Integrating AI Agents into BIM for Architectural Design Support: A Torsional Eccentricity Assessment Case
5th International Graduate Research Symposium | IGRS’26 , İstanbul, Türkiye, 11 - 13 Mayıs 2026, ss.1148-1152, (Özet Bildiri)
- Yayın Türü: Bildiri / Özet Bildiri
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Sayfa Sayıları: ss.1148-1152
- Van Yüzüncü Yıl Üniversitesi Adresli: Evet
Özet
The role of human-machine interaction in the design process was defined as “partnership” by Negroponte (1970); Bernstein (2025) states that this process has evolved beyond representing design with artificial intelligence to a stage capable of digital reasoning. Currently, Large Language Model (LLM)-based AI agent technology is advancing human-machine interaction beyond the level of mere partnership. Building automated compliance checking (ACC) processes are evolving from static rule-based automation to LLM-based agent architectures that can autonomously interpret regulations and independently query Building Information Modeling (BIM) data (Chen et al., 2024; Ying and Sacks, 2024). Code control systems are most effective when integrated into the design development process, ensuring that errors are detected and corrected during the design phase rather than after the design is complete (Eastman et al., 2009). Türkiye's location in a region with high seismic risk necessitates the consideration of earthquake effects in the early stages of design. The 2018 Turkish Building Seismic Code (TBDY 2018) defines the eccentricity criterion (e ≤ L/5) to limit plan irregularities. However, although this criterion is directly related to architectural plan organization, it is mostly evaluated during the engineering analysis phase. Architects do not have tools that allow them to directly predict the effects of plan decisions on seismic behavior in the early design phase. Behaviors related to plan organization, such as torsion, are usually identified during the structural analysis phase; this leads to revisions and interdisciplinary coordination difficulties in later stages.