Pedestrian-Level Cooling Performance Index and Climate-Adaptive Layout of Tropical Commercial Plots Under Constant Development Intensity

1. Introduction

With the acceleration of global urbanization, high-density urban development has brought about economic agglomeration effects while simultaneously triggering numerous environmental concerns. Wind, a critical natural element governing urban microclimates, plays a pivotal role in pollutant dispersion, thermal comfort regulation, and ultimately, public health and the functional quality of urban spaces [1]. Current scholarship has investigated this subject from multiple dimensions, with research trajectories primarily categorized into three areas: the interplay between wind environment and building layout, its role in thermal mitigation, and its impact on pedestrian-level comfort.

Early explorations into the wind environment–building layout nexus primarily employed wind tunnel experiments, focusing on the aerodynamic effects of isolated building forms. Stathopoulos (1985) demonstrated through wind tunnel tests that corner-cut designs in high-rise buildings can significantly reduce the area of high-wind zones at ground level, with this effect becoming more pronounced as building height increases [2]. The advent of computational fluid dynamics (CFD) in the 21st century shifted the focus toward simulations of complex building clusters. Hu and Wang (2005) observed that while CFD simulations align well with wind tunnel results under conditions of regular layouts and uniform heights, predictive accuracy diminishes with significant height variations, underscoring the methodological challenges posed by complex urban morphologies [3]. Using Large Eddy Simulation (LES), Aristodemou et al. (2018) revealed that high-rise clusters in London drastically alter surrounding airflow structures, creating pollutant concentration zones with limited dispersion [4]. Extending the analysis to the vertical dimension, Lu et al. (2022) found that building height distribution patterns significantly affect vertical wind speed and pressure, indicating that uniform-height layouts are more conducive to overall wind comfort [5]. From a design perspective, Guo et al. (2022) compared Chinese Feng Shui layouts with modified layouts at Prince Kung’s Palace in Beijing, validating that nature-responsive building arrangements significantly optimize courtyard ventilation during summer [6]. Furthermore, Zhao et al. (2023) investigated non-enclosed atrium spaces, uncovering how indoor morphology influences airflow penetration and temperature distribution [7]. Collectively, these studies trace an evolution from static analyses of solitary forms to systematic, dynamic investigations of complex urban fabrics and their vertical ventilation effects. Nevertheless, a significant lacuna remains in the literature: the characterization of wind environments across different land-use types and a systematic consideration of urban adaptation strategies for tropical regions.

The link between the wind environment, urban cooling, and the heat island effect is a core issue of sustained academic interest. At the macro-scale, Ngarambe et al. (2021), based on nine years of meteorological observation data from Seoul, quantitatively revealed the significant impact of wind speed on urban heat island intensity, finding that increased wind speeds lead to a decrease in intensity, which is notably higher in dense building areas than in sparse ones [8]. At the micro-scale, Davtalab et al. (2020) demonstrated in the context of Iran that areas with vegetation cover exhibit significant reductions in air temperature, mean radiant temperature, and physiological equivalent temperature compared to non-vegetated areas, indicating complex interactions between wind, vegetation, and the underlying surface [9]. Research by Chan and Chau (2021) on urban parks in Hong Kong further revealed pronounced seasonal variations in this relationship; factors such as surrounding building layout and park size were found to be more critical for thermal comfort in summer, while building spacing played a more significant role in winter [10]. Similarly, Deng and Wong (2020) systematically analyzed the impact of street aspect ratios and orientations on outdoor thermal comfort through simulations of street canyon scenarios in Nanjing’s central business district, discovering a significant correlation between canyon geometry and physiological equivalent temperature [11]. At a broader urban scale,Yang et al. (2020) and Zeng et al. (2024) utilized multi-source remote sensing data and the Local Climate Zone (LCZ) framework, respectively, to verify the positive correlation between building morphology and heat island intensity from a broader perspective of urban form [12,13]. These studies collectively indicate that research on wind environments and urban cooling effects has advanced from macro-scale meteorological statistics to micro-scale local climate regulation, with an increasing focus on multi-factor coupling mechanisms. However, most existing studies focus on single-season or macro-scale analyses. Consequently, there is a dearth of empirical evidence to guide layout optimization that simultaneously addresses the dual objectives of summer cooling and winter wind control.

Pedestrian-level wind comfort is of paramount importance, as it is directly intertwined with the usability of urban public spaces. Kubota et al. (2008) conducted wind tunnel experiments across 22 residential neighborhoods in Japan, revealing a robust correlation between the building coverage ratio and the mean wind speed ratio at the pedestrian level [14]. Li et al. (2012) corroborated this in the context of South China, finding that a increase in building coverage leads to a linear decrease of 0.1 in the wind speed ratio [15]. As the field advanced, Iqbal and Chan (2016) examined the effects of spacing, orientation, and wind incidence angles within cruciform high-rise clusters, discovering that vortices formed under oblique winds can paradoxically enhance wind comfort [16]. Furthermore, Jin et al. (2017) and Ma and Chen (2020) performed systematic classifications and simulations of residential areas in Tianjin, China respectively; their findings highlighted significant variations in wind environments across different layout patterns and identified key morphological factors such as frontal area density and mean building height [17,18]. More practice-oriented, Zhang and Zhang (2021) identified appropriate building arrangements for cold-climate cities based on their findings in Shenyang [19]. Recently, Feng et al. (2022) integrated UAV-based field measurements with numerical simulations to extend the research scope into three-dimensional vertical space, identifying building height variation as the most critical factor influencing wind speed ratios across different elevations [20]. Collectively, this research domain has matured from quantitative analysis of singular morphological indicators to multi-factorial, multi-scale systemic evaluations. Yet, a significant research gap persists regarding the correlation between spatial layout configurations and potential cooling efficacy under conditions of constant development intensity.

Focusing on the tropical island city, this study investigates the impacts of building quantity and spatial enclosure forms on the pedestrian-level wind environment and potential cooling efficacy. The analysis is based on typical summer and winter wind characteristics and specifically examines commercial plots under the constraint of constant development intensity. This study constructs a Pedestrian-level Cooling Performance Index (PLCPI) by employing a combined weighting method that integrates the Analytic Hierarchy Process (AHP) and the Entropy Weight Method (EWM), thereby enabling a quantitative comparison and ranking of the cooling performance of various layout configurations. The primary contributions of this study are twofold: First, it proposes a novel dual-season adaptive evaluation framework that balances summer cooling requirements with winter wind environment control, thereby filling a regulatory gap in quantitative wind environment research for commercial plots in tropical regions. Second, it reveals a relationship between building quantity and cooling performance under the constraint of constant development intensity, identifying optimal and suboptimal layout forms to provide robust decision support for the meticulous design of urban commercial plots.

2. Study Design

2.1. Research Precondition and Research Framework

Drawing on the existing literature and the central aims of this study, this paper establishes its foundational preconditions and overall research framework (For details, see Figure 1). Prior work offers two lines of evidence supporting a positive correlation between wind speed—and its amplification factor—and pedestrian-level cooling. Direct evidence comes from Feng et al. (2022), Kubota et al. (2008), and Li et al. (2012), who, through field measurements, wind tunnel tests, and numerical simulations, have quantitatively linked building morphology to pedestrian-level wind speed ratios [14,15,20]. Indirect evidence is provided by Jin et al. (2017) and Ma and Chen (2020), who identified building density, frontal area density, and mean building height as key morphological determinants of the pedestrian wind field [17,18]. These findings implicitly validate the mechanism by which building layout regulates thermal comfort through its influence on wind speed. Collectively, the evidence indicates that higher pedestrian-level wind speeds enhance convective heat exchange at the body surface, thereby producing a cooling effect. Accordingly, the operating precondition of this study is that wind speed and its amplification factor are positively correlated with cooling performance at the standard pedestrian height of 1.5 meters.

On this basis, the study turns to an unexplored context—commercial plots in tropical regions. Under a fixed development intensity, it systematically investigates how building count (one to four units) and spatial enclosure form shape the pedestrian-level wind environment and its attendant cooling efficacy. To enable a quantitative comparison among configurations, we introduce the PLCPI. This composite index integrates summer and winter wind speeds and amplification factors through a linear weighted sum (Figure 1). The weights are determined by a hybrid approach that combines the AHP with the EWM [21,22]. This approach preserves the subjective judgment prioritizing summer cooling in tropical island cities while utilizing data dispersion to calibrate the weights, thereby ensuring scientific rigor and rationality.Wind environment simulations encompassing 65 layout scenarios—varying building counts and enclosure types—were performed using the Vent2026 software (Version 20260303(SP1)).

2.2. Case Study Selection

Haikou sits at the northern edge of Hainan Island, China’s southernmost province, on the low-latitude tropical fringe. The city experiences a classic tropical maritime monsoon climate, marked by year-round heat and abundant rainfall, with long, intensely hot, and humid summers. The annual mean temperature hovers between 23 °C to 28 °C, yielding generally poor outdoor thermal comfort [23]. These hygrothermal conditions create a pressing demand for ventilation-driven cooling in outdoor spaces, making Haikou an ideal setting in which to examine how building layout can harness wind for thermal relief.

As a tropical island city undergoing accelerated urbanization, Haikou has seen sustained increases in the development intensity of its commercial plots. Under such land constraints, how to improve the outdoor wind environment through refined building layout has become a tangible challenge for local urban design and planning practice. Choosing Haikou as the case study provides a realistic climatic context and a validation setting for the PLCPI model proposed in this research; its findings can also serve as a reference for commercial plot development in climatically similar regions. Figure 2 presents the geographical location of Haikou, the immediate surroundings of the site, and the key development parameters. The site is bordered by undeveloped land, so the influence of neighboring buildings on the layout under study can be excluded (Figure 2(b)). The plot has a floor area ratio of 1.6, a site area of 2,500 m², a building coverage of 40%, and a green space ratio of 40% (Figure 2(c)).

This study operates under the precondition of constant development intensity. That is, key indicators—including building storeys, height, total floor area, building coverage, a minimum fire-separation distance of 6 meters, floor area ratio, and green space ratio—are held fixed. Within these constraints, the permitted total construction area is divided into one to four building blocks, following the standard dimensional conventions of commercial land use (for details, see Figure 3). This subdivision strategy respects the real-world restrictions of high-density urban development while capturing the spatial diversity achievable within a limited repertoire of layout types.

The building configurations on the site are classified into four groups, labeled A through D, with numerical suffixes distinguishing the layout variants within each group (for details, see Figure 3); for instance, B-2 denotes the second layout variant for the two-building condition. The designated layouts also fall into four typological categories: detached layout, row layout, perimeter layout, and courtyard layout. The detailed classification is as follows:

Group A: One building on site, comprising 4 layout variants (A-1 to A-4). All four are detached layouts.

Group B: Two buildings on site, comprising 31 layout variants (B-1 to B-31). Among these, B-1 to B-19 are row layouts, and B-20 to B-31 are perimeter layouts.

Group C: Three buildings on site, comprising 28 layout variants (C-1 to C-28). Among these, C-1 to C-16 are row layouts, and C-17 to C-28 are courtyard layouts.

Group D: Four buildings on site, comprising 2 layout variants (D-1 and D-2). Both are perimeter layouts.

2.3. Data Collection

In urban microclimate research, the pedestrian-level wind environment is a core determinant of outdoor thermal comfort and cooling efficacy. For tropical island cities, the combination of high temperatures, high humidity, and intense solar radiation makes natural ventilation the principal means of improving outdoor thermal conditions. During summer, elevated wind speeds around buildings effectively remove heat and moisture from the body surface, markedly improving thermal comfort; conversely, excessive wind speeds in winter can cause wind discomfort, discouraging the use of outdoor spaces. Building cluster design must therefore reconcile summer cooling demand with winter wind comfort, pursuing an integrated optimization that is adaptive to both seasons.

Accordingly, this study adopts building count and spatial enclosure form as independent variables, and selects maximum pedestrian-level wind speeds and their amplification factors in summer and winter as the core evaluation metrics. Through a multi-indicator weighted synthesis, these metrics capture the integrated outdoor cooling performance of alternative spatial layouts under tropical island climatic conditions (For details, see Table 1).

The mesh resolution in wind environment simulation directly affects computational accuracy and efficiency: an overly fine mesh incurs prohibitive computational cost, whereas an overly coarse mesh compromises precision. A rational meshing strategy therefore applies differentiated settings to distinct regions of the computational domain. The refinement scheme adopted in this study is detailed in Table 2. The near-field ground mesh is refined with a subdivision level of 4.000m, while the far-field ground mesh remains coarser at 8.000m. Additionally, informed by the case study context, the power-law exponent for the wind speed gradient profile is set to 0.16, with a gradient wind height of 350m. The standard k-ε turbulence model is employed for the numerical simulation of the flow field.

According to the wind environment data provided by the Design Code for Heating, Ventilation and Air Conditioning of Civil Buildings (GB50736-2012), the prevailing summer wind in Haikou is southerly at an average speed of 2.700 m/s, while the winter wind is predominantly northeasterly at 3.100 m/s (for details, see Table 3) [24]. All subsequent wind environment calculations and simulation analyses for each layout configuration are based on the baseline values presented in Table 3.

3. Model Construction and Weight Calibration

3.1. Model Construction

Computational Fluid Dynamics (CFD) is adopted in this study to resolve the wind field (equation 1). Within the computational domain, the governing equations for mass, momentum, and energy conservation are established and solved using the SIMPLE algorithm as follows:

$${\displaystyle \frac{\partial \left(\rho \varphi \right)}{\partial t}}+div\left(\rho U\varphi \right)=div\left({\Gamma }_{\varphi }grad\varphi \right)+S_{\varphi }$$

(1)

To quantify the impact of alternative building spatial configurations on pedestrian-level wind speed and cooling efficacy, the PLCPI is constructed. Its formulation follows three principles: summer cooling takes precedence, winter wind environment is controlled, and multiple indicators are synthesized. In tropical island cities, the summer season is prolonged and intensely hot, and outdoor thermal comfort is highly sensitive to wind speed. Accordingly, summer wind speed and its amplification factor are assigned dominant weights in the index. Although the demand for cooling subsides in winter, winter wind speed and its amplification factor are assigned lower weights to avert discomfort from strong winds. This weighting embodies a dual-season adaptive design logic—promoting ventilation in summer while shielding against strong wind in winter. A linear weighted synthesis method integrates four indicators—maximum summer wind speed, summer amplification factor, maximum winter wind speed, and winter amplification factor—into a single composite index that enables quantitative comparison and ranking of cooling performance across layout configurations:

$${PLCPI}_i={\alpha }_1{V}_{s,max}+{\alpha }_2{R}_{s,max}+{\alpha }_3{V}_{w,max}+{\alpha }_4{R}_{w,max}$$

(2)

$$\sum {\alpha }_i=1,{\alpha }_1,$$

(3)

In equation (2) and (3), PLCPI_i is the Pedestrian-level Cooling Performance Index for layout type i; V_s,max is the maximum wind speed at pedestrian level during summer (m/s); R_s,max is the maximum wind speed amplification factor at pedestrian level during summer; V_w,max is the maximum wind speed at pedestrian level during winter (m/s); R_w,max is the maximum wind speed amplification factor at pedestrian level during winter; α₁ to α₄ are the respective weighting coefficients.

3.2. Calibration of Comprehensive Weight

Determining the weighting coefficients is a critical step in constructing the PLCPI model. To ensure scientific rigor and robustness, this study adopts a combined weighting strategy that couples the AHP with the EWM. This approach integrates professional judgment—which reflects the specific cooling demands of tropical island cities—with the intrinsic information entropy of the simulated data to calibrate the final weights.

Within the AHP 1–9 scale framework, the judgment matrix was logically assigned based on the core climatic characteristic of tropical island cities: strong cooling demand in summer and weak cooling demand in winter. Taking summer wind speed as the reference (assigned a value of 1), it was judged “slightly more important” (value 2) relative to the summer amplification factor due to its direct dominance over ventilation efficiency. Relative to winter wind speed, it was judged “demonstrably more important” (value 6) because summer demand significantly outweighs winter demand. Compared with the winter amplification factor, it was judged “very strongly more important” (value 7). Based on these pairwise comparisons, the judgment matrix was constructed as shown in Table 4.

Indicator weights were calculated using the geometric mean method, and a consistency check was performed via the maximum eigenvalue. The results yielded a λmax of 4.003, a Consistency Index (CI) of 0.001, and a Consistency Ratio (CR) of 0.001 (<0.100), confirming that the judgment matrix satisfies consistency requirements and that the subjective weights are reasonable. The AHP-derived subjective weights are summarized in Table 5. Concurrently, the Entropy Weight Method was applied for objective weighting, where the weight of each indicator is determined by its degree of variation: greater variation implies greater information content and, consequently, a higher weight.

Preliminary data normalization was performed on the dataset. With m layout samples (m=65) and n evaluation indicators (n=4), an original data matrix was established. Because all indicators are positive (higher values indicate better cooling performance), min–max normalization was applied. Based on the quantified sample data, the objective weights for each indicator were calculated and are presented in Table 6. The objective weights reveal that summer indicators—particularly the wind speed amplification factor—exhibit higher data dispersion and carry greater information, which is consistent with empirical observations that diverse layouts markedly influence the summer wind environment.

Comprehensive weights were derived by integrating the subjective and objective weights using a multiplicative synthesis method (Table 7). In the final weight combination, summer wind speed and its amplification factor receive weights of 0.558 and 0.252, respectively, yielding a combined weight of 0.810. This aligns with the design priority of summer cooling in tropical island cities. The two winter indicators account for a combined weight of only 0.190, ensuring that winter wind conditions remain manageable without distorting the overall cooling assessment. The summer amplification factor carries a lower weight (0.252) than the summer maximum wind speed (0.558). Given that the amplification factor is a dimensionless ratio, it complements the absolute wind speed within the PLCPI index, and together they provide a robust representation of ventilation and potential cooling efficiency.

4. Results

4.1. Pedestrian-Level Wind Speed

All 65 building spatial layout configurations were imported into the Vent2026 for analysis, yielding pedestrian-level wind speed data for both summer and winter (Figure 4 and Figure 5). Across the 65 configurations, the sample mean of maximum pedestrian-level wind speed in summer is 2.124 m/s (blue dashed line in Figure 4), while the winter counterpart is 2.702 m/s (green dashed line in Figure 5).

The sample maximum of summer pedestrian-level wind speed is 2.640 m/s, recorded for layout B-7 (Group B, row layout). Among the 65 configurations, 41 exceed the sample mean in summer. The sample minimum is 0.630 m/s, observed for layout B-5 (Group B, row layout), yielding a range of 2.010 m/s from the maximum. Additionally, layout B-1 (Group B, row layout) registers a notably low summer value of 1.010 m/s, differing from the maximum by 1.630 m/s.

Figure 5 presents the simulated maximum winter pedestrian-level wind speeds for all 65 configurations. The sample maximum reaches 3.100 m/s, attained by layout C-17 (Group C, courtyard layout). Of the 65 configurations, 38 surpass the sample mean in winter. The sample minimum is 1.390 m/s, corresponding to layout C-10 (Group C, courtyard layout).

When examined by group and layout type, Group A’s detached layout yields the highest within-group means for both seasons—2.459 m/s in summer and 2.915 m/s in winter—whereas Group C’s row layout records the lowest means, 1.798 m/s and 2.511 m/s, respectively (Table 8). Within Group A, only layout A-2 exceeds the group mean in summer (2.630 m/s), while layouts A-1 (3.050 m/s) and A-4 (3.060 m/s) surpass the winter mean.

According to the simulation results summarized in Table 8, the perimeter layout in Group B produces higher within-group mean wind speeds in both summer (2.334 m/s) and winter (2.776 m/s) than the row layout (2.290 m/s and 2.726 m/s, respectively). This indicates that, under the two-building condition, the perimeter layout outperforms the row layout in terms of wind speed and potential pedestrian-level cooling.

Similarly, within Group C, the courtyard layout yields higher within-group mean wind speeds (summer: 1.908 m/s; winter: 2.776 m/s) than the row layout (1.798 m/s; 2.511 m/s), suggesting superior ventilation and cooling potential under the three-building condition. Group D, which consists solely of a perimeter layout, records within-group means of 2.440 m/s in summer and 2.700 m/s in winter. Notably, the summer mean trails only that of Group A’s detached layout, with a marginal difference of merely 0.019 m/s.

4.2. Pedestrian-Level Wind Speed Amplification Factor

Overall, the sample mean of the maximum summer wind speed amplification factor across all 65 configurations is 1.066, compared with 1.181 in winter, indicating that zones of elevated wind speed in winter offer greater potential cooling benefit than those in summer.

Figure 6 displays the simulated summer amplification factors. The sample maximum of 1.320 is shared by five configurations: A-2 (Group A, detached layout) and B-7, B-8, B-16, and B-17 (Group B, row layout). A total of 41 configurations exceed the summer sample mean.

The sample minimum of the summer amplification factor is 0.320, recorded for layout B-5 (Group B, row layout), producing a range of 1.000. Layout B-1 (Group B, row layout) also falls well below the mean at 0.510, a deviation of 0.810.

The winter amplification factor (Figure 7) attains a sample maximum of 1.360 at layout C-17 (Group C, courtyard layout). Of the 65 configurations, 36 surpass the winter sample mean. The sample minimum of 0.610 occurs at layout C-10 (Group C, courtyard layout).

Table 9 presents the within-group means of the amplification factor by season. Group A’s detached layout exhibits the highest values (summer:1.253; winter:1.275), while Group C’s row layout registers the lowest (summer:0.903; winter:1.097). Within Group A, only layout A-2 surpasses the group mean in summer (1.320), whereas layouts A-1 (1.330) and A-4 (1.340) do so in winter. Based on Table 9, Group B’s perimeter layout again outperforms the row layout, with higher within-group mean amplification factors in both summer (1.173vs.1.151) and winter (1.213vs.1.192), further corroborating the enhanced ventilation and cooling potential of the perimeter configuration under the two-building condition.

4.3. Quantification Results of PLCPI

Based on the PLCPI model in equation (2) and the comprehensive weights (Table 7), the indicator data presented in Figure 4 to 7 were substituted to calculate the PLCPI values for all 65 layout configurations. The quantitative results are displayed in Figure 8.

In Figure 8, Group A (one building, four detached layouts) yields PLCPI values between 2.301 and 2.419. Layout A-2 achieves the highest value (2.419), while A-3 records the lowest (2.301). This indicates that even under the single-building condition, differences in building form and orientation measurably affect the pedestrian-level wind environment.

Group B (two buildings, 31 layouts: 19 row and 12 perimeter) exhibits the widest PLCPI range among the four groups, ranging from 1.115 to 2.456. Layout B-8 attains the highest value (2.456)—the overall optimum—whereas B-5 registers the lowest (1.115), the poorest performance. The within-group range of 1.341 underscores that, under the two-building condition, the spatial enclosure form directly governs ventilation efficiency and exerts a decisive influence on cooling performance.

Group C (three buildings, 28 layouts: 16 row and 12 courtyard) produces PLCPI values between 1.184 and 2.375. The maximum occurs at C-17 (2.375) and the minimum at C-22 (1.184). Overall, Group C’s PLCPI values are generally low; even its best-performing layout, C-17, is surpassed by numerous configurations in Groups A and B. This suggests that the three-building condition poses greater challenges for ventilation organization, constraining overall cooling performance.

Group D (four buildings, two perimeter layouts) yields PLCPI values of 2.297 (D-1) and 2.244 (D-2), which are closely comparable, with D-1 marginally superior. Despite the limited number of layout variants, both configurations exhibit stable cooling performance with negligible within-group variation.

Averaging the PLCPI values of all 65 layouts by group yields the group-mean PLCPI (Table 10), which reflects the overall cooling performance at different building counts. The results reveal a clear non-linear relationship between building quantity and mean PLCPI. Group A (one building) achieves the highest group mean (2.367). When the count increases to two, the means of Group B’s row and perimeter layouts decline to 2.187 and 2.228, respectively. At three buildings, Group C’s means drop to the lowest values among all groups (row layout: 1.825; courtyard layout: 1.968). However, at four buildings, Group D’s mean rebounds to 2.271—notably higher than Group C and even slightly above Group B. This non-monotonic trend demonstrates that adding buildings does not inevitably degrade cooling performance: the three-building condition marks a performance trough, while the four-building condition outperforms both the two- and three-building scenarios. Under the constraint of constant development intensity, spatial enclosure form and ventilation organization thus emerge as more critical determinants of cooling performance than building count alone.

Group A’s PLCPI mean (2.367) is the highest among the four groups, indicating that the fewest buildings achieve the best cooling performance. This advantage stems from the fact that the wind environment around a single building is governed primarily by its own form and orientation, free from inter-building sheltering and interference, and hence simpler to organize and control.

The PLCPI means of Group B’s row (2.187) and perimeter (2.228) layouts rank third overall. Although the means are moderate, the extreme within-group variation highlights the high sensitivity of cooling performance to layout configuration under the two-building condition.

Group C’s PLCPI means are the lowest among all groups, reflecting the weakest overall cooling performance. This suggests that, under the three-building condition, spatial enclosures readily form semi-enclosed or fully enclosed structures that severely impede natural ventilation in summer.

Group D’s PLCPI mean (2.271), derived from perimeter layouts, exhibits a narrow range and high stability due to the limited number of layout variants. This indicates that, with four buildings, the spatial organization converges toward a stable ventilation structure, thereby attaining better average cooling performance than Groups B and C.

5. Conclusions

Focusing on the tropical island city of Haikou, this study addresses the optimization of building layouts for commercial plots under the constraint of constant development intensity. A Pedestrian-level Cooling Performance Index was constructed, and a combined weighting method integrating the AHP and the EWM was employed to quantitatively evaluate the effects of 65 spatial layout configurations on the pedestrian-level wind environment and cooling efficacy.

The principal findings are as follows:

First, when considering single-season simulation data in isolation, the best summer cooling performance is achieved by layout B-7, a row layout, with a maximum summer wind speed of 2.640 m/s and a maximum summer amplification factor of 1.320. In winter, the best performance is observed for layout C-17, a courtyard layout, where the maximum winter wind speed reaches 3.100 m/s and the amplification factor attains 1.360.

Second, no linear correlation exists between the number of building units and cooling performance. Group A’s single-building detached layout yields the highest mean PLCPI of 2.367, representing the optimal spatial subdivision mode. Group C’s three-building row layout records the lowest mean PLCPI of 1.825 and should be adopted with caution. Group D’s four-building perimeter layout sees the mean PLCPI rebound to 2.271, demonstrating high stability.

Third, when the building count is held constant, the spatial enclosure form is the decisive determinant of cooling performance. Group B (two buildings) exhibits the widest internal PLCPI range: the best layout, B-8, reaches a PLCPI of 2.456, surpassing the single-building optimum A-2 (2.419), while the poorest layout, B-5, records only 1.115. This confirms that, through precise spatial organization, dispersed layouts can outperform even a centralized single building in cooling efficacy.

This study makes two primary contributions. First, it proposes a dual-season adaptive evaluation framework that balances summer cooling with winter wind control, thereby filling a research gap in the quantitative wind environment analysis of commercial plots in tropical regions. Second, it reveals a non-linear relationship between building count and cooling performance under constant development intensity, identifies optimal and suboptimal layout forms suited to tropical island climates, and provides a scientific basis for the meticulous design of urban commercial plots in Haikou.

This study is subject to certain limitations. The experimental design is based on idealized site boundaries and assumes constant development intensity, without accounting for real-world complexities such as surrounding building shading or terrain variations. Future research could incorporate a broader set of realistic urban environmental variables and utilize field measurement data to further validate and calibrate the PLCPI model.