The research aims to identify and integrate the fragmented and heterogeneous data needed for road slope design by Industry Foundation Classes (IFC) extension to improve interoperability and to formalize and automate the overall process of road slope design. A prototype has been developed for implementation. The research scope has been further extended to design optimization with the support of an improved data exchange process.

- 

Developing a formal schedule evaluation methodology for resource constrained hard rock tunnel projects.

- 

Developing Q-learning-based schedule optimization methodology for hard rock tunnel projects.

• 

A methodology is proposed to assess the energy savings of dynamic passenger adaptive thermal control strategies in subway stations.

• 

The methodology integrates building energy modeling and passenger flow simulation.

• 

A prototype is developed to implement the methodology using AnyLogic and EnergyPlus.

• 

The methodology is applied to a Hong Kong subway station.

This research proposes a BIM-based railway track facility management framework. The first step is to integrate and visualize the multi-source information into IFC-based openBIM environment. Then the deterioration process and inter-relationship of track defects will be determined by machine learning algorithms. Finally, the facility management plans will be optimized.

With the growth of deep learning, graph convolutional networks (GCNs) excel in skeleton-based action recognition. However, GCN models often rely on body connections, not ideal for complex tasks. This study introduces a topological graph approach based on body segments. A multiple-input streams attention (MISA) network enhances GCNs with comprehensive input graphs, integrating motion data and attention blocks. Experiments on the Construction Motion Library (CML) dataset achieved approximately 84.94% accuracy, showcasing the method's superiority.

City-scale energy consumption simulations usually use the LOD2 format. However, the model contains little information and the accuracy of the calculation results is questionable. This study proposes a LOD2ES model based on IFC and also provides a top-down modeling method, which allows LOD2ES to retain a large amount of information while reducing modeling time and cost.

Despite the rapid emergence of BIM-based VR applications, no evaluation framework specialized for the technologies exists in the construction industry. After extensively reviewing existing studies and interviewing experts, the research team proposed an evaluation framework for BIM-based VR applications, which consists of 3 stages, 5 areas, 14 criteria, and 29 metrics and focuses on the design phase of the projects. To assess the usefulness of the framework, the team applied it to five BIM-based VR applications using a BIM-based design project in Hong Kong.

· 

Understand the BIM market in Hong Kong with a baseline of BIM adoption

· 

Identify key hurdles for BIM adoption

· 

Recommend strategies and actions for CIC and the industry to advance BIM implementation

· 

stablish a benchmarking methodology for Hong Kong’s BIM adoption in the futureconsistent results for comprehensive evaluation criteria and metrics in a quantitative and flexible manner