Intelligent digital twins for assessing and predicting bridge road traffic demands


This project develops digital twin bridge models which predict the current and future compliance and maintenance condition of bridges in response to realistic simulated traffic loads and real-world sensing data, using a probabilistic modelling approach.

Road bridges are a vital part of transportation networks, forming crucial links in natural bottleneck locations and enabling the continual flow of people and goods into, out of, and across cities. However, the analysis used for design and maintenance planning of this community-critical infrastructure is typically carried out using static models and assuming generalized traffic patterns. This analysis represents only peak loading scenarios and does not reflect the spatial and temporal variations in real-world traffic loads. The resulting uncertainty in load prediction can lead to in overengineering in bridge design as well as sub-optimal maintenance planning. Furthermore, as current analysis techniques model only maximal loads, they cannot be used to predict the maintenance condition of bridges due to fatigue from repeated loading and unloading of the bridge over time.

This research addresses these limitations by developing intelligent digital twins which can simulate the response of a bridge to realistic traffic loading scenarios. These digital twin models combine two primary elements: (i) a traffic simulation model which exploits detailed traffic count and weigh-in-motion data to generate time-dependent traffic loadings, and (ii) a detailed structural model which predicts the compliance and maintenance condition of a bridge for different maximal and cyclic loading patterns. The intelligent digital twin is intended to be generalizable to any bridge or network of bridges for which relevant data exists. This will enable these models to be used within an integrated approach to study infrastructure vulnerability and multi-hazard risk management.

Principal investigators Tim Hillel, Mathew Sjaarda, Eloise Hautecoeur
Sponsor EPFL ENAC Research Clusters
Period 2020-2022
Laboratories TRANSP-OR, RESSLab, CNPA