This section outlines the comprehensive methodology employed to simulate aerosol dispersion and evaluate ventilation strategies in an indoor office environment. The study begins with a description of the simulation geometry and domain setup, including diffuser configurations and occupant modeling. It then details the numerical methods and simulation parameters used to solve airflow and particle transport using an Eulerian–Lagrangian approach. Subsequent subsections present the mesh generation and refinement strategy, grid independence analysis using Richardson Extrapolation and GCI, and the governing equations of the continuous phase (air) and discrete phase (droplets). The particle injection modeling, including droplet sizing, thermal properties, evaporation dynamics, and turbulence dispersion, is discussed in detail. The section concludes with validation procedures comparing simulation outputs to established literature and experimental benchmarks. Together, these elements ensure the robustness, accuracy, and physical fidelity of the simulation framework.

Simulation geometry and setup

The simulation will take place in a workplace located in Alexandria, Egypt with an already tested and published research for airflow in various air ceiling diffusers⁹. The physical model used in this study represents office space with specific dimensions. The office has a height of 3 m (y), a width of 5 m (x), and a length of 5.12 m (z) (Fig. 1) provides a three-dimensional visual representation of the room being analyzed, illustrating the positions of employee mannequins within the space. The human sitting mannequin is 1.5 m sitting height with a 320 mm² mouth opening, the room setting as illustrated in showing two sitting human bodies. The coughing personnel will be sitting in the room marked as the cough source (Fig. 1). In this model, the square air diffuser was used in simulation (Fig. 1). The model development and simulation were carried out using ANSYS Fluent (Version 19.1), a widely validated computational tool capable of performing both qualitative and quantitative analyses of airflow and heat transfer in enclosed environments. The existing HVAC system⁹ served as the foundation for this model. The three-dimensional (3D) geometry of the model was created using ANSYS-Design Modeler, providing precise control over the placement of diffusers. The numerical setup in this study is adapted for two different air conditioning configurations, aiming to analyze how airflow patterns influence the transport and dispersion of cough-generated droplets within an office environment, and their potential role in increasing infection risk. To examine so, the Exhaust diffusers will be replaced by supply diffusers and vice versa as shown in Fig. 2, contrasting the effect of the parallel exhaust diffusers in the center of the workplace (Case # 1) to the location of the exhaust diffuser above the staff sitting in the office (Case # 2).

Numerical setup and simulation parameters

To increase the study’s comprehensiveness, researchers assess exposure dangers using aerosol concentration as a parameter. Indoor airflow and droplet transmission are simulated using an Euler-Lagrangian model. While the droplets that people exhale are seen to be separate phases, the air is thought to be a continuous phase. The simulation technique may be loosely separated into two primary parts. The Lagrangian model is used to predict the trajectory of cough droplets after the time-dependent Euler model has solved interior flow fields. Moreover, this simulation anticipates droplet evaporation by considering mass and heat transport between droplets and the surrounding air^25,26.

ANSYS Meshing was employed to generate the computational grid and proved to be a reliable tool for achieving high-quality mesh control. The mesh (Fig. 3) was constructed based on the actual geometric proportions of the office layout, as depicted in Fig. 1 ensuring geometric fidelity between the simulation domain and the real-world environment.

To improve local accuracy and better resolve flow features, targeted mesh refinement was applied in critical regions, as shown in Fig. 4, including:

• The mouth opening of the coughing mannequin, where sharp velocity gradients and droplet injection occur.

• The vicinity of supply and exhaust diffusers, where directional airflow and turbulence are most prominent.

• The surface of the human body, where recirculation zones and droplet deposition are likely.

These refinements allowed for accurate resolution of airflow and droplet dynamics in key areas of interest, thereby enhancing the physical realism and reliability of the simulation results.

ANSYS Fluent was utilized to perform the simulations based on the Eulerian–Lagrangian framework, due to its comprehensive suite of boundary condition options and advanced physics models capable of accurately representing heat transfer and fluid flow. The simulation setup included the specification of velocity inlets, pressure outlets, and wall boundaries, along with the definition of fluid properties such as viscosity, density, and thermal conductivity. All enclosing surfaces, walls, ceiling, and floor were modeled as adiabatic boundaries, implying that no heat flux was exchanged through these surfaces. This assumption allowed the simulation to maintain thermally isolated conditions, facilitating a more controlled assessment of airflow behavior and droplet dispersion within the indoor environment.

Boundary conditions were retrieved from the previously published work by authors⁹ to set up simulation in the ANSYS-FLUENT application, the conditioned fresh air was introduced to the supply grilles as 1 m/s inlet velocity and 22 °C to as a low-speed setting for the air conditioning unit. The exhaust had an outflow zone set. The human body’s surface temperature, measured at 32 °C²⁷. At the supply diffusers (velocity inlets), turbulence intensity and hydraulic diameter were specified as 5% and 0.5 m, respectively. Exhaust diffusers (Pressure outlets) were treated with zero-gauge backpressure and backflow turbulence intensity of 5%. All solid surfaces were treated as no-slip, adiabatic walls with enhanced near-wall treatment applied to resolve boundary layer effects.

To ensure accurate near-wall treatment, the mesh was refined such that the dimensionless wall distance (y⁺) values remained predominantly in the range of 1 < y⁺ < 5, which is within the recommended bounds for the use of enhanced wall treatment²⁸ in combination with the standard k-ε turbulence model. This resolution allows the solver to adequately resolve the viscous sublayer and ensures fidelity in capturing near-wall flow phenomena. The second-order upwind scheme is employed to discretize the equations related to momentum, energy, k, and \(\varepsilon\). To accurately capture the pressure gradient at the boundary, the “PRESTO!” scheme is used to discretize the pressure term. The algebraic equations are solved using a coupled algorithm.

The transient simulation was conducted over a duration of 10 s, which was selected to capture the initial dispersion phase of respiratory droplets following a cough event, an interval known to encompass the highest exposure risk due to concentrated particle release and dominant airflow effects. This approach is consistent with similar transient CFD studies in the literature^29,30, which analyzed airborne droplet dynamics within comparable timeframes. While longer simulations may provide insights into long-term mixing or viral decay, the present study focuses on early-stage aerosol transport and ventilation influence, which are critical for evaluating exposure risk and mitigation strategies.

A time step of 0.01 s was selected to adequately resolve the highly transient dynamics of the cough jet and droplet injection, which exhibit rapid acceleration and turbulent mixing within the first second of release. This time resolution ensures accurate capture of flow development and particle trajectories without compromising numerical stability. The selected time step is consistent with prior studies^29,31 on respiratory events, which employed similar temporal resolutions for modeling cough-induced airflow and aerosol dispersion. A time step sensitivity check was conducted by comparing simulations with values of 0.005 s, 0.01 s, and 0.02 s. The results showed less than 2% variation in key parameters such as peak cough jet velocity and particle travel distance, confirming that the selected time step of 0.01 s offers an appropriate balance between temporal accuracy and computational efficiency.

The convergence criterion for the simulation is defined as the energy relative residual reaching 10^–7 and the relative residuals of other variables reaching 10^–4. Convergence is typically achieved with 20 iterations for each timestep.

Mesh sensitivity and grid independence analysis

To evaluate numerical accuracy and eliminate discretization-related uncertainties, a mesh sensitivity analysis was performed using three systematically refined grids comprising approximately 0.3 million, 1.2 million, and 4.8 million hexahedral cells. These grids were generated with consistent refinement ratios to maintain geometric and topological integrity across simulations. The analysis was centered on a representative monitored quantity specifically, the peak velocity magnitude in the vicinity of the cough jet source chosen for its sensitivity to mesh resolution. The mesh sensitivity of the numerical solution was assessed using the Richardson Extrapolation method and the Grid Convergence Index (GCI), following the guidelines proposed by Roache (1999)³², to ensure the reliability and mesh-independence of the simulation results.

The peak velocity results obtained were 16.3 m/s, 16.0 m/s, and 15.9 m/s for the coarse, medium, and fine meshes, respectively. The apparent order of convergence (\(p\)) was calculated as 0.79, using Richardson Extrapolation:

where \({\phi }_{1}\), \({\phi }_{2}\), and \({\phi }_{3}\) represent the monitored quantity (peak velocity) on the fine, medium, and coarse grids, respectively, and \(r\) is the uniform grid refinement ratio. Once the order of convergence was obtained, the Richardson Extrapolated Value \({\phi }_{ext}\) was computed as:

This provided an estimate of the solution on an infinitely fine grid, yielding \({\phi }_{ext}\) = 15.85 m/s. Furthermore, the Grid Convergence Index (GCI) was calculated for the finest mesh as:

The resulting GCI value of 0.39% confirms that the numerical solution is mesh-independent and within an acceptable margin of discretization error for engineering accuracy.

Based on this evaluation, the 4.8 million-cell mesh was selected for all subsequent simulations, as it offered a balance between computational cost and solution fidelity. Figure 5 presents the convergence behavior of peak velocity as a function of mesh resolution, along with the extrapolated reference value.

Euler model (air model)

As mentioned earlier, the simulation process is divided into two phases. The first phase is the flow of air model. In this study, the standard k-ε turbulence model was utilized to simulate the airflow field within the computational domain along with the enhanced wall treatment³³. The Reynolds-Averaged Navier–Stokes (RANS) equations were employed to model the mean flow behavior, while turbulence effects were captured using the two-equation k-ε model³⁴, which has been extensively validated for applications in indoor airflows and ventilation studies. The governing equations for incompressible turbulent flow are given as follows:

• The continuity equation:

  $$\frac{\partial \rho }{{\partial t}} + \nabla \cdot \left( {\rho {\text{u}}} \right) = 0$$

  (4)

• The momentum equation:

  $$\frac{{\partial \left( {\rho {\text{u}}} \right)}}{\partial t} + \nabla \cdot \left( {\rho {\text{uu}}} \right) = – \nabla P + \nabla \cdot \left( {\mu_{{{\text{eff}}}} \nabla {\text{u}}} \right) + \rho {\text{g}}$$

  (5)

where \(\rho\) is air density, \(\text{u}\) is velocity, \(P\) is pressure, and \({\mu }_{\text{eff}}\)​ is the effective viscosity.

The standard k-ε model introduces two additional transport Eqs. ³⁴ to model turbulence:

where G_k and G_b are the kinetic energy of turbulence produced by the mean velocity gradients and buoyancy and Y_M The variable dilatation of compressible turbulence contributes to the total dissipation rate. As for G_1ε, G_2ε and G_3ε considered as constants.

The standard k-ε turbulence model was selected in this study due to its well-documented applicability in indoor airflow modeling and its proven performance in mechanically ventilated environments. This model offers a computationally efficient means of resolving large-scale turbulent structures, making it particularly suitable for room-scale simulations of airflow and aerosol dispersion. Given the study’s focus on evaluating ventilation configurations and the transport of airborne particles in a high-flow indoor setting, the k-ε model provides a practical balance between numerical accuracy and computational cost. Previous studies^35,36,37,38 have shown that the k-ε model performs adequately in predicting airflow distribution and contaminant transport under similar conditions. Moreover, the validation of airflow fields and droplet dispersion patterns in “HVAC and cough particle validation” section confirms that the simulated results align well with experimental observations, supporting the reliability of the turbulence modeling approach used.

Nevertheless, we acknowledge that alternative Reynolds-Averaged Navier–Stokes (RANS) models may offer enhanced accuracy, particularly for capturing flow separation, near-wall effects, and anisotropic turbulence. Models such as the realizable k-ε, RNG k-ε, and especially the SST k-ω have been shown to outperform the standard k-ε in complex geometries or scenarios involving particle deposition. For instance, Gao and Niu³⁹ demonstrated that SST k-ω yields more accurate predictions of particle deposition in turbulent channel flows. Similarly, Zabihi et al.⁴⁰ successfully employed Euler–Lagrange URANS simulations to capture the transient dispersion of aerosols indoors, highlighting the benefit of refined turbulence closure models. Ai and Melikov⁴¹ further advocate for advanced turbulence modeling when studying exposure in proximity to contaminant sources. While the standard k-ε model remains appropriate for the current room-scale evaluation in a mechanically mixed environment, future work may incorporate these advanced models to better resolve localized airflow phenomena, near-wall dynamics, and model sensitivity in more complex or stratified indoor conditions.

Particle transportation and sizing (cough model)

As for the lagrangian model (cough model), the focus is on examining the droplets released during coughing. Data related to cough measurements were adapted from Experimental benchmark by Duguid⁴², that conducted measurements of the size distribution of cough droplets using a microscope. It was reported that the velocity of the expelled airflow during coughing is approximated by combining multiple sine functions, based on experimental data obtained from a spirometer that utilized a Fleish type pneumotachograph. The maximum velocity of the airflow is recorded as 15.9 m/s.

Droplets released by coughing range in size from 1 to 1000 μm. We make use of two Rosin–Rammler distributions to fit the droplet diameter distributions of 1 to 50 μm and 50 to 1000 μm (Fig. 6), respectively, due to the variation of cough droplets diameter. \({Y}_{d}\) is labelled in the Rosin–Rammler distribution function as the mass fraction of droplets having a diameter larger than \(d\). Furthermore, according to Rosin and Rammler⁴³, there is an exponential connection between \({Y}_{d}\) and \(d\). Va

where \(\overline{d }\) represents the mean droplet diameter, and \(n\) denotes the spread parameter, assigned values are represented in Table 1. Solid cone injection method was utilized in front of the mouth weighing 7.7 mg^44,45. Injection duration was 0.5 s with a total of 35,000 parcels. According to the overall mass and Rosin–Rammler distribution, each parcel reflects a certain number of droplets.

The angle between the cough airflow and the horizontal is 27.5°, and the droplet parcel injection velocity is 6.6 m/s^29,46. The path that cough droplets take is examined using the discrete phase model (DPM). The non-steady motion of individual droplet packets is seen by the application of Newton’s second law and the consideration of the effects of drag force, gravity, and buoyancy^47,48. According to Nicas et al.⁴⁹, the concentration of non-volatile components in cough droplets is 1.8%, meaning that the equilibrium diameter of the droplet nucleus is approximately 26% of its initial size. The particles were not handled in bulk; instead, each droplet parcel represented a group of droplets, ensuring computational efficiency while maintaining accurate dispersion dynamics. The injected droplets were assigned an initial temperature of 310.15 K (37 °C), representing the average temperature of exhaled respiratory fluid.

To accurately model the behavior of airborne droplets emitted from coughing, a Eulerian–Lagrangian approach was employed, where the airflow field was simulated using the Reynolds-Averaged Navier–Stokes (RANS) equations with a k-ε turbulence model²⁸, and individual droplet trajectories were tracked using the Discrete Phase Model (DPM) in ANSYS Fluent. The path of cough droplets was investigated using the discrete phase model (DPM). Using Newton’s second law and accounting for the effects of drag force, gravity, and buoyancy, each droplet parcel is observed for unstable motion⁵⁰

where \(\overrightarrow{{u}_{p}}\) is the droplet velocity, \({\tau }_{r}\) is the droplet relaxation time, which depends on the diameter and density of the droplet, \(\overrightarrow{g}\) is the gravity acceleration, and \(\rho\) is the droplet density.⁴⁸.

The simulation accounted for multiple physical mechanisms influencing airborne particle motion, ensuring an accurate representation of aerosol transport. Gravity and buoyancy effects were incorporated, where larger droplets experienced rapid gravitational settling, while smaller aerosols remained suspended due to their interaction with turbulent airflow patterns. Turbulent dispersion was modeled using the Random Walk Model (Stochastic Tracking Approach), allowing particle trajectories to fluctuate in response to local airflow variations⁵¹.

where \(\zeta\) is a normally distributed random number, and \(k\) is the turbulent kinetic energy.

The eddy interaction time⁵¹ used in the turbulence dispersion model was defined as

where \(k\) is the local turbulent kinetic energy, and \(\epsilon\) is the local rate of dissipation of turbulence.

This formula ensures that particle dispersion responds dynamically to local turbulence levels, with higher turbulence resulting in shorter-lived eddies and greater particle fluctuation. This interaction time governs the lifetime of an eddy before a droplet encounters a new eddy, simulating the effect of turbulent diffusion.

A two-way coupling strategy was employed to allow momentum and energy exchange between the droplet and airflow phases. This approach is particularly suitable for transient events with high local particle concentrations, such as coughing, and has been adopted in similar respiratory droplet studies⁵². The Taylor Analogy Breakup (TAB) model was activated in the DPM setup to account for secondary droplet breakup caused by aerodynamic deformation during the cough event, enhancing the accuracy of droplet size evolution in turbulent airflow.

Evaporation handling and phase change

To accurately simulate droplet evaporation, the DPM evaporation model²⁸ was employed, where mass transfer was calculated based on latent heat exchange, surrounding air temperature, and relative humidity levels. The mass transfer of water from droplets to the surrounding air was governed by the following equation:

where \(m\dot{}\) is the mass evaporation rate (kg/s), \({k}_{m}\) is the mass transfer coefficient (m/s), \(A\) is the droplet surface area (m²), \({C}_{s}\) is saturation vapor concentration at the droplet surface (kg/m³) and \({C}_{\infty }\) is the ambient vapor concentration (kg/m³).

The mass transfer coefficient (\({k}_{m}\)) was determined using the Sherwood number correlation, given by:

where \(Sh\) is Sherwood number determined by Ranz-marshall correlation for droplet evaporation, \(D\) is the binary diffusion coefficient of water vapor in air (m²/s) and \(d\) is the droplet diameter.

The heat exchange between the droplet and the surrounding air was modeled using the heat balance equation:

where \({h}_{v}\)​ is the latent heat of vaporization (J/kg), \(h\) is the convective heat transfer coefficient (W/m²K), \({T}_{\infty }\) is the ambient air temperature (K) and \({T}_{d}\) is the droplet temperature (K).

This model ensures that droplet size continuously decreases due to evaporation, leading to the formation of residual airborne droplet nuclei that can remain suspended for extended periods. The evaporation rate was influenced by temperature, humidity, and airflow conditions, accurately replicating real-world aerosol transformation dynamics. The numerical implementation of this model allowed for precise tracking of droplet lifetime, helping to assess the risk of airborne pathogen transmission in indoor office environments. Surface interaction dynamics were also considered, with walls, floors, ceilings, and human surfaces modeled using a trap boundary condition, meaning deposited droplets adhered upon contact. On the other hand, supply and exhaust vents were assigned escape boundary conditions, allowing airborne particles to exit the computational domain when entrained within the airflow pathways. By integrating these transport and deposition mechanisms, the model effectively captured the complex behavior of respiratory droplets, including their suspension, evaporation, and surface interactions, ensuring a realistic assessment of airborne pathogen transmission in office environments.

Influence of thermal plumes and respiratory cycles on aerosol dispersion

In indoor environments, aerosol dispersion is governed by a combination of factors, including mechanically induced airflow, body-generated thermal plumes, and respiratory flow cycles. While the present study focuses on aerosol dynamics driven by mechanical ventilation, it is important to contextualize the role of thermal plumes and inhalation–exhalation cycles based on current scientific understanding. Thermal plumes arise from convective heat transfer between the human body and the surrounding air, generating upward buoyant flows that may affect the transport of airborne particles in the immediate vicinity of an occupant. However, the influence of these plumes is strongly dependent on the ambient airflow conditions. Several studies have demonstrated that under low-velocity ventilation regimes such as displacement or natural ventilation thermal plumes can significantly impact the direction and concentration of aerosol transport near the breathing zone. For instance, Zong et al.⁵³ highlighted that thermal plumes increase exposure risk by drawing particles from lower regions into the breathing zone when background airflow velocities are below approximately 0.25 m/s.

Conversely, in mechanically ventilated environments with higher supply airflow velocities, the effect of thermal plumes is substantially diminished. Yang et al.⁵⁴ showed that ventilation velocities at or above 0.25 m/s are sufficient to disrupt thermal plumes and limit their influence on overall airflow patterns. Similarly, Ai and Melikov⁴¹ concluded that in well-mixed indoor environments driven by mechanical ventilation, forced convection dominates particle transport, thereby minimizing the contribution of buoyancy-driven flows. These findings are further supported by Zabihi et al.⁵⁵, who demonstrated through CFD simulations that while thermal plumes can elevate particles locally, the broader dispersion and removal of aerosols are primarily governed by mechanical ventilation. Given that the airflow velocities modeled in our study exceed these critical thresholds, the exclusion of thermal plumes is justified and consistent with established literature.

The continuous inhalation–exhalation cycles of occupants also play a role in shaping airflow fields and particle transport, particularly in short-range transmission scenarios. However, their influence is largely confined to the near-field region (within approximately 1 m of a source), as shown by Zong et al.⁵³ and others. Although prior studies have incorporated respiration dynamics into CFD models, their impact becomes less significant at the room scale under active mechanical ventilation. In this study, a discrete high-momentum cough event is used as the representative emission source to evaluate aerosol transport across space. Given the focus on room-scale ventilation performance rather than close-contact exposure, the exclusion of breathing cycles is considered appropriate for the study objectives.

Although thermal plumes and respiratory cycles may affect localized aerosol dynamics under specific conditions, their impact is considerably reduced in high-flow, mechanically ventilated environments such as the one examined in this work. Given the elevated airflow velocities modeled well above thresholds where buoyancy-driven or respiratory effects become significant, their exclusion is unlikely to alter the overall findings related to ventilation strategies and aerosol dispersion. Nevertheless, their importance in proximity-based transmission and stratified airflow conditions is acknowledged, and future studies may incorporate these factors to expand the applicability of airflow and exposure assessments to a broader range of indoor environments.

HVAC and cough particle validation

The validation process was conducted in three sequential stages to ensure the accuracy and reliability of the simulation results. First, the HVAC airflow field was validated based on previously published experimental data from past research⁹, where simulated airflow patterns were compared with on-site velocity measurements. Second, the cough jet dynamics were validated by comparing the simulated trajectory and behavior of the cough cloud with experimental data on human expiratory events. Finally, the droplet transport and evaporation behavior were validated using two references: the Wells evaporation–falling curve for assessing vertical displacement trends, and Ugarte-Anero et al. (2022) for time-resolved diameter decay, confirming the evaporation and dispersion accuracy over the 10-s simulation period.

The DPM model has received extensive validation, performing well in forecasting droplet mobility and deposition. Nonetheless, the velocity of air greatly affects how droplets travel. Prior to examining the droplet lifespan released during a cough, we confirm the cough airflow phase’s validity. Agrawal and Bhardwaj developed a mathematical formulation for predicting the distance traveled by a cough cloud in an enclosed area, building upon experimental findings⁵⁶. The validation process was conducted by tracking the movement distance of the cough cloud front (Fig. 7), utilizing 1 μm diameter particles to estimate the advancement of the cough cloud over time. These particles were expelled from the mouth and closely followed as they interacted with the surrounding airflow, ensuring accurate representation of the cloud’s dispersion dynamics.

Figure 8, a representative snapshot of the simulated cough cloud is presented, illustrating particle dispersion at 1 s and capturing the cloud measurement process over a 10-s timeline for validation purposes. The cloud was measured from the point of expulsion (mouth region) to the farthest detected particle, ensuring a comprehensive evaluation of its propagation.

The validation results demonstrated strong agreement between the computational model and the mathematical reference model (Fig. 9). The maximum observed variation between the simulation and the mathematical model⁵⁶ was 11%, indicating a high level of accuracy in the numerical approach. Furthermore, the discrepancy between the models decreased significantly after 3 s, suggesting that the CFD simulation effectively captures the transient behavior of the cough cloud as it stabilizes over time. These findings reinforce the reliability of the current model in accurately predicting airborne particle transport and dispersion.

To further validate the accuracy of the droplet dynamics modeled in this study, the simulated vertical displacement of droplets over a 10-s period was compared against the classical Wells evaporation–falling curve, originally proposed by Wells (1934)⁵⁸. The Wells curve outlines the relationship between droplet size and settling behavior in indoor environments, highlighting that smaller droplets (< 10 μm) remain suspended due to rapid evaporation and air resistance, while larger droplets (> 50 μm) tend to settle quickly under gravity. In our simulation, representative droplets with initial diameters of 1, 5, 10, 20, 50, and 100 μm were tracked using the Discrete Phase Model (DPM). Their vertical displacement was recorded after 10 s, and the results were compared with scaled reference values derived from the Wells curve. As shown in Fig. 10, the simulation results exhibited good agreement with the theoretical trend: larger droplets displayed significant downward motion, while smaller droplets showed minimal displacement.

In addition to the conceptual validation based on the Wells evaporation–falling curve, which verifies the overall fate of droplets across different size ranges, a more detailed, time-resolved validation was performed by comparing our simulation with the study by Ugarte-Anero et al.¹⁹ (Fig. 11). In their CFD analysis, a saliva droplet with an initial diameter of 100 μm was tracked over a 10-s period in a mechanically ventilated indoor environment. Their findings showed a progressive reduction in droplet diameter to approximately 28 μm at 10 s, with key intermediate values of ~ 82 μm at 1 s, ~ 60 μm at 3 s, and ~ 42 μm at 5 s, reflecting the influence of evaporation and convective airflow. In our simulation, droplets of the same initial size followed a closely similar trend, decreasing to approximately 32 μm at 10 s. This close agreement between both datasets validates the evaporation kinetics and thermodynamic modeling implemented in the current study. Moreover, both simulations highlight the combined effects of droplet shrinkage, airflow entrainment, and gradual sedimentation, reinforcing the physical accuracy of our Lagrangian discrete phase model and the realism of the indoor environmental conditions simulated.

Although the present study employed a simulation time of 10 s, this duration was carefully selected to capture the initial dispersion phase of cough-generated aerosols—an interval that encompasses the most critical period for short-term exposure risk in mechanically ventilated environments. Previous studies, such as those by Li et al.²⁹ and Zhao et al.³⁰, have demonstrated that the dominant mechanisms influencing aerosol transport—including jet momentum, turbulent entrainment, evaporation, and near-field dispersion—occur within the first few seconds following a respiratory event. Therefore, the selected timeframe offers sufficient resolution to evaluate the immediate influence of ventilation strategies on aerosol behavior.

The accuracy of this approach has been validated against well-established experimental and theoretical benchmarks, including the Wells evaporation–falling curve⁵⁸ , droplet diameter decay data from Ugarte-Anero et al.¹⁹, and cough cloud propagation characteristics described by Agrawal and Bhardwaj⁵⁷. These comparisons confirm that the model reliably captures the key physical processes relevant to short-term aerosol dynamics.

While it is acknowledged that longer simulation durations and advanced modeling techniques such as Large Eddy Simulation (LES) can provide enhanced insight into long-term suspension, room-scale mixing, and re-entrainment phenomena, such approaches come with substantial computational demands. Given the need to compare multiple full-scale ventilation scenarios, the use of a validated Eulerian–Lagrangian framework coupled with RANS turbulence modeling was deemed appropriate. Future work may extend this modeling framework using LES and longer simulation windows to capture the broader temporal dynamics of aerosol behavior in more complex or naturally ventilated indoor environments.