| Title |
A Study on the Development of an Embodied Carbon Estimation Model through IFC Automatic Generation and λ-Correction Based on Building Input Information |
| Authors |
강찬혁(Kang, Chan-Hyeok) ; 정보경(Jung, Bokyung) ; 이태규(Lee, Tae-Kyu) ; 권영철(Kwon, Young-Cheol) ; 최창호(Choi, Chang-Ho) |
| DOI |
https://doi.org/10.5659/JAIK.2026.42.3.251 |
| Keywords |
Building information modeling; Industry foundation classes; Embodied carbon; Life cycle assessment; Building attributes; Cohort approach; Machine learning |
| Abstract |
This study presents a framework for estimating embodied carbon in buildings by integrating automated IFC generation with A-correction. Six
basic inputs were used to generate IFC models: gross floor area, number of floors, floor height, structural type, year of completion, and
building use. Preset values for material thickness, density, and surcharge rates were applied, and Environmental Product Declaration data were
linked to calculate baseline emissions. The baseline IFC results captured only 5 to 20 percent of actual values, indicating systematic
underestimation. To improve accuracy, three correction approaches were tested. The first was a Scaled method using a global X factor. The
second was a Cohort-based approach using K-nearest neighbors. The third applied machine learning through LightGBM regression. The
analysis was conducted on 304 buildings, with 260 used for training and 44 for testing. The Scaled method showed poor performance, with
R equals 0.084 and MAPE equals 69.0 percent. The Cohort-based method demonstrated limited stability, with R equals 0.454 and MAPE
equals 113.1 percent. In contrast, the machine learning model achieved the highest accuracy, with R greater than or equal to 0.803 and
MAPE equals 23.3 percent, while maintaining stable predictions across different building types and scales. These results demonstrate that
embodied carbon can be reasonably estimated for existing buildings using minimal input data and without design drawings. Combining BIM
IFC automation with AI-based correction significantly improves the reliability of life cycle assessment and supports carbon-neutral decision
making. |