Introduction
Digital twins are computational models of infrastructure and construction assets which help to answer questions like “how does this bridge flex in the wind?”. This project will develop the mathematical theory and computational techniques underpinning future statistical digital twins that will provide seamless fusion of sensor data, computational data and expert knowledge. The applications driving the project include the condition monitoring of structural assets in particular rail, vehicular, and pedestrian bridges.
Explaining the science
Whereas the approach currently adopted in engineering practice uses different, and incoherent, mathematical approaches, where mathematical, numerical, computational and statistical issues are addressed in isolation, this project takes an innovative, comprehensive and integrated view of the quantification and propagation of uncertainty. What is urgently needed is a natural cohesive representation of uncertainty that is coherently propagated and appropriately calibrated. The ambition to unify the characterisation and propagation of uncertainty through all four stages of digital twin development will ensure resolution of these serious issues, as motivated by condition based monitoring and maintenance of engineering structures.
Project aims
The vision for this project is to formalise and comprehensively address major challenges at the frontier of the engineering, mathematical and computational sciences to deliver a complete pipeline of uncertainty propagation, essential for mathematical and statistical models underpinning digital twins of infrastructure in the built environment.
Applications
This project will have rapid and far-reaching influence on design, construction, commissioning and operation of structural assets, in particular rail, vehicular, and pedestrian bridges. Beyond this specific application domain, it will be transformative in many different fields of engineering science and will impact practice widely. Propagation of quantified uncertainty is of immediate importance to all data-driven engineering modeling and prediction within the UK, and internationally, to ensure that risk, analytics and decision-making is informed and guided by sound principled reasoning under uncertainty.