| Download | - View final version: AI-enhanced beam-column elements for efficient system-level fire analysis (PDF, 983 KiB)
|
|---|
| DOI | Resolve DOI: https://doi.org/10.4224/40004123 |
|---|
| Author | Search for: Ji, Wei1; Search for: Cashell, Katherine A.2; Search for: Li, Guo-Qiang3; Search for: Usmani, Asif1; Search for: Jiang, Liming1 |
|---|
| Affiliation | - The Hong Kong Polytechnic University
- University College London
- Tongji University
|
|---|
| Format | Text, Article |
|---|
| Conference | SiF 2026, 14th International Conference on Structures in Fire, May 18-21, 2026, Kingston, Ontario, Canada |
|---|
| Subject | fire safety analysis; hybrid simulation; artificial intelligence; numerical simulation |
|---|
| Abstract | Accurately predicting the behaviour of structural systems in fire remains challenging, costly, and time-consuming for complex buildings. Computational simulation has become the primary tool for investigation, while physical experiments are mainly conducted for validation purposes. However, system-level numerical analysis remains difficult as it is hard to simultaneously capture both localised component behaviour and global load redistribution within the structural system. This paper proposes an AI-enhanced beam-column element for structural fire analysis. The proposed element integrates a trained AI model into the conventional beam-column formulation to enhance the accuracy of nonlinear structural response prediction under fire conditions. The AI model is implemented as a dedicated element in OpenSees and can be utilised in the same manner as a standard beam-column element without requiring additional modelling procedures. This element enables efficient modelling of the nonlinear behaviour of fire-affected members without resorting to computationally intensive shell or solid elements. By representing fire-affected members using AI-enhanced elements and incorporating them into a system-level finite element model consistent with the building prototype, the proposed framework becomes adaptable to diverse building configurations and fire scenarios. |
|---|
| Date published | 2026 |
|---|
| Publisher | National Research Council Canada |
|---|
| Copyright statement | |
|---|
| In | |
|---|
| Language | English |
|---|
| Peer reviewed | Yes |
|---|
| Export citation | Export as RIS |
|---|
| Report a correction | Report a correction (opens in a new tab) |
|---|
| Record identifier | 7d576c27-73cb-40d9-a57f-4efe340e7c09 |
|---|
| Record created | 2026-11-03 |
|---|
| Record modified | 2026-11-13 |
|---|