AI-Assisted Clash Management for Sustainable Healthcare Construction: A Case Study Using Revit, Autodesk Construction Cloud, and Navisworks

Clash detection using Building Information Modelling (BIM) is now a key part of design coordination in complex construction projects. Finding clashes early reduces rework, which saves materials, cost, and embodied carbon. However, the workflow is still fragmented. Models are created in one platform, shared in another, and checked in a third. Clash reports must then be reviewed manually, which takes a long time and often gives inconsistent results, affecting both project performance and sustainability.

This study examines these gaps and tests whether a general-purpose large language model (LLM) can help classify and prioritise clashes without any programming skills or model training. A qualitative single-case study design is used, based on a healthcare facility project (mental health unit) with dense building services and strict spatial and safety requirements.

Discipline models were created in Revit, shared and coordinated through Autodesk Construction Cloud (ACC), and checked in Navisworks Clash Detective. The exported clash reports were then classified by an LLM into relevance and severity categories. The workflow was assessed in three ways: consistency of clash results between platforms, agreement between the LLM and manual engineering review, and time saved in the review process.

Early findings show that cloud-based and desktop clash detection give overlapping but different results, and that LLM-assisted classification may greatly reduce manual review effort. Gaps remain in clash data standards, severity definitions, and workflow integration.

The study proposes a simple three-step workflow: run consistent clash detection across platforms, add accessible AI-assisted classification to the review process, and track decisions through clash-to-issue resolution in ACC. This gives practitioners practical guidance to improve coordination, cut rework-related waste, and deliver healthcare projects more sustainably.