In the assessment of existing structures, it is uncommon to consider a track record of the structural performance of the structure itself or similar structures. However, the structure's proven strength in service could play a significant role, along with the performance of similar structures in the population. Because the population track record does not apply in the design of new structures, it is not encountered in design standards. An assessment that does not incorporate the track record may conclude insufficient structural reliability whilst, in reality, the reliability is satisfactory. In the suggested approach, information obtained from laboratory experiments is combined with the track record in a Bayesian way to assess a structure's reliability. As a case study for this article, the reliability of the connection strength between wide slab floor elements is considered. Although laboratory tests indicate poor connection strength, the track record indicates just one failure and many well-performing floors. It is found that considering the time-dependent nature of structural reliability is vital for understanding how proven strength develops from the completion of the structure to its usage today. The number of similar objects in the population that show satisfactory performance is varied and is shown to have a significant effect when its number grows. The presented method and case study show that reliability assessments incorporating a track record enable more accurate structural reliability predictions for existing structures.

In the assessment of existing structures, such as bridges and viaducts, reliability requirements are used to decide if a structure is sufficiently safe, even when subjected to degradation. The reliability requirements may be expressed in different ways but should ultimately result in similar reliability performance of the structure. Most of the current assessment rules follow from a maximum allowable failure probability, depending on the reliability class, within a fixed period of time (the reference period: e.g. 15, 30, 50 or 100 years). A reliability requirement expressed using a fixed reference period fits the design of new (to be built) structures, but it is problematic for existing or temporary structures. In the case of existing structures, when for instance the design life has passed, the assessment should be flexible with respect to the expected or desired remaining life. Other factors, such as deterioration or changing loads, may also call for an assessment with a smaller time period since the failure rate could increase with time which makes a reliability requirement for a longer period less useful. Using fixed reference periods therefore could lead to suboptimal solutions where the economy and environment are unnecessarily hurt. Reliability requirements formulated on an annual basis provide a solution to this problem. In this article, reliability analyses of several typical bridges are performed to quantify the requirements in such a way that assessments based on annual reliability result in performance similar to the current practice. Special attention is paid to minimising and highlighting the cases where the annual requirements may lead to a trend that breaks with the current reliability requirements.

In the assessment of existing infrastructure performing only a desk study is often not sufficient to determine the structural reliability of a bridge or viaduct. For concrete structures gathering field data by performing a proof load test offers detailed information about the structural performance. However, the relation between the magnitude of the load and the structural reliability is not immediately clear. In the present study the challenges in determining the target load and the uncertainties that require attention are described. An approach is presented that addresses the time-dependent character of the structural reliability, the need for accurate stop-criteria, the knowledge level and spatial uncertainty. It is shown how both past trafic loads and a proof load test may contribute to the proven strength of a structure. The described methodology provides a starting point towards a flexible approach for proof load testing in which structure-speciic information and requirements are considered.

While reliability methods have already been widely adopted in civil engineering, the efficiency and robustness of finite element-based reliability assessments of quay walls are still fairly low. In this paper, the reliability indices of structural and geotechnical failure modes of two real-life quay walls are determined by coupling probabilistic methods with finite element models, taking into account a large number of stochastic variables. The reliability indices found are within the range of the targets suggested in the design codes presently in use. Nevertheless, neglecting model uncertainty and correlations between stochastic variables leads to an underestimation of the probability of failure. In addition, low sensitivity factors are found for time-independent variables, such as material properties and model uncertainty. Furthermore, the results are used to reflect on the partial factors used in the original design. Important variables, such as the angle of internal friction, are subjected to a sensitivity analysis in order to illuminate their influence on the reliability index. Port authorities and terminal operators might be able to use the findings of this paper to derive more insight into the reliability of their structures and to optimise their service life and functionality, for example by deepening berths or increasing operational loads.