Building disaster-ready underground infrastructure with AI digital twins

What’s the problem? 

Urbanised coastal cities have high dependence on underground infrastructure – the critical tunnels and pipelines that carry water, power and transport. But in regions exposed to earthquakes and floods, such as western India, these hidden assets face multiple and compounding hazards. Damage is hard to detect and track below ground, and failures can disrupt essential services for the entire community. 

What’s UQ’s innovation? 

A UQ-India team has developed an AI-enabled digital twin of underground infrastructure – a detailed geometric virtual replica of a real tunnel that engineers could utilise to inspect, evaluate and track deformation and damage over time by a non-intrusive approach. The twin is built from mobile LiDAR scans, which use lasers (Near-Infrared Light) to record the exact 3D shape and surface condition of the structure. AI turns those scans into a physics-based model accurate enough to analyse the current state of the structure and how it would behave under future hazards probabilistically. The geospatial data anchors it in the real-world condition, alongside the hazards it faces such as nearby faults or flood-prone ground. Automation allows this to be done consistently and at scale, while conventional, manual processing of these data consumes immense amounts of time and skilled manpower. The validated approach equips the decision-making authorities to evaluate the asset against multiple hazards by computationally modelling the structure with earthquakes, floods or their compounding effects. To produce risk-informed intelligence, it guides tunnel monitoring and maintenance decisions transparently. 

What’s the impact? 

The central innovation and development from this project is the Resilience Dashboard, enabled by Digital Twin, that gives decision-makers early risk insights about the tunnel asset. It aims to support the predictive maintenance, strategic upgrades and emergency planning for coastal tunnels and pipelines. By pairing local geotechnical information about the site with advanced tools, the engineers can design safer underground assets and minimise service disruptions, improving the resilience of water, power and transport infrastructure to natural hazards. The risk information and asset evaluation data from the Resilience Dashboard gives all stakeholders and owners – government and urban planners –  objective recommendations to inform decisions. It includes: which asset needs urgent attention, where to invest, which assets to upgrade first, and how to set standards. Focused initially on earthquake- and flood-prone zones in western India, the framework is designed to scale to vulnerable coastal regions worldwide. 

Led by

• Abhinav Dewangan 
  PhD Researcher, School of Civil Engineering  

UQ PROJECT TEAM 

• Dr Jurij Karlovšek
  School of Civil Engineering 

Published 15 September 2026

99 more AI innovations at UQ