AI‑Driven Modelling and Simulation for Sustainable Transformation of University Campuses

University campuses are complex environments comprising buildings, landscapes, and infrastructure that collectively determine sustainability performance. Many campuses were developed before sustainability principles became central to architecture, planning, and management, resulting in challenges such as inefficient energy systems, aging assets, and limited adaptability to environmental and social change. In line with global sustainability commitments, including the Talloires Declaration and Australasian Campuses Towards Sustainability (ACTS Inc.), universities are increasingly expected to lead by example through innovation in their built environments.

Most current sustainability initiatives remain descriptive rather than predictive or experimentally validated. There is a growing need for AI-driven modelling combined with real simulation to dynamically assess, test, and optimise campus sustainability performance. The Principal Supervisor has already contributed for Deakin Waurn Pond campus through GIS‑based flood modelling (Liu & Li, 2017) and solar-potential visualisation (Li & Liu, 2018). Building on this foundation, this research integrates architectural design, spatial planning, and management strategies with AI, mathematical modelling, and real-time simulation to enable data-driven decision-making.

The project will assess building performance using machine learning models for energy consumption prediction, genetic algorithms to optimise retrofits, and reinforcement learning to simulate occupant behaviour and operational strategies. Predictive models will prioritise retrofits based on lifecycle carbon, cost, and functional performance. At the campus scale, GIS and network modelling will examine morphology, connectivity, mobility, and environmental dynamics, while real simulation scenarios test walkability, passive cooling, stormwater resilience, and biodiversity integration. A digital twin will integrate real-time sensor data for dynamic monitoring and scenario testing.

Complementary quantitative and qualitative research on governance, management, and user behaviour will ensure AI-driven modelling and simulation outputs align with practical institutional strategies. This research contributes to campus planning, design, and operations by providing tools to evaluate spatial and environmental systems, advancing simulation-based design and retrofit methods, and introducing AI-supported operational simulations for maintenance, energy use, and resource allocation.

The proposed research spans three years. In the first half of Year 1, literature review, campus data collection, and identification of sustainability indicators will be conducted. In the second half of Year 1, preliminary AI and mathematical models will be developed, including machine learning algorithms for building performance. In the first half of Year 2, real simulations of individual buildings will be validated using sensor data, while the second half of Year 2 extends simulations to the campus scale and integrates reinforcement learning and optimisation algorithms. In the first half of Year 3, a digital twin and AI-driven decision-support system will be developed alongside qualitative analysis of governance and user behaviour. The second half of Year 3 will focus on optimisation, validation, dissemination of findings, and thesis preparation.

This PhD research will provide valuable insights into the future trajectory of university campuses and their evolution to meet sustainability challenges, particularly for Deakin and IITM. It will also reinforce their position as global leaders in campus sustainability. The project is expected to produce at least five peer-reviewed papers in SCImago Q1 journals, contributing significantly to both academic and practical understanding of sustainable campus development.

Words: 499

References:

Liu, C. & Li, Y. (2017). GIS-based Dynamic Modelling and Analysis of Flash Floods Considering Land-use Planning. International Journal of Geographical Information Science, 31(3), 481–498. https://doi.org/10.1080/13658816.2016.1207774

Li, Y. & Liu, C. (2018). Revenue Assessment and Visualisation of Photovoltaic Projects on Building Envelopes. Journal of Cleaner Production, 182, 177–186. https://doi.org/10.1016/j.jclepro.2018.01.128