SABER-BIM: A Component-Level Adaptive Lightweighting Framework for Digital Twin BIM Models.

Area of Science:

• Digital Engineering
• Computer-Aided Design (CAD)
• Geometric Modeling

Background:

• Lightweighting Building Information Modeling (BIM) models for digital-twin applications is challenging due to the heterogeneity of components and the need to balance geometric reduction with engineering tolerances.
• Existing simplification methods often use uniform ratios or heuristics, failing to adapt to varying component complexity and detail.
• Learning-based approaches can be adaptive but struggle with enforcing and auditing engineering constraints.

Purpose of the Study:

• To introduce SABER-BIM (Semantic-Geometric Co-driven Adaptive Budget Estimation and Reduction for BIM), a novel approach for lightweighting BIM models.
• To formulate model simplification as a component-level face-budget allocation problem that respects semantic and geometric properties.
• To develop an auditable and deployable pipeline for adaptive BIM model simplification.

Main Methods:

• SABER-BIM predicts target face counts for individual BIM components based on Industry Foundation Classes (IFC) types and geometric descriptors.
• A global scaling mechanism ensures adherence to a user-specified overall face budget.
• Component budgets are executed using a robust geometric backend, such as Quadric Error Metrics (QEM), supported by an offline pseudo-ground-truth procedure for constraint satisfaction.

Main Results:

• SABER-BIM demonstrates more effective budget allocation compared to existing methods under identical global constraints.
• The method improves stability in controlling geometric error.
• Engineering usability is enhanced through semantically aware tolerance and mesh-validity constraints.

Conclusions:

• SABER-BIM provides an auditable and deployable solution for lightweighting BIM models, crucial for digital-twin applications.
• The semantic-geometric co-driven approach enables adaptive simplification that respects engineering requirements.
• This method advances the state-of-the-art in BIM model optimization for efficient digital engineering workflows.