Damage diagnosis of existing reinforced concrete structures

Many typical defects existing in some of reinforced concrete structural elements such as honeycomb, crack, scaling and strength deduction for concrete, and corrosion caused decreased section area for steel members. These defects are often caused by improper construction management and maintenance, overloading, environmental impact, disaster, fatigue and so forth. The development of defects certainly weaken the structures and reduce the expected life time of structures. Consequently, diagnosis and repair in time for the structures in order to provide the safety for the people is the most important task of our civil engineers. The purpose of this study is try to establish a feasible and efficient diagnosing model for reinforced concrete structures by using of displacement time history of the existing structures and back-propagation neural network technique to assess the severity and location of defects. This paper present the theoretical analysis of a simply-supported reinforced concrete beam in specified size (i.e., rectangular cross section and 4 meter span) with assumed defects by a finite-element program is applied to generate training and testing examples which are needed for neural network assessing task. Examples are generated according to the displacement time history of the defected beams due to a dynamic force at the center of the beam. The results of the damage classification from testing examples show this model is extremely sensitive in diagnosing damage processes in existing reinforced concrete structures.