vyzn's Reference Model - the technical bridge from BIM to Simulation & Analysis

The challenge

Although Building Information Modeling (BIM) has become increasingly established in construction planning, many of the underlying 3D building models, especially in IFC format, exhibit considerable structural and semantic deficits. Common problems are:

• Missing or inconsistent information on rooms, areas, materials or component types
• Heterogeneous data quality
• Overlapping, duplicate or non-uniquely assignable elements
• Limited semantic depth: The model lacks explicit information on the functional role or logical relationship of the components. For example, it remains unclear whether a component is part of the thermal envelope or how rooms are topologically connected. In most cases, the room definitions are missing entirely.

These structural and semantic deficits make building automated workflows – for example for deriving energy or comfort simulations, CO₂ balances or circularity evaluations – extremely labor-intensive and error-prone. In practice, a direct transition from the 3D building model to simulation & analysis is not possible without an upstream, manually performed data cleaning, completion and structuring.

Our approach

To close this gap, we developed the vyzn Reference Model. Using geometric processing, topological analysis and AI-supported algorithms (#DeepModelProcessing), we extract and structure from the BIM models exactly the information that is actually relevant for simulation-based planning processes. This creates an abstracted, consistent and high-quality building model: the vyzn Reference Model. It forms the basis for all downstream analyses, from thermal simulations through energy efficiency, cost calculations and structural analysis to CO₂ accounting.

Through the automated derivation of relevant building structures, the manual effort for data cleaning is drastically reduced and the quality of planning is significantly improved.

Technical core principles:

• Geometric & topological analysis: From the building model, we extract precise geometric structures such as surfaces, room boundaries, building envelopes and openings. Through topological evaluation, we reconstruct relations such as adjacencies, zone and story affiliation, or affiliation with functional areas (e.g. circulation zones, elevator shafts, thermal envelope). This makes it possible to automatically derive logical units such as buildings, zones or circulation cores.
• AI-supported completion: Missing or inconsistent information, such as component classifications, opening assignments or functional roles, is detected using trained models and plausibly supplemented or corrected. This creates a complete, structured building dataset with high semantic depth.
• Semantic model enrichment: The reference model is uniformly typed, logically named and supplemented with derived attributes, for example on thermal relevance, eBKP-H classifications, room function or zonal assignment. It thus offers an ideal basis for automated analyses, simulation-based optimization and AI-based predictive models.

The vyzn Reference Model - the key to the next level of digital building planning

The vyzn Reference Model paves the way for a new quality in digital building planning: structured, complete and simulation-ready models: automatically created through Deep Model Processing. Instead of labor-intensive data cleaning by hand, we enable a robust and traceable basis for analyses, evaluations and optimizations.

From the 3D model to a well-founded decision-making basis: in minutes instead of weeks.

Through targeted geometric, topological and semantic processing, an error-prone, incomplete BIM model becomes an intelligent digital twin – machine-readable, simulation-ready and ideal as a starting point for further digital workflows.

With the vyzn Reference Model, we create the technological bridge between construction planning and data-based decision-making: efficient, automated and scalable.