Overspray Containment Using an Air-Curtain Spray Hood in High-Pressure Airless Spray Coating with CFD Simulation and Experimental Validation

Abstract

High-pressure airless spray coating can atomize high-viscosity, high-solids coatings without compressed air and is widely used for large-scale anticorrosion applications, but robotic operation often produces substantial overspray that increases material waste, environmental burden, and lowers deposition efficiency. In this work, air-curtain blowing is investigated as an overspray control strategy for wall-climbing robotic airless spraying. A validated CFD framework was established using the realizable k–ε turbulence model coupled with a discrete-phase model (DPM) to simulate particle atomization, transport, impact, and escape, and to examine the effects of blowing angle and gap distance on the flow field and particle trajectories. Overspray performance was quantified using the wall deposition rate, hood collection rate, and particle escape rate. Experiments using a transparent spray hood with a mass collection system were conducted to validate the numerical predictions. The CFD results captured the measured trends in deposition and escape across the tested conditions. Among the evaluated parameters, a 60° blowing angle provided the most effective overspray reduction by redirecting particles toward the target surface. Overall, combining CFD analysis with experimental validation offers a practical methodology for designing and optimizing air-curtain systems to improve coating efficiency in automated high-pressure airless spray applications.

1. Introduction

High-pressure airless spray coating is widely adopted in industries such as automotive, shipbuilding, steel construction, and large storage tanks due to its high transfer efficiency, strong adhesion, and reduced paint waste compared with conventional air spraying. By atomizing paint solely through liquid pressure, without compressed air, this method achieves fine droplet formation and uniform coating application, making it suitable for large-area and high-viscosity coating tasks. Despite these advantages, overspray remains a persistent challenge. In complex geometries or enclosed spaces, airflow disturbances can cause paint particles to escape from the spray zone, leading to material loss, environmental pollution, and unstable film quality.

To address overspray and improve working safety, wall-climbing spray robots have been developed as an alternative to manual coating in hazardous or elevated environments. These robots ensure consistent coating quality, reduce human error, and minimize operator exposure to harmful vapors. For overspray control, integrating spray hoods with air-curtain blowing and exhaust systems has become a promising approach. The blowing structure forms a directional airflow barrier to block or redirect particles that might escape from the sides of the hood, while the exhaust system removes suspended particles from the spray zone. The effectiveness of these systems depends on the optimization of the blowing angle, flow rate, and hood geometry, as well as on the interaction between the blowing and exhaust flows.

This study investigates overspray control strategies for high-pressure airless spraying on tank surfaces using a wall-climbing spray robot, as shown in Figure 1. Both experiment testing and CFD simulations are performed with a semi-circular spray hood to evaluate the impact of blowing angles and exhaust conditions on wall deposition, hood attachment, and particle escape. The experiment data serve as a basis for validating the CFD model, ensuring that simulated airflow patterns and particle trajectories closely match realistic conditions. This combined approach enables a deeper understanding of particle behavior and supports the development of more efficient air-curtain designs for overspray mitigation.

In industrial-scale coating operations, overspray control is of considerable practical importance because it directly influences coating material consumption, deposition efficiency, environmental emissions, and operational safety. This issue becomes even more critical in large-area applications and confined working environments, where uncontrolled particle escape can lead to significant material loss and airborne contamination. Therefore, developing an effective overspray containment strategy is not only beneficial for improving coating performance, but also necessary for enhancing process sustainability and reducing environmental burden. In addition, because spray transport involves complex multiphase flow behavior, the combined use of CFD analysis and experimental validation is essential for obtaining reliable insight into particle dispersion and for evaluating the effectiveness of air-curtain-based containment designs under realistic operating conditions. Recent studies have also emphasized the importance of experimental measurement techniques for industrial-scale multiphase flow systems [1] and numerical investigations for understanding spray characteristics under different process conditions [2]. More recent studies have further extended spray-coating research toward detailed droplet distribution analysis, practical dispersion reduction device design, and experimentally validated numerical modeling of atomization and spray transport. For example, Li et al. [3] applied CFD-DEM to investigate droplet behavior in an airless spray-coating process, while Park et al. [4] evaluated a paint particle dispersion reduction device for airless spraying. In addition, Chen et al. [5] combined numerical modeling with experimental validation to analyze near-nozzle atomization behavior, further demonstrating the value of CFD-based approaches for obtaining physical insight into spray transport and validation.

Previous research provides the scientific foundation for this work. Domnick et al. [6] used Phase Doppler Anemometry (PDA) to observe atomization processes in airless spraying, identifying the transformation from a liquid sheet to ligaments and droplets, with droplet size strongly affected by pressure and nozzle geometry. Ye et al. [7] studied fan-shaped sprays, highlighting how droplet collisions, coalescence, and rebound influence particle size distribution and coating efficiency, and emphasizing the importance of accurate droplet initial conditions in CFD models. Chen et al. [8,9,10] and Yang et al. [11] applied CFD to predict coating thickness and uniformity on curved and moving surfaces, demonstrating the method’s value in design optimization. Fogliati et al. [12] compared turbulence models, finding that the realizable k–ε model better reproduces particle distribution and spray shape, making it well-suited for simulating airflow in blowing and exhaust structures.

The novelty of the present study lies in establishing a validated CFD–experiment framework for overspray containment analysis in high-pressure airless spray coating with an air-curtain spray hood. In addition to model validation, the effects of hood-to-wall distance, blowing flow rate, and blowing angle are systematically evaluated using wall deposition, hood collection, and particle escape as performance indicators, thereby providing practical guidance for overspray control design in automated coating systems.

2. Theory

2.1. Airflow Simulation Theory

In fluid flow analysis, a fixed control volume is used to describe the physical behavior of the system, governed by three fundamental laws: the continuity Equation (1) for mass conservation, and the momentum Equations (2)–(4). In high-pressure airless spray simulations, airflow inside the hood is treated as a viscous, incompressible, transient 3D flow, primarily affected by the high-speed paint jet from the nozzle. Solving these governing equations is essential for accurately capturing airflow direction and velocity distribution:

$${\displaystyle \frac{\partial \rho }{\partial t}}+{\displaystyle \frac{\partial (\rho u)}{\partial x}}+{\displaystyle \frac{\partial (\rho v)}{\partial y}}+{\displaystyle \frac{\partial (\rho w)}{\partial z}} = 0,$$

(1)

$$\rho F_x-{\displaystyle \frac{\partial p}{\partial x}}+\mu ({\displaystyle \frac{{\partial }^{2}u}{\partial x^2}}+{\displaystyle \frac{{\partial }^{2}u}{\partial y^2}}+{\displaystyle \frac{{\partial }^{2}u}{\partial z^2}})+{\displaystyle \frac{1}{3}}\mu {\displaystyle \frac{\partial }{\partial x}}({\displaystyle \frac{\partial u}{\partial x}}+{\displaystyle \frac{\partial v}{\partial y}}+{\displaystyle \frac{\partial w}{\partial z}})=\rho {\displaystyle \frac{\mathrm{D}u}{\mathrm{D}t}},$$

(2)

$$\rho F_y-{\displaystyle \frac{\partial p}{\partial y}}+\mu ({\displaystyle \frac{{\partial }^{2}v}{\partial x^2}}+{\displaystyle \frac{{\partial }^{2}v}{\partial y^2}}+{\displaystyle \frac{{\partial }^{2}v}{\partial z^2}})+{\displaystyle \frac{1}{3}}\mu {\displaystyle \frac{\partial }{\partial y}}({\displaystyle \frac{\partial u}{\partial x}}+{\displaystyle \frac{\partial v}{\partial y}}+{\displaystyle \frac{\partial w}{\partial z}})=\rho {\displaystyle \frac{\mathrm{D}v}{\mathrm{D}t}},$$

(3)

$$\rho F_z-{\displaystyle \frac{\partial p}{\partial z}}+\mu ({\displaystyle \frac{{\partial }^{2}w}{\partial x^2}}+{\displaystyle \frac{{\partial }^{2}w}{\partial y^2}}+{\displaystyle \frac{{\partial }^{2}w}{\partial z^2}})+{\displaystyle \frac{1}{3}}\mu {\displaystyle \frac{\partial }{\partial z}}({\displaystyle \frac{\partial u}{\partial x}}+{\displaystyle \frac{\partial v}{\partial y}}+{\displaystyle \frac{\partial w}{\partial z}})=\rho {\displaystyle \frac{\mathrm{D}w}{\mathrm{D}t}}.$$

(4)

2.2. Droplet Impact Theory

In the discrete-phase model, particles interact with the wall upon impact. Depending on the impact energy, particles may undergo adhesion, spreading, or splashing. The impact energy is defined in Equation (5):

$$E^2 = {\displaystyle \frac{\rho V_r^2{D}_D}{\sigma }}({\displaystyle \frac{1}{\min (\frac{h_0}{\left(D_D,1\right)} + \frac{{\delta }_{\mathit{bl}}}{D_D})}}),$$

(5)

where $\rho$: liquid density, $V_r$: particle wall relative velocity, defined as $V_r^2 ={ (V_p - V_w)}^2$, where $V_p$ and $V_w$ represent the velocities of the particle and the wall, $D_D$: particle diameter, $\sigma$: liquid surface tension, $h_0$: wall film thickness, and ${\delta }_{\mathit{bl}}$: boundary layer thicknessdefined as:

$${\delta }_{\mathit{bl}} = {\displaystyle \frac{D_D}{\sqrt{\mathit{Re}}}},$$

(6)

and the Reynolds number is defined as:

$$R_e = {\displaystyle \frac{\rho V_r{D}_D}{\mu }}.$$

(7)

Based on the definition of impact energy, droplet splashing can lead to the formation of a wall film. In the present simulation conditions, no obvious non-physical behavior associated with the wall film treatment was observed, even when the film thickness became very small.

When the nondimensional impact energy is below 16, particles adhere to the wall and match its velocity. During spreading, their motion follows a wall-jet model, which approximates inviscid radial flow.

In general DPM formulations, particle rebound may occur under certain conditions and is typically modeled using the restitution coefficient. The rebound behavior is described based on impact energy, with the restitution coefficient defined in Equation (8):

$$e = 0.993 - 1.76{\theta }_I + 1.56{\theta }_I^2 - 0.49{\theta }_I^3,$$

(8)

where ${\theta }_I$ is the angle between the particle and the wall at impact.

Splashing occurs when the particle impact energy exceeds a critical threshold, producing secondary droplets. Their size, velocity, and direction are randomly sampled from experimentally derived distributions. If no splashing occurs, the model remains inactive. Droplet sizes are sampled from a cumulative probability distribution function (CPDF) based on Mundo et al. [13], using a Weibull distribution. The function is given by:

$$f\left(D_D\right) = 2{\displaystyle \frac{d}{{D_D}^2}}\mathit{exp}[-({\displaystyle \frac{d}{D_D}}{)}^2].$$

(9)

This equation gives the probability of forming a secondary droplet of diameter d. The distribution peaks at $D_D ={ d}_{\mathit{max}}/\sqrt{2}$, where $d_{\mathit{max}}$ is the maximum droplet diameter. To maintain physical consistency at higher Weber numbers, the distribution follows O’Rourke’s expression [14], in which the droplet size distribution is characterized by a representative droplet diameter d₀ determined based on the atomization model implemented in the DPM framework. The resulting size distribution is:

$$D_D = {\displaystyle \frac{d_{\mathit{max}}}{d_0}} = \mathrm{MAX}({\displaystyle \frac{E_{\mathit{crit}}^2}{E^2}},{\displaystyle \frac{6.4}{\mathit{We}}},0.06).$$

(10)

The middle term describes droplet size distribution at low Weber numbers, where splashing occurs under low impact energy. O’Rourke found that, under high-energy impacts, the droplet size rarely falls below 0.06. The Weber number, defined by particle diameter and relative velocity, represents the ratio of inertial to surface tension forces and is given by:

$$\mathit{We} = {\displaystyle \frac{\rho V_r^2{D}_D}{\sigma }}.$$

(11)

In Equation (6), $D_D$ represents the particle diameter. To perform diameter-based analysis based on experimental data, the cumulative probability distribution function (CPDF) must be used. This cumulative function can be evaluated using Equation (12):

$$F\left(D_D\right) = 1 - \exp [-{\displaystyle \frac{d^2}{D_D}}].$$

(12)

The probability distribution is bounded between 0 and 1. To determine the particle diameter, Equation (12) must be inverted, and random sampling within this range is performed using the CPDF. This leads to a droplet diameter expression. For the ith secondary droplet formed by splashing, the diameter is given by:

$$d_i = D_D\sqrt{-\ln (1 - c_i)},$$

(13)

where $c_i$ is the ith random number uniformly distributed between 0 and 1, used for sampling from the cumulative probability distribution function.

Once the secondary droplet size is known, its occurrence probability can be estimated using the distribution function to calculate the number of splashed droplets. The total number also depends on the total splashed mass, which is obtained from mass conservation. Based on Mundo’s experiments, this mass is proportional to the square of the impact energy. The droplet mass fraction $m_s$ is calculated using Equation (14):

$$m_s = \left\{\begin{array}{c}1.8 \times { 10}^{-4}\left(E^2 - E_{\mathit{crit}}^2\right),{ E}_{\mathit{crit}}^2 <{ E}^2< 7500\\ 0.70,{ E}^2 > 7500\end{array}.\right.$$

(14)

Studies have shown that under typical spray conditions, impact energy often exceeds the splashing threshold, with up to 70% of the droplet mass lost through splashing. To estimate the total number of splashed droplets, mass conservation is applied, meaning the total mass of generated secondary droplets must match the fraction of mass removed by splashing. The corresponding equation is:

$${\displaystyle \frac{\rho \pi }{6}}{\mathrm{N}}_{\mathrm{tot}}\sum _{\mathrm{n}=1}^{{\mathrm{N}}_{\mathrm{parcels}}}\left(f_n{d}_n^3\right) = m_s{m}_0,$$

(15)

where $m_0$: total mass of particles impacting the wall, and $N_{\mathit{tot}}$: total number of particles contacting the wall.

After determining the total number of splashed droplets, their velocity distribution must be specified. To match Mundo’s experimental observations, a Weibull probability density function is used to describe the velocity component in the normal direction. The corresponding expression is:

$$\mathrm{f}\left({\displaystyle \frac{V_{\mathit{ni}}}{V_{\mathit{nd}}}}\right) ={ [{\displaystyle \frac{b_v}{{\theta }_v}}({\displaystyle \frac{\frac{V_{\mathit{ni}}}{V_{\mathit{nd}}}}{{\theta }_v}})}^{b_v-1}]\exp [-{\left({\displaystyle \frac{\frac{V_{\mathit{ni}}}{V_{\mathit{nd}}}}{{\theta }_v}}\right)}^{b_v}],$$

(16)

where $b_v$ and ${\theta }_v$ are defined as:

$$b_v =\left\{\begin{array}{c} 2.1, {\theta }_I < 50°\\ 1.10 + 0.02{\theta }_I, {\theta }_I > 50°\end{array}\right.,$$

(17)

$${\theta }_v=0.158e^{0.017{\theta }_I}.$$

(18)

The tangential velocity component $V_{\mathrm{ti}}$ is calculated based on the reflection angle ${\theta }_s$:

$${\theta }_s = 65.4 + 0.226{\theta }_I,$$

(19)

$$V_{\mathit{ti}}={\displaystyle \frac{V_{\mathit{ni}}}{\tan ({\theta }_s)}}.$$

(20)

The total kinetic and surface energy of the splashed droplets should not exceed the initial energy of the impacting particle. Based on this, the following energy conservation equation can be derived:

$$\begin{gathered} {\displaystyle \frac{1}{2}}\sum _{\mathrm{i}=1}^{{\mathrm{N}}_{\mathrm{parcel}}}\left(m_i{V}_i^2\right) + \pi \sigma \sum _{\mathrm{i}=1}^{{\mathrm{N}}_{\mathrm{parcel}}}\left(N_i{d}_i^2\right), \\ ={\displaystyle \frac{1}{2}}{m}_d{V}_d^2\left(m_i{V}_i^2\right) + \pi \sigma \left(m_d{d}_i^2\right) - E_{\mathit{crit}}, \end{gathered}$$

(21)

where $E_{\mathit{crit}}$ is the critical energy.

The normal and tangential velocity components of the secondary droplets are given by:

$${V^{\prime }}_{\mathit{ni}} = \sqrt{K}{V}_{\mathit{ni}},$$

(22)

$${V^{\prime }}_{\mathit{ti}}=\sqrt{K}{V}_{\mathit{ti}}.$$

(23)

3. Finite Element Simulation

3.1. Spray-Coating Analysis Procedure

This section outlines the simulation procedure for airless spray coating in Fluent, as illustrated in Figure 2. The process includes geometry setup, meshing, material and model settings, and boundary condition definition. Simulation results are validated by comparison with experimental data.

3.2. Geometry

This study describes the geometric setup used to analyze particle dispersion behavior under different airflow and exhaust conditions in high-pressure airless spraying. All models adopt the same basic configuration, including a hemispherical hood with a diameter of 58.6 cm, a vertical spraying distance of 32 cm, and a 40 cm gap between the hood and the wall.

In addition, a rectangular block measuring 20 cm × 5.3 cm × 4 cm is added on each side of the hood to simulate the inclined blowing slits applied on the left and right sides during the experiments. The differences among the geometric models are summarized in

Table 1. Geometrical configurations of spray models with different blowing and exhaust conditions.

Model	Blowing Angle	Exhaust Type	Exhaust Angle	Exhaust Size	Blowing Slot Size
A	None	None	-	-	-
B	45°	None	-	-	20 cm × 0.5 cm
C	45°	Slit	30°	20 cm × 1 cm	20 cm × 0.5 cm
D	60°	Slit	30°	20 cm × 1 cm	20 cm × 0.5 cm

, and the corresponding spraying model diagrams are shown in

Table 2. Spray model.

Model	Z Direction	X Direction
A
B
C
D

. In Fluent, all boundary regions are assigned consistent names (e.g., left blow, right exhaust, and upper inlet), as illustrated in Figure 3.

3.3. Mesh Generation

The computational mesh is generated using ANSYS Fluent Meshing 2023 R1, which is designed specifically for CFD applications in Fluent. To ensure accuracy near critical regions, Figure 4 shows a local refinement box (8 cm × 32 cm × 30 cm), which is defined around the nozzle and wall area. A poly-hexcore scheme is applied for mesh generation, providing a balance between computational efficiency and accuracy.

In addition, a grid independence test was carried out using Model A under a representative spray condition with a spray distance of 30 cm and a sprayer flow rate of 11.1 g/s. Three mesh densities with different element sizes were examined, while all other simulation settings were kept the same. The comparison was performed using the wall flow rate, hood flow rate, and particle escape rate as the evaluation indicators. As shown in

Table 3. Grid independence test under a representative spray condition.

Maximum Element Size	Minimum Element Size	Number of Cells	Wall Flow Rate (g/s)	Hood Flow Rate (g/s)	Escape Flow Rate (g/s)
0.012	0.004	284,145	7.16	2.52	1.42
0.01	0.003	445,963	7.28	2.45	1.37
0.008	0.002	638,420	7.29	2.44	1.37

, only minor differences were observed as the mesh was refined, and the variation between the meshes with 445,963 and 638,420 cells was particularly small. Therefore, the mesh containing 445,963 cells was selected for the subsequent simulations, considering both computational accuracy and computational cost. The final mesh contains approximately 445,963 cells, with a mesh quality of around 0.22, as shown in Figure 5. A cross-sectional view of the mesh structure is presented in Figure 6.

3.4. Material

This simulation involves the definition of material properties for the fluid, solid, and inert particle. The simulation assigns air as the continuous phase and water as the discrete phase, while high-carbon steel is used to represent the wall. The steel is modeled with a density of 7810 kg/m³, which corresponds to typical high-carbon steel containing approximately 0.6% to 1.0% carbon. This material setup reflects realistic surface conditions during spray impact. Key properties for all phases, including density, viscosity, and surface tension, are listed in

Table 4. Material properties.

Material Type	Material	Density	Viscosity (kg/m∙s)	Droplet Surface Tension (N/m)
Fluid	Air	1.225	1.7894 × 10−5	-
Solid	High-carbon steel	7810	-	-
Insert particle	Water	998.2	0.001003	0.0719404

.

Water is used as a paint substitute in experiments and simulations due to its similar density and viscosity to some low-viscosity coatings, allowing representative spray behavior while avoiding VOC emissions and environmental pollution. It also reduces cost, cleaning effort, and clogging risks, improving safety and repeatability.

3.5. Numerical Simulation Settings

In the general settings, a pressure-based solver is selected because high-pressure airless spraying involves transient flow and droplet formation. Transient conditions are enabled, and gravitational acceleration is applied in the Z direction as −9.81 m/s² to reflect realistic droplet motion under gravity. At the start of the transient simulation, the surrounding openings were set to atmospheric pressure, the wall was treated as a stationary wall, and the spray injection was activated at t = 0 s according to the flat-fan atomizer settings. The spray duration was set to 1 s to represent the actual spraying condition.

In the viscous model settings, the realizable k–ε model is selected because the present flow field involves jet-like flow, recirculation, and strong wall interaction inside the spray hood, which are characteristic features of high-pressure airless spray coating. In addition, this model provides a reasonable balance between numerical robustness and predictive capability for industrial turbulent flow simulations coupled with discrete particle tracking. Previous spray-related CFD studies have also shown that the realizable k–ε model can provide reliable predictions of spray shape and particle distribution, and it was therefore adopted in the present study. For the near-wall region, standard wall functions were employed. The closure coefficients of the turbulence model were kept as the default values provided in ANSYS Fluent, and no additional modification of the turbulence model constants was introduced in the present study.

The continuous- and discrete-phase models are configured within the Models panel in Fluent. The discrete-phase settings are illustrated in Figure 7. To simulate the interaction between air and particles, continuous-phase coupling and unsteady particle tracking were enabled in the DPM settings. This corresponds to a two-way coupling strategy between the airflow and the dispersed droplets, in which the continuous-phase influences droplet motion through the local flow field, while the dispersed phase feeds back to the continuous phase through source terms associated with particle transport. In the present study, the droplets were treated within the Fluent DPM framework. Therefore, unlike CFD–DEM coupling, particle–particle contact mechanics were not resolved, and the coupling was limited to airflow–particle interaction and wall interaction behavior during the transient spraying process. Besides, in the physical model settings, as shown in Figure 8, the stochastic collision model is enabled to account for random droplet interactions caused by turbulence. Additionally, the breakup process is modeled using the stochastic secondary droplet (SSD) model, which simulates the formation of smaller droplets due to shear and aerodynamic forces. This helps better represent the secondary atomization in high-pressure airless spray coating.

After configuring the main DPM settings, an injection is then created to define the jet source and apply the flat-fan atomization model. A new injection named injection-0 is created, with the injection type set to a flat-fan atomizer model to simulate the fan-shaped spray in high-pressure airless coating. This atomizer type generates elliptical or elongated droplets that spread in a flat pattern, matching the characteristics of a typical fan spray. The particle type is set to inert, representing non-reactive water droplets in the simulation. Key parameters, such as origin coordinates, spray angle, and direction, are defined to represent the actual spraying conditions, as summarized in

Table 5. Flat-fan atomizer parameter settings.

Parameters	Value
X center (mm)	0
Y center (mm)	280
Z center (mm)	0
X virtual origin (mm)	0
Y virtual origin (mm)	280.56
Z virtual origin (mm)	0
X fan normal vector (mm)	1
Y fan normal vector (mm)	0
Z fan normal vector (mm)	0
Mass flow rate (g/s)	11.8
Start time (s)	0
Stop time (s)	1
Spray half angle (deg)	25
Orifice width (mm)	0.3
Flat-fan sheet constant	3
Atomizer dispersion angle	6

. The flat-fan settings ensure that the spray shape, impact area, and droplet distribution closely reflect real-world behavior.

Finally, the governing equations were solved in ANSYS Fluent using a pressure-based transient solver. The airflow field was computed with the realizable k–ε turbulence model, while particle transport was simulated using the discrete-phase model with continuous-phase coupling and unsteady particle tracking. The spray process was simulated for 1 s with a timestep size of 0.0002 s, resulting in 5000 timesteps in total. Up to 20 iterations were performed within each timestep to ensure stable convergence of the transient solution.

4. Experiment Setup

4.1. Experiment Equipment

In practical spraying, many paint particles escape from the spray region and reach the surroundings, causing waste and pollution. To better understand particle behavior and improve overspray control, this study establishes an experiment platform in collaboration with a local company. The system simulates high-pressure airless spraying and uses a transparent hemispherical hood with a diameter of 58.6 cm to observe airflow and particle motion inside the hood. The setup is shown in Figure 9.

A 525 model airless nozzle is used, with a spray angle of 50° and an orifice diameter of 0.0635 cm. The spray distance is fixed at 32 cm. A high-pressure water pump supplies the central pressure for paint delivery. Blowing and exhaust systems are installed on both sides of the hood and operate at 500 L/min to evaluate how airflow affects particle dispersion. Figure 10 shows the nozzle and pump.

Table 6. Experiment equipment and operating parameters.

Equipment	Parameters
Wall	High-carbon steel
Spray nozzle	525 model
	Spray angle 50°
	Sprayer diameter 0.0635 cm
Spray hood	29.3 cm × 29.3 cm
Spray pressure	150 psi
Blowing	500 L/min
Exhaust	500 L/min

lists the key equipment and operating parameters. The adjustable distance between the hood and the wall allows tests under various conditions to ensure reliable simulation validation.

4.2. Spray Process Settings

Before spraying, two beakers are placed in front of and behind the wall to collect paint. The one near the wall captures paint attached to the surface, while the rear one collects paint that escapes or reflects off. Some droplets rebound and enter the hood region, where they are carried by airflow into the lower gutter and finally collected in the rear beaker. Paint trapped on the wall also drains down to the front gutter and enters the front beaker.

Since dispersed particles are hard to measure directly, this study uses the paint mass collected in the beakers to represent deposition and overspray. The paint weight per unit time in the rear beaker indicates the amount of escaped paint. By comparing wall deposition, attached paint, and overspray under different conditions, the results help evaluate simulation accuracy and provide valuable data for model validation.

5. Results and Discussion

5.1. Effect of Hood-to-Wall Distance on Particle Dispersion and Model Validation

This section investigates how the gap between the hood and the wall affects particle dispersion during spraying. Three gap distances, 1 cm, 2 cm, and 4 cm, are tested to observe variations in particle flow behavior, as shown in Figure 11. Model A presents the geometric model used in this section.

The boundary name selection definition is shown in Figure 3. In the simulation, the inlets on both sides beneath the hood are set as pressure inlets with 0 Pa, representing atmospheric pressure, allowing airflow in and out of the hood region. For the discrete phase, these inlets are defined as escaped, meaning particles exiting through these boundaries are no longer tracked, simulating particle loss from the hood-to-wall gap. The wall is defined as a stationary wall with a wall film condition for the discrete phase to model paint film formation after particle impact. The hood surfaces are set as trap boundaries, representing particle capture by the hood system. The full boundary setup is summarized in

Table 7. Boundary condition settings for the analysis of the effect of hood-to-wall distance on particle dispersion.

Boundary Name	DPM Boundary Condition Type	Flow Boundary Type	Setting
Left inlet	Escaped	Pressure inlet	0 Pa
Right inlet	Escaped
Spray hood	Trap	-	-
Wall	Wall Film	-	Stationary
Sprayer	Flat-fan atomizer model	Mass flow rate	11.8 g/s

.

Taking the 4 cm case as an example, particles reach the wall by 0.004 s and exhibit behaviors such as rebound, splashing, adhesion, and sliding, which are determined by the impact energy defined in Equation (5). Due to the nature of airless spray, larger droplets with higher kinetic energy are generated, resulting in a strong wall-directed flow after impact. As shown in Figure 12, by 0.012 s, some particles begin escaping the hood along the wall surface. This is further illustrated in Figure 13, where vortex structures are observed near the spray region due to the high-speed jet drawing in surrounding air.

According to

Table 8. Simulated total mass of water mist in each region.

Boundary Name	DPM Boundary Condition Type	Mass Flow Rate (g/s) (Percentage)
		1 cm	2 cm	4 cm
Inlet	Escaped	0.874 (7.4%)	1.488 (12.6%)	2.566 (21.7%)
Spray hood	Trapped	3.404 (28.8%)	2.504 (21.2%)	1.073 (9.1%)
Wall	Wall film	7.522 (63.7%)	7.811 (66.2%)	8.161 (69.2%)
Total	-	11.8

, with a 1 cm hood-to-wall gap, most particles are either deposited on the wall or collected by the hood, leaving only 7.4% as dispersed particles. When the gap increases to 2 cm, wall deposition slightly rises to 7.811 g, while hood collection drops to 2.501 g, indicating that more particles escape—dispersed mass increases to 12.6%, nearly twice that of the 1 cm case. At 4 cm, dispersion becomes more severe, reaching 2.566 g, and hood collection further decreases to 1.073 g. This suggests that over half the particles captured initially by the hood are now lost to the surroundings.

These results indicate that a smaller gap between the hood and the wall improves containment and reduces dispersion, while a larger gap weakens airflow control, allowing more particles to escape.

To evaluate the effectiveness of the hood in suppressing particle dispersion, four spray distances (30 cm, 32 cm, 35 cm, and 40 cm) were tested in the experiments, corresponding to hood-to-wall gaps of 2 cm, 4 cm, and 7 cm. Without the hood, the overspray ratio ranges from 26.9% to 33.4%, indicating that nearly one-third of the spray escapes into the environment. At closer distances, high particle momentum leads to more rebound and overspray. This rebound behavior is quantitatively described by the restitution coefficient in Equation (8), which depends on the impact angle. As the distance increases, particles lose energy due to air resistance, resulting in higher deposition but still significant dispersion.

After installing the hood, the overspray is notably reduced. For example, at a 30 cm spray distance, the escaped mass drops from 3.71 g/s (33.4%) to 1.37 g/s (12.3%), while a portion of the paint (22.1%) is collected by the hood. Similar trends are observed at other distances. The optimal suppression effect is seen at medium spray distances (30–35 cm), where the hood effectively collects stray particles.

Table 9. Water mist in different regions under varying hood gaps and installation conditions.

Spray Distance (cm)	Hood Presence	Sprayer Flow Rate (g/s)	Hood Distance (cm)	Wall Flow Rate (g/s) (Percentage)	Hood Flow Rate (g/s) (Percentage)	Escape Flow Rate (g/s) (Percentage)
30	X	11.1	-	11.1	-	3.71 (33.4%)
	O		2	7.28 (65.6%)	2.45 (21.1%)	1.37 (12.3%)
32	X		-	7.55 (68%)	-	3.55 (32%)
	O		4	7.75 (69.8%)	1.02 (9.2%)	2.34 (21%)
35	X		-	7.78 (70.1%)	-	3.33 (29.9)
	O		8	8.23 (74.1%)	0.57 (5.1%)	2.3 (20.7%)
40	X		-	8.11 (73.1%)	-	2.99 (26.9%)

summarizes the experiment results, showing that the hood reduces overspray by 10–20% on average. However, as the gap increases, hood collection efficiency declines, and dispersion becomes more sensitive to environmental factors like airflow.

To validate the developed CFD model, simulation results were compared with experimental measurements under two spray conditions, corresponding to hood-to-wall gaps of 2 cm and 4 cm. The comparison was based on wall deposition, hood collection, and particle escape, which were used as the quantitative validation metrics. Both setups include a spray hood, and the results are evaluated based on wall deposition, hood collection, and particle dispersion.

As shown in

Table 10. Simulation and experiment validation.

Spray Distance (cm)	Hood Distance (cm)	Method	Sprayer Flow Rate (g/s)	Wall Flow Rate (g/s) (Percentage)	Hood Flow Rate (g/s) (Percentage)	Escaped Flow Rate (g/s) (Percentage)
30	2	Simulation	11.8	7.811 (66.2%)	2.501 (21.2%)	1.488 (12.6%)
		Experiment	11.1	7.28 (65.6%)	2.45 (22.1%)	1.37 (12.3%)
32	4	Simulation	11.8	8.161 (69.2%)	1.073 (9.1%)	2.566 (21.7)
		Experiment	11.1	7.75 (69.8%)	1.02 (9.2%)	2.34 (21%)

, the simulation and experimental results show consistent trends and good agreement for all three indicators. For the 30 cm spray distance condition, the differences between simulation and experiment were 0.6%, 0.9%, and 0.3% for wall deposition, hood collection, and particle escape, respectively. For the 32 cm condition, the corresponding differences were 0.6%, 0.1%, and 0.7%. Overall, all differences remained within 1%. These results further demonstrate that the proposed model can reasonably reproduce the measured particle distribution behavior under the investigated spray conditions.

These results suggest that the simulation model reliably predicts particle behavior and dispersion patterns under realistic spray conditions, making it a valid tool for evaluating overspray control. Minor discrepancies are likely due to experiment uncertainties, such as flow fluctuations, environmental effects, or measurement tolerance.

5.2. Effect of Blowing Flow Rate on Particle Dispersion

Model B presents the geometric model used in this section. To further reduce particle dispersion, a side-blowing structure is added. This setup aims to create an air curtain by adjusting the blowing flow rate, helping to prevent particles from escaping through the sides of the hood.

Figure 3 illustrates the naming of each boundary. The boundary conditions are summarized in

Table 11. Boundary condition settings for the analysis of the effect of blowing flow rate on particle dispersion.

Boundary Name	DPM Boundary Condition Type	Flow Boundary Type	Setting
			Simulation 1	Simulation 2
Upper inlet	Escaped	Pressure inlet	0 Pa
Down inlet
Left inlet
Right inlet
Left blow		Flow inlet	250 L/min	500 L/min
Right blow
Spray hood	Trap	-	-
Wall	Wall Film	-	Stationary
Sprayer	Flat-fan atomizer model	Mass flow rate	11.8 g/s

. For the continuous phase, the two lower side inlets of the hood are set as pressure inlets with a gauge pressure of 0 Pa. The corresponding DPM boundary condition is set as escaped, meaning particles leaving from these boundaries are considered lost from the domain. The wall is defined as a stationary wall, and the DPM condition is set to wall film to simulate particle deposition after impact. The hood surface is defined as trap, indicating that particles entering this area are collected. For the blowing inlets, the boundary type is set to mass flow inlet. Two blowing flow rates are tested: 250 L/min (approximately 5.1 g/s) and 500 L/min (approximately 10.2 g/s), with the latter being twice the former. The DPM condition at the blowing inlets is also set to escaped, indicating no particle reflection or trapping occurs at those locations.

Table 12. Comparison of total water-mist mass under different blowing flow rates based on simulation results.

Boundary Name	DPM Boundary Condition Type	Mass Flow Rate (g/s)
		Simulation 1	Simulation 2
Upper inlet	Escaped	0.5591	0.8346
Down inlet	Escaped	0.6993	0.9387
Left inlet	Escaped	0.4218	0.1499
Right inlet	Escaped	0.4224	0.1321
Left blow	Escaped	0	0
Right blow	Escaped	0	0
Spray hood	Trap	1.3773	1.5951
Wall	Wall film	8.318	8.148
Total	-	11.8

shows the water-mist mass distribution across different regions for Simulations 1 and 2. In Simulation 1, the total escaped mass through the upper and side inlets is 2.101 g/s (17.8%), while in Simulation 2, it is 2.055 g/s (17.4%), showing only a slight decrease with a higher blowing flow rate.

Figure 14 and Figure 15 present the airflow vectors. In Simulation 1, the low-speed side flow slightly forms a “wind wall”, but its blocking effect is weak due to its limited velocity. In Simulation 2, although the speed increases with higher flow, a stronger wind wall effect is not observed. This suggests that simply increasing flow may not effectively reduce dispersion and may even disrupt particle paths, leading to more particles escaping.

As for the hood, the captured mass increases from 1.3773 g/s to 1.5951 g/s, showing about 2% improvement. This indicates that some particles that would otherwise escape are redirected into the hood.

Wall deposition remains nearly unchanged at around 69% in both cases (8.318 g/s and 8.148 g/s), suggesting that blowing flow has little impact on final deposition. This is mainly influenced by spray angle and hood design rather than airflow disturbance before impact.

Table 13. Comparison of total water-mist mass under different blowing flow rates with experimental validation.

Boundary Name	DPM Boundary Condition Type	Mass Flow Rate (g/s) (Percentage)
		Simulation 1	Simulation 2	Experiment 1	Experiment 2
Inlet	Escaped	2.1026 (17.82%)	2.0553 (17.42%)	2.275 (20.49%)	2.3241 (20.94%)
Spray hood	Trap	1.3773 (11.67%)	1.5951 (13.51%)	1.182 (10.65%)	1.2492 (11.25%)
Wall	Wall film	8.318 (70.49%)	8.148 (69.05%)	7.643 (68.86%)	7.5267 (67.8%)
Total	-	11.8		11.1

compares simulation and experimental results under two different blowing flow rates. The simulation shows a slight decrease in particle dispersion as the flow increases, from 2.10 g/s (17.82%) to 2.06 g/s (17.42%). Experimental results show a similar trend, though the change is less pronounced. For hood attachment, the simulation indicates an increase from 11.67% to 13.51% as the blowing flow rises, while experiments show a smaller change from 10.6% to 11.25%, still following the same trend. In terms of wall deposition, both simulations and experiments show only a slight reduction with higher blowing flow, suggesting that wall deposition is not significantly affected. Overall, both the simulation and experimental results show the same trend, indicating that increasing the blowing flow can reduce overspray and enhance hood collection, although the improvement is limited. This consistency further supports the reliability of the developed CFD model in predicting particle behavior under different blowing flow conditions.

5.3. Effect of Blowing Angle on Particle Dispersion

Based on previous simulation and experimental results, increasing the blowing flow rate and adjusting the exhaust angle can help reduce particle dispersion to some extent, but the improvement remains limited. To further enhance overspray control, this section explores alternative process conditions by varying the blowing angle, as shown in Figure 16. The geometrical configurations used in this study are based on Model C and Model D, as listed in Table 1.

Figure 3 shows the boundary naming, and the settings are summarized in

Table 14. Boundary condition settings for the analysis of the effect of blowing angle on particle dispersion.

Boundary Name	DPM Boundary Condition Type	Flow Boundary Type	Setting
Upper inlet	Escaped	Pressure inlet	0 Pa
Down inlet
Left inlet
Right inlet
Left blow		Flow inlet	500 L/min
Right blow
Exhaust		Flow outlet	1000 L/min
Spray hood	Trap	-	-
Wall	Wall Film	-	Stationary
Sprayer	Flat-fan atomizer model	Mass flow rate	11.8 g/s

. For the continuous phase, all four sides of the domain are set as pressure inlets with a pressure of 0 Pa to represent ambient air. The exhaust ports are defined as mass flow outlets with a flow rate of 10.2 g/s per side (equivalent to 500 L/min) to simulate the extraction process.

The spray nozzle is assigned to a mass flow rate of 11.8 g/s, matching the experiment condition at 150 psi. The wall is set as a stationary wall with wall film enabled for the discrete phase to model paint deposition. The spray hood is also defined as a wall, with the discrete-phase condition set to trap to simulate particle capture. All remaining openings, including inlets, blowing ports, and exhaust outlets, use the escaped condition to allow particles to exit the domain. The side-blowing ports are defined as mass flow inlets with a flow rate of 10.2 g/s per side.

Table 15. Simulated total mass distribution of water mist in each region under different blowing angles.

Boundary Name	DPM Boundary Condition Type	Mass Flow Rate (g/s)
		45°	60°
Upper inlet	Escaped	0.7589	0.6831
Lower inlet	Escaped	0.8811	0.7332
Left inlet	Escaped	0.1531	0.0803
Right inlet	Escaped	0.1329	0.08627
Left blow	Escaped	0	0
Right blow	Escaped	0	0
Exhaust	Escaped	0.0777	0.0759
Spray hood	Trap	1.6149	2.001
Wall	Wall film	8.181	8.14
Total	-	11.8

shows that the blowing angle significantly affects the particle distribution. When the blowing angle increases from 45° to 60°, the amount of overspray escaping from the lower and side inlets decreases noticeably. For example, the mass flow at the left and right inlets drops from 0.1531 g/s and 0.1329 g/s to 0.0803 g/s and 0.08627 g/s. This indicates that the steeper blowing angle helps form a stronger air curtain, guiding particles back into the hood rather than allowing them to escape.

As shown in Figure 17 and Figure 18, a 60° blowing angle produces a more complete air-curtain effect than 45°, effectively redirecting particles toward the hood. Additionally, the mass of particles attaching to the hood increases from 1.6149 g/s to 2.001 g/s, while the wall deposition remains nearly unchanged at around 69%. The amount of paint captured by the exhaust ports remains low in both cases (below 1%), indicating limited suction influence in this configuration.

Table 16. Total mist mass distribution.

Boundary Name	DPM Boundary Condition Type	Mass Flow Rate (g/s) (Percentage)
		45°	60°
Right blow	Escaped	1.926 (16.3%)	1.5829 (13.4%)
Exhaust	Escaped	0.0777 (0.6%)	0.0759 (0.6%)
Spray hood	Trap	1.6149 (13.7%)	2.001 (17%)
Wall	Wall film	8.181 (69.3%)	8.14 (69%)
Total	-	11.8

summarizes the total particle mass distribution. The total overspray decreases from 1.926 g/s (16.3%) at 45° to 1.583 g/s (13.4%) at 60°, while hood attachment improves from 13.7% to 17%. These results show that increasing the blowing angle reduces overspray and improves particle capture on the hood, without affecting wall deposition. Thus, adjusting the blowing angle is an effective strategy for improving overspray control and overall coating efficiency.

6. Conclusions

This study investigated overspray control in high-pressure airless spraying using a wall-climbing spray robot. Both numerical simulation and experiments were carried out to systematically analyze the effects of hood design and airflow configuration. Since overspray directly affects coating efficiency, material waste, environmental burden, and operational safety in industrial applications, establishing a reliable evaluation method for overspray containment is necessary. In this work, a validated CFD–experiment framework was established to evaluate particle transport behavior under different hood and airflow conditions. Based on both simulation and experiment, the main conclusions are summarized as follows:

• The developed CFD model demonstrated high accuracy after experimental validation and reliably predicted particle motion and overspray behavior in high-pressure airless spraying.
• The integration of a spray hood, particularly with reduced hood-to-wall spacing, effectively decreased overspray and limited particle escape into the surrounding environment.
• Both numerical and experimental results confirmed that increasing the blowing flow rate significantly reduced the overall overspray rate, highlighting its effectiveness in particle suppression.
• The blowing angle played a decisive role in overspray control. At 60°, the airflow confined most particles within the hood region, leading to a substantial reduction in overspray and achieving the objectives of this study.
• The main contribution of the present work lies in establishing a validated CFD–experiment approach for systematically evaluating overspray containment in high-pressure airless spray coating. The results provide practical guidance for hood design and airflow optimization and can serve as a useful basis for future research on automated industrial coating systems and overspray mitigation.

Overall, the present study showed that the combined use of CFD simulation and experimental validation can provide a practical basis for evaluating overspray behavior in high-pressure airless spraying. The findings help clarify how hood design and airflow conditions influence particle deposition and escape and can, therefore, support the improvement of spray hood design and overspray control in automated industrial coating applications.