CFD Simulation Supported Development of Wind Catcher Shape Topology in a Passive Air Conduction System (PACS)

Abstract

New studies and reports are published on a daily basis about the dangers of climate change and its main causes: humanity’s constantly growing population, the built environment and resource consumption. The built environment is responsible for approx. 40% of the total energy consumption, and a significant part comes from maintaining an appropriate indoor comfort environment by heating ventilation and air conditioning. Though contemporary studies have achieved a wide knowledge about natural ventilation and passive air conducting systems (PACS) and their applicability, further investigations are necessary to deepen the aerodynamic topology of air conducting building structures’ shape properties. Hence, in our current research we conducted a series of tests applying different wind catcher geometries. The methodology of this work is based on the authors’ previous work, where passive air conduction systems were compared with different airflow directions via computational fluid dynamic simulations (CFD). After finding the better performing PACS (a downdraught system), this research evaluates whether further improvements in ventilation efficiency are possible due to the aerodynamic shaping of the roof integrated inlet structures. Four different wind catcher geometries were examined to determine the most advantageous dimensional settings in the natural ventilation system’s given boundaries. After multiple series of basic and developed calculation runs, diverse shape designs of the passive air conduction inlet (PACI) were examined, including wind deflector geometries. The initial reference wind catcher’s air change rate was increased by approx. 11%. The results deliver the potential measure of improvements achievable in the aerodynamic shape design of structures under identic conditions of the same building domain. As a consequence, more sophisticated natural ventilation structural solutions will be possible in more operation cost- and performance-effective ways.

1. Introduction

Nowadays, one of our biggest challenges is to reduce the energy demands of our modern lifestyle. New warnings, analysis and predictions are published on a daily basis by scientists, trying to emphasize the necessity of change [1]. Although the world’s governments and international institutions have attempted to provide answers to this evident problem [2,3], the created guidelines and regulations—as a basis of the legislation—have not achieved the required level of energy consumption reduction [4]. If we do not want to wait for authorities to force the transition to a sustainable future, researchers need to find and propose energy costs and time-saving solutions for industry and private stakeholders.

The built environment is responsible for a significant fraction of global energy consumption. Based on different estimations it can reach up to 40%, taking into account the total life cycle [5]. The active heating, ventilating, and air conditioning (HVAC) of our interior spaces can use between 50 and 70% of this energy [6]. In the last decade, numerous papers have successfully proved the applicability of natural ventilation (NV) and hybrid (combined natural and mechanical ventilation) as an alternative to the exclusively active system-driven building operation. Chen et al. [7] created a database about the yearly potential of NV solutions in different meteorological zones, whereby in some climates NV can be used almost during the whole year, achieving approx. 45% energy conservation. Khan et al. [8] reviewed the different options and aspects of NV from traditional structures to modern examples, and created a wide base of knowledge for Passive Air Conduction System (PACS) design. During the development of scientific methods, numerous techniques became available for researchers in aerodynamic topics and for ventilation system designers. Beside modern measuring tools, wind and water tunnels, Computational Fluid Dynamics (CFD) simulations are the most widespread device for airflow modeling and analyzing [9]. Maneuvered by educated and experienced engineers and scholars, it is able to replace the previous options and save time, energy and financial expenses during the aerodynamic assessments.

In this study, the investigation focus is set on the fresh air inlet structure and its formal design parameters, which originated as wind catchers from the ancient Iranian, Middle-East and Egyptian territories. Bahadori [10] mentioned them first, and offered this option for consideration as a new/old option instead of mechanical ventilation (MV). The main element is a vertical structure—a wind tower—that has a ‘wind catcher’ geometry on top. It can harvest the predominant wind flow (one-sided wind catchers) or capture the wind from all directions (two or more-sided wind catchers) and force the air to flow down to the interior, thus providing fresh air and improving indoor air quality (IAQ). Hughes [11] and Jomehzadeh [12] write about the history and development of wind catchers in a thorough way by reviewing geographical locations and climate aspects of vernacular wind towers, together with contemporary examples from an architectural and scientific point of view. Finally, they suggest the application of wind catchers in modern PACS, and they support this idea with the detailed evaluation of numerous modern wind catcher geometries. Dehghani et al. [13] also wrote about the evolution of wind catchers. The work not only collects the available data and offers it as a base knowledge for the next generation, it creates a new special industrial wind catcher geometry based on their own conclusions. Heidari et al. [14] investigated the vernacular earthen architecture of Iran using CFD simulations. The paper classified the houses by their typology. Based on the results, a guideline was created that can help to avoid unwanted air movements, positively affect IAQ and optimize air change per hour (ACH). Badran [15] collected information about the traditional Jordanian wind towers, whose heights are between 4 and 15 m. It is stated that the towers already provide enough fresh air for a two-story building if the towers are only 9 m high, thus saving construction materials and time. In addition, based on the results, with evaporative cooling a temperature decrease of approximately 9 K is possible in the towers in the Jordan valley. Mohamadabadi et al. [16] discussed the aerodynamic operation of a rectangular wind catcher with six sub-sections under different wind incident angles. The different sections conduct the airflow up or down simultaneously, depending on the wind direction and the evolving pressure zones around the towers. The research used wind tunnel model experiments and CFD simulations to evaluate the towers. It pointed out that the combined towers, without any wind dependent control, can have a short-circuit on the bottom, where the entering fresh air immediately leaves the interior without mixing with the exhaust air. Zarmehr et al. [17] took an ancient Persian water reservoir as a reference model in their research. In a complex geometry analysis, seven different modifications were tried out and compared. The most promising version achieved 15 m/s of air velocity inside the towers, and produced approximately 8 K cooling, showing that it is possible to improve the vernacular base model. However, many of the alterations did not enhanced the PACS’s ventilation performance. Calautit et al. evaluated the cooling potential of single-sided [18] and multi-directional [19] wind catchers with integrated heat transfer devices with CFD simulations. The outcomes of the paper were positive: the single-sided geometry achieved 8–12 K cooling with only a 15% decrease of air velocity, and the multi-directional wind catcher reached a similar level of effectiveness. Long et al. [20] proposed a complex PACS in a visitor center of a National Park in Utah, USA. The PACS contained wind catchers, Trombe-walls, elevated exhaust (outlet) points and evaporative cooling, etc. A total of 67% in energy savings has been achieved, and visitors think about the wind catchers as an amenity of the building. This is important, because the integration of these techniques into architectural concepts is crucial to make NV more popular among architects as well as contractors and occupants. Balabel et al. [21] investigated the inlet opening angles effect on a partial-cylinder wind catcher geometry and different locations of the tower in a given building. Their findings show that the higher opening angles can achieve better results, and the wind catcher has a better performance between 20.55—37.5% when it is behind the building due to the incoming wind, compared to the other versions. Haghighi et al. [22] proposed a combination of a wind catcher and a solar absorption chiller integrated in a two-story building. The system could cool down the internal air by 10–20 K. Beside the cooling potential, an urban scale investigation was also conducted by combining three similar buildings related to the occurring wind direction. The results were analyzed considering also geometrical differences, e.g., the lower towers produced less ACH. Chew et al. [23] approached the topic from a different way: in an urban environment, different wind catcher geometries were integrated to the tops of buildings to conduct the airflow down from the roof level to the street canyons. After multiple alterations in water channel model measurements, the improved design of a wind catcher with a closed sidewall enhanced the maximum near-ground wind speed in the canyons by four times. Alwetaishi et al. [24] could reach a 12 °C temperature reduction in the region of Saudi Arabia inside an underground room via evaporative cooling, which are commonly implemented with wind catchers.

These studies proved that wind catchers were not only successfully used in the ancient era, but they are also able to satisfy higher needs, and can be combined with modern techniques to achieve a higher, optimal level of IAQ. However, the use of wind catchers is only a guiding principle of a PACS. The determination of the right geometry based on the local functions, building shape, structure and dimensions, urban texture and meteorological data is essential for further performance improvements and savings in financial-, energy-, environment and time resources. Afshin et al. [25] described the behavior of a traditional two-sided wind catcher under different wind directions. Based on the CFD results, the air stream direction changed in the tower if the wind incident angle is higher than 55° (0° = orthogonal to the wind catcher’s inlet). If the wind direction was parallel (90°) to the openings, the wind catcher reached the highest ACH, because in this case both of the tower sections extracted exhaust air from the interiors, therefore the ventilation’s cross section was doubled. Benkari et al. [26] applied two simple wind catcher shapes to a semi-opened atrium. The version with a curvy transaction to vertical walls delivered a better ACH performance, and it was successful in providing fresh air for the atrium’s whole area. Ansar et al. [27] combined wind catchers with solar chimneys, and the PACS’s different parameters were systematically changed (e.g., cross section). According to the conclusions, not all geometrical change improved the performance, and the danger of frictional losses threats is highlighted if the shape design is not created with proper blending and smoothing. Sarjito et al. [28] compared different wind cowl variations with CFD simulations. The most effective version with an extended baffle had a 79.4% mass capture efficiency, which was five times higher than the weakest option. Alsailani et al. [29] conducted a comprehensive parametric CFD study of wind catcher topology by investigating the thirty-three variation of six properties. The results showed approximately 23% in difference between the best and worst scenario, and it was raised further to 29% when guide vanes were implemented to the model. Varela-Boydo et al. [30] had also parametrically examined the shapes of wind catchers. Nevertheless, in this study the parameters were investigated in a combined way. There were 174 scenarios of 28 versions solved in CFD simulations. The wind catchers were based on traditional two-sided towers (one side is an inlet and the other is an outlet), therefore the short-cut problem mentioned by Mohamadabadi [16] is possibly active in these cases too. Nonetheless, very detailed and gap-filling insights were created, and was offered as a ‘how to do and not to do’ guideline for wind catcher engineers.

Among the available research, two principal trends prevail: one part of the studies consider concrete, complex building use, climate and urban circumstances (e.g., tests in the wind direction), but the tested ventilation structure domain has mostly a low level of detail and the number of the examined cases are rather low. The other trend includes investigations without exact, detailed building and neighbor structures and further boundary conditions (e.g., tests in only one wind direction), but the examined ventilation structure domain has a higher level of resolution, and the number of cases are higher as well. All research work models the relatively small building and neighborhood areas in the form of simplified geometries due to the high level of fluid mechanical complexity, mesh generation issues, computation capacity and time requirements.

During the design stage of a prototype winery, boutique hotel and panorama restaurant facility, appropriate literature data were found on the proven functionality and workability of PACS. However, sufficient information about the NV structures precise sizing, scaling and geometrical parameters, in particular given that building boundaries were not available. According to the quality and quantity of the gained results from literature, the outlining knowledge about the shape design of ventilation structures is apparently incomplete. There is a scientific need for building a wide and thorough knowledge about wind catcher topology, which functions as a guideline in such situations. The purpose of this study is to expand this field with new versions and results along with the reviewed authors. Therefore, in a previous study of the authors [31], PACS versions with different air conduction directions were proposed and compared in a winery industry building project. A downdraught (DD) PACS including a wind catcher inlet and an updraft (UD) PACS with a ‘Venturi’ outlet tower were compared. The results suggested that on a site with no prevailing wind situation, the DD wind catchers are more favorable against the uni- or bidirectional UD ‘Venturi’ towers, since the DD system possesses an omnidirectional behavior. Although the study offered a solution to decide between the different PACSs, and other valuable information was found to describe the behavior of a DD and UD PACS, it has not helped to specify the detailed dimensions of the system’s ‘engines’: the wind catcher inlet and the ‘Venturi’ outlet. However, the following questions remained open: Which detailed geometry versions offer a better solution for the chosen PACS? How and what kind of topology of natural ventilating structures should be chosen for a DD PACS in an industrial building to achieve the highest possible ACH, and thus the best cooling and natural IAQ for the wine process and storage technology? The aim of this particular study is to reply to these questions, which are still not satisfactory answered by the reviewed literature. At the same time, reasonable structural and architectural design aspects are to be considered during the shape modelling and optimization process.

2. Materials and Methods

2.1. Shape Design Test Variations of the PACS Intlet Structure

According to the results of the previous study, a DD wind catcher PACS should have been integrated in the new winery building in Villány, Hungary. The prototype building and the PACS is currently in the final stage of the implementation at the construction site (Figure 1), described in detail in [31].

The territory is ranked as a Cfb-temperate oceanic climate, based on the Köppen-Geieger climate classification [32]. Since no meteorological data were collected on site, the Meteonorm ^® [33] database were chosen to generate wind distribution data for the local wind speeds. Because the values did not suggest any exclusive prevailing wind direction, an omnidirectional PACS was selected as a solution for future integration into the test reference building. The structure of the PACS is a DD system which includes a wind catcher inlet and a wind deflector outlet.

The PACS was integrated in the final architectural plan. The sections in Figure 2 presents the location of the in- and outlets and chimney structure. The inlet structure is integrated in the flat roof terrace of the restaurant in the form of a wind catcher to collect the incoming wind, and force fresh air down to the interior. The stale air is sucked out by the ‘Venturi’ tower, where the special wind deflector geometry on top of the tower should create a low-pressure zone just above the outlet opening. The indicated airflow directions, velocities and volume flow rates are based on the CFD results in the previous study [31]. An inlet structure’s shape should be defined so that it is able to provide the highest supply of airflow and thus ACH in the two-story high production hall and the cellar in the basement. The position of the investigated wind catcher is indicated in the sections as well.

Figure 3 proposes the basic boundary conditions for the wind catcher setup. All cases have a rectangular cross section with two separators through the diagonals; hence, four sections are created, regulated (opened or closed) by a Building Management System (BMS), depending on the incoming wind’s direction. The rectangular chimney section comes from another study of the authors [34], where high performance ventilation towers were achieved in a real-world industry building with an 2000 m³ technology hall space. According to the conclusions, the tower sections (5.2 × 4.9 m) were over-dimensioned (undesired deaccelerating vortices were generated), and thus in the 1200 m³ winery spaces the chimney section was generally reduced. A particular size was designed as 1.80 × 1.80 m together with the rectangular layout. The chimneys were positioned from a functional aspect, as well to offer easy integration into the space arrangement and the load-bearing reinforced concrete structure of the building. In addition, the extremely close arrangement of the towers to each other created a weakening ventilation capacity due to the slipstream effect of the chimneys. Therefore, in this particular case, only two chimneys are considered to provide a higher distance between the openings. The height of the wind catcher is as high as possible to fit into the maximum allowed roof integrated structure dimensions. The walls of the wind catchers were modeled as infinite thin walls, because many modern materials are suitable to create very thin (just a couple of cm thickness) and thus aerodynamically ‘two-dimensional’ shapes.

For optimal air inlet design, the transition of the entering air movement is crucial, i.e., as little as possible turbulence and vortices are desired to avoid unwanted contra-productive currents against the air supply’s current direction. Four different geometries were selected based on the work of Ford et al. [35], which should operate sufficiently. The geometry variations are presented on this. The first case is the ‘empty’ wind catcher (base case (Figure 4a). The second is the ‘circle’ (Figure 4b), where the radius of the curved inwards and downwards deflecting surface is the height of the inlet openings (the right side of the figures). The third one has a ‘parabolic’ curve (Figure 4c), while the fourth version was equipped with the same wind catcher topology as the third, but the tower’s cross section was modified as a hyperbolic bell (Figure 4d). The parabola was chosen in the third version’s modification according to Ford’s design source book [35], and Pearlmutter et al. [36] proposed the parabolic shape as a better aerodynamic option. However, their work was done in a tower with a 3.75 m diameter. This means that the cross section’s area was approximately 3.5 times larger, so for more universal insights, the possible generalization of the results can be advantageous. The ideal curves were represented by segments (the left side of the figures), in order to enable easier volume discretization as well as practicality for the construction.

The investigation of the wind catchers was conducted in a simplified version of the original building, as the complicated façade and roof (Figure 5a) are reduced to a cubic shape (Figure 5b). Thus, less modelling and calculation time is needed, and a more generic outcome is acquired. Neighboring buildings and vegetation details are replaced with the mathematical modelling of the environment’s boundary layer: a custom-developed wind profile generator calculated the typical wind circumstances with a modified roughness constant [37,38]. The connection between inside and outside with the two towers and the sizes, proportions and geometries of the spaces and chimneys are similar to the CFD model of our previous winery study. The only difference is that the production and storage rooms in the basement are merged in the current model to gain the more generic insights. The connections to the previous study are necessary for the easier validation of the CFD simulation method, since it has been already performed correctly.

The ventilation outlet chimney consists of the same dimensions and shape as described in [31] in the validated CFD model. The only difference is the geometrical simplifications of the building envelope. Both openings have the same cross section. The height of the exhaust tower’s top opening is settled 3.0 m higher than the inlet opening of the supply structure. This is in order to enable as much exposure as possible to the approaching wind in case of the most disadvantageous direction, i.e., wind coming from the direction of the wind catcher.

2.2. Simulations and Model Set Up

The wind catcher geometries were modeled and simulated via CFD software. The CFD simulations are modifiable in terms of cost and time, and the basic infrastructure and its maintenance is more cost-effective as well. This is why it is more widespread than wind tunnel model experiments or in-situ measurements [9,39].

Firstly, a virtual wind tunnel domain was created. Multiple guidelines are proposed to dimension CFD models [40,41,42]. Usually, the basic unit of an atmospheric domain is the height of the building [H]. In this series of simulations, the domain sizes are presented in Figure 6. Since the investigated building is symmetrical, only the three main wind directions’ calculation is required: cross (orthogonal to the building’s longitudinal axis), diagonal and longitudinal (parallel with the axis). The building longitudinal axis is equal to the connection axis between the two chimneys. Due to the approaching wind direction cases, the calculated domain became asymmetrical.

The created volume is required to be discretized by the Finite Volume Method (FVM), whereby in the appropriately densely created cell nodes, the main field variables will be calculated in the CFD simulations. The precision of the results depends on two important aspects of the modelling. First, the CFD results, and the methods that tends to achieve precise results should have been validated with manual measurements or with an already validated literature comparison. This study follows the second option, whereby the authors previous works [31,34] were taken as reference models, including real world measurements, and the representative sizes are shown in

Table 1. General cell sizes of the simulated meshes.

Region	Validated Mesh [34]	Detailed Domain [31]	Simplified Domain (Current Research)
Total domain size	500 m × 500 m × 100 m	200 m × 200 m × 80 m	105 m × 135 m × 60 m
Atmospheric	6 m	4 m	4 m
Macro environment	3 m	2 m	2 m
Micro environment	2 m	1 m	1 m
Near building walls	0.5 m	0.3 m	0.3 m
Towers/openings	0.3 m	0.1 m	0.1 m
Interior	0.2 m	0.25 m	0.25 m

. It can be clearly seen that the previous study’s validated mesh sizes (detailed domain) and particular mesh domain dimensions agree. Only the total domain size is smaller in the new research, because here the three wind directions were solely investigated with accordingly smaller upwind areas [37,38]. The cell size diversion ranged between 1.5 and 2.5 of the model’s discretization.

The second important aspect of reliability is the ‘grid independency’, where the quality of the used grid is determined by the widely accepted guidelines of Celik et al. [43] Three different mesh were generated with 353,622; 776,196 and 1,696,657 cells, respectively. The fine-grid convergence index is 1.04% and 0.19% for the medium and the fine mesh (Figure 7). The medium grid was selected for further work, because the deviance is below an acceptable level. Appendix A discusses the details of the grid independency test’s steps.

The simulations were carried out in the ANSYS^® Fluent 17.2 software (Ansys, Canonsburg, PA, USA). The suitable turbulence model is crucial for respectable CFD simulations, as stated by Peng et al. [44]. They invited 19 CFD engineer teams to solve the same problem with their own turbulence modelling method, and the paper found 150% deviances between the different solution methods. Based on the experiences of the authors and other fellow researchers [16,45,46,47], the Reynolds-averaged Navier-Stokes (RANS) equations with the SST k-Ω turbulence model was selected for the CFD. Since the assessment of the wind catchers’ performance is based on the ACH, only isothermal calculations were made and no energy equations were solved. One case contained 1500 iterations in the steady state and 1200 iterations in transient mode (120 s, and 1 s has 10 iterations). The transient mode was necessary only in few cases for convergence, but for unified solutions and comparisons, every case was solved with the same settings. The atmospheric wind profile was generated in Fluent, with User Defined Functions (UDF), based on the work of Balogh et al. [37,38]. The air velocity at the height of 10 m was 2.73 m/s (coming from the Meteonorm^® database) and the gradual vertical change of the velocity was determined by the local agricultural land (some houses, hedgerows and fences) which were modeled with a uniform roughness constant coded into the UDF.

3. Results

3.1. Test Wind Catchers with Various Inlet Geometries

The four selected cases were investigated with three different wind directions, which meant a total of 12 scenarios. The three directions are described in Section 2.2. The ventilation performances were evaluated separately in each of the directions, but the final evaluation was made in dependency of the averaged mean values, since an omni-directional PACS is the origin of this study (provided from a previous investigation [31]. The wind catcher of this system should work acceptably under all wind situations.

Figure 8 shows the flow fields in the 12 cases of all geometries in the horizontal sections around the wind catchers, Figure 9 represents the local airflow patterns above the roofs and Figure 10 presents the same cases in the vertical sections. Firstly, the flow fields show great similarity in all cases, whose consequence in ACH can be tracked in Figure 11.

The diagram clearly shows that there are no significant differences in ACH in the same wind direction cases, only between different wind direction cases is there a deviance to observe of up to approx. 2 ACH. The best performance values were obtained with the cross direction in all four cases, followed by the ‘longitudinal’ direction with approx. 1 ACH decrease and the ‘diagonal’ cases with approx. 2 ACH decrease. The reason behind this phenomenon is that in the parallel longitudinal case, the roof behind the wind catcher generates frictional losses (Figure 9), which deaccelerate the air movement at the outlet chimney’s top opening, thus the ACH becomes less effective. In contrast, higher velocities arise when the currents approach from the orthogonal direction to the building, because there is no obstacle in the proximity of the arriving air. However, in the diagonal cases (as in the cross version as well), the approaching airflow velocity of 3 m/s could be reached, less fresh air entering and ACH evolves in these cases. This is due to the wind incident angle orthogonal to the wind catchers’ divider walls, and hence the air bypasses the geometry. This is contrary to the other two situations, where the diagonal walls direct the incoming fresh air to the center of the tower.

By calculating the average values of the three main sample wind directions, the generic behavior of the omnidirectional inlet can be characterized. This value is in all cases approximately 7.50 h⁻¹. The only shape with a poorer mean performance is the version of ‘hyperbola’, although the deviance is not coming from the flow field’s change, rather the narrowed cross section area weakens the ventilation rate. Figure 10j–l distinctly shows the special shape in the chimney shaft has helped to decrease the forming of detached vortices on the entering side and in the shaft, thus the higher air velocity evolved inside the geometry. Similar to this, Alwetaishi et al. [48] has better air velocity performances in innovative designs of wind catchers with curved towers. However, the narrowed chimney section area was a stronger force for ACH manipulation, therefore a lower value (6.45 h⁻¹) is achieved here.

3.2. Effects of Deflectors

In order to increase the ‘wind collecting’ mechanism of the walls, the diagonal deflectors were extended for further tests. Two candidates were selected from the best performing three models. The simplest (and most cost-efficient) version is the ‘empty’ as well as the curved version (‘parabola’). In Figure 12, the walls have been extended in a way where the inlet’s flat roof became 1.00 m wider in all perimeter directions, enlarging the receiving inlet area of the new geometry by 211%. The original size of the diaphragms possessed 1.27 m × 2.10 m, with a surface area of 2.67 m² in the ‘empty’ version, and it became 2.68 m × 2.10 m (surface area 5.64 m²) in the new shape.

In the new cases, unexpected results were obtained (Figure 13): both geometries changed their ventilation behavior with the new deflectors. Thanks to the increased deflector area, the diagonal cases improved significantly by balancing the disadvantages of the shape: the ACH raised from 6.50 h⁻¹ to 7.70 h⁻¹.

While the cross-direction’s performance did not change appreciably, the longitudinal one features a considerable decrease in ACH from 7.50 h⁻¹ to 6.30 h⁻¹. The source of the relapse is presented by Figure 14b,d in form of a backflow circulation, immediately in front of the wind catcher’s inlet opening. The vortex is generated by the wider deflectors (sidewalls), i.e., the towers opening is not able to sufficiently take all the extra captured airflow. Based on this phenomenon, and due to the frictional losses along the roof (see Section 3.1), the velocities and thus the ACH decreased by 16.4 %. The problematic vortices are noticeable in the cross-directions as well (Figure 14a,c), but the faster air velocities can balance out the restraining effect of the recirculation.

3.3. Modified Deflector Design

While the performance in the cross-direction remained (together with the average ventilation rates—See Figure 13), and the diagonal direction improved, in the longitudinal direction the decreased ACH made further modification on the deflectors necessary. Figure 15 presents the modified topology, where the diagonal sidewalls’ bottom part was diminished. The receiving inlet area of the new geometry, and the new surface area of 4.15 m² in the new shape was reduced by 46%. The created ‘gap’ should allow the congested air to bypass the geometry, and lower the chance of arising backflow vortices around the inlet opening.

Figure 16 visualizes the resulting airflow, and Figure 17 displays the obtained ACH performance of the wind catcher development. The accomplished ventilation rates were significantly better compared to the initial topologies, delivering an average advancement of 11.5%. The developed deflectors not only improved the average performance, but they produced in both cross, diagonal and longitudinal directions a visible enhancement of 9.6%, 22.1% and 20.2% to the worst cases, respectively. The average ‘parabolic’ (8.34 h⁻¹ ACH) and ‘empty’ (8.16 h⁻¹ ACH) inlet structure versions’ ventilation efficiency was improved by 9.1% and 11.8%, respectively. The diagonal scenarios did not bring any change to their previous version. Fundamentally, it is a good outcome, since the improvement in the diagonal direction was already achieved in the previous test. Instead of hindering the evolution of the ACH rates by the smaller deflector surface, they rather ensured a greater inflow, as the counterproductive recirculating vortices could laterally bypass the structure.

4. Discussion

The tested tower has smaller dimensions compared to general wind catchers [11,13,49], as well as compared to a previous study of the authors [34], where the ventilation chimney cross section was oversized (5.20 × 4.90 m) and hence a counterproductive lateral distribution of the airflow’s kinetic energy was developed in the tower. Due to the downsized chimneys’ cross section area, the lateral vortices are not present in the flow field of the chimneys anymore. Hence a more efficient ventilation performance was achieved (Figure 18).

The unexpected poor performance from the ‘hyperbola’ (Figure 4d) version is coming from the reduced cross-sectional area of the tower. However, it is easy to identify in the results (Figure 8j–l) that the hyperbola shape decreased the aerodynamically ineffective area in the chimney shaft. In this way, the higher inner velocities and less detached vortices were generated in the tower. However, the smaller cross section became contra productive, and overwrote the mentioned improvement. Nevertheless, the ‘hyperbola bell’ should not be devoted after these results. A more focused study on this promising inner chimney shaft design could be conducted on the synchronization of the hyperbola’s parameters and the correct use of this version [50].

In the case of wind direction diagonal to the longitudinal axis of the building, the geometries generally performed poorly. The reason behind this phenomenon is the orthogonal position of the deflectors to the coming wind direction, as opposed to the other two situations, where the wind catchers’ inner walls collect fresh air with a 45° incident angle on both sides. The high incident impinge angle on the deflectors’ surface enables evolve turbulences which bypass the inlet structure instead of entering the ventilation opening (Figure 8b,e,h,k). To avoid this problem, further modifications should be considered, such as multiple (e.g., eight) sectioned tower [11], or more specialized deflector wall shapes as this study’s results suggests, or even moving (automated) deflectors according to the incoming wind current incident angles (wind direction independency) [8].

In case of the longitudinal direction, the negative behaviors of the passive towers—mentioned by Mohamadabadi’s study [16]—could be avoided by arranging the chimneys a greater distance from each other in a particular wind direction. The gained ACH values are still the lowest in this direction, however this is solely due to the roof structure caused by frictional forces, which are more or less unavoidable in a building situation.

The concept of extended deflector walls of 1 m was intended to capture more incoming fresh air in the wind catcher. It was applied to the ‘empty’ and ‘parabola’ shapes to follow their effect, not just on the general ACH values, but also to see the relations between the alternating geometries. The first test unexpectedly did not significantly modify the averaged performance, but the new PACS element behaved differently in the three directions. In the longitudinal situation the performance dropped, and the visualization of the flow field helped to recognize the generated vortices ahead of the wind catcher’s opening. The wider deflectors captured more air at the front of the tower opening, but the inlet opening’s area remained the same size, hence a backflow formed from the congested air, and it restrained the ventilation by bypassing the wind catcher’s deflectors. This resulted in the reduction of incoming fresh air into the inlet tower, and hence less ACH performance. In addition, due to roof generated deaccelerating frictions, in this wind direction the ventilation velocities were lower. The recirculation vortices evolved in the cross direction as well, but as was the case in the first shape test, the cross-directed wind could remain faster around the wind catcher. Therefore, the bigger momentum of the incoming air could balance out the negative effect of the vortex, and there was no negative change in the ventilation performance. Under the diagonal wind incident angle, where the capturing surface was missing in the starting geometries, the deflectors achieved a significant improvement. The obtained results create cohesive experiences about wind catcher geometries with contemporary studies [29,30].

The inclined reduction of the deflectors’ shape was proposed to enforce air flow to bypass the wind catcher at its plinth zone (and hence reduce recirculation and ACH reduction), if the resistance from the inlet openings is raised critically. The ACH values increased to almost 11.5% compared to the initial models. Both cross and longitudinal directions performed better, proving that the forming of restraining vortices were reduced with the proposed geometry.

Limitations

The simulations were solved within an isothermal model, meaning that no energy equations were calculated. The performance of NV is always influenced by the outdoor temperature, occupancies, solar heat sources and other pressure differences. The consideration of these factors is to be carried out in the next research steps. In the future, it is favorable to examine the role and relation of thermal buoyancy to these wind catcher variations and other topics of IAQ, such as draught rate, relative humidity or air contaminant distribution.

The target of this paper was to find a maximized ACH-producing wind catcher topology under strict boundary conditions. During this optimization process, excessive high velocities of exiting air from the inlet tower were not considered as a comfort factor. This is due to the fact that the winery technology hall is a non-continuous occupied space (i.e., during the wine harvesting season, it is used for approx. 1.5 months a year; otherwise it is for non-continuous use). During occupied periods, automated dampeners manipulate the incoming airflow rate to ensure comfortable conditions. However, NV solutions need to serve incredibly complex demands with a large number of variables (including wind velocity and direction dependency, temperature and humidity dependency and an anti-draught operation). Thus, to map out this complicated synchronized operation of a PACS is a difficult future task.

The generated atmospheric wind profile represented a rural, agricultural (mostly empty) site. Hence, the research cannot give insights about the behavior of the selected topologies in a dense city or other complicated environment with an unstable, more turbulent flow. Additionally, due to the sparsely occupied agricultural hall, further investigation in more intensely occupied spaces (such as an office atrium, event hall, sport hall, etc.) would provide useful insights. The modeled building function gives information about only one segment of buildings (agricultural, industrial), and research in other building typology shows great potential to complete the knowledge on NV systems.

5. Conclusions

Scientific research repeatedly emphasizes the urgent need to reduce the energy consumption in our built environment. The application of NV solutions can result in significant energy savings, and in certain climate zones the maintenance is almost possible during the whole year. The available work in the field of PACS geometry optimization shows experiences where NV options have extraordinary complicated aerodynamic properties. Hence, it is inevitable to conduct further research about the PACS’s topology to deepen the general knowledge about their aerodynamic behavior.

This paper connects to a previous work of the authors [31], where the direction of the air conduction of the PACS was investigated and a DD system was developed as a multi-directional solution for NV. This study tries to answer to the following questions: Is it possible to enhance the ACH rate of the already developed DD PACS and, if yes, to what content? How should the inlet structure’s shape look like in the previously developed DD PACS, in order to provide as efficient a wind capturing as possible? The relation to the previous work [31] was favorable, because within these tests a previous CFD model was adequately compared with the validated CFD model of an in-situ measured, real-world building [34]. Thereby, the experiences in CFD simulation techniques of the authors’ previous studies were applied here. The main implications of this research are to successfully expand the knowledge in the field, and create a good foundation for further studies about wind catchers:

• The shape and size of the building body around the wind catcher can highly influence the flow field around the wind catcher and its air capturing performance. If the in- and outlet structures of the PACS are aligned in line with each other, the cross directed wind—unaffected from the building body—can enter with highest efficiency into the wind catcher structures’ inlet. In the longitudinal direction, the roof surface behind the wind catcher generates frictional losses and notably lower velocities at the exhaust tower’s outlet, and hence there is a decrease of the vent performance. Therefore, weaker ventilation is present under such circumstances in every construction situation. This lower vent performance can, however, still provide appropriate or even higher vent rates (in one particular study an ACH of 7.5 h⁻¹ was provided, an 846% greater performance as required in EN 15215 [51].
• The developed small-sized wind catchers (smaller than generic, traditional wind catchers [11,13,49]) possess a floor area-related PACS ratio of 2.43% and a volumetric PACS ratio of 5.55%. This cross section represents a small-scale inlet structure solution, in contrast to a previous measured real-world project [34] with a floor area-related PACS ratio of 10% and a volumetric PACS ratio of 30%. A future classification of the diverse historical and contemporary PACSs could rationalize design and development of such ventilation systems.
• The topology development confirms that small-scale wind catchers are less responsive to different shape changes regarding their role in the ACH rate distribution of the PACS. The very similar ventilation air flow values (7.50 h⁻¹ ACH) of most shapes (three basic geometries) in this study supplement the Pearlmutter et al. [46] outcome: while ventilation efficiency rates show similar differences between different inlet geometries, by downscaling the cross section of an inlet chimney, the wind catcher’s ventilation performance becomes less sensible to different geometry types in contrast to larger chimney cross sections.
• A V-shaped layout of vertical deflector walls positively influence the air-capturing ability of a wind catcher’s inlet structure facing the approaching wind direction (in one particular study, the cross direction). If the incident angle of the coming wind flow to the deflector surfaces increases, the ventilation rate decreases (particularly with 2 h⁻¹ ACH) due to unwanted impinging turbulences.
• A wind direction independent, roof-integrated ventilation inlet structure is developed for a DD PACS. Through significant upsizing, the wind catchers’ deflector walls (in this study by 211%) by simultaneously keeping the same inlet opening dimensions, improvements in ventilation rate occur when the wind incident angle to the deflector surfaces are high, i.e., in this study a significant increase of ACH is only in certain wind directions (almost 100% vent increase in the diagonal direction). If the capturing surfaces are in a lower incident angle with the coming wind, counterproductive turbulences enforce reduced ACH performance. In order to gain improvements in all directions, the enlarged deflectors’ lower part should be trimmed back to the initial inlet opening edge. In this way, significant improvement could be achieved in all wind directions (in this study, a mean increase of approx. 1 ACH is achieved).
• Even with enlarged and optimized capturing deflector walls, the shape independent vent behavior of wind catcher versions remains. The ‘parabola + 1 m’ version had only a slightly better mean performance than the ‘empty + 1 m’ wind catcher versions (2.1%), but the small improvement cannot possibly compensate the difficulty of the more complicated construction.
• Even if dimensions of a wind catcher are specified by general experiences, and those should positively influence the performance, sometimes unexpected behavior can occur, because aerodynamic systems react in a complex way to design changes. This is in accordance with the conclusions of Varela-Boydo et al. [30] and Zarmehr et al. [17]. To fundamentally understand the pressure difference-driven PACS, a wide and thorough knowledge is essential, where the parameters are investigated organically.

6. Outlook

The ‘Parabola + modified deflector’ version was chosen by the project’s contractor as the most efficient alternative among the tested inlet structure shapes. Though this version provides a marginal improvement contrary to the ‘empty + modified deflector’, the more characteristic architectural appearance was preferred as an integral part of the company’s corporate identity. Figure 19 contains photos of the wind catcher under construction in the prototype winery building that will be accessible for long-term monitoring, including in situ, real-world comfort and energy measurements. The airflow velocities will be measured via a constant temperature anemometer (CTA). The monitored building is intended to serve as a validation testbed for the CFD simulations, as well as to ensure a well-functioning operation of the PACS.

For future research, a challenge will be the complex investigation of geometry parameters where the dimensions are not separately tested, but their permutations are tested as well. The difficulties come from the numerous generated variations. If only a simple wind catcher is taken for examination, at least four basic parameters must be calculated (e.g., height, width, tower section separation by deflectors, geometrical shape of the inlet surface, etc.), and for the acceptable fine trends, a further three to four variations of each parameter should be calculated. It means that a simple series of CFD simulations start with a minimum of $4^4$ = 256 cases, consuming a huge amount of calculation time. Despite the amount of permutations, in order to find the financial and energy efficient wind catcher version for a large variety of situations, the mapping of the aerodynamic optimum is indispensable in the future.

Valuable information is missing in the field about the topology’s influence on IAQ values. PACSs are intended to improve the indoor comfort of our spaces. Therefore, the performance of a wind catcher should not only be measured by volume flow, but other aspects, such as temperature, humidity, draft effect, contaminants, etc. should also be considered. It is necessary to analyze these variables of the scientifically selected geometries in order to locate difficulties and suggest new modifications if needed, to avoid local discomfort.

In the base model of this research, the stale air was exhausted by a ‘Venturi’ tower, with help of a so-called Venturi plate that was located on top of the tower. It created a low-pressure zone just above the outlet’s opening, which helped to extract exhaust air from the interiors. A further investigation could also provide valuable information about the use of the Venturi-effect in PACS, specifically because it is a less recognized tool from NV engineers.

The above-mentioned geometry-optimizing future research tasks should generate a generic dataset system of shape dimension parameters, in order for engineers to enable an efficient ventilation design.