Digital-Twin-Oriented Surrogate Modelling of Architectural Identity and Thermal Performance in Benghazi A Semi-Physical, Evidence-Informed Framework for Multi-Objective Design-Space Exploration
DOI:
https://doi.org/10.65405/wtr3w927Keywords:
Digital Twin; semi-physical surrogate model; machine learning; HistGradientBoosting; SHAP; permutation importance; AIPI; architectural identity; Benghazi; Monte Carlo; Pareto; Golden Ratio.Abstract
This study develops a Digital-Twin-oriented computational framework for exploring relationships between selected architectural-envelope characteristics, thermal performance, and a limited evidence-informed proxy of architectural identity in Benghazi. The framework integrates documented historical geometric evidence with a semi-physical thermal formulation, tree-based machine-learning surrogates, permutation importance and SHAP analysis, and a 50,000-point Monte Carlo design-space exploration followed by Pareto analysis. HistGradientBoosting achieved an independent-test R² of 0.98940, RMSE of 0.29227 °C, and MAE of 0.23646 °C, while five-fold cross-validation produced a mean R² of 0.98815 ± 0.00116. Wall thickness and external temperature were the dominant thermal predictors within the current formulation. The documented geometric evidence from the Sidi Khrebish/Berenice context showed substantial variability, with a weighted empirical mean ratio of 1.37453 (SD = 0.24723); the Golden Ratio (φ ≈ 1.618) was therefore retained as an independent comparative geometric reference rather than embedded in the base AIPI. The multi-objective exploration produced a computational Pareto set of 16 non-dominated solutions under standardized conditions of 35 °C outdoor temperature and 800 W/m² solar irradiance. A computational compromise was selected with wall thickness of 0.592982 m, WWR of 0.104893, arch ratio of 1.415982, predicted indoor temperature of 26.546933 °C, AIPI of 98.604171/100, and independent Golden Ratio proximity of 87.512504/100. These results remain exploratory and model-dependent; they represent surrogate fidelity to the defined semi-physical formulation rather than field validation of real buildings.
Downloads
References
Historical visual and archival sources are additionally documented in Sections 12.1–12.4 so that image ownership, licensing, attribution, and evidentiary use remain traceable.
Michaelides, D. (1998). Excavations at Sidi Khrebish Benghazi (Berenice), Volume IV, Part 1: The Mosaics and Marble Floors. Tripoli: Department of Antiquities, in collaboration with the Society for Libyan Studies. Supplements to Libya Antiqua, 5. Open-access edition: https://www.jstor.org/stable/jj.14962420. DOI: 10.2307/jj.14962420.
Pye, E. (1974). The painted wallplaster from House H, Sidi Khrebish. Libyan Studies, 6, 10–12. https://www.cambridge.org/core/services/aop-cambridge-core/content/view/F5C7023E2C5FF7A37BB467ACBA8A22EF/S0263718900000546a.pdf.
Safour, A. A., & Elmazek, E. M. (2024). Architectural Identity of Benghazi City Between Tradition and Modernity: Case Study—Urban Centre of Benghazi. In Urban Narratives: Exploring Identity, Heritage, and Sustainable Development in Cities (pp. 47–57). Springer. https://doi.org/10.1007/978-3-031-48517-6_5.
Safour, A. A., Elmazek, E. M., Benmoussa, Z. A., & Alteer, N. A. (2025). Potential of Rehabilitating Benghazi Historical Buildings—Challenges and Strategies. ARCHive-SR, 9(4), 1–24. https://doi.org/10.21625/archive-sr.v9i4.1180.
Coccoli, C., Arenghi, A., Tanghetti, F., & El-Rida, A. (2024). Knowing is Saving: The Italian Architecture in Libya and the Case of Benghazi Cathedral. In Conservation of Architectural Heritage (CAH): Developing Sustainable Practices (Vol. 1, pp. 391–402). Springer. https://doi.org/10.1007/978-3-031-33222-7_32.
Shahzad, M., Shafiq, M. T., Douglas, D., & Kassem, M. (2022). Digital Twins in Built Environments: An Investigation of the Characteristics, Applications, and Challenges. Buildings, 12(2), 120. https://doi.org/10.3390/buildings12020120.
Mousavi, Y., Gharineiat, Z., Karimi, A. A., McDougall, K., Rossi, A., & Gonizzi Barsanti, S. (2024). Digital Twin Technology in Built Environment: A Review of Applications, Capabilities and Challenges. Smart Cities, 7(5), 2594–2615. https://doi.org/10.3390/smartcities7050101.
AlBalkhy, W., Karmaoui, D., Ducoulombier, L., Lafhaj, Z., & Linner, T. (2024). Digital twins in the built environment: Definition, applications, and challenges. Automation in Construction, 162, 105368. https://doi.org/10.1016/j.autcon.2024.105368.
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., & Duchesnay, E. (2011). Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research, 12, 2825–2830.
Lundberg, S. M., & Lee, S. I. (2017). A Unified Approach to Interpreting Model Predictions. Advances in Neural Information Processing Systems, 30, 4765–4774.
American Society of Overseas Research (ASOR) Cultural Heritage Initiatives. (2017). Incident Report LHI 17-0013: Ottoman Municipal Building, Souq al-Hout, Benghazi, Cyrenaica, Libya. In ASOR CHI Appendices. https://www.asor.org/wp-content/uploads/2019/09/ASOR_CHI_July17_Appendices.pdf.
World Monuments Fund. (2022). Benghazi Historic City Center. https://www.wmf.org/monuments/benghazi-historic-city-center.
Media and web sources are documented separately in Sections 14.1–14.3 to distinguish archival/visual evidence from peer-reviewed and scholarly references.
Gad, A. F. (2021). PyGAD: An intuitive Python library for building the genetic algorithm and training machine learning algorithms. arXiv preprint arXiv:2106.06158.
