Project Details
Duration: 24 months (2024-2026)
Funding: Zukunft Bau Federal Research Initiative

This interdisciplinary project develops digital methods to integrate automatic architectural floorplan generation with spatial timber load-bearing systems for urban rooftop extensions. By combining evolutionary algorithms and machine learning with structural engineering principles, the research aims to create resource-efficient solutions that maintain architectural quality in dense urban environments.

Key Aspects:

Spatial Structural System

• Uses 3D arrangement of wall-like timber elements as load-bearing system
• Integrates necessary partition walls into structural concept
• Reduces material-intensive distribution layers common in modular construction

AI-Assisted Workflow and Procedural Generation

• Employs machine learning trained on Neufert 4.0 floorplan database and seeks to develop it further to tackle the exact apartment mix distribution in the project
• Develops "construction grammar" ruleset for wall placement
• Evaluates the resulting designs to provide qualitative floorplans

Funding

Supported by the German Federal Ministry for Housing, Urban Development and Building through the Zukunft Bau program.

Project Team
Prof. Dr.-Ing. Matthias Beckh (TU Dresden)
M.A. Architekt Patrick Schäferling (TU Dresden)
Prof. Dr.-Ing. Reinhard König (BUW)
Dr. Martin Bielik (BUW)
M.Sc. Luyang Zhang (BUW)
M.Sc. Architektin Iuliia Osintseva (BUW)

Partners:

• Technical University Dresden (Lead)
• Bauhaus University Weimar
• GWG München (Housing Association)