Investigating the Impact of Wind Tower Geometry on Ventilation Efficiency in Semi-Enclosed Spaces: A Comprehensive Parametric Analysis and Design Implications

Abstract

Passive building ventilation features, such as wind towers, can help meet rising cooling and ventilation demands in hot, arid regions. However, most prior studies rely on scaled models or isolate single design parameters, limiting holistic insight. This study conducts a full-scale, validated computational fluid dynamics (CFD) parametric analysis of wind tower geometry and its impact on ventilation efficiency in semi-enclosed spaces. Five geometric properties are investigated: tower shape, roof type, number of shafts, separator height, and number of louvres. Additionally, the sensitivity of the optimal configuration to wind speed, wind direction, and louvre orientation is assessed. Results from 88 CFD cases highlight strong interactions among design parameters and show that straight towers with curved roofs consistently perform best. Compared with a tower with six shafts, a flat internal roof, and downward-facing louvres, an optimized tower with four shafts, a convex internal roof, and upward-facing louvres increases airflow rate by a factor of 2.7 and occupied-zone air velocity by 45%, underscoring the importance of holistic geometric optimization.

1. Introduction

Since the advent of heating, ventilation, and air-conditioning (HVAC) systems, buildings have increasingly shifted away from passive ventilation and cooling toward mechanical systems. With the rise in global interest in sustainable and green building designs and low- and zero-carbon building trends, designers and engineers are now reinvestigating vernacular, traditional, and passive techniques that can meet performance and comfort needs while reducing energy consumption, emissions, and pollution as the traditional HVAC systems account for nearly 40% of global building energy use and contribute to about 20% of worldwide CO₂ emissions, making them a major driver of environmental impact [1]. Using natural ventilation techniques based on wind pressure and air buoyancy to reduce the reliance on mechanical ventilation systems has been a research focus for many researchers [2,3].

One such solution is the wind tower. Wind towers, also known as wind towers and Malqafs, have been featured in ancient Egyptian architecture and have emerged in vernacular architecture across the Middle East and North Africa (MENA) and the Persian Gulf Regions [4]. In these hot, dry regions, the design of wind towers aimed to direct cooler air from building roofs into usable spaces (i.e., breathable zones) [4]. The towers utilize the pressure difference (ΔP) between their sides (i.e., windward and leeward) and the roof heights as higher-pressure areas, directing air flow from these areas into indoor spaces as lower-pressure areas, thereby providing ventilation and cooling [5]. Additionally, vernacular towers in historical buildings often featured impressive architecture and demonstrated considerable engineering ingenuity, given the local climate, building materials, and occupant needs [6].

In modern times, wind towers have attracted significant interest in engineering and architectural literature. Modern wind towers have emerged in several locations, including the MENA region, Europe, and the USA [7]. The modern use of these wind towers has also led to heightened interest in studying their effectiveness and applications across various climates, as well as the impact of design parameters on performance. Existing research has confirmed that a tower’s design characteristics considerably affect its ability to move and direct air into built spaces.

While some researchers aimed to study the effect of some design parameters such as shaft height, shaft cross-section shape, and number of opening sides [8,9], there is a knowledge gap in comprehensive parametric studies that on full-size wind tower models that consider the simultaneous effect of the wind towers’ characteristics on their performance. Additionally, the definition and measurement of ventilation performance for these passive natural ventilation elements remain unclear.

This paper presents the results of a comprehensive parametric study of a full-scale, validated numerical model of a wind turbine. The main aim is to develop an understanding of the interplay among different wind-tower design features, including under-investigated parameters such as roof shape, the length of shaft separators, and louvre characteristics. In addition, the study establishes two KPIs for wind-tower performance evaluation: tower complexity and airflow rate. Finally, the airflow velocities in the occupied spaces were evaluated against ASHRAE Standard 55-2023 [10] to assess thermal comfort criteria.

1.1. Literature Review

Wind towers have been used in hot, arid, and humid regions, with many designs influenced by the local architecture and climate. However, all wind towers share the same design objective: to improve building passive ventilation and cooling. Different wind tower designs emerged across the MENA and Persian Gulf regions, influenced by local architecture, building techniques, and materials (Dehghani-Sanij, Soltani, and Raahemifar 2015) [4]. Egypt is one of the first locations where the usage of wind towers was documented. Evidence of Malqafs (i.e., single-sided wind towers) was found on papyri from ancient Egypt dating back more than 3500 years [4], and later, in Coptic architecture, wind catchers were evident since the 12th century AD [11]. Wind towers were also extensively used in Islamic architecture across the MENA region, and many examples still stand today, such as the Muhib Ad-Din Ash-Shaf’i Al-Muwaqqi house in Cairo, Egypt, built in the 14th century [12]. In the modern era, many wind towers have emerged and have been integrated into sustainable design projects. In the Middle East, large-scale wind towers were integrated into the design of Masdar City (UAE), built in 2010, and the Khalifa Stadium in Doha, Qatar, built in 2017 [13]. Globally, wind towers were also used at the Royal Chelsea Hospital in London and the Zion National Park Visitors Center [14,15].

Table 1. Typologies of traditional wind towers found in the Middle East [4].

	Egypt	Persian Gulf	Iraq	Pakistan	Afghanistan
Most Prevailing Wind Tower Shape

presents typological differences between towers in the MENA and Persian Gulf regions, which were strongly influenced by local climate, with the inlet design particularly affected [16]. In Egypt, the unidirectional Malqaf utilizes a high-altitude scoop to capture consistent northern winds while avoiding southern winds, which are often high in temperature and laden with sand. Conversely, in Afghanistan, one-sided towers are built with a low profile and thick masonry to catch the high-velocity winds. In Pakistan, towers are characterized by significant height and diagonal orientation to reach faster breezes above the dense urban canopy. The multi-directional Badgirs of the Persian Gulf employ internal shaft partitions to ensure a constant pressure loop for both intake and exhaust, regardless of shifting wind directions [16].

Interest in wind towers as passive ventilation solutions has led to numerous research articles on them. The available literature can be divided into four main categories: (1) full-scale experimental field studies, which experimentally use full-scale towers to study their performance and optimization potential. These are generally limited due to the complexity and cost of such work. (2) Scaled experimental studies, which use scaled-down models of towers to study their performance and optimization, are usually tested in wind tunnels. (3) Computational Fluid Dynamics (CFD) studies aim to explore methods to model and simulate wind towers or optimize tower performance numerically. (4) Combined studies, which use a combination of experimental and numerical techniques for their investigation. The following paragraphs will provide an abridged overview of the recent outcomes in these categories. Thereafter, the review will focus specifically on studies that have investigated the effects of different tower design parameters on wind tower performance.

1.1.1. Experimental Studies of Wind Towers and Their Parameters

A considerable body of experimental research over the past decades has investigated wind tower performance under varying design and environmental conditions. These studies have provided valuable empirical data on parameters such as geometry, wind speed, and wind direction, forming the foundation for subsequent numerical and optimization approaches. However, despite their contributions, many of these investigations exhibit methodological limitations that restrict their applicability to real-world buildings and diverse climatic contexts.

Table 2. Summary & Limitations of Experimental Studies on Wind Towers.

Research	Scaled Model/Full-Scale	Parameters Studied & Main Findings	Limitations
[17]	Full-scale (real building)	Air velocity, temperature, and solar radiation: Conducted in-site measurements at Lary House (Yazd, Iran) over a 24 h summer period to evaluate the performance of a simple traditional wind tower.	Limited to a single day of monitoring, narrow temporal coverage, and limited sensor locations, reducing generalizability across climates and seasons.
[18]	Scaled Model	Inlet shapes and shaft heights (3.5, 5.5, and 7.5 m): Performed wind tunnel tests on multiple tower variations to quantify how height and inlet geometry specifically evaluate and influence airflow rates.	Scaled experiments cannot fully capture urban wind dynamics; the focus was primarily on mass flow rate, with little attention to thermal comfort or indoor conditions.
[19]	Scaled Model (1:40)	Angular orientation and solar effects: Used a moving test stand to study a one-sided traditional Iranian wind tower. Investigated how rotation and sun exposure affect the internal airflow distribution.	Small-scale modelling does not accurately replicate real turbulence, buoyancy, or boundary layer effects; therefore, its accuracy for real buildings is limited.
[20]	Scaled Model	Novel dual-channel rotary scoop design: Tested a new aerodynamic configuration in a scaled setup; findings showed improved airflow capture and enhanced potential for thermal comfort.	Innovative design, but findings are limited by scaling; performance under real wind conditions and variable turbulence remains uncertain.

summarizes key insights from the available body of knowledge.

For instance, in situ measurements of a traditional wind tower at Lary House, Yazd, employing thermocouples and anemometers to capture temperature and velocity profiles under prevailing summer winds were described in [17]. While this study provided an early benchmark for understanding the microclimatic behavior of wind towers, its reliance on a single day of monitoring and limited sensor distribution restricts the generalizability of its findings to other climatic conditions or longer-term variations.

Similarly, wind tunnel experiments to assess the influence of different shapes (i.e., testing one-round side and multi-sided variations) and different shaft heights (i.e., 3.5, 5.5, and 7.5 m) on airflow rates entering the occupied area were reported in [18]. Although their parametric approach provided insights into how wind tower location affects ventilation potential, the use of scaled models inherently simplifies external flow conditions and may not capture complex wind–building interactions present in real urban environments. Furthermore, the study primarily focused on mass flow rate without integrating thermal comfort outcomes, leaving a gap in assessing practical building performance.

Scaled experimental setups, investigating airflow distribution within a one-sided wind tower inspired by traditional Iranian designs, were presented in [19]. Their apparatus allowed exploration of the influence of angular orientation and solar radiation on ventilation performance. However, the use of a 1:40 scale model introduces significant limitations, as reduced-scale physical models cannot fully replicate the turbulent flow structures, thermal buoyancy effects, and interaction with atmospheric boundary layers that occur at full scale. Consequently, while their findings highlight general airflow patterns, the accuracy of such results in predicting real building performance remains restricted.

More recently, an innovative dual-channel rotary scoop wind tower, demonstrating improved airflow capture and indoor thermal comfort in scaled experiments, was reported in [20]. This work highlights the potential of adaptive geometries to respond dynamically to prevailing wind directions. Nonetheless, as in earlier studies, reliance on scaled models rather than full-scale applications raises questions about how these results translate to real buildings under varying wind speeds and turbulent boundary-layer conditions.

1.1.2. Numerical Studies of Wind Towers and Their Parameters

Several studies have used numerical and parametric methods to study wind towers.

Table 3. Summary & Limitations of CFD Studies on Wind Towers.

Research	Scaled Model/Full-Scale	Parameters & Key Findings	Limitations
[21]	Scaled model (1:10)	Wind speeds $(0.5-5m/s)$ and incidence angles $(0°-90°)$: CFD simulations, validated by wind tunnel tests, evaluated velocity and pressure distribution. Findings confirmed how these variables dictate air movement in both the shafts and the occupied area.	Scaled modelling may not fully capture full-scale flow dynamics or urban wind interactions.
[9]	Full-scale application	Tower location and spatial positioning: Used CFD at a train station in Aqaba, Jordan, to validate that specific positioning is the primary driver for optimizing fresh air delivery and natural ventilation.	Focused mainly on spatial positioning of towers; limited analysis of geometry parameters such as shaft height, cross-section, or louvers.
[19]	Scaled Model	Passive rotary thermal wheels (20 and 32 blades): Investigated crossflow towers via CFD and wind tunnel testing. Found that the rotary device configuration directly alters airflow velocity, volume flow rate, and temperature.	Narrow focus on rotary devices rather than broader geometric or environmental factors; limited exploration of multi-parameter effects.
[22]	Scaled Model	Cross-sectional geometry (Plus-blade vs. others): Applied CFD to evaluate temperature reduction. Identified the square wind tower with a plus-blade form as the most effective configuration for indoor cooling.	Focused only on cross-sectional geometry; did not investigate the combined influence of other parameters like height, louvers, or orientation.
[23]	Scaled Model	Inlet velocity and prevailing wind direction: Conducted CFD using Normalized Catching Efficiency (NCE) as the metric. Proved that effectiveness depends heavily on the interaction between wind speed and direction.	Relied on a single performance indicator (NCE); did not consider multi-parameter interactions or comfort-related outcomes.
[14]	Full-scale application	Airflow rate and Air Change Rate (ACH): Studied the Doha Stadium using CFD. Established that wind towers can achieve specific natural ventilation benchmarks required for large-scale sports venues.	Focused on stadium-specific case; results may not generalize to residential contexts; limited analysis of geometry variations.
[8]	Scaled model (conceptual/parametric)	Occupied area and ASHRAE metrics: Proposed a shift to the occupant zone as the critical evaluation area. Emphasized that comfort metrics must align with ASHRAE standards to be architecturally relevant.	Provided a conceptual shift in performance assessment, but did not conduct comprehensive multi-parameter simulations on real building cases.

summarizes the relevant literature. A CFD study employed a 1:10 scaled wind tower to evaluate velocity and pressure distribution within the shafts and occupied area, with results validated against wind tunnel experiments across a range of wind speeds (0.5–5 m/s) and incidence angles (0–90°) [21]. While their integration of experimental and numerical analysis strengthens the credibility of the findings, reliance on scaled modelling introduces uncertainties in capturing full-scale flow phenomena and urban wind interactions.

Researchers combined wind tunnel testing with CFD modelling to investigate airflow in a crossflow wind tower incorporating passive rotary thermal wheel devices with 20 and 32 blades [19]. Similarly, another study [21] examined the effects of rotary devices on fresh-air velocity and flow distribution. These studies demonstrated the potential of hybrid wind tower designs to improve ventilation rates, yet the focus remained narrowly on mechanical inserts rather than broader geometrical parameters or their interaction with environmental conditions.

CFD analysis by [9] validated a wind tower design for a train station in Aqaba, Jordan, and explored different tower locations to optimize performance. While this application-driven approach highlights the adaptability of wind towers in large semi-enclosed spaces, the study primarily focused on spatial positioning and did not investigate detailed design variables, such as shaft height, opening area, or louvre configurations.

CFD was used in [22] to compare indoor and outdoor temperatures, concluding that a square wind tower with a plus-blade form was most efficient in reducing indoor air temperature. However, the study focused exclusively on cross-sectional form while disregarding the combined influence of other critical parameters, such as tower height and orientation. Normalized catching efficiency (NCE), a dimensionless metric to evaluate the aerodynamic performance of the inlet (the ratio of the volumetric airflow rate captured by the tower inlet to the theoretical airflow rate that would pass through the tower’s inlet area if there were no aerodynamic resistance), was employed by [23] as a performance metric, linking tower effectiveness only to inlet velocity or prevailing wind direction. While these approaches provide clarity on specific design sensitivities, they fail to capture the multi-parameter interactions that determine overall ventilation effectiveness.

More recent works, such as those in the Doha Stadium project, have shifted toward practical performance indicators, including airflow rate and air change rate (ACH) [14], aligning with international comfort standards such as ASHRAE Standard 55 [10]. The occupied area was emphasized as the primary focus for evaluating ventilation performance [8]. These contributions mark a step toward standardization of assessment methods; however, ambiguity persists in selecting performance metrics, especially in semi-enclosed or occupied spaces, making it difficult to benchmark and compare results across studies.

1.1.3. Relevance and Gap Identification

The reviewed literature demonstrates considerable progress in evaluating wind tower performance through experimental and CFD-based approaches. It is important to note that most recent literature that have explore wind tower design parameters have opted to use computational methods, such as CFD, instead of theoratical or analytical methods due to the complexity of the problem (i.e., the complex interaction between turbulent wind boundary layers, non-linear pressure coefficients around varied roof shapes, and the internal resistance of complex louvre systems).

Experimental investigations have yielded valuable insights into airflow distribution, shaft geometry, and environmental influences. However, many of these studies often rely on scaled models and simplified measurement techniques, which limit their ability to capture full-scale flow dynamics and thermal comfort outcomes [17,18,19]. CFD studies, in contrast, have provided more detailed analyses of airflow behavior and performance metrics [8,21,22,23]. Nevertheless, most of these efforts remain fragmented, with a predominant focus on single design parameters in isolation, such as shaft height, cross-sectional form, or orientation, without considering their combined effects. Moreover, inconsistencies in performance indicators further complicate cross-study comparisons, leaving gaps in both methodological rigor and practical applicability.

The present research addresses these gaps by conducting full-scale CFD simulations of a building equipped with a wind tower. Unlike earlier works that isolated one parameter at a time, this study adopts a multi-parameter approach, simultaneously investigating the combined effects of multiple geometrical parameters, wind direction, and velocity. By emphasizing the occupied area as the critical zone for occupant comfort [8] and applying consistent comfort-oriented performance indicators, this research aims to move beyond fragmented analyses toward a more holistic framework for wind tower optimization. Thus, the novelty of this study lies in (1) its comprehensive parametric investigation of full-scale, validated wind tower model, (2) emphasis on the occupied zone as the critical area for evaluating comfort and apply consistent, ASHRAE 55-2023 [10] aligned performance indicators, and (3) investigates the impact of key, yet often neglected, design features like roof shape and louvre characteristics. Thus, the outcomes will identify the most effective design configuration and also establish a methodological reference for evaluating passive ventilation strategies in real residential contexts.

2. Materials and Methods

This research aims to apply best practices reported in the literature to numerically model a wind tower case and validate the model against available experimental data. The paper aims to fill a knowledge gap by conducting a multi-parameter analysis to investigate how wind tower design characteristics affect performance. Specifically, the paper investigates five geometry design parameters: (1) tower shape, (2) roof type, (3) length of inner vanes, (4) the number of shafts and tower ratio, and (5) louvres’ density. In addition, a sensitivity study is conducted for three parameters: (1) the louvre direction, (2) wind direction, and (3) wind speed.

In doing so, the research also provides a definition and methodology for calculating the performance of these passive ventilation elements in semi-enclosed spaces. The following subsection presents this methodology in detail.

2.1. Wind Towers Physics

The physics of wind towers depends on the nature of wind motion and its fluid properties, such as density and pressure, which change. As previously mentioned, wind towers are designed to face the prevailing wind direction to maximize airflow for distribution within the built spaces. This ensures that the tower has access to more substantial, fresh air, which can help improve its performance [24]. Two main forces are affected by the airflow path inside the building [25]: (A) Mechanical forces, due to the pressure difference (ΔP) between the inlet, usually windward, and the outlet, typically leeward (Wind Driven Flow) [4,20] and (B) Buoyancy or thermal forces: due to temperature variation, higher density colder air typically moves downward inside the building to displace the hotter air inside the building, which will exit through the opening in the leeward direction (Buoyancy Driven Flow) [26,27]. However, previous research has suggested that buoyancy and thermal effects can be neglected in the simulations because the wind forces at this time of the year in Iran are the dominant force for natural ventilation [28], and also the wind tower in this research was used as a natural ventilation tool [8], not as a solar chimney [29].

2.2. Numerical Modelling Approach

A three-dimensional CFD simulation is performed using ANSYS Fluent Version 2023 R1 [30] to model the tower. The standard k-ϵ turbulence closure equations were identified as the most appropriate [28,31]. Best practices based on [32], along with additional assumptions, are summarized below:

• The flow is incompressible since compressibility effects can be neglected at relatively low airflow speeds.
• The flow is steady due to insignificant changes in airflow inside and outside the wind tower.
• Body forces may be neglected.
• Buoyancy-driven flow due to temperature change may be neglected because the wind tower used as a natural ventilation tool [8] is not a solar chimney [29].

With these assumptions, the steady Reynolds-averaged Navier–Stokes (RANS) equations can be solved to predict airflow in the domain. RANS are shown in Equations (1)–(4) below for the continuity equation (conservation of mass), momentum equation (conservation of momentum), and standard (k-ϵ) turbulence closure model [31].

$$\nabla .\overrightarrow{u}=0,$$

(1)

$$\rho \left(\overrightarrow{u}.\nabla \overrightarrow{u}\right)=-\nabla p+\nabla .\left(\left(\mu +{\mu }_t\right)\left(\nabla \overrightarrow{u}+{\left(\nabla \overrightarrow{u}\right)}^T\right)\right)+{\overrightarrow{F}}_B,$$

(2)

$$\left(\nabla .\left(\overrightarrow{u}k\right)\right)=\nabla .\left[\left(\mu +{\displaystyle \frac{{\mu }_t}{{\sigma }_k}}\right)\nabla k\right]+P_k-\rho ϵ,$$

(3)

$$\rho \left(\nabla .\left(\overrightarrow{u}ϵ\right)\right)=\nabla .\left[\left(\mu +{\displaystyle \frac{{\mu }_t}{{\sigma }_{ϵ}}}\right)\nabla ϵ\right]+C_{1ϵ}{\displaystyle \frac{ϵ}{k}}\left(P_k\right)-C_{2ϵ}\rho {\displaystyle \frac{{ϵ}^2}{k}},$$

(4)

where

• $ϵ$ is the rate of dissipation of turbulent kinetic energy, m²/s³.
• ${\overrightarrow{F}}_B$ is the body force vector, the force acting on a unit volume of a body, N/m³.
• $k$ is the turbulent kinetic energy, m²/s².
• $p$ is the static pressure, Pa.
• $P_k$ is the Turbulent Production Term, m²/s³.
• $\rho$ is the density, kg/m³.
• $\mu$ is the dynamic viscosity, Pa∙s.
• ${\mu }_t$ is the eddy viscosity, Pa∙s.
• ${\sigma }_k,{\sigma }_{ϵ},C_{1ϵ},C_{2ϵ},C_{\mu }$ are the constant coefficients, with these values (1, 1.3, 1.44, 1.92, and 0.09), respectively [32].
• $\overrightarrow{u}$ is the velocity vector, m/s.

2.3. Numerical Modelling Validation

The Mortaz House, located in Yazd’s city center, Iran, was used as a case study to validate the CFD modelling in this study. The house is a traditional residence in the historic center situated within a compact low-rise urban fabric. The house (50 × 25 × 7 m) features a four-sided wind tower measuring 6.05 m × 3.45 m and a total height of 14 m. Internally, the tower is divided into six vertical shafts (i.e., the tower has separators that divide it into 6 shafts), each with an area of roughly 1.5 m². Its cap features vertical louvres. The tower is connected to a semi-enclosed space (Talar) with an arched roof and open to a courtyard on the other side, as shown in Figure 1. The Talar, also known as Iwan in other geographies, is a semi-enclosed, high-ceilinged porch that serves as the main summer living area. This Talar is the focus of the analysis since it is the occupied zone requiring natural cooling. The north wind airflow enters the wind tower from the top and flows down to the lower building’s spaces through an opening gate to the occupied area.

The case has been previously investigated experimentally and numerically [28]. The published experimental results were used to validate the numerical model of this study. Ultrasonic wind velocity sensors were installed in four of the six (labelled A, B, C, and E). The sensors had a working range of 0–20 m/s and were installed in the middle of the shaft cross-section, above the wind tower outlet, at a height of 2 m (i.e., 5.35 m above ground level). A local weather station (wind sensors and data logger) was located 20 m away from the tower on the roof of the Mortaz House. This study focuses on the four primary shafts (A, B, C, and E) that were monitored as they provided sufficient resolution to validate the inlet/outlet flow patterns.

The experimental wind velocities and weather conditions were measured at hourly intervals for 24 h on 3 November 2014 [28]. The air velocity ranged from 0.28 to 2.5 m/s, with the highest recorded in the afternoon and the lowest in the evening and night. The wind direction inside the shafts of the wind towers was also different, with the air in shafts A and B flowing downward into the building (named inlet shafts), and the air in shafts C and D flowing upwards outside the building (named outlet shafts). The average velocity in each of the four shafts is shown in

Table 4. Average air velocity inside the four shafts in the wind tower reported in [28].

Shaft	A	B	D	E
Velocity (m/s)	0.58 (down)	0.37 (down)	0.75 (up)	0.51 (up)

. The mean air velocity recorded by the weather station was 1.5 m/s from the North at 11 m above ground on top of the arched roof of the Talar space.

2.3.1. Geometric Modelling

The full-scale house with dimensions (50 × 25 × 7 m) with its wind tower (6.05 × 3.45 × 14 m) with its six shafts and openings were modeled geometrically on Rhinoceros 3D (Rhino 7) [33] and then exported to SolidWorks, 2022 [34], as shown in Figure 2. before being imported to ANSYS Design Modeler Version 2023 R1 [30] to create the simulation domain with 350 m in X-direction (100 m before the model and 250 m after it), 200 m in Y-direction (100 m right & and left of the model), and 100 m in the z direction above the house [28], as shown in Figure 3. These values exceed the recommended values for single building models by available guidelines [35]: top boundary, lateral and inlet boundaries > 7H (recommended 5H), outlet distance > 17H (recommended 10H), where H is the height of the tower 14 m, and the domain frontal area is 21,400 m², and the model frontal area is 392 m²; the blockage ratio is about 1.8% (recommended below the 3%).

The Air domain boundary conditions (shown in Figure 2) with four different types are:

1.

Inlet Boundary Condition: A logarithmic wind profile was applied to represent ‘town and village’ terrain (Category III), following the standard log-law, with wind blowing from the North direction. The wind velocity was set to 1.5 m/s at a reference height of 11 m above ground level (roughness length (z₀) of 0.5 m and friction velocity of 0.2 m/s). In ANSYS Fluent Version 2023 R1 [30], the turbulence parameters were specified using the Intensity (5%) and turbulent viscosity ratio (10), corresponding to Turbulent kinetic energy $(k=0.0084m^2/s^2)$ and the rate of dissipation of turbulent kinetic energy $(ϵ=0.044m^2/s^3)$

2.

Outlet Boundary Condition: The pressure outlet was set to atmospheric or zero-gauge pressure to allow air to enter and exit the ambient domain.

3.

No Slip Wall: A no-slip wall with zero velocity components was used for the ground and the building walls

4.

Symmetry Boundary Conditions: Because the air domain was sufficiently large in all directions from the building (The Mortaz House), symmetry boundary conditions were used for the ceiling and side walls, with zero normal velocity only [36].

2.3.2. Meshing

The ANSYS Fluent Version 2023 R1 [30] meshing tool was used to generate tetrahedral elements with clustered elements near the walls, enabling the prediction of velocity distribution, as illustrated in the boundary layer in Figure 4.

To ensure the internal aerodynamics of the Mortaz House were captured, the mesh was divided into three main domains: (1) an outer domain with a large element size, (2) an inner domain with a smaller element size, and (3) a Building domain with the smallest element sizes (Clustering mesh). To clarify the resolution of internal flow features,

Table 5. Mesh models with different mesh sizes.

Case #	Outer Domain Element Size [m]	Inner Domain Element Size [m]	Wind Tower Element Size [m]	Total Number of Elements
1	15	6	1.5	162,242
2	12.5	5	1	329,973
3	10.5	4.25	0.725	498,626
4	10	4	0.7	555,687
5	9.5	3.5	0.65	701,935
6	8.5	3	0.6	938,232
7	7.5	2.5	0.5	1,451,691

provides a quantitative comparison of element sizes across the three domains, highlighting significant mesh refinement within the wind-tower shafts and the internal talar region relative to the far-field surroundings. As shown in Table 5, the transition from the far field to the building interior results in a significant increase in spatial resolution, with the finest model elements as small as 5% of the size of the largest outer-field elements. This ensured that the intake louvres and internal shaft partitions were resolved with sufficient density to depict the flow physics, while the outer domain developed the atmospheric boundary layer without unnecessary computational cost.

To capture the boundary layer near solid surfaces, an inflation layer is used with the standard wall function with (30 < y+ < 300) in a log-law layer, with a first-layer height (y_H = 0.8 mm) that can be calculated by the following relations that depend on the reference length (L), wind velocity (U), wind density ($\rho$), wind dynamic viscosity ($\mu$), and the required ($y^{+}$) value [32]. After calculating the first-layer height, the other two parameters required to create the inflation layer are the number of layers (N = 10) and the growth ratio (G = 1.2), as shown in the following equations and in Figure 5.

$$R{e}_L={\displaystyle \frac{\rho UL}{\mu }}\mathrm{ }\mathrm{ }\mathrm{ }\mathrm{ }\mathrm{ }\mathrm{ }{C}_f={\displaystyle \frac{0.026}{R{e}_L^{1/7}}}\mathrm{ }\mathrm{ }\mathrm{ }\mathrm{ }\mathrm{ }\mathrm{ }{\tau }_w={{\displaystyle \frac{1}{2}}C}_f\rho U^2\mathrm{ }\mathrm{ }\mathrm{ }\mathrm{ }\mathrm{ }{u}^{\mathrm{*}}=\sqrt{{\displaystyle \frac{{\tau }_w}{\rho }}}$$

$$y_H=2{\displaystyle \frac{y^{+}\mu }{\rho u^{\mathrm{*}}}}$$

Seven mesh models were developed to study mesh convergence (see Table 5), with Case 1 the coarsest, containing approximately 160,000 elements, and Case 7 the finest, with approximately 1.45 million elements.

2.3.3. Model Setup

According to the experimental description of [28], the wind blew from the North, and the tower faced Northeast. The prevailing winds in the region blow predominantly from the North and Northwest, which the wind tower is specifically oriented to capture; secondary winds also come from the Southeast, as seen in the wind rose in Figure 6. Thus, the geometry was rotated by 45° as shown in Figure 6. As discussed in the theoretical modelling section, the airflow can be considered 3D incompressible, with a steady RANS standard k-ϵ turbulence model and a standard wall function.

A SIMPLE scheme with a second-order upwind spatial discretization method was employed for pressure, momentum, energy, and turbulence parameters, along with a pseudo-time method, to facilitate the steady-state solution. Convergence residuals were less than $<{10}^{-4}$ for all parameters. A 20-core server with 16 GB RAM was used to run the following simulations.

2.3.4. Mesh Independence Study

In the numerical modelling, two error calculation methods were used to select the most appropriate mesh. The Relative Root Mean Square Deviation Error (RRMSE), as defined in Equation (5), was used to calculate the absolute error for each mesh. The Grid Convergence Index (GCI) [37], as defined in Equation (6), was used to calculate the relative error between two different mesh cases.

$$RRMSE={\left[{\displaystyle \frac{{\sum }_{i=1}^N{\left({X_{num}}_i-X_{E{x}_i}\right)}^2[\left({X_{num}}_i-X_{E{x}_i}\right)/X_{E{x}_i}]}{N}}\right]}^{1/2}/{\displaystyle \frac{{\sum }_{i=1}^N{X}_{E{x}_i}}{N}}$$

(5)

where

• $X_{nu{m}_i} is the numerical simulation shafts velocities data$.
• $X_{E{x}_i} is the Experimental Shafts velocities data$.
• N is the number of shafts.

$$GC{I}_{i,i+1}=F_s \mathrm{\ast } \left|\epsilon \right|/(r^p-1)$$

(6)

where

• r is the mesh elements ratio.
• F_s is the Factor of safety = 1.25.
• $\epsilon is the error=X_i-X_{i+1}/X_i$.
• p is the order of convergence = $ln\left(X_{i+1}-X_i/X_i-X_{i-1}\right)/ln\left(r\right)$.

2.3.5. Model Validation

Based on the results shown in

Table 6. Mesh models results vs. Experimental data results with RRMSE & GCI.

Velocity, m/s	Case #1 (162,242)		Case #2 (329,973)		Case #3 (498,626)		Case #4 (555,687)		Case #5 (701,935)		Case #6 (938,232)		Case #7 (1,451,691)		Exp Data
Shaft A	0.24		0.43		0.52		0.52		0.59		0.56		0.55		0.58
Shaft B	0.15		0.33		0.43		0.4		0.42		0.41		0.4		0.37
Shaft D	0.3		0.4		0.56		0.63		0.66		0.7		0.7		0.75
Shaft E	0.26		0.3		0.49		0.49		0.49		0.5		0.49		0.51
RRMSE	59.3%		39.5%		18.9%		12.6%		9.5%		6.1%		6.2%
$\mathit{G}\mathit{C}{\mathit{I}}_{\mathit{i},\mathit{i}\mathbf{+}\mathbf{1}}$		63.4%		43.7%		1.25%		3.67%		0.3%		0.8%

, the smallest GCI is between mesh models #5 and #6, with an error of 0.3%. The RMSE for mesh model #5 is less than 10%. A comparison between the numerical results and the experimental data of [28] is presented in Table 6, demonstrating the suitability of mesh model #5 for predicting air velocity within the four shafts. Therefore, it was concluded that mesh model #5 with 701,935 elements was deemed the most suitable for the study (Mesh details shown in Table 5).

Given the complexity level of the Mortaz house geometry and the large computational domain used, the parametrization of the tower conducted in this research has not significantly altered the generated mesh properties, warranting additional mesh independence investigations. This is clear since the altered tower geometry parameters only affected a small portion of the building domain, specifically the smallest element sizes (Clustering mesh), and did not significantly alter the overall setup of the simulation or the generated mesh. However, if significant changes are made to the building’s geometry or layout in future studies, further mesh investigations will be required to ensure the accuracy and reliability of the results.

Figure 7 shows the pressure contours around The Mortaz House, with higher values in front of the building and some suction inside the open areas, such as the courtyards, which explains the recirculation in the velocity contours. The lowest pressure values occurred at the back of the wind tower, resulting in outward airflow. A distinct high-pressure zone is observed at the windward tower head (Figure 7), driving airflow into the windward shafts. Figure 8 shows the velocity vectors around The Mortaz House with detailed eddies and recirculation flow inside the interior structure, as well as detailed vectors inside the wind tower’s shafts. As the air travels down the tower, a slight reduction in velocity is observed due to internal friction. The streamlines in Figure 8 show a well-defined circulation pattern within the courtyard, with the tower serving as both an inlet and, through the chimney effect, an outlet in the leeward shafts.

2.4. Parametric Study and Data Analysis

This section uses the validated CFD model of the wind tower to study its performance under varying operating and design conditions. The following sections outline the performance evaluation methodology and parametrization process, and present and discuss the key findings.

2.4.1. Simulation Cases

Figure 9 presents the overall models that were simulated in this research for the five design parameters (shape, roof type, number of shafts, and shaft separator’ height, and louvre density) and three sensitivity study parameters—Louvre direction, and wind speed and wind direction—a total number of 88 cases were completed: 72 cases for design parameters and 16 cases for sensitivity analysis as shown in Appendix A, Appendix B and Appendix C.

2.4.2. Parametrization Process

The parameters selected for this study were identified through a review of available literature as the most influential geometric features [38,39,40,41,42,43]. Figure 9A visualizes differences in the tower design and parameters, and Figure 9(B) presents the parametric simulation process that yielded the 88 cases. The selection of these features was based on their prominence in both traditional and modern designs:

• Wind Tower Shape: Two main shapes emerged as prominent for towers: straight and tapered/pyramidical [38]. Thus, these two shapes were selected. The straight tower has a shape ratio of 1:1 between the roof and base, and the tapered tower has a ratio of 0.75, as shown in Figure 9A.
• Roof Type: Two main roof types were selected using a similar process: straight and curved [39], as shown in Figure 9A. The curved roof is characterized by a flat exterior top and a convex internal geometry.
• Number of shaft divisions and tower shaft ratio: Two designs were selected: the original design of The Mortaz House, which included six shafts, and another design with the tower containing only four shafts. The quadrant type is more common in modern towers [40]. The number of shafts also changed the tower-shaft ratio: the 4-shaft tower had a 1:1 ratio, and the 6-shaft tower had a 2:3 ratio, as shown in Figure 9A.
• Wind Tower Separators length: The vanes or separators’ lengths inside the tower shaft were considered in the literature. Sometimes, the vanes are shorter than the cap, stop at the middle of the cap (Half CAP or HC), stop at the tower’s cap (CAP or C), or continue through the shaft (Full CAP or FC) [41]. These three conditions are shown in Figure 9A.
• Wind Tower Louvres Density (Number of Louvres per meter): Horizontal louvres are used for parametric cases. The louvres are modelled with a standard angle of 45° downward [42]. However, the density of the louvres appeared to vary significantly in the towers investigated in the available literature [43]. Thus, three conditions were selected: three louvres/m/m, six louvres/m/m, and nine louvres/m/m, as depicted in Figure 9A.
• Louvre directions: The original cases studied the tower with downward-facing louvres. This direction is reversed to an upward-facing angle in the sensitivity study, as illustrated in Figure 9A.

Regarding the location and surroundings, the study specifically utilized a validated simulation of the Mortaz House, including a semi-enclosed space (Talar) open to a courtyard. Although the validation experiments [28] were conducted on-site, the building was modeled without its surrounding urban context to isolate the aerodynamic influence of the tower’s geometry from random turbulence and the shielding effects of neighboring structures. This approach ensures that the performance gains observed are a direct result of the geometric optimizations rather than site-specific external interference. Furthermore, this simplification allowed for the concentration of computational resources on a high-density mesh within the internal shafts and occupied zones (as shown in Table 5), which is critical for capturing the internal pressure gradients across 88 parametric cases. Also, the study relied on the steady-state RANS approach and isothermal conditions, as they are computationally efficient for large-scale parametric studies and allow for isolating aerodynamic performance from buoyancy-driven effects. While this neglects thermal stratification, it provides a consistent framework for comparing the synergistic effects of multiple design features across 88 simulations. Neglecting buoyancy forces enables a focused quantification of how geometry affects mechanically (wind)-driven ventilation, which is the primary driver in the investigated scenario. While isothermal conditions neglect effects that could enhance or oppose ventilation, our findings for louvre and shaft performance are in good agreement with previous studies [43,44,45], providing confidence in the model’s predictions for these key components.

2.4.3. Wind Sensitivity Study

In addition to the parametric analysis mentioned above, the best-performing cases were further analyzed across varying wind speeds and directions.

• Wind speed: In this study, the reference speed was 1.5 m/s, as reported in the original experiment [28]. For the sensitivity study, the speed is adjusted by 25% to both speed up and slow down, investigating the effect of speed on performance, as illustrated in Figure 9.
• Wind Direction: The reference wind direction in this study is true North, as reported in the original experiment [28]. For the sensitivity study, the direction is rotated by 45° East and West to investigate the effect of wind direction on performance, as illustrated in Figure 9.

Figure 10 shows the investigation volume and area while Figure 11 presents the process used to analyze the results. First, all design and wind parameters were run, and then results exceeding the comfort conditions in the occupied space were excluded. To visualize and compare the remaining models, some design parameters were kept constant, while others were varied, enabling a visual comparison and the selection of the best cases to test against the sensitivity analysis.

2.4.4. Performance Evaluation Methodology

The main performance index used in the study to compare wind towers’ performance under different scenarios is their ability to ventilate the spaces connected to them. In the case of The Mortaz House, this value was computed using the total volume flow rate of air from the wind tower’s open gate into the Talar as shown in Figure 10. ANSYS Fluent Version 2023 R1 [30] was used to compute this value by indicating the opening as a surface. The analysis was designed to systematically evaluate the influence of geometric design parameters on ventilation performance, maintaining statistical rigor throughout the investigation. Given that air flows through this opening in multiple directions (i.e., inwards and outwards), the reported value considers the sum of the air flow rate at the occupied area. Thus, the net volume flow rate entering the Talar was selected as the parameter to assess the wind tower’s efficiency and effectiveness in providing ventilation and improving indoor air quality (IAQ) inside the occupied area, as seen in Figure 10. Another key performance indicator is the airflow velocities in the occupied zones or occupied area. Ensuring they do not exceed 0.8 m/s, as per ASHRAE Standard 55, was important [10]. However, since the Talar space exceeds the usable and occupied height, the research defines the occupied area within the space as between 0.075 m and 1.8 m above the floor, representing the readily occupied and used area within the space ventilated by the tower. While local comfort targets may vary, ASHRAE Standard 55 was adopted as a standardized global benchmark to ensure that optimized passive designs are relevant and can meet modern occupant needs, address existing ambiguities in performance metrics for semi-enclosed spaces, and facilitate cross-study comparisons.

Data quality control procedures were implemented to ensure the integrity of the CFD simulation results. This included verification of physical constraints, specifically ensuring non-negative flow rates for forward flow conditions and identification of performance states based on ASHRAE Standard 55-2023 criteria [10]. Cases were systematically categorized into three distinct performance categories: Acceptable (forward flow with maximum velocity ≤ 0.8 m/s), Exceeds Velocity Limit (forward flow with maximum velocity > 0.8 m/s), and Reverse Flow (negative flow rate indicating air extraction from the occupied space). Only cases that met both criteria of forward flow and velocity compliance were included in the optimization analysis to ensure the practical applicability of the findings.

Statistical analysis employed a hierarchical approach to evaluate the significance of parameters and interactions. One-way Analysis of Variance (ANOVA) was conducted for each geometric parameter to test the null hypothesis that all group means were equal. The F-statistics were calculated from the between-group and within-group variance components, and significance was assessed at an α level of 0.05. To quantify the practical significance of each parameter, the effect size was calculated using eta-squared (η²), which represents the proportion of total variance in volume flow rate explained by each design parameter.

Prior to conducting ANOVA, the homogeneity of variance assumption was verified using Levene’s test. For parameter interaction analysis, post hoc pairwise comparisons were conducted to identify specific configuration combinations that differed significantly. Each unique combination of parameter levels (e.g., Straight tower with Curved roof, Tapered tower with Normal roof) was treated as a distinct group. Tukey’s Honestly Significant Difference (HSD) test was applied to all pairwise comparisons among these configuration combinations, controlling for the family-wise error rate. Significance levels were determined at α = 0.05, with p < 0.001 indicating highly significant differences, p < 0.01 indicating very significant differences, and p < 0.05 indicating significant differences. This approach enabled the identification of specific design configurations that performed significantly better or worse than others, providing actionable guidance for optimal design selection beyond assessing individual parameter effects.

To investigate linear relationships between design parameters and performance outcomes, correlation analysis was performed using both Pearson and Spearman correlation coefficients. Design parameters were encoded numerically to facilitate correlation analysis: binary parameters (shape, roof type) were coded as 0–1 variables, ordinal parameters (shaft configuration) were assigned sequential values, and continuous parameters (louvre density, separator length) were represented by their actual numerical values.

Sensitivity analysis was conducted to quantify the impact of design modifications on ventilation performance. For louvre direction sensitivity, paired comparisons were conducted between base configurations (downward-facing louvres) and modified configurations (upward-facing louvres). The sensitivity index was calculated as the percentage change in volume flow rate relative to the base configuration, providing a standardized measure of parameter influence across different geometric configurations.

To facilitate performance-complexity trade-off analysis, a design-complexity metric was developed to quantify the relative constructability and cost implications of different wind-tower configurations. The complexity score was calculated as a weighted sum of geometric design parameters, with each parameter assigned a value based on its contribution to construction difficulty, material requirements, and fabrication precision. The scoring system assigned values as follows: tower shape (Straight = 0, Tapered = 1), roof type (Normal = 0, Curved = 1), shaft configuration (4 quadrants = 0, 6 quadrants = 1), louvre density (3/m = 0, 6/m = 1, 9/m = 2), and separator length (Half cap = 0, Cap height = 1, Full tower = 2). The total complexity score for each configuration ranged from 0 (simplest: straight tower, normal roof, 4 quadrants, 3 louvres/m, half-cap separators) to 7 (most complex: tapered tower, curved roof, 6 quadrants, 9 louvres/m, full-tower separators). This metric enabled the construction of Pareto frontiers to identify configurations offering optimal trade-offs between ventilation performance and design complexity, providing practical guidance for design decision-making that balances performance objectives with economic and constructability constraints.

3. Results

3.1. Effects of Geometric Parameters

The parametric investigation revealed significant variations in wind tower performance across different geometric configurations, with volume flow rates ranging from −0.453 m³/s (reverse flow) to 1.224 m³/s among acceptable configurations (Figure 11). Statistical analysis identified three parameters with statistically significant effects on ventilation performance (p < 0.05): Tower Shape, Separator Length, and Shaft Configuration. Figure 11 also showed that experiments with the Straight Tower, Separator Length of full Cap, and 4 shafts consistently outperformed other designs.

Tower shape emerged as the most influential parameter (F = 15.14, p = 0.0003, η² = 0.248), explaining approximately 25% of the variance in ventilation performance. Straight towers demonstrated superior performance with a mean flow rate of 0.519 ± 0.319 m³/s compared to tapered towers at 0.244 ± 0.135 m³/s. The superior performance of straight towers can be attributed to maintaining a consistent flow area throughout the tower height, thereby minimizing flow separation and pressure losses associated with area changes in tapered designs. The constant cross-section preserves the momentum of the incoming wind stream, facilitating more efficient pressure-driven ventilation. This finding aligns with fundamental principles of fluid dynamics, where sudden or gradual changes in area introduce additional resistance and potential flow instabilities. Thus, external aesthetic tapering should be decoupled from internal shaft geometry to maintain efficiency.

Shaft configuration showed the second strongest effect (F = 11.51, p = 0.0014, η² = 0.200), with four-shaft towers achieving significantly higher flow rates (0.505 ± 0.285 m³/s) than six-shaft configurations (0.258 ± 0.216 m³/s). The advantage of four-shaft over six-shaft configurations reflects an optimal balance between flow distribution and resistance. While increased partitioning might theoretically improve flow uniformity, the additional internal surfaces introduce friction losses that outweigh the benefits of distribution. The four-shaft design appears to provide sufficient flow guidance while minimizing internal resistance, particularly important given the relatively low driving pressures typical of natural ventilation systems.

The separator length configuration demonstrated a significant influence (F = 10.10, p = 0.0027, η² = 0.180), although the effect was more complex than anticipated. Separators extending to cap height achieved optimal performance (0.499 ± 0.315 m³/s), whereas both shorter (half-cap) and longer (full-length) configurations showed reduced effectiveness. Notably, full-length separators consistently induced reverse flow conditions, effectively converting the wind tower from a ventilation device into an exhaust chimney. The complex behavior of separator length effects reveals competing mechanisms in wind tower operation. Separators extending to cap height optimize the capture and redirection of wind flow without creating excessive internal recirculation zones. Shorter separators fail to adequately separate inlet and outlet flows, leading to short-circuiting and reduced net ventilation. Conversely, full-length separators create such strong flow separation that they induce stack-effect dominance, reversing the intended flow direction and converting the tower into an exhaust device.

3.2. Parameter Interaction Effects

Analysis of parameter interactions revealed complex synergistic effects that significantly influenced overall tower performance (Figure 12). The bar charts demonstrated that optimal performance requires careful consideration of parameter combinations rather than independent optimization of individual features.

The shape–roof interaction analysis revealed that curved roofs provided performance benefits primarily for straight towers (mean difference = 0.086 m³/s), while showing a negative effect on tapered towers (mean difference = −0.045 m³/s). The superior performance of the curved roof is likely due to reduced flow separation, compared to flat-roof configurations.

The density of the louvre and the separator length exhibited a particularly notable interaction pattern. For separators extending to cap height, increasing louvre density from 3/m to 9/m improved performance by 63%, which is in line with previously published literature [43,44]. In contrast, for half-cap separators, the same increase in louvre density resulted in a 27% reduction in performance. This reversal in the effects of louvre density highlights the importance of considering parameter interactions in design optimization [43]. However, only the interactions between different separator lengths and the 9/m louvre density were found to be statistically significant in our post hoc pairwise comparisons.

The interaction effects observed between parameters necessitate a holistic approach to wind tower design optimization. The synergistic relationship between tower shape and roof curvature suggests that these features should be considered as a coupled system rather than independent design choices. Similarly, the reversal of louvre density effects depending on separator configuration highlights the inadequacy of single-parameter optimization approaches [44].

3.3. Performance Optimization Landscape

The performance landscape analysis revealed distinct optimization regions characterized by trade-offs between design complexity and ventilation effectiveness (Figure 13). The contour plot of louvre density versus separator length ratio revealed a performance peak at 9 louvres/m with separators at the cap height, achieving flow rates exceeding 1.0 m³/s. Pareto frontier analysis identified seven non-dominated configurations representing optimal trade-offs between performance and design complexity. The frontier exhibited a characteristic convex shape, with diminishing performance returns as complexity increased beyond a score of 4. The optimal configuration, which achieved maximum performance (1.224 m³/s), had a moderate complexity score of 5, suggesting that excessive design complexity does not necessarily translate to improved performance.

The Pareto frontier analysis provides practical guidance for design decision-making by explicitly quantifying the trade-off between performance and complexity. The identification of a complexity threshold beyond which performance gains diminish offers valuable economic insights, suggesting that moderate complexity designs can achieve near-optimal performance while maintaining constructability and cost-effectiveness. Furthermore, practical implementation challenges and costs of the Pareto-optimized designs should be considered.

3.4. Sensitivity Analysis Results

Louvre direction sensitivity analysis revealed consistent performance improvements when the louvre orientation was reversed from downward to upward facing (Figure 14a). Across all straight-shape with flat-roof tested configurations, upward-facing louvres increased volume flow rates by an average of 100 ± 14%, with the effect being most pronounced for low-density louvre configurations. However, for the tapered tower, the upward louvres reduced the flow rate, leading to reverse flow in both the 3 m and 6 m configurations. This negative effect is consistent with findings for airfoil-shaped louvers, where flow separation can occur at high angles of attack [45]. The consistent performance improvement achieved through louvre reversal represents a simple yet effective optimization strategy that requires minimal design modification. This finding challenges conventional wind tower designs, which typically employ downward-facing louvres, presumably for rain protection. The performance benefits of upward-facing louvres suggest that alternative rain-protection strategies, such as extended caps or drainage systems, may be warranted to capitalize on these benefits.

The wind speed and direction sensitivity analysis (Figure 14b) illustrates the impact of varying wind conditions on the volumetric flow rate for three primary configurations: Straight-Cap Height, Tapered-Cap Height, and Straight-Full Length. The green bars represent the reference case, which is a standard wind speed (V_ref) from the North. The orange and light blue bars illustrate the effect of wind speed variations, with a 25% increase in wind speed resulting in a significant increase in flow rate for the Straight, Cap Height configuration (from 0.883 m³/s to 1.329 m³/s), and a 25% decrease in wind speed causing a decrease in flow. The changes in volumetric flow rate are smaller than would be expected if the flow were strictly proportional to the square of the wind speed, suggesting that, at higher velocities, aerodynamic losses from internal shaft friction and flow separation at the louvres begin to diminish the tower’s overall capture efficiency. The purple and pink bars illustrate the effect of changing wind direction to 45° West (NW) and 45° East (NE), respectively, showing that directional changes can also have a significant impact on performance for both straight and tapered tower configurations. Notably, the North-East (NE) wind direction (pink bars) shows enhanced performance across configurations compared to the reference North wind, indicating that the internal shaft partitions and roof geometry can capture flow from this orientation. The Straight-Full Length configuration is unique in that its negative values indicate a reverse flow, a phenomenon amplified by the 25% faster wind condition.

3.5. Top Performing Configurations

Analysis of the top-performing configurations revealed consistent design patterns that optimize ventilation performance. All ten highest-performing towers featured straight shapes, with nine incorporating curved roofs and eight utilizing four-shaft configurations (Figure 15). The optimal configuration combined all favorable features: a straight shape, a curved roof, four shafts, a 9/m louvre density, and cap-height separators, achieving a volume flow rate of 1.224 m³/s while maintaining maximum velocities below the ASHRAE comfort threshold.

Parameter frequency analysis among top configurations confirmed the dominance of specific design features. Straight towers were present in 90% of top performers, while curved and flat roofs were each found in 50% of the same group. Four Shafts and Cap Height also appeared in 80% of the top configurations. The consistency of these patterns across high-performing designs provides clear guidance for practical implementation.

4. Discussion

This paper presents a comprehensive parametric analysis for the validated numerical model of the Mortaz house in Iran. The multi-parameter investigation of wind tower performance included some underexplored design features, such as cap design, louver features, and separator height. The effects of wind speed and direction are also adequately studied to address the gap in the literature, bringing the total number of parameters of this research to eight geometric parameters and three sensitivity parameters, which were investigated through more than 88 CFD simulations. Unlike previous studies that isolated single design features, this research simultaneously examined the combined effects of multiple geometric and environmental parameters using a validated, full-scale numerical model. The results of the CFD simulations were analyzed, and only feasible design configurations that met the target performance requirements of ASHRAE Standard 55 [10] were used in the final comparison to identify the best-performing designs.

The parametric investigation revealed significant variations in wind tower performance, with an optimal configuration achieving a volume flow rate of 1.224 m³/s among acceptable designs. Statistical analysis confirmed that tower shape (p < 0.001), shaft configuration (p < 0.01), and separator length (p < 0.05) were the most influential parameters, collectively accounting for over 60% of the variance in ventilation performance.

The study found that straight towers performed better than tapered designs, a finding attributed to their constant flow area, which helps preserve the incoming wind’s momentum and minimizes pressure drops associated with flow separation and area changes. Contrary to a common assumption, curved roofs provided performance benefits only for straight towers but negatively affected tapered towers. The analysis of shaft configuration showed that four-shaft designs achieved significantly higher flow rates than six-shaft designs, highlighting an optimal balance between flow distribution and internal resistance. Separators extending to the cap’s height achieved optimal flow rates. In contrast, both shorter (half-cap) and longer (full-length) configurations reduced effectiveness. Notably, models with separators extending the full length of the shaft consistently induced reverse flow, acting as an exhaust device and exceeding the ASHRAE Standard 55 [10] recommended indoor air velocity. The study also found that upward-facing louvres yielded higher volume flow rates in straight wind towers, challenging the traditional use of downward-facing louvres. However, this effect was not universal: upward louvres reduced flow rate in tapered towers. These findings underscore the critical importance of considering the synergistic effects of multiple design parameters rather than optimizing each feature in isolation.

Although this result is significant, the study has several limitations. The numerical model, while validated against experimental data from the Mortaz House with reasonable agreement (RRMSE < 10%), had limitations and relied on several key assumptions. First, full-scale field experiments are often constrained by complexity and cost. Thus, available data for validation were limited to single-day monitoring benchmarks, as seen in [17,28]. The choice of the steady Reynolds-Averaged Navier–Stokes (RANS) system and the standard k-ε turbulence model followed established best practices and was identified as the most appropriate for predicting wind-driven ventilation in this context. The steady-state RANS approach, while computationally efficient for parametric studies, cannot capture transient phenomena such as wind gusting or thermal stratification effects that may influence real-world performance. Furthermore, specifying a constant 5% turbulence intensity at the inlet simplifies the complex, time-varying atmospheric boundary layer characteristic of urban environments. This value was selected to ensure streamwise profile stability and to provide a conservative baseline for aerodynamic performance, consistent with the available data [28]. Actual urban turbulence levels typically exhibit higher and more stochastic fluctuations that could influence local pressure coefficients.

Furthermore, the assumption of isothermal conditions neglects the interaction of buoyancy with wind-driven ventilation. The neglect of these thermal forces is justified by the study’s focus on isolating geometric performance during wind-driven conditions. Given the complexity of the Mortaz house geometry and the large computational domain, the parametrization of the tower conducted in this research did not significantly alter the overall mesh properties generated. This is clear because the altered tower geometry parameters (including louvers) affected only a small portion of the building domain. However, the authors acknowledge that if more significant changes to the geometry are made in future studies, further mesh investigations will be required.

Regarding the building model in isolation, the authors acknowledge that the original validation experiments [28] were conducted on-site within a complex urban fabric. However, the decision to exclude surrounding buildings in the parametric phase was a deliberate methodological choice to isolate the tower’s aerodynamic influence from random turbulence and shielding effects from neighboring structures, while still achieving a validated model. This approach ensures that the identified performance trends are a direct consequence of internal and external geometric optimizations rather than site-specific external interference. Furthermore, this simplification enabled the concentration of computational resources on a high-density mesh within the internal shafts and occupied zones to capture fine-scale pressure gradients across the 88-case design space.

The study focused on a specific building configuration, a semi-enclosed space (Talar) that opens to a courtyard, which may limit the generalizability of the findings to other architectural contexts. However, this building exemplifies a typical climate-responsive typology for the region, and this research provides a demonstrated methodology that can now be applied on a case-by-case basis to other contexts. Wind towers integrated with fully enclosed spaces or multi-story buildings may exhibit different optimal configurations due to altered pressure distributions and flow patterns. Furthermore, for several geometric parameters, only two discrete options were investigated. While this identifies high-impact features, this binary choice may limit the generality of the conclusions. Additionally, the single wind direction assumption, while aligned with the prevailing conditions at the validation site, does not account for variable wind directions that could affect tower performance, where, for example, improved performance was observed with NE winds. This underscores the need to optimize wind towers as integrated systems rather than by isolated features.

Nevertheless, the study establishes a transferable methodological framework for full-scale optimization that can be applied to diverse architectural contexts and climate zones. Future work should investigate these variable conditions. However, despite these limitations, our findings on louver performance are in good agreement with previous studies on louver configuration and density [43,44], providing confidence in the model’s predictions for these key components. Several future directions for research are required to build on this study. These include examining the application of geometric parameters to complex buildings that employ modern architecture, featuring multiple floors and rooms. This would result in a more in-depth examination of the efficiency of traditional wind towers, building on this parametrization study. Numerically, the integration of transient CFD simulations using Large Eddy Simulation (LES) can provide insights into temporal performance variations and the impact of atmospheric turbulence on ventilation rates.

Furthermore, coupling aerodynamic optimization with thermal analysis would enable the evaluation of wind towers’ cooling effectiveness beyond simple ventilation metrics. Ultimately, field studies of optimized wind tower designs in various climatic zones would provide crucial validation of CFD simulations and identify site-specific factors that affect performance. Such studies should include long-term monitoring of ventilation rates, indoor air quality parameters, and occupant comfort metrics to comprehensively assess the practical benefits of geometric optimization.

Based on this work, we recommend that future research expand to examine the effectiveness of wind towers in other building configurations, including multi-story buildings and modern architectural layouts. It is also important to combine and expand the possible design parameters to investigate how different tower geometric configurations respond to changes in wind speed and direction. Experimental investigations, where possible, on full-scale modern towers can provide valuable resources for researchers to validate their modeling work. Finally, accounting for the impact of dynamic wind conditions and thermal effects (i.e., buoyancy forces) in transient computational models would provide additional value and insights for operations under real-life conditions.

Finally, several practical recommendations can be made. Straight wind towers with curved roofs, four shafts with separators, and dense upward-facing louvres should be used, ensuring that the cap height is optimized. These towers should be oriented toward the prevailing wind direction to achieve optimal performance. Unless the tower is specifically designed to function as a chimney for extracting air from indoor spaces, the shaft separators should not extend the full shaft length to avoid exceeding comfortable air velocities in occupied zones. In addition, incorporating features such as flow control at the tower outlets can allow users to balance the volumetric flow rate with indoor air velocity, thereby maintaining thermal comfort while ensuring compliance with relevant standards.

5. Conclusions

This study used a validated full-scale CFD model of the Mortaz house to systematically quantify how wind tower geometry and operating conditions affect ventilation performance. By varying eight geometric and three sensitivity parameters in more than 88 simulations, the analysis showed that tower shape (p < 0.01), shaft configuration (p < 0.01), and separator length (p < 0.05) together explain over 60% of the variance in airflow.

Among the ASHRAE Standard 55 [10] compliant cases, the optimal configuration was a straight tower with a curved roof, four shafts, separators extending to the cap height, and dense upward-facing louvres, achieving a volume flow rate of 1.224 m³/s. It is important to note that the superior performance of straight towers was attributed to their constant flow area, which minimizes pressure losses, and to the lower number of shafts tested (i.e., four shafts), which provided a better balance between flow distribution and internal resistance. The results also highlighted that these parameters interact in non-trivial ways; for example, curved roofs and upward-facing louvres enhance performance only for straight towers, while having an adverse effect on tapered ones. Notably, models with separators extending the full length of the shaft consistently induced reverse flow, acting as an exhaust device and exceeding the ASHRAE Standard 55 [10] recommended indoor air velocity. This underscores the need to optimize wind towers as integrated systems rather than by isolated features.

The work is subject to several limitations, including the use of steady-state, isothermal RANS simulations, a single prevailing wind direction, and a specific semi-enclosed building configuration (RRMSE < 10%), which constrain the generality of the findings. Furthermore, for several geometric parameters, only two options were investigated. While the results are specific to the semi-enclosed configuration of the Mortaz House, this study establishes a transferable methodological framework for full-scale optimization that can be applied to diverse architectural contexts and climate zones. Nonetheless, the present results provide actionable guidance for practice: straight, curved-roof wind towers with four shafts, optimally sized separators that do not span the full shaft length, and controllable, dense upward-facing louvres oriented toward prevailing winds offer a robust baseline for achieving comfortable, wind-driven ventilation in hot, arid climates.

Table 7. Summary of research and practical recommendations.

Research Recommendations	Practical Recommendations
Expand the research to consider the effectiveness of wind catchers in other building configurations, including multi-story buildings and modern architectural layouts.	Use straight wind towers with curved roofs, four shafts with separators, and dense louvres facing upwards, ensuring the caps are optimized in height.
Combine and expand the possible design parameters to investigate how different tower geometric configurations respond to changes in wind speed and direction.	The towers should be positioned directly to face the prevailing wind direction to achieve optimal performance.
Expand the effort to include full-scale experimental testing on modern towers, ensuring the availability of experimental data for further validation.	Unless the tower is designed to act as a chimney to remove air from spaces, avoid extending shaft separators to their full length, to prevent exceeding comfortable air speeds in breathable spaces.
Include dynamic wind conditions and thermal effects (i.e., buoyancy forces) in transient computational models to simulate the operation of the towers under real-life conditions.	Incorporate features, such as flow control at tower outlets, to enable users to balance volume flow rates with indoor air velocity, ensuring comfort and compliance with standards.

summarizes the key research and practical recommendations. Future research should, therefore, extend this parametrization to modern architecture, incorporate transient LES-based simulations that account for buoyancy and dynamic wind effects, and validate optimized designs through field measurements of airflow and indoor comfort.