Managing, handling and modelling uncertainty in structural engineering codes
September 30, 2014

Observations made during the workshop

Uncertainty modelling plays a crucial role in the development and evaluation of structural engineering codes. The current structural Eurocodes apply a so-called semi-probabilistic approach with calibrated partial factors which are based on structural reliability methods of Level II (most often using First-Order Reliability Methods, so called FORM analyses, or a simplified approach) and a set of underlying probabilistic models for basic variables.

The distillation of consistent, accurate and practically applicable structural engineering codes is far from trivial and needs careful consideration. In this regard, during this workshop a group of 5 international speakers were invited to give a lecture in order to clarify the basic ideas behind the uncertainty modelling in current structural engineering codes and their applicability in practice. As a result of the discussions raised at the end of the workshop a set of observations is made here, which clarify potential future research perspectives to be undertaken.

Based on the observation that different target safety levels are suggested in different European and international standards, it was suggested that the ambiguity related to the observed scatter in the proposals should be investigated in more detail as well as the meaning of calculated safety levels in relation to occurrence frequencies of structural failures. Also the incorporation of societal preferences as well as knowledge with respect to acceptable risks considered in other research domains such as nuclear industry, airplane industry, etc. for the determination of the target safety level plays an important role and more research efforts should be directed towards an interdisciplinary approach in this regard.

With respect to a semi-probabilistic safety format based on the use of partial factors, it was observed that still further research efforts are needed to come up with coherent proposals for several structural engineering problems such as (i) the partial factor based design and assessment with respect to fatigue, (2) the assessment of ageing structures with or without deterioration and (3) the design and assessment of offshore and nearshore engineering structures, especially with respect to the partial factor based treatment of the loads. Also the effect of hidden safeties is to be investigated in more detail.

Further, the extrapolation of the current Eurocode methodology based on a semi-probabilistic partial factor approach towards mechanical engineering problem formulations is not trivial and needs further research and development due to differences such as the type of loading to be considered, the treatment of large series in production processes, etc.  It was however observed that the general theory of structural reliability based design still seems applicable and there is a potential to use reliability-based semi-probabilistic design formats in other disciplines such as mechanical engineering.

Finally, also the balance and duality between increasing complexity and conservatism in design guidelines was discussed. More research efforts should be directed to the determination of application boundaries and associated accuracies with respect to the selection of the level of detailing of calculations and investigations for design or assessment. In this regard, also the awareness should be increased with respect to the fact that codified design is sometimes limiting advances in analysis approaches for engineering and it is important to recognize the importance of expanding the knowledge of future engineers with respect to full-probabilistic approaches through educational efforts. With respect to the latter we however have to cope with the fact that a strong background on probability theory and statistics is a prerequisite.

 

Prof. Dr. D. De Buyst
Prof. Dr. R. Caspeele
Prof. Dr. L. Taerwe
Prof. Dr. J. Van Dyck