Many legacy engineering structures in the UK, such as centuries-old bridges, are at significant risk from environmental threats. In recent years, multiple such structures have failed due to catastrophic flood events, with damage likely to other structures even under less severe conditions. Therefore understanding and modelling these risks effectively is essential to ensuring the resilience of UK infrastructure and ultimately national prosperity.
Explaining the science: Modelling environmental interaction
Models are needed to understand the interaction between engineering structures and environmental hazards. In the case of historic and ageing structures, this modelling effort is challenging for multiple reasons. One is that it is difficult to gain information about the original design, construction, and intended lifespan of structures that can be many centuries old. Another is that the degenerated state of the structure is difficult to determine. Lastly, the two-way interaction between structures and the natural environment is typically not described by systematic, statistically robust data.
These issues result in an approximate and assumptive modelling process, reliant on judgement and expertise, leading to a high amount of uncertainty in any interaction and risk model produced. Current approaches attempting to quantify this uncertainty are often ‘ad-hoc’ and outdated by modern statistical standards.
This project is addressing the urgent need to develop rigorous statistical protocols for the modelling of legacy engineering structures and their interaction with environmental hazards. Such models need to involve robust uncertainty quantification for the parameters and inputs involved.
The models produced by this work will allow for improved decision-making in the management of ageing structures and promote the future resilience of these natural and man-made systems.
The work is being driven by collaborations between engineers, earth scientists, and statisticians, across the data-centric engineering programme at the Turing.