mechanics through several steps of fatigue load cycles. As a result, remaining fatigue life is numerically estimated from non-perfect data. This data assimilation procedure is verified with wheel-running test of reinforced concrete (RC) slab specimen. Concrete cracking patterns on the lower face is firstly used as concrete damage data. Then, it is found that other types of damage data such as compressive fracture is able to be used in data assimilation. Three dimensional data of elastic wave velocity analyzed by acoustic emission tomography is tried to estimate the degree of compressive fracture in this study. Although cracking is not explicitly taken into account, the small number of load repetition automatically generates internal cracks over the volume of analysis domains, and the remaining life of the slabs is successfully estimated. Multi scale analysis enables to estimate fatigue life, however, it takes computational time because fatigue loading is simulated directly in cyber space. In order to achieve a quick deterioration-magnitude assessment of RC decks, two evaluation methods are proposed. First is a predictive correlation between the remaining fatigue life and the cracks density. Another is an artificial neural network (ANN) model. Both assessment methods are built commonly by numerical results of multi scale analysis with thousands of artificial random crack patterns to cover all possible ranges since the variety of the real crack patterns on site is more or less limited.
