An AI-Enabled Multiscale Computational Platform for Evaluating Fire Performance of Flame-Retardant Mass Timber Compartments

The global construction sector faces an urgent challenge: reducing its carbon footprint to combat climate change. Engineered wood products (EWPs), such as cross-laminated timber, offer a sustainable alternative to concrete and steel by storing carbon and enabling circular economy practices. However, their widespread adoption in tall buildings is hindered by one critical issue - fire safety. Timber is combustible, and current fire protection strategies rely heavily on costly, destructive full-scale tests, which limit innovation and confidence in design. This project proposes a transformative solution: a high-fidelity multiscale fire modelling and risk assessment framework that integrates advanced computational techniques with experimental validation. At the material level,Reactive Molecular Dynamics (MD) simulations will reveal how timber and flame-retardant coatings behave under extreme heat, uncovering the chemical pathways that govern charring, gas emissions, and self-extinguishing potential. These insights will be used to extract accurate thermal degradation kinetics and material properties. At the structural level, these data will feed into Computational Fluid Dynamics (CFD) models to simulate realistic compartment fires in mass timber buildings. This approach will capture complex interactions between heat, smoke, and structural elements under various ventilation and loading conditions—critical for predicting flashover and burnout scenarios. By coupling MD and CFD, the framework will bridge the gap between microscopic chemistry and large-scale fire dynamics, enabling unprecedented predictive accuracy. Finally, the project will leverage machine learning (ML) to develop a rapid fire risk assessment tool. This tool, designed as a Building Information Modelling (BIM)plug-in, will allow architects and engineers to instantly evaluate fire performance indicators—such as temperature, smoke layer height, and radiation—during the design phase. This capability will empower performance-based fire safety design and accelerate the safe adoption of timber construction in Hong Kong and beyond.By combining cutting-edge simulation, experimental validation, and AI-driven design tools, this research will deliver a robust, scalable platform for fire safety engineering.The outcome will not only enhance confidence in sustainable timber construction but also contribute to global efforts in decarbonising the built environment.