Comparative Analysis of Generative AI-Augmented Machine Learning Models for Embodied CO₂ Prediction in Multi-Story Structural Systems

Comparative Analysis of Generative AI-Augmented Machine Learning Models for Embodied CO₂ Prediction in Multi-Story Structural Systems

Danish AHMED, Muhammad AJMAL, Hend Walid A ALOTAIBI, Tahar AYADAT, Saidur Rahman CHOWDHURY, Andi ASIZ, Sani I. ABBA

Abstract. The building sector contributes significantly to global carbon emissions, particularly through embodied carbon associated with structural materials such as concrete, steel, and timber. This study presents an integrated structural artificial intelligence framework for predicting embodied CO₂ emissions in multi-story buildings. Three structural systems, Cross-Laminated Timber (CLT), steel-concrete, and hybrid CLT-steel-concrete, were modeled at 5, 10, 25, and 50 stories using ETABS and verified in accordance with the Saudi Building Code. Material quantities were extracted to calculate embodied CO₂ emissions based on established emission factors and carbon storage effects of CLT. Due to the limited size of the original dataset (12 instances), three generative artificial intelligence techniques, Large Language Model (LLM), Gaussian Mixture Model (GMM), and Kernel Density Estimation (KDE), were employed to expand the dataset to 100 samples. Five regression models were trained and tested using a 70/30 split. Results indicate that LLM and GMM preserved the statistical structure of the original data and achieved near-perfect predictive performance (R² = 0.999), while KDE showed comparatively lower generalization accuracy. The findings demonstrate that generative AI can effectively overcome small-data limitations and support reliable embodied carbon prediction in sustainable structural design.

Keywords
Embodied CO₂ emissions, Cross-Laminated Timber, Steel-Concrete Structures, Generative Artificial Intelligence, Gaussian Mixture Model, Kernel Density Estimation, Large Language Model, Machine Learning

Published online 10/5/2026, 11 pages
Copyright © 2026 by the author(s)
Published under license by Materials Research Forum LLC., Millersville PA, USA

Citation: Danish AHMED, Muhammad AJMAL, Hend Walid A ALOTAIBI, Tahar AYADAT, Saidur Rahman CHOWDHURY, Andi ASIZ, Sani I. ABBA, Comparative Analysis of Generative AI-Augmented Machine Learning Models for Embodied CO₂ Prediction in Multi-Story Structural Systems, Materials Research Proceedings, Vol. 72, pp 156-165, 2026

DOI: https://doi.org/10.21741/9781644904336-14

The article was published as article 14 of the book Smart and Sustainable Architecture and Urban Resilience

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 license. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

References
[1] Hart, J., D’Amico, B., & Pomponi, F. (2021). Whole-life embodied carbon in multistory buildings: Steel, concrete and timber structures. Journal of Industrial Ecology, 25(2), 403-418. https://doi.org/10.1111/JIEC.13139
[2] Risha, A., Bezabeh, M. A., Rogers, C., Feng, H., & Salenikovich, A. (2025). Life cycle carbon assessment of reinforced concrete, structural steel, and mass-timber buildings in Canada. Canadian Journal of Civil Engineering, 52(3). https://doi.org/10.1139/cjce-2025-0107
[3] Becker, I., Anderson, F., & Phillips, A. R. (2023). Structural design of hybrid steel-timber buildings for lower production stage embodied carbon emissions. Journal of Building Engineering, 76, 107053. https://doi.org/10.1016/j.jobe.2023.107053
[4] M. Ajmal, R. Ismaeel, M. Ismail, D. Ahmed and T. Ayadat, “Seismic Assessment of Reinforced Concrete Building Strengthened with CFRP Sheets using Non-Linear Static Approach, Materials Research Proceedings, Vol. 48, pp 123-130, 2025 https://doi.org/10.21741/9781644903414-14
[5] N. Mohammed, S. Abdulla, D. Ahmed, M. Ajmal, A. Asiz, “Analyzing the differences between the impacts of Wind and Earthquakes on base shear and drift in diagrid structures”, Materials Research Proceedings, Vol. 48, pp 11-19, 2025 https://doi.org/10.21741/9781644903414-2
[6] Afshar, A., Nouri, G., Ghazvineh, S., & Hosseini Lavassani, S. H. (2024). Machine-learning applications in structural response prediction: A review. Practice Periodical on Structural Design and Construction, 29(3). https://doi.org/10.1061/ppscfx.sceng-1292
[7] Aydın, Y., Bekdaş, G., Nigdeli, S. M., Isıkdağ, Ü., Kim, S., & Geem, Z. W. (2023). Machine Learning Models for Ecofriendly Optimum Design of Reinforced Concrete Columns. Applied Sciences, 13(7), 4117. https://doi.org/10.3390/app13074117
[8] S. Dernayka, D. Ahmed, M. Ajmal, A. Asiz and T. Ayadat, “Comparison of wind induced response of high-rise buildings with reinforced concrete and cross laminated timber”, The Fourth European and Mediterranean Structural Engineering and Construction Conference, 2022
[9] M. Ajmal, D. Ahmed and A. Asiz, “Structural Performances of Hybrid Tall Buildings Designed Using Ultra-Lightweight Floor Slabs”, 6th International Conference on Structural Engineering, Mechanics and Computational, SEMC 2016
[10] Askarinejad, P., & Behnia, B. (2024). Decarbonizing tall building structures: Implementing machine learning at the early-stage of design process. Preprints. https://doi.org/10.20944/preprints202408.2029.v1
[11] Patel, S., Patel, U., Nanavati, J., & Patel, A. (2024). Analyzing generative models for realistic data augmentation across modalities and applications. In 2024 11th International Conference on Computing for Sustainable Global Development (INDIACom) (pp. 1601–1606). IEEE. https://doi.org/10.23919/INDIACom61295.2024.10498685
[12] M. Ajmal, D. Ahmed, S. N. L. bt Taib, T. Ayadat, A. Hasan, and S. I. Abba, “Neural network prediction of raft foundation bearing capacity on stone column-reinforced soft clay using PLAXIS 3D generated data,” Proc. 9th IET Smart Cities Symposium, 2025, (In publication)
[13] M. Aimal, D. Ahmed, M. S. Alam, N. J. Alzayani, R. Ismaeel, and S. I. Abba, “Compressive strength prediction of reactive powder concrete through artificial neural networks,” in Proceedings of the 2025 International Conference on Innovation and Intelligence for Informatics, Computing, and Technologies (3ICT), 2025. (In publication)
[14] Ahmed, D., Ayadat, T., Asiz, A. (2020), Design and Performance of High-Rise Structure Using Ultra-lightweight Cross Laminated Timber Floor System, Proc. of International Structural Engineering and Construction- Emerging Technologies and Sustainability Principles, 7(2), ISSN: 2644-108X.
[15] Chen, Y., Yan, Z., & Zhu, Y. (2023). A unified framework for generative data augmentation: A comprehensive survey. arXiv preprint. https://doi.org/10.48550/arxiv.2310.00277