Computational Methods in Vernacular Passive Cooling: A Comparative Review and the Research Gap in Iranian Windcatcher-Qanat Systems

Computational fluid dynamics (CFD), parametric modeling, building-energy simulation, and machine learning (ML) increasingly translate climate-responsive vernacular architecture into quantifiable models. Yet these methods are distributed unevenly across vernacular traditions, making it difficult to distinguish mature areas of computational building-science research from specific openings for further work. This paper presents a structured comparative review of computational studies associated with five courtyard and passive-cooling traditions: Chinese siheyuan courtyards; Miao and Southwest China stilt dwellings; Mediterranean courtyard houses; Nigerian and West African courtyard houses; and Iranian badgir (windcatcher), qanat, and courtyard systems. Searches were updated through August 2026, and the reviewed evidence was coded by computational method, performance or design target, study scale, use of learned/data-driven models, and degree of system coupling. The comparison shows the broadest methodological range in Chinese and Mediterranean courtyard research, where simulation and parametric methods now coexist with machine-learning prediction and optimization. The Miao/stilt and West African evidence bases remain more limited and are dominated by simulation, rule-based parametric, or generative approaches. Iranian windcatcher research occupies an intermediate position: CFD studies are well established, and recent work has introduced CFD-trained artificialneural-network prediction, neural-network optimization, machine-learning classification, and AI-assisted heritage analysis. However, the data-driven physical-performance studies identified remain windcatcher-only or single-building in scope. No study identified in this review applies ML or another learned surrogate model to the coupled thermal-fluid behavior of a badgir and qanat operating as an integrated passive-cooling system. By defining that gap narrowly, the paper reframes the next computational step for Iranian vernacular cooling as a systems-level building-science problem and establishes a basis for future simulation, dataset construction, and performance optimization.