Probabilistic property associated with debonding of CFRP-bonded steel plate under bending load is theoretically discussed by the use of a probabilistic model based upon a spatially random differential equation. Further, a statistical experiment for debonding of CFRP-bonded steel plate under bending load is performed to verify the theoretically estimated probabilistic property of debonding. First, a system of statically equilibrium equations is formulated to describe axial and shear forces of a steel plate bonded by a CFRP plate. Next, it is extended to a probabilistic model consisting of a system of spatially random differential equations where spatially random variation of the adhesive thickness is introduced as a random field, whose solution leads to a spatially random variation of a principal stress of the adhesive. A Monte Carlo simulation technique is then applied to estimate the probability of debonding under bending load based upon the proposed probabilistic model, where the debonding of adhesive is supposed to occur when the maximum principal stress exceeds a critical value. Further, statistical experiment is also performed to clarify the probability that the debonding of CFRP-bonded steel plates occurs under bending load. Finally, the experimental result is compared with the probability of debonding estimated through the Monte Carlo simulation. The result shows that the probability of debonding obtained through the statistical experiment can be quite well reproduced by the Monte Carlo result based upon the proposed probabilistic model.
