ABSTRACT
ABSTRACT A large portion of existing hydraulic structures in Germany have an age of more than 70 years and about 25% of them are already past their intended design working life of 100 years. An efficient life-cycle management is necessary to manage these large concrete structures safely and economically. Inspections and material testing form an important part of this process, as they enable an improved assessment of the condition and properties of the structure and its material. Traditionally, the limited data available from these tests is used to fit probability distributions of material parameters; characteristic values for the design are then obtained from the fitted distributions. Spatial correlation between the measurement locations or different material layers are typically neglected. In this contribution, we propose to model the spatial variability of the material parameters with random fields. We use the available data from local measurements to update the distribution of the random fields with Bayesian analysis. We then calculate the structural reliability by application of subset simulation, an adaptive sampling approach. We show that through the employed random field modeling approach, a more detailed statement about the condition of a structure can be made by using not only the measurements of the material properties, but also the information about their spatial location. The results show that consistent modeling of the spatial variability of the concrete materials can potentially increase the reliability estimate of large hydraulic structures when measurement information is included.
