Design and Optimization of Divider Head Geometry in Air-Assisted Metering Devices for Enhanced Seed Distribution Accuracy

Abstract

Achieving precise seed distribution is essential for optimizing crop yields and agricultural productivity. This study examines the impact of divider head geometry on seed distribution accuracy in pneumatic air seeder systems using rapeseed, wheat, and rice. Three custom-designed divider heads—funnel distributor (A1), closed-funnel distributor (A2), and cone-shaped distributor (A3)—were developed for an eight-furrow opener seeding system, each featuring eight outlets per opener. Bench tests at air pressures of 3, 3.5, 4, 4.5, 5, and 5.5 kPa and speeds of 4 and 5 km/h revealed significant variations in seed distribution accuracy among the designs. The A2 distributor demonstrated the lowest coefficient of variation (CV) across all seed types: 4.3%, 2.6%, and 6.95% for A1, A2, and A3 in wheat, respectively; 4.5%, 3.4%, and 6.2% in rice, respectively; and 0.3%, 0.1%, and 1.0% in rapeseed, respectively. Seed types also significantly influenced feed rate uniformity, with average CVs of 2.91% for rapeseed, 3.85% for rice, and 4.90% for wheat. CFD-DEM simulations validated the superior performance of the A2 distributor by analyzing flow fields and velocity distributions, showing reductions in CVs by 19.09–54.55% compared to A1 and A3. Thus, the A2 distributor was identified as the optimal design, significantly improving seeding uniformity across all seed types. In conclusion, this study provides critical insights for redesigning seed drill distribution heads to minimize turbulence in the seed–air mixture transport, enhancing seeding uniformity and increasing crop yields and agricultural productivity.

1. Introduction

Sowing is a vital process in the cultivation of cereal crops; it mainly affects plant germination, uniformity, and overall crop yield [1,2]. This process involves the precise placement of seeds in the soil to ensure optimal growth conditions, which include the correct depth, spacing, and soil contact [3,4,5], achieving uniform germination, maximizing yield potential by using suitable sowing methods, and ensuring the efficient use of resources such as water, nutrients, and sunlight [6,7]. Seeding uniformity is a pivotal metric for assessing a planter’s sowing performance [8,9]. The precision seed meter is the key component in achieving precision mechanized sowing. It can be classified as mechanical or pneumatic based on its working principle [10,11]. Air seeders are agricultural machines equipped with an air-delivery system in which airflow is used as the carrier to complete the process of seed conveying and seed guiding, achieving uniform seed distribution [12,13,14,15]. According to ASAE/ASABE S506 [16], air drill seeders are pieces of machinery equipped with a centralized hopper for seed containment and a volumetric metering system for precise seed measurement and distribution [17]. The key part of pneumatic seeders is the air-assisted centralized seed-metering device [18,19]. Through centralized seeding and pneumatic conveying methods, the air-assisted centralized system can effectively address the challenges of filling and clearing seeds at high speed and has the advantages of reducing the rate of seed damage and increasing adaptability for crops of various particle sizes [20,21]. Therefore, air-assisted centralized seeding technology is one of the effective methods for achieving high-performance seeding [22,23,24]. Within this system, the subsequent division of the air–seed flow occurs in the divider head [25,26,27], and then the seed should move along feed tubes into the soil.

Divider head geometry primarily affects the distribution accuracy of the air–seed mixture flow in air-assisted centralized seed-metering devices [8,28,29]. To analyze the performance of the distributor head, Yatskul et al. [25] studied the effect of divider head geometry and operating conditions on seed distribution accuracy in air seeders. It was found that increasing the working width of sowing implements reduced distribution accuracy; key factors affecting distribution included air velocity, material flow rate, outlet conditions, and structural elements like pipe elbows and tower configurations. Using a high-speed camera, they identified parameters critical for designing more accurate divider heads. Lei et al. [26] studied how distribution head geometry affects air seeder distribution accuracy; they used computational fluid dynamics (CFD) and the discrete element method (DEM) to monitor seeds and distribution pattern results, where distributor head geometry can affect seed distribution uniformity. Gierz and Markowski [30] investigated the effect of distribution head tilt and different diffuser variants on the evenness of seed sowing in pneumatic seed drills. It was discovered that adjusting the angle of the distribution head from 0° to 10° significantly improved the sowing quality for oat seeds.

In addition to the researchers mentioned earlier, Kumar and Durairaj [8] studied the importance of head geometry in the distribution performance of air seed drills. This research clearly shows a direct relationship between the design of the divider head and the effectiveness of seed dispersal. These findings emphasize the need for innovation and improvement in air seeder technology to boost agricultural production. Bourges and Medina [31] performed a numerical simulation model for the drill air seeder distributor head. Further, Guichuan et al. [24] studied how various seed distributor geometries impact seeding uniformity in pneumatic centralized systems across three structure types: M-type, T-type, and Y-type. Additionally, they examined the influence of outlet pipe angle, top cover cone angle, and fillet radius on seed distribution. This provides valuable information on how to improve the design of pneumatic seed distribution systems for better and more even seeding. Hu et al. [19] conducted a numerical study on the motion characteristics of rapeseeds in the two-phase flow of the distribution head of an air-assisted centralized seed-metering device based on coupled simulations using computational fluid dynamics (CFD) and the discrete element method (DEM); they investigated the effects of various structural and working parameters on seed motion characteristics and airflow field.

In describing the characteristics of a flow field, computational fluid dynamics (CFD) is an effective tool for simulating fluid flow and revealing the mechanism of fluid phases with sustainable development [32,33]. CFD-DEM gas–solid coupling has been used to study the movement of the flow of the seed–air mixture inside the pneumatic centralized system [19,24,34,35,36,37]. FLUENT simulates the flow field, and EDEM is used to build the seed model. The movement law of the seeds flowing inside the pneumatic centralized system can be simulated by combining the two software. Compared to traditional flow simulation, CFD-DEM can obtain simulation results closer to the actual seed metering, so that the accuracy of the results is easier to verify.

The CFD-DEM coupling approach has been used extensively to study fluid flow and particle motion behavior, both of which are relevant to several industrial applications such as pneumatic particle motion [33,38,39], circulating fluidized beds [40,41], and air screen cleaning devices [33,39]. The research demonstrated that simulations implementing the linked DEM-CFD methodology accurately represent and correlate closely with experimental findings.

Numerous studies have explored the impact of distributor head geometries and airflow dynamics on seed distribution in pneumatic seeders. However, many have focused on traditional designs, such as funnels and cones, without fully exploring innovative geometries or integrating CFD-DEM simulations to model airflow and seed movement.

Table 1. A comparative table of previous works and the novelty of our current work.

Study	Objective	Design/Methodology	Key Findings	Limitations	Novelty of Current Work
Kumar and Durairaj [8]	Analyze head geometry’s role in seed distribution.	Experimental testing with different head shapes	Head geometry directly affects seed uniformity.	Limited to standard geometries and seed types.	Expands the analysis to three advanced designs (A1, A2, A3) across multiple seed types (wheat, rice, rapeseed).
Bourges and Medina [31]	Numerical simulation of air-seed distribution in air drills.	CFD simulations	Focused on the impact of air speed on distribution.	Did not include experimental validation or new geometry designs.	Performs CFD-DEM simulations with real-world validation, highlighting the superior performance of A2.
Yatskul et al. [25]	Investigate the effect of divider head geometry on seed distribution accuracy.	CFD simulations and experimental testing	Identified the impact of different geometries on distribution uniformity.	Focused only on basic funnel and pipe geometries.	Introduces barrier-based designs (A2) for improved seed flow control.
Lei et al. [26]	Study seed motion characteristics in pneumatic systems.	CFD-DEM coupled simulations	Found that CFD-DEM coupling improves prediction of seed distribution.	Did not explore barrier effects in distributor heads.	A2 design with barriers minimizes turbulence for better uniformity.
Gierz and Markowski [30]	Study distribution head tilt and diffuser variants.	Experimental testing and simulations	Found that tilting the distribution head improves sowing quality.	Only focused on oat seeds, limited analysis of geometry.	Introduces multi-seed analysis (wheat, rice, rapeseed) and new geometric configurations (A1, A2, A3).

summarizes key studies in this area, highlighting their objectives, methods, findings, and limitations. Our work distinguishes itself by combining CFD-DEM simulations and introducing barrier-based distributor head designs (A2) to address issues like turbulence and inconsistent seed flow, offering a novel approach to optimizing seed distribution.

This study introduces a novel approach to optimizing seed distribution in pneumatic air-seeding systems by employing CFD-DEM simulations and testing barrier-based distributor head designs (A2); this work advances this by integrating CFD-DEM coupling to simulate both airflow dynamics and seed behavior, providing a more accurate prediction of seed placement. This study develops distributor heads with different aerodynamic configurations—funnel distributor (A1), closed-funnel distributor (A2), and cone-shaped distributor (A3)—and evaluates their relative performance in uniformly metering seeds of varying geometries to furrow openers. Local Chinese varieties of wheat (Lei xiaoling932), rice (Quanyou737), and rapeseed (Huayouze62) were considered in the study. The designs (A1, A2, A3) differ in managing the air–seed mixture’s momentum and flow patterns.

The A2 barrier-based design is a significant innovation that reduces turbulence and improves seed flow consistency, issues that have not been sufficiently addressed in prior research. This design stabilizes the air–seed mixture, enhancing seed distribution accuracy and reducing common problems like seed damage and uneven placement. In summary, the novelty of this study lies in the combined use of CFD-DEM simulations and the introduction of barrier-based geometries, offering new insights into optimizing pneumatic seeders.

2. Materials and Methods

2.1. Development of Seed Distributor Heads

Three distributor heads with different aerodynamic configurations were explicitly developed for this investigation. These designs were intended to integrate into a pneumatic system with eight furrow openers.

2.1.1. Funnel Head Distributor (A1)

The funnel distributor in Figure 1 was designed with an inverted truncated cone section, closed at the broader end by a concave disc. The air–seed mixture enters centrally and is divided into eight equal parts through outlet ports positioned symmetrically around the disc’s periphery. These ports are spaced at 45° and 90° intervals, allowing for a balanced distribution. The funnel structure is intended to reduce the velocity of the air–seed mixture as it expands before distributing the seeds into the tubes. The outlets are connected to the vertical air plenum, where air from the blower directs the mixture.

2.1.2. Closed-Funnel Head Distributor (A2)

With the same parameters as those of the funnel distributor design, the closed-funnel distributor in Figure 2 features an inward-tapering funnel that terminates in a closed apex. While the funnel’s shape and angle control the seed flow, the key element in this design is the set of strategically placed barriers near the outlet ports. The main role of these barriers is to regulate the movement of the air–seed mixture, directing it smoothly and evenly toward the eight symmetrically positioned outlets.

2.1.3. Cone-Shaped Distributor (A3)

The cone-shaped distributor has a downward-facing conical structure to guide the air–seed mixture toward the outlets. The distributor in Figure 3 features eight evenly spaced outlets arranged symmetrically around the periphery, which divides the mix into separate streams. The bottom of the cone is connected to a vertical air plenum, where air from the blower ensures seed movement and distribution to the seed tubes.

2.2. EDEM-CFD Model Establishment

The ANSYS Fluent 19.0 (CFD) and EDEM software version 2018 software were used to perform the DEM-CFD gas–solid coupling simulations. The main objective was to examine the impact of using different distributor device structures on seed migration and the effect of barriers inside the distributor.

Table 2. Computational parameters used in the simulations.

Item	Parameter	Value
EDEM software v. 2018	Fixed time steps (s)	5 × 10−6
	Target save interval (s)	1 × 10−3
	Gravity (m s−2)	9.81
CFD software v. 19.0	Time step size (s)	5 × 10−4
	Time step (s)	1 × 10−3
	Airflow density (kg m−3)	1.225
	Velocity (Pa s−1)	15 (rapeseed); 24 (wheat and rice)

details the simulation parameters. The head distribution device module was imported from Creo.5 parametric software. The entry point for the air–seed mixture was configured at the inlet of the seed pipe, while the outlet served as the exit for the mixture.

Figure 4 shows the DEM-CFD coupling simulation process. The CFD solver solves the airflow field, and the DEM-CFD coupling method calculates the fluid drag force, gravity, and buoyancy acting on the particles. The particles’ motion state is simulated within each time step. If the simulation time has not reached the set value, the airflow field and particle motion state are iteratively calculated within each time step.

To simulate the airflow field with a seed volume fraction of below 10%, the Eulerian–Lagrangian approach was utilized in the DEM-CFD coupling simulation. The forces acting on the particles were simulated using the free-stream equation for drag and the Saffman and Magnus lift, as derived from theoretical analysis. The turbulent nature of the gas flow was represented using the standard k-ε turbulence model within the ANSYS Fluent 19.0 software to simulate the continuous airflow field. The Hertz–Mindlin (no-slip) contact model was applied to the EDEM simulation. The casing material in the simulation was assumed to be ABS plastic (acrylonitrile butadiene styrene copolymer). The material properties of ABS and seeds, and the interaction of their mechanical properties, are detailed in

Table 3. Mechanical interaction properties between seeds and ABS for EDEM.

Parameter		Rapeseed	Wheat	Rice	ABS
Three axes diminution, mm3		2 × 2 × 2	6 × 3 × 3	10.2 × 3 × 2
Density kg/m3		1060	1350	1125	1060
Poisson’s ratio		0.25	0.42	0.30	0.394
Shear modules/pa		1.1 × 107	5.1 × 107	1.01 × 108	8.96 × 108
Collision recovery coefficient	Seed–seed Seed–ABS	0.60	0.42	0.50
		0.75	0.60	0.001
Static friction coefficient	Seed–seed Seed–ABS	0.50	0.35	0.50
		0.30	0.40	1.0
Dynamic friction coefficient	Seed–seed Seed–ABS	0.01	0.05	0.01
		0.01	0.05	1.03

, while the airflow field parameters are summarized in Table 1. Due to the minor time step required for EDEM compared to CFD, the time steps for EDEM and CFD were set to 5 × 10⁻⁶ s and 1 × 10⁻³ s, respectively, with the total simulation spanning 8.0 s.

2.3. Numerical Model of Seed Gas Delivery

2.3.1. Development of a Numerical Model for Seed Gas Delivery Based on Air Pressure

To effectively estimate seed gas delivery in pneumatic seeding systems, a complete numerical framework is required to account for the dynamic behavior of seeds as they interact with the airflow. This model applies fluid dynamics and particle mechanics principles, particularly emphasizing the forces applied to seeds due to pressure differentials in the pneumatic system. A detailed model formulation, including the relevant assumptions and governing equations, is provided below.

To simplify the interaction in the seed–air mixture, the model assumes that the seed is a rigid particle in the airflow. Therefore, when the airflow is stable and compressible, it leads to constant seed movement and predictable velocity changes. The seed’s interaction with the pneumatic tube is also expected not to induce deformation, keeping it safe. The forces impacting the seed’s mobility are drag from the airflow and gravity, allowing the model to concentrate on these dominant components for precise behavior prediction.

The motion of the seed in the pneumatic system is governed by Newton’s Second Law of Motion, which states that the net force acting on an object is equal to its mass multiplied by its acceleration. In this context, it is expressed as in Equation (1).

$$m{\displaystyle \frac{dv}{dt}}=F_{drag}+F_{pressure}-mg$$

(1)

where m is the mass of the seed, kg, dv/dt is the acceleration of the seed, m/s², F_drag is the drag force, N, F_pressure is the force due to the pressure gradient, N, and mg is the gravitational force acting downward, N.

The drag force (F_drag) experienced by the seed as it travels through the pneumatic system is expressed in Equation (2).

$$F_{drag}={\displaystyle \frac{1}{2}}{C}_d\rho A{v}^2$$

(2)

where C_d is the drag coefficient, ρ is the air density, kg/m³, A is the cross-sectional area of the seed, m², and v is the velocity of the seed relative to the airflow, m/s.

The force exerted on the seed due to the pressure differential (ΔP) in the pneumatic system can be expressed as follows:

$$F_{pressure}=A∆p$$

(3)

where A is the seed’s cross-sectional area, m², and ΔP is the pressure difference across the seed, Pa.

Based on Equations (2) and (3), the equation of motion governing the acceleration of the seed can be expressed as follows:

$$m{\displaystyle \frac{dv}{dt}}={\displaystyle \frac{1}{2}}{C}_d\rho A{v}^2+A∆P-mg$$

(4)

This equation encapsulates the seed’s dynamic behavior as it travels through the pneumatic delivery system. It highlights the balance among aerodynamic drag, pressure-induced force, and gravity.

2.3.2. Seed–Gas Delivery Ratio

The particle–gas mass ratio (ψ) is commonly utilized to quantify the relative amount of the particle phase in the gas phase. It is defined as the ratio of the particle mass flow rate to the gas mass flow rate.

$$\psi ={\displaystyle \frac{G_p}{G_g}}$$

(5)

where G_p and G_g are the particle and gas mass flow rates, respectively. According to actual agronomic requirements, the rapeseed and wheat seed G_p ranges from 0.63 to 1.43 g/s and from 10.36 to 23.31 g/s, respectively. The particle–air mass ratio (ψ) ranges from 0.1 to 1.0, as the air–solid flow is in a dilute phase.

2.3.3. Inlet Airflow Tube Diameter

The gas mass flow rate affects particle and airflow velocity and pressure in the tube. The particle motion in the airflow field was ultimately changed. Such a change could be expressed as follows:

$$G_g={\rho }_g{\displaystyle \frac{\pi D^2}{4}}{v}_2$$

(6)

where ρ_g = 1.205 kg/m³ is the air density. Combining Equations (5) and (6), the following is obtained:

$$D=\sqrt{{\displaystyle \frac{4G_p}{\pi \psi {\rho }_g{v}_2}}}$$

(7)

Gp and ψ are 23.31 g/s and 0.75, respectively, in Equation (7). Equation (7) indicates that the air outlet diameter (D) is 41.44 mm. Considering the tubes on the market, the air diameter of the inlet and outlet tube (D) was revised to 42 mm.

2.4. Experiment Design

2.4.1. DEM-CFD Coupling Simulation Experimental Experiment

In order to investigate the influence of different configurations on the seed distribution at eight seed outlets in the distribution device, a single-factor experiment was conducted. The distribution devices were simulated during the experiment, with the airflow velocity at the inlet configured to 13 m s⁻¹ for rapeseed and 24 m s⁻¹ for wheat and rice. Based on the seeding rate and seed flow within the distribution devices, the factory was generating 500 rapeseed seeds, 700 wheat seeds, and 550 rice seeds at the seed pipe inlet per second. The simulation ran for 8 s, with the mass of the seeds at the outlets recorded during the initial 6 s to evaluate how changes in distributor geometry affected the uniformity of seed distribution across the outlets. According to the measured physical and mechanical parameters mentioned by [42], the manual filling of the seed model was established through the EDEM software, as shown in Figure 5. The CFD software was used to establish the seed distributor simulation model, as shown in Figure 6.

The quality of seed metering was evaluated using the coefficient of variation (CV) of seeding uniformity, following the guidelines provided in the GB/T 9478-2005 [43] standard for seeder operation quality. The formula for calculating the coefficient of variation of seeding uniformity is expressed in Equation (8):

$$\left\{\begin{array}{c}x={\displaystyle \frac{{\sum }_{i=1}^n{x}_i}{n}}\\ s={\displaystyle \frac{\sqrt{{\sum }_{i=1}^n{(x_i-x)}^2}}{n-1}}\\ V={\displaystyle \frac{S}{x}}\times 100\%\hfill \end{array}\right.$$

(8)

where x_i is the weight of the seeds at the outlet i, g, x is the average weight of the seeds, g, s is the standard deviation of seeding uniformity, V is the coefficient of variation of seeding uniformity, %, and n is the number of outlets (in this study, n = 8).

2.4.2. Description of Distributor Head’s Bench Test

The testing setup used a particular test bench to assess distributor heads in air-drill seeders. This bench’s primary objective was to enable controlled tests that evaluate the effects of different variables on the distribution heads’ pneumatic seed transport system. The design replicated the essential functionality of an air-drill seeder’s pneumatic system while simplifying it for laboratory conditions. Figure 7 shows that the configuration allowed for the evaluation of a single distributor head during each trial.

Key components of the test system included a centrifugal fan, a seed hopper, and a centralized seed meter, mechanically connected to a gear motor via a chain and controlled by a variable frequency drive (VFD). For accurate data collection, the distribution heads (Figure 1, Figure 2 and Figure 3) were attached to outlet ducts that guided the seeds into plastic containers. This setup was based on techniques developed in earlier research by Scola et al. [44] and Allam and Wiens [45], who modified their methods to assess seed dispersion performance under various operating circumstances.

Table 4. Factors influencing distributor head performance.

Type of Seed	Recommended Sowing Rate kg/ha	Air Pressure, Kpa	Seed Feed Rate with the Air Plenum, Sowing Rate, Distributing System, g/min Travel Speed, km/h
			4 km/h	5 km/h
Wheat	187.5	3.0, 3.5, 4.0, 4.5, 5.0, 5.5	2000	3100
Rice	60	3.0, 3.5, 4.0, 4.5, 5.0, 5.5	600	1000
Rapeseed	6	3.0, 3.5, 4.0, 4.5, 5.0, 5.5	60	100

provides an overview of the variables affecting distributor head performance.

The experiment employed a factorial, completely randomized design, with data analysis conducted using the Design-Expert version 12 statistical software. Variance (ANOVA) was analyzed to assess the statistical significance of the treatment effects on seed distribution performance. The means were arranged in descending order to compare treatment means, and the CV between the different treatment combinations was calculated.

The slightest variation was identified as the most consistent, making it the optimal choice for performance. This robust statistical approach ensured a thorough and precise evaluation of the factors influencing seed distribution, enabling reliable conclusions about the system’s performance.

3. Results and Discussion

3.1. Analysis of DEM-CFD Coupling Simulation Experimental Results

The velocity distribution and flow field of the A1, A2, and A3 distributors are shown in Figure 8. Horizontal and vertical sections of the outlet velocities were analyzed, respectively. The results indicate that the flow field velocity at the outlets of the A2 distributor is more uniform than that of A1 and A3, and the A2 distributor achieves a lower coefficient of variation (CV) in feed uniformity.

Table 5. Simulation results of distribution uniformity for the three types of distributors.

Crop	Type of Distributor
	A1 CV (%)	A2 CV (%)	A3 CV (%)
Rapeseed	0.46	0.25	0.95
Wheat	3.09	2.5	5.5
Rice	3.75	3.1	4.5

shows the calculated CV values for the three distributors. For rapeseed, the CV of A2 is reduced by 45.65% and 73.68% compared to A1 and A3, respectively. The A2 distributor also demonstrates improved uniformity for wheat, with a CV of 2.5%, which is lower than that of A1 (3.09%) and A3 (5.5%). This indicates a CV reduction of 19.09% compared to A1 and of 54.55% compared to A3. Similarly, for rice, the CV of A2 is 3.1%, representing a reduction of 17.33% and 31.11% compared to A1 (3.75%) and A3 (4.5%), respectively.

The velocity contours and seed particle distributions show that A1 and A3 exhibit turbulence and an uneven flow distribution across the outlets. This leads to significant differences in feed rates among the outlets, resulting in higher CV values. In contrast, the A2 distributor minimizes turbulence, allowing the rapeseed–air, wheat–air, and rice–air mixtures to pass through with greater uniformity and lower velocity differences across the outlets. Additionally, the A2 distributor reduces the accumulation of seed particles at the outlets compared to A1 and A3, further enhancing distribution consistency.

The simulation results suggest that the A2 distributor’s optimized structure effectively balances airflow and reduces pressure loss, ensuring a more uniform feed rate. Consequently, the A2 distributor achieves better distribution uniformity than A1 and A3, as reflected in the lower CV values across rapeseed, wheat, and rice.

3.2. Analysis of Bench Test Experimental Results

3.2.1. Influence of Wheat–Air Mixture on Distributing Performance

This study analyses the impact of air pressure and speed on the performance of the wheat–air mixture’s three distributor head designs: A1, A2, and A3. The results shown in Figure 9 illustrate a significant variation in feed rates across the different head geometries when exposed to changes in air pressure and speed.

The ANOVA outputs in

Table 6. Analysis of variance on distributed air pressure with air–wheat mixture.

Source	df	Sum of Squares	Mean Square	F-Value	p-Value
Model	287	16,944.5	59.04	10.34	<0.0001	significant
A—Head	2	366.28	183.14	32.07	<0.0001	**
B—Air pressure	5	210.79	42.16	7.38	<0.0001	**
C—Speed	1	12,275.26	12,275.26	2149.37	<0.0001	**
D—Outlet	7	306.14	43.73	7.66	<0.0001	**
AB	10	297.18	29.72	5.2	<0.0001	**
AC	2	442.08	221.04	38.7	<0.0001	**
AD	14	394.82	28.2	4.94	<0.0001	NS
BC	5	70.23	14.05	2.46	0.0322	NS
BD	35	502.85	14.37	2.52	<0.0001	**
CD	7	71.18	10.17	1.78	0.0886	NS
ABC	10	327.95	32.8	5.74	<0.0001	**
ABD	70	644.05	9.2	1.61	0.002	NS
ACD	14	133.3	9.52	1.67	0.0584	NS
BCD	35	246.77	7.05	1.23	0.1701	NS
ABCD	70	655.6	9.37	1.64	0.0014
Pure Error	576	3289.59	5.71
Total	863	20,234.08

Coefficient of variation = 4.9%. ** Significant at the 1% level. NS = non-significant.

confirm that air pressure and speed influence the distribution performance, with speed showing a powerful effect.

Notably, A2 performs better in maintaining a consistent feed rate across varying speeds and air pressures, a phenomenon that is apparent in the coefficient of variation (CV) among the feed rates of the three head distributors: 4.3% for A1, 2.6% for A2, and 6.9% for A3. This suggests that its design promotes more stable distribution. This is evident in Figure 9, where A2 exhibits lower feed rate fluctuations than A1 and A3, whose feed rates fluctuate more under certain conditions, especially at lower air pressures. A1 is particularly sensitive to these fluctuations, showing more significant variability in feed rates at reduced air pressures.

Figure 10 shows the feed rate variability across the outlet tubes for the different head geometries A1, A2, and A3. It demonstrates A2’s superior uniformity across the eight outlets compared to A1 and A3. This implies that the structure of A2 eliminates turbulence and adequately balances airflow, resulting in the wheat–air mixture being distributed more uniformly. In addition, A1 and A3 show more variation in feed rates across the eight outlets. In detail, A3 achieves a higher overall feed rate and lower uniformity, while A1 shows considerable fluctuations, which could negatively impact applications requiring consistent distribution. It was confirmed that the head distributor with a barrier effectively minimizes outlet-to-outlet variation, making it the most efficient design for ensuring uniform wheat distribution, even with a slightly lower overall feed rate. This consistency is essential in applications where uniform feed rates are crucial.

3.2.2. Influence of Rice–Air Mixture on Distributing Performance

The ANOVA table (

Table 7. Analysis of variance on distributed air pressure with air–rice mixture.

Source	df	Sum of Squares	Mean Square	F-Value	p-Value
Model	287	1.87 × 105	652.62	80.92	<0.0001	significant
A—Head	2	33,563.29	16,781.64	2080.69	<0.0001	**
B—Speed	1	1.13 × 105	1.13 × 105	14,034.54	<0.0001	**
C—Pressure	5	1047.95	209.59	25.99	<0.0001	**
D—Outlet	7	8623.73	1231.96	152.75	<0.0001	**
AB	2	3503.09	1751.54	217.17	<0.0001	**
AC	10	1203.4	120.34	14.92	<0.0001	**
AD	14	16,268.31	1162.02	144.07	<0.0001	**
BC	5	781.24	156.25	19.37	<0.0001	**
BD	7	510.98	73	9.05	<0.0001	**
CD	35	1456.48	41.61	5.16	<0.0001	**
ABC	10	440.23	44.02	5.46	<0.0001	**
ABD	14	1018.94	72.78	9.02	<0.0001	**
ACD	70	3894.36	55.63	6.9	<0.0001	**
BCD	35	488.76	13.96	1.73	0.0064	**
ABCD	70	1307.27	18.68	2.32	<0.0001	**
Error	576	4645.69	8.07
Total	863	1.92 × 105

Coefficient of variation = 3.85%. ** Significant at the 1% level.

) reveals that all main factors, namely A—Head, B—Speed, and C—Pressure, significantly affect the distribution of the rice–air mixture. Furthermore, A—Head shows a remarkably high F-value (2080.69, p < 0.0001), indicating that the head design plays a crucial role in the system’s overall performance.

In Figure 11, similar to the wheat–air mixture, A1 and A3 show more pronounced variability in feed rates as a function of speed and pressure. At the same time, A2 demonstrates greater consistency, with less fluctuation across different conditions. In relation to A1 and A3, the feed rate for the 5 km/h speed condition increases significantly at higher speeds and air pressures.

Furthermore, the coefficient of variation (CV) among the feed rates across the eight outlets through the different distributors is 4.5% for A1, 3.4% for A2, and 6.2% for A3, showing that A2 exhibits a lower CV compared to A1 and A3, signifying better uniformity in distribution across the outlets. Figure 12 also proves that A2 maintains a more consistent feed rate across all outlets, with less deviation in performance compared to A1 and A3, indicating a more uniform distribution of rice–air mixture among the outlets.

3.2.3. Influence of Rapeseed–Air Mixture on Distributing Performance

For rapeseed, the ANOVA table (

Table 8. Analysis of variance on distributed air pressure with air–rapeseed mixture.

Source	df	Sum of Squares	Mean Square	F-Value	p-Value
Model	287	1602.91	5.59	31.26	<0.0001	significant
A—Head	2	74.12	37.06	207.42	<0.0001	**
B—Speed	1	1136.67	1136.67	6361.86	<0.0001	**
C—Pressure	5	37.92	7.58	42.45	<0.0001	**
D—Outlet	7	55.1	7.87	44.06	<0.0001	**
AB	2	26.16	13.08	73.2	<0.0001	**
AC	10	19.22	1.92	10.76	<0.0001	**
AD	14	194.18	13.87	77.63	<0.0001	**
BC	5	2.61	0.5213	2.92	0.013	NS
BD	7	1.39	0.1988	1.11	0.3534	NS
CD	35	8.59	0.2455	1.37	0.0774	NS
ABC	10	13.45	1.34	7.53	<0.0001	**
ABD	14	3.05	0.218	1.22	0.2557	NS
ACD	70	14.23	0.2034	1.14	0.2174	NS
BCD	35	4.48	0.128	0.7164	0.8878	NS
ABCD	70	11.74	0.1677	0.9384	0.6199	NS
Error	576	102.91	0.1787
Total	863	1705.82

Coefficient of variation = 2.91%. ** Significant at the 1% level. NS = non-significant.

) results indicate that the A—Head shows a significant influence on feed rate variation (F-value = 207.42, p < 0.0001), implying that the design of the distributor head plays a crucial role in determining the overall efficiency of the distribution system.

Figure 13 shows that A1 and A3 experience a relatively wider range of feed rates compared to A2. A2 maintains a more stable and consistent performance. A2 demonstrates the least variation in feed rates across the different pressure levels and speeds, as proven by the nearly flat trend lines in Figure 13. In turn, this proves that A2 provides a more uniform distribution of rapeseed, reducing the variability caused by air pressure and speed fluctuations. The variation in feed rates across the eight outlets is an important indicator of how uniformly the rapeseed is distributed. An analysis of the coefficient of variation (CV) for A1, A2, and A3 revealed values of 0.3%, 0.1%, and 1%, respectively, among the outlets, illustrating that A2 has the lowest variation in feed rate compared to both A1 and A3, indicating a more consistent performance across outlets.

As shown in Figure 14, A2 has a higher degree of uniformity, with closely clustered feed rates across the outlets. This is proven by the minimal spread of the data points, suggesting that A2 provides more consistent airflow and material distribution. This, in turn, indicates that the head distributor with the barrier is more effective in reducing airflow turbulence, improving feed rate accuracy across the outlets. This uniformity is essential for efficient distribution, as it ensures that each outlet receives the same material.

3.3. Discussion

The results obtained from this study highlight the impact of different distributor head geometries (A1, A2, A3) on the precision and uniformity of seed distribution in air-assisted seeding systems. Our study demonstrates that modifying the design of the distributor head significantly improves seed placement accuracy, reducing the coefficient of variation (CV). The findings align with previous studies that suggest the importance of optimizing the distributor head for improved seed distribution.

The sensitivity analysis reveals that air pressure and distributor head design significantly impact seed distribution accuracy, with air pressure having the most profound effect. Speed also influences distribution, but its impact is comparatively minor. The interaction between air pressure and distributor head designs further enhances seed uniformity, with optimal results observed at medium air pressures and under specific distributor head configurations.

Our study’s findings are consistent with those of Yatskul et al. [25] and Kumar et al. [8], who also identified that alterations to the distributor head design could reduce the variation in seed placement. In particular, our A2 model, with its barrier-based geometry, shows significant improvements in seed distribution uniformity compared to the conventional funnel design (A1). This design addresses issues related to turbulence and uneven air–seed mixing which have been highlighted in previous studies as significant causes of poor seeding performance.

These findings are consistent with the results of Yatskul et al. [25], who reported higher CV values for similar crops under conventional distributor designs. For instance, their study showed a CV value of 4.5% for wheat using a traditional distributor design, a value which is higher than the 2.5% observed in our study for wheat with the A2 design. Similarly, Lei et al. [26] found CV values of approximately 4% for rice when using standard distributors, whereas our A2 design improved this to 3.1%.

In this study, the coefficient of variation (CV) values are used to assess the uniformity of seed distribution across different distributor head designs (A1, A2, A3) for rapeseed, wheat, and rice. The CV values observed in this study indicate significant improvements, especially with the A2 design, which demonstrates reduced CV values compared to A1 and A3 for all the crops tested. The CV for rapeseed is 0.46% for A1, 0.25% for A2, and 0.95% for A3, with the A2 design showing a 45.65% reduction compared to A1 and a 73.68% reduction compared to A3. Similarly, for wheat, the CV values are 3.09% for A1, 2.5% for A2, and 5.5% for A3, with A2 showing a 19.09% reduction compared to A1 and a 54.55% reduction compared to A3. For rice, the CV values are 3.75% for A1, 3.1% for A2, and 4.5% for A3, with A2 showing a 17.33% reduction compared to A1 and a 31.11% reduction compared to A3.

In addition, our results indicate a clear correlation between the air pressure and speed settings and the seed distribution quality. These findings are consistent with those of Gierz and Markowski [30], who found that airflow velocity significantly influences the uniformity of seed sowing.

However, unlike previous studies, our research goes further by integrating the impact of barrier designs in the distributor’s head, offering a novel approach to mitigate the effects of air pressure fluctuations during seed transport. This could prove particularly beneficial in varied field conditions, as Hu et al. [19] observed, where uneven terrain exacerbates seed distribution issues.

The practical implications of this work are far-reaching. For farmers, improving seed uniformity can lead to better crop establishment, reduced seed competition, and, ultimately, higher yields. In terms of agricultural machinery manufacturers, integrating these optimized distributor heads into existing pneumatic seeders could significantly improve their performance, reducing seed wastage and ensuring more efficient seed placement. This is especially critical for modern agricultural practices, where precision is key to maximizing productivity and sustainability.

While our study focused on the impact of distributor head geometry, we acknowledge the potential for integrating machine learning algorithms in future work to predict optimal seeding conditions in real time. This would allow further refinement of the seed distribution process, especially under varying field conditions.

4. Conclusions

Several factors, including head geometry, input air pressure, seed feed rate, and seed properties, significantly influence head distributors’ feed rate and distribution accuracy. The research findings highlight the following:

• The A2 distributor consistently outperforms A1 and A3 across all seed–air mixtures (wheat, rice, and rapeseed) regarding distribution uniformity and stability due to its barrier design, which effectively mitigates turbulence.
• A2 achieved the lowest coefficients of variation (CV): 2.6% for wheat (compared to A1′s 4.3% and A3′s 6.9%), 3.4% for rice (compared to A1′s 4.5% and A3′s 6.2%), and 0.1% for rapeseed (compared to A1′s 0.3% and A3′s 1%).
• CFD-DEM simulations demonstrate significant CV reductions for A2 compared to A1 and A3—45.65% and 73.68% for rapeseed, 19.09% and 54.55% for wheat, and 17.33% and 31.11% for rice, respectively.
• The optimized structure of the A2 distributor minimizes turbulence, reduces velocity differences and pressure loss, and prevents seed particle accumulation, ensuring consistent and uniform feed distribution.
• Further research using finite element analysis (ANSYS) is recommended to study airflow dynamics, refine head configurations, and enhance distribution efficiency for practical applications.

This study establishes A2 as the most efficient distributor, offering superior uniformity and precision, which is critical for optimizing seeding efficiency and supporting uniform crop establishment.