Considerations on the reliability of service life predictions using a probabilistic approach

Abstract
Designing for durability of reinforced concrete structures using a probabilistic approach is becoming increasingly popular among consultants. This paper will address a number of pitfalls in probabilistic calculations by identifying the most critical model parameters and demonstrating the influence of statistical distribution of these parameters on the outcome. Especially, the so-called age factor and critical chloride content play a dominant role in service life predictions. A procedure will be proposed to quantify these parameters from existing structures to achieve an improved prediction of residual service life. In addition, it will be demonstrated that probabilistic service life modelling may be misused by manipulation of the input values, shifting the risks for premature maintenance to the owner.