HVAC System Topology-Spatial Graphs from Real BIM Models

An anonymized topology-spatial graph dataset for HVAC systems in BIM-based operation and maintenance. It contains public local graph fragments extracted from real BIM HVAC models, preserving component categories, MEP system labels, local spatial attributes, and explicit connection edges. The dataset was created by pinhaohuang7 and was last updated on 2026-05-26.

Use Cases

• Train graph neural networks for HVAC component classification based on preserved component categories and MEP system labels.
• Develop models for predicting system connectivity or fault propagation based on the explicit connection edges between components.
• Analyze spatial relationships within mechanical systems for layout optimization using the local spatial attributes.
• Benchmark algorithms for automated information extraction and knowledge graph construction from BIM models.

Strengths

• Data is derived from real Building Information Modeling (BIM) HVAC models, suggesting practical relevance.
• Graphs preserve multiple structural and semantic features including component categories, MEP labels, spatial attributes, and connection edges.
• Dataset is anonymized, which may help with privacy concerns for real-world building data.

Limitations

• Description metadata is limited; actual data quality, scale, and completeness require manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and dataset size are unknown, which may limit suitability assessment for large-scale model training.

Provenance

Source

Real BIM HVAC models, anonymized and processed into graph fragments.

Collection Method

Extracted from real BIM models, likely through an automated or semi-automated parsing process.

Freshness

Last updated 2026-05-26 06:14:45; freshness should be verified.