|
Development of an Electric Variable Air Assist System for Apple Orchard Sprayers
Hongyoung Jeon1,*, Heping Zhu1
Published in Journal of the ASABE 67(4): 853-864 (doi: 10.13031/ja.15853). 2024 American Society of Agricultural and Biological Engineers.
1 Application Technology Research Unit, USDA Agricultural Research Service, Wooster, Ohio, USA.
* Correspondence: hongyoung.jeon@usda.gov
The authors have paid for open access for this article. This work is licensed under a Creative Commons Attribution NonCommercial NoDerivatives 4.0 International License https://creative commons.org/licenses/by nc nd/4.0/
Submitted for review on 10 October 2023 as manuscript number MS 15853; approved for publication as a Research Article by Associate Editor Dr. Reza Ehsani and Community Editor Dr. Harold Thistle of the Machinery Systems Community of ASABE on 29 March 2024.
Mention of company or trade names is for description only and does not imply endorsement by the USDA. The USDA is an equal opportunity provider and employer.
Citation: Jeon, H., & Zhu, H. (2024). Development of an electric variable air assist system for apple orchard sprayers. J. ASABE, 67(4), 853-864. https://doi.org/10.13031/ja.15853
Highlights
Abstract. Air assist for apple orchard spray applications is a necessity to deliver pesticides to target crops. However, conventional orchard sprayers use axial fans to provide air assists which are generally designed for tall crops. Thus, these systems have very limited capabilities to control airflows to match canopy densities. An electric air assist system (EAAS) was developed with an electric fans, pulse width modulation (PWM) controllers, a custom-design air channel, and a 400-Ah LiFePO4 battery to address such limitations. The system could ramp up its airflow to nearly 80% of its maximum in 2.5 s, while it took approximately 4 s to reach 100% airflow from no air assist. The EAAS provided airflow of 11.3 and 5.3 m s-1 at the fan outlet and 1 m away, respectively, with 100% duty cycles (DC) of PWM. It was capable of modulating its average air velocities from 0 to 11.3 ms-1 by changing DCs. Spray coverage samples of constant rate applications with different air assist levels were collected from the behind apple tree canopies throughout the 2023 growing season to characterize the potential for off-target spray drift. A multivariable regression model for controlling DCs was developed with the spray coverage data and leaf area index (LAI) to minimize the off-target drift. The EAAS was a promising approach to developing more advanced air assist spraying systems to enable the adjustment of both air and liquid flows based on canopy characteristics to maximize spray depositions on intended targets while minimizing spray drift.
Crop protection product (CPP) application is a key production process to protect crops from disease or insect damage and reduce the weed population. The weed control spray application is a standardized process as it generally applies a uniform rate in unit area. However, CPP applications for tree crops to control disease or insects are complicated because their size and foliage density are continuously changed during the time that requires CPP applications. Such variations are observed even in the same row (Colaço et al., 2019).
One of the approaches to addressing these issues for CPP applications is real-time variable rate applications by varying spray outputs based on crop sizes and foliage densities. Research progress in real-time variable rate spray applications with ultrasonic sensors (Giles et al., 1988; Gil et al., 2007; Stajnko et al., 2012; Jeon and Zhu, 2012; Maghsoudi et al., 2015), a LiDAR (Light Detection and Ranging) system (Chen et al., 2012; Yan et al., 2019), or a stereo vision system (Jeon and Zhu, 2022; Román et al., 2023) has been continuously made to match sprayer outputs with crop canopy sizes since the benefits from the technologies are obvious in terms of reducing CPP uses. The benefits from these technologies, for example, are the reduction of the off-target movements of CPP sprays into the environment (Chen et al., 2012; Nackley et al., 2021) and the saving of labor and operation costs by treating more acreage with the same volume of tank mix compared to conventional sprayers (Warneke et al., 2021). As a result, a commercially available laser-guided variable-rate spray control system was developed as a retrofit onto existing sprayers for growers to reduce pesticide waste and safeguard the environment (Zhu, 2023). However, this new system can only control liquid outputs but not airflow to match crop canopy characteristics.
Although variable-rate technologies for tree crop sprayers have made substantial technological advancements (Zhu, 2023), their air assist system part transporting spray droplets stayed unchanged (Fox et al., 2008). Common air assist systems are equipped with an axial fan driven by PTO (power take-off) at a constant RPM during spray applications, which discharges a consistent air volume to the area that is being sprayed regardless of the conditions of crop canopy sizes and foliage density. This is problematic for tree crop CPP applications, as spray applications with uniform air assist to non-uniform canopy sizes and densities may result in large variations in spray depositions or spray loss beyond the canopies due to consistent over- or under-air-assisted sprays. In addition, one of major issues with uniform air assisted spray occurs during applications for low canopy density crops, resulting in over air assist, which have negatively affects spray depositions (Marucco et al., 2008). Furthermore, excessive air assist for tree crop spray would prompt unintended off-target movements (Zhu et al., 2006; Gu et al., 2014) and reduce spray coverage of fine spray droplets on plants (Salas et al., 2022). Prior research pointed out that adjusting air assist intensity enhanced spray deposition and reduced spray drift compared to applications without the adjustment in vineyard and apple orchard applications (Landers and Muise, 2010). The need for airflow adjustment to increase spray depositions was also highlighted in Pascuzzi et al. (2017), who tested vineyard spray applications with airflows of air assist at 2.43 and 5.71 m3 s-1 for spray deposition differences in two different growth stages. Their results indicated that spray applications have more spray depositions with airflow of 2.43 m3 s-1 in the full flowering stage and with airflow of 5.71 m3 s-1 in the berry touch growth stage, which suggests different air assist intensities are necessary in different growth stages of the vine. In addition, Miranda-Fuentes et al. (2015) tested spray applications in high-density olive tree orchards with air assist at 6.15, 8.90, and 11.93 m3 s-1. Their results indicated the medium airflow (8.90 m3 s-1) achieved the highest spray deposit on the canopy, although olive trees had high-density canopies with a leaf area density of 6.04 m2 m-3.
These evidences demonstrate that the air assist system in a sprayer for tree crops should be adjustable for different tree crops. However, conventional sprayers generally have two settings for adjusting the intensity of air assist. Research efforts have been made to address this issue, and a common approach to addressing this issue is to adjust the air assist levels of sprayers by manually adjusting the inlet size of an axial fan. For example, Marucco et al. (2008) blocked the fan inlet with a wood disc to control airflow to the outlets, and they showed the air velocity discharged was proportional to the opening area. Another example is Gu et al. (2014), who mounted an iris damper on the axial fan inlet to control airflow by changing the iris damper opening, and they showed the canopy's external and internal airflow could be varied by adjusting the iris damper opening. Although previous research showed the adjustment of the fan inlet opening could lead to variable air assists, the approaches would not be feasible for real-time variable air assists since these approaches required manual adjustments of the air inlet opening.
Furthermore, Pai et al. (2009) used an air deflector with a servo motor to adjust its position to regulate airflow at the outlets as a variable-rate air-assist system. Their system could adjust airflow based on the height, volume, and foliage density of a tree. However, the system was incapable of controlling fan speed, which often resulted in over-air assist. Another approach to controlling the intensity of air assist is Khot et al. (2012), who added louvres to the air outlets of an orchard sprayer with an axial fan. They added two louvers with linear actuators on both sides of air outlets to adjust the opening area of the outlets, which can adjust exit air speeds approximately from 5 to 28 m s-1 although the speed of the adjustment was undisclosed. The sprayer with adjustable louvre positions was tested in a small citrus (approximately 2 m high) orchard for spray depositions at a travel speed of 4.5 km h-1 when the louvres were fixed in fixed positions to produce 40%, 70%, and 100% of air assist. Their study showed that full capacity air assist had spray depositions compatible with those with 70% air assist, which indicated 100% air assist was not necessary for those trees. Holownicki et al. (2017) also developed a prototype variable air assist system that used an electric motor to run an axial fan in a chamber. The fan blew air into the chamber, and the air jets were discharged through a vertical opening slot to provide air assist to spray for better penetration. Their system could change the air volume and directions by adjusting impeller blade angles or rotational speed, and the performance of their air assist system could match the performance of the commercial air assist system, although it required high power (1.9–12.6 kW per fan) and did not test with a field sprayer for its effects on spray deposition or coverage.
Recently, Mahmud et al. (2022) mounted an iris damper coupled with a stepper motor and a microcontroller to control the inlet diameter of a sprayer’s axial fan from 0.34 m to 0.80 m to regulate its airflow. They used the canopy detection results of a LiDAR sensor to determine the iris opening. Their air assist system reiterated that controlling the opening of the air inlet could regulate the airflow of the air assist system for tree crop applications, and regulating airflow based on the canopy density could reduce spray off-target movement. However, their control system for regulating iris damper opening had a relatively slow response time, so they determined the air assist intensity from the first 10 trees to be used for a row or a block.
Prior research work showed potential paths to developing variable air assist systems for conventional sprayers and their benefits, although relevant research work is still at a preliminary stage. However, most prior work was focused on developing an attachment or apparatus to add to existing sprayers to provide variable-rate air assist to spray. Although such approaches have the potential to improve the air assist performance of existing sprayers, adjusting their intensities is inherently limited as existing air assist systems are not designed for variable air assists. Therefore, the objectives of this research were to develop a prototype of electric air assistance system for a new tree crop sprayer and to evaluate its performance and characteristics for variable air assisted spray application.
Materials and Methods
Fan, Controller, and Duct Design
An electric pulling fan (VA18-AP71/LL-59A, Spal Automotive Srl, Correggio, Italy) was used to control the airflow of the air assist system. The fan diameter is approximately 397-mm with a rated maximum airflow of 3430 m3 h-1 (Spal Automotive Srl, 2022). The fan was selected based on its airflow and preliminary field assessments in an orchard with mature apple trees (over 30 years old) similar to commercial apple trees in Ohio. A pulse width modulation (PWM) stepper motor driver (HiLetgo Technology Co. Ltd., Shenzhen, China) coupled with a microcontroller (Arduino Uno Rev3, Arduino Co., Somerville, MA, US) controlled the fan speed by changing duty cycles (DC) of PWM signals. A LiFePO4 battery (rated voltage: 12.8 volts direct current, Redodo LiFePO4 battery, 400 Ah, Shenzhen Maicheng Technology Innovation Co., Ltd., Shenzhen, China) provided the power to the system.
The fan was mounted at the center of a square steel frame (406-mm (width (W)) × 406 mm (height (H))), and two duct parts, a straight duct and a reducer, were connected to the frame in series, and its air velocities at the opening were measured while running it at 100% DC. Two different widths of the openings, 152- and 305-mm, were selected to accommodate the spray nozzle head with air assist. Both openings had the fan, a 152-mm straight duct, and a 305-mm reducer (with either a 152-mm or a 305-mm opening) in series. Air velocities from the fan with the straight duct (the opening of 406-mm) were used only as a reference configuration for air velocities. All measurements were made at the upper (305-mm above the bottom of the opening), the middle (203-mm above the bottom), and the lower positions (102-mm above the bottom) of the opening.
After determining the opening width, two fan and duct configurations were selected based on air velocities after evaluating various combinations of the reducers and straight duct. The first configuration was the fan, 305-mm reducer, and 152-mm straight duct in series (configuration 1), and the other was configured with the fan, 305-mm straight duct, and 305-mm reducer (configuration 2) in series. Air velocities and their uniformity at their openings for both configurations were evaluated at distances of 0 m (at the opening), 0.25 m, 0.5 m, 1.0 m, and 1.5 m from the horizontal center of the opening, as the distance of 1.5 m was an approximate distance between the EAAS outlet and outer surface of the tree canopy considering the size of tree canopies, the length of the EAAS duct, and the distance between trunks and the center of the sprayer. In addition, air velocities in lateral positions (±100-mm laterally from the center with a 50-mm increment for the distances of 0.25 and 0.50 m and ±300 mm laterally from the center with a 100-mm increment for the distances of 1.0 and 1.5 m) were measured to determine the uniformity of air assistance near the center. The measurement locations for air velocities remained the same throughout the experiments.
An air velocity meter (9545-A, TSI Inc., MN, USA) was used to measure air speeds, and the fan operated at 100% DC (always on) for approximately 5 s to stabilize fan speed before starting air speed measurements. The meter acquired air speeds for 5 seconds, and an average of each measurement was recorded. The air velocities at each position were measured 10 times under indoor conditions without air movement. During the measurements, the path of airflow was clear of any obstructions to prevent any potential airflow resistance. The measurements were taken at indoor temperatures ranging from 18.5 °C to 24.7 °C.
Fan Response, Power Consumption, and Variable Airflow Characteristics
Fan response time was evaluated by measuring the time that the electric air assist system (EAAS) reached its maximum air speed from a complete stop after powering the fan with 100% DC for 10 times. The EAAS was controlled by the PWM driver, which took PWM signals with 100% DC from the microcontroller. A pre-calibrated Wind Sensor Rev. P (Modern Device Co., Brooklyn, NY, USA) was placed at the horizontal center of the lower opening position of the EAAS to acquire air velocities as the position had more airflow than the other positions (fig. 1). The sensor outputs and TTL (Transistor–Transistor Logic) signals were acquired by a high-speed data acquisition system (USB-204, Measurement Computing Co., Norton, MA, USA) and a custom designed Visual Basic.NET application (Visual Studio 2017, Microsoft Co. Ltd., Redmond, WA, USA) at 1000 Hz for 10 s. In addition, a True-Root Mean Square (TRMS) clamp meter (375FC, Fluke Co., Everett, WA, USA) acquired current consumptions of the EAAS at 4 Hz while the EAAS was starting and operating to measure the power use of the system. The air velocities of the EAAS were also measured in the aforementioned condition while operating it with DCs from 10% to 100% to determine a range of air velocities from the EAAS. Air velocities were measured with the air velocity meter for 5 s at the horizontal center of the opening in the upper, middle, and lower positions of the opening after the fan operation was stabilized, typically 5 s after powering the EAAS. All measurements were made at 0 m (at the opening) and 1 m away from the opening and repeated 10 times.
Canopy Pass Through Spray Coverage
Sprayer and Electric Air Assist System
A prototype sprayer with the spray delivery system (Jeon and Zhu, 2012) and a vertical boom was used for this experiment. The boom had five spray nozzles (XR11006, TeeJet Technologies, Glendale Heights, IL, USA) with a nozzle space of 0.41-m (fig. 2), and each nozzle had an EAAS behind it. A 2.1-m high air duct with a 305-mm reducer and a 152-mm straight duct was made and integrated into the sprayer to direct air assist from the EAAS. Each fan was connected to the PWM driver, which controlled the air velocities of the EAAS by DCs of PWM signals from a microcontroller (Seeeduino Nano, Seeed Studio, Shenzhen, Guangdong, China).
 |
| Figure 1. A calibration curve for Wind Sensor Rev. P to measure air speeds. |
 |
| Figure 2. Experimental tower sprayer with an electric air assist system. |
Canopy Pass Through Spray Coverage
Three apple trees were selected from an apple orchard in Horticultural Research Unit 2 of The Ohio State University in Franklin township, Wayne County, OH (fig. 3) to collect canopy pass through spray coverage (CPTSC). CPTSC was also defined as spray coverage of spray passing through a canopy being potentially drifted away. Selected trees were untrained, 8-year-old golden delicious (Malus domestica ‘golden delicious’) as they had representative size, shape, and foliage density in the orchard. A stationary cross aluminum frame was placed approximately 30.3 cm behind the canopies of each tree, and each frame had five water-sensitive papers (WSP) to collect CPTSCs during the tests. WSPs were positioned at the top, middle, and bottom of the horizontal center of the canopies, one on the following edge of the canopy and the other between the center and leading edge of the canopy, to have various canopy conditions before spray droplets reached WSPs.
The sprayer equipped with the EAAS sprayed local tap water through five spray nozzles (XR11006) at 275 kPa while a tractor (John Deere 6310S, Deere and Co., Moline, IL, USA) carried the sprayer at a travel speed of 6.1 km h-1 to apply the water at a spray volume of approximately 400 L ha-1. The spray volume was approximately twice as much as the recommended spray volume of a commercial fungicide (Syngenta Crop Protection LLC, 2021) to have more air assist spray available to pass through the canopies. Five levels of air assist, DC of 0% (all fans were off), 30%, 50%, 70%, and 100% (all fans were fully on), were selected for the spray application to measure CPTSCs from different levels of air assist over the 2023 growing season. Expected average air velocities at the opening of the EAAS with DC of 0%, 30%, 50%, 70%, and 100% were approximately 0 (0 m3 h-1), 3.4 (745.4 m3 h-1), 5.9 (1300.0 m3 h-1), 8.5 (1885.9 m3 h-1), and 11.3 (2510.7 m3 h-1) m s-1, respectively. The distance between the center of the sprayer and the tree trunk was approximately 2.7 m, and the DCs of the EAAS were changed manually after each test.
After each test with the five levels of air assist, WSPs were dried, collected, and kept in a paper envelope. Dried WSPs were scanned at 600-dpi resolution and processed by DepositScan (Zhu et al., 2011) to determine the spray coverage of each WSP. In addition, the leaf area index (LAI) of each tree was measured with a canopy analyzer (LAI-2200C, Li-Cor Biosciences, Lincoln, NE, USA) after each test to characterize tree canopy density. A total of six tests were carried out to collect CPTSCs over the 2023 growing season, from no canopy stage (14 April 2023) to a fully developed canopy stage (29 June 2023). Figure 4 shows the canopy conditions of three trees over the growing season, including average wind speed and direction. The weather conditions were recorded by a weather station nearby. The average wind speeds ranged between 0.2 and 0.9 m s-1 from WSW (253 degrees) to SE (140 degrees), and the ranges of average temperatures and relative humidities during the tests were from 2.5 to 17.8 °C and 64.0% to 77.9%, respectively. A multivariable linear equation with average CPTSCs and LAIs was developed using JMP (Ver. 16.2.0, SAS Institute Inc., Cary, NC, USA) to predict DCs for the EAAS for attaining intended CPTSCs (potential spray drift risk) at a given LAI of the apple trees.
 |
| Figure 3. A schematic of the apple orchard used for collecting canopy pass through spray coverage discharged from the experimental variable air assist tower sprayer. Drawing is not to scale. |
 |
| Figure 4. Canopy and leaf density conditions of three trees at the time of tests to collect canopy pass-through spray coverage, including average leaf area index of trees 1, 2, and 3, and average wind speed (in m s-1) and direction (W: west; S: south; E: east) during the test. |
Results and Discussion
Fan, Controller, and Duct Design
Electric fans with different duct configurations had wind speeds ranging from 3.4 to 12.6 m s-1 at different positions of the outlets (table 1). Generally, the fan blew substantially more air in the upper and lower positions than in the middle position, regardless of opening widths and duct configurations, likely because the motor at the center of the fan limited the airflow in the middle. In addition, having a reducer in the configuration increased air velocities compared to the configuration without the reducer. For example, the fan generated an average air velocity of 5.8 m s-1 from the reference configuration (no duct on the fan); however, with a reducer, the average air velocities were increased to 11.3 and 7.8 m s-1 for 152- and 305-mm openings, respectively. The fan with a 305-mm opening had similar airflow, with an average air speed increase of approximately 34.5% compared to the reference configuration. The fan with a 152-mm opening generated a higher average air velocity (95% increase); however, its airflow was reduced to approximately 73% of the reference configuration, likely due to flow restriction from the reducer. The fan with the reducer with a 152-mm opening had more uniform air velocities with less than a 12% coefficient of variation (CV) of average air velocities in three positions compared to the other configurations (CVs ranging from 21.3% to 36.7%).
| Table 1. Average air speeds (m s-1) from electrically driven fan with different duct openings. | |||
| Duct Openings (mm) | |||
| 152 | 305 | 406[a] | |
| Upper position | 12.6 (0.6)[b] | 9.1 (2.0) | 7.5 (5.6) |
| Middle position | 9.9 (1.5) | 5.9 (4.4) | 3.4 (13.6) |
| Lower position | 11.5 (1.5) | 8.4 (5.4) | 6.4 (1.32) |
| Average | 11.3 (12.0) | 7.8 (21.3) | 5.8 (36.7) |
| Average airflow (m3 h-1) | 2513 | 3481 | 3449 |
[a] Reference configuration. Fully open with a 305-mm straight duct only without a reducer. [b] CV (%) of air speed measurements are presented in parentheses. | |||
Two duct configurations with an opening width of 152-mm (refer to configurations 1 and 2) had average air velocities of 10.6 to 3.9 m s-1 and 11.3 to 4.5 m s-1, respectively, when the distances from the opening were 0 to 1.5 m (table 2). The data suggest that the EAAS with both configurations could deliver air assists with air velocities ranging from 3.9 m s-1 to 4.5 m s-1 for spray applications near the outer canopy surfaces of trees. Both configurations had more airflow from the upper or lower position at the opening; however, vertical airflow along the opening became more uniform as it was further away from the opening. The highest variations of air velocities in lateral positions from the center were observed at a near distance (0.25 m), and the variations were reduced when the measurement distance was further away from the opening. When air velocities were measured at 1.5 m away from the opening, the average air velocities were down to 37.1% and 40.0% from the opening for configurations 1 and 2, respectively. The data show that configuration 2 had less variations in air velocities, with a higher average air velocity (approximately 0.6 m s-1 higher) at 1.5 m away from the opening, a likely distance between nozzles and tree canopies during tree crop spray applications. This suggests that better and more uniform spray penetrations could be achieved with configuration 2 over configuration 1. Thus, configuration 2 was selected for the EAAS prototype to evaluate for air assist performance.
| Table 2. Average air speeds (m s-1) from electrically driven fan with Two duct configurations with same opening. | ||||||
| Distance From the Opening (m) | ||||||
| 0 | 0.25 | 0.5 | 1.0 | 1.5 | ||
| Configuration 1 (305-mm straight, and 305-mm reducer duct) | Upper position | 12.3[a] (0.7)[b] | 7.3 (57.8) | 6.2 (28.9) | 3.4 (51.1) | 3.1 (38.3) |
| Middle position | 8.4 (4.1) | 7.8 (20.0) | 6.2 (16.9) | 4.0 (45.5) | 3.9 (29.4) | |
| Lower position | 11.1 (1.0) | 7.6 (50.4) | 7.3 (49.0) | 4.6 (46.8) | 4.7 (30.1) | |
| Average | 10.6 (15.7) | 7.6 (44.9) | 6.7 (34.3) | 4.0 (49.2) | 3.9 (36.1) | |
| Configuration 2 (305-mm reducer and 152-mm straight duct) | Upper position | 13.1 (1.0) | 7.5 (62.3) | 7.1 (17.1) | 4.3 (45.4) | 4.3 (30.8) |
| Middle position | 8.5 (5.0) | 7.7 (36.3) | 7.6 (14.3) | 4.2 (41.8) | 4.4 (22.5) | |
| Lower position | 12.5 (1.0) | 8.5 (44.1) | 9.6 (23.7) | 4.7 (50.0) | 4.9 (24.6) | |
| Average | 11.3 (18.4) | 7.9 (48.4) | 8.3 (26.7) | 4.4 (46.3) | 4.5 (26.6) | |
[a] Average air speeds measured from different lateral points at upper, middle, and lower positions except for the measurements at the opening (0 m), where the average measurements were from the center of each position. [b] CV of air speed measurements from different lateral positions. | ||||||
Fan Response, Power Consumption, and Variable Airflow Characteristics
Figure 5a shows the average air velocity changes at the opening of the EAAS. The air velocity measurements between replications had less than 3% CV. Air velocities from the EAAS were stabilized after approximately 4 s (around the 5 s mark) after powering it (the high TTL signal at the 1 s mark indicates the powering of the EAAS). After powering the EAAS, its air velocities started to increase after having the power for 200 ms, and its air velocities increased much faster during the initial 2 s. For instance, its average air velocity increased from 0 to 10.3 m s-1 during the initial 2 s (from 1 s to 3 s mark), and the increase was less than 4 m s-1 for the next 2 s (from 3 s to 5 s mark). In addition, powering the EAAS for more than 4 s led to a slight increase in air velocities; however, the increases were relatively small (air velocity increase of approximately 1.2 m s-1).
Figure 5b shows an example of current consumption of the EAAS for 5 s (time marks from 1 s to 6 s) when it powered from the stop with a DC of 100%. High start-up current uptake (up to 33 A (startup power of 422.4 W)) of the EAAS was observed immediately from the start. Its current uptake started decreasing after approximately 750 ms, and its operational current, approximately 17.4 A, was observed after 2.5 s from the start. This trend of current consumption was related to air velocity changes: the air velocity increase of the EAAS by start-up current was more than twice the air velocity increase by operational current. Since the EAAS draws a large amount of the startup current, the power source for the EAAS must be able to provide a high current for the initial few seconds to ensure the proper start of the EAAS. In addition, although the startup current was approximately twice as high as the operational current, it was responsible for increasing power consumption of the EAAS by approximately 22% if it operated for 5 s continuously. The impact of high startup current in the power consumption of the EAAS would decrease as the operation time increased. This implies that the EAAS with one fan could run at 100% DC continuously for over 20 h with a 400Ah LiFePO4 battery when it is fully charged. Modulating the EAAS could increase its power consumption due to its high start-up current, although modulation frequencies would be a critical factor.
 |
| (a) Average air velocity changes of the EAAS opening up on starting. |
 |
| (b) Current consumption of the EAAS operating with 100% DC. |
| Figure 5. Air velocity and acceleration changes of the lower position of the EAAS opening after starting the fan. |
Average air velocities from the EAAS were from 0.9 to 11.3 m s-1 at 0 m away from the opening and from 0.3 to 5.3 m s-1 at 1 m away from the opening for DCs of 10% to 100%, respectively, and the standard deviations of the measurements were from 0.01 to 1.21 m s-1 (measurement CVs of 0.63% to 18.51%) (fig. 6). Air velocities from the EAAS increased as DCs increased. For example, every 10% increase in DC for the EAAS increased the average air speeds by approximately 1.16 m s-1 and 0.56 m s-1 (approximately 44.6–47.6% of the air velocities at the opening), at 0 m and 1 m away from the opening, respectively. At the opening, vertical variation of air velocities was observed mainly because the upper and lower positions had more airflow than in the middle, regardless of DCs. However, the variations decreased substantially at 1 m away from the opening, which suggests that although there were variations in airflow at the opening, the variations would diminish near tree canopies.
Canopy Pass Through Spray Coverage
Figure 7 shows examples of CPTSCs on WSP 2, 3, and 4 from tree 3 by spray applications with different intensities of air assist, which were acquired from the back of the thickest canopy part. Thus, spray droplets with sufficient air assist would have spray coverage on WSPs. CPTSCs varied by air assist intensities from the EAAS. For example, almost no CPTSCs (0.06%–0.44%) were observed on WSPs when there was no air assist except a few small droplets that might be drifted by ambient wind in the field. However, when the air assist intensities from the EAAS went up by increasing DC to 50%, the CPTSCs increased to 4.3%–11.6%. Airflow from the EAAS with 70% and 100% DCs increased the CPTSCs further to 6.2%–24.2% and 45.1%–60.1%, respectively. This indicates that the EAAS provides sufficient air assist and does not need to run at 100% DC for the apple orchard where the tests were carried out, as it would move too many spray droplets passing through tree canopies, which might lead to spray off target movement or more loss to the ground.
 |
| (a) Air velocities at the opening of the EAAS. The error bars indicate standard deviations of measurements. |
 |
| (b) Air velocities at 1 m away from the opening of the EAAS. |
| Figure 6. Air velocities of the EAAS at different locations of the opening with duty cycles from 10% to 100%. |
 |
| Figure 7. Canopy pass through spray coverage (numbers in parentheses) on WSP 2, 3, and 4 (Middle canopy) on tree 3 on 5 May 2023 with various levels of air assisted spray. |
Figure 8 shows examples of spray plume movements after they left the nozzles with various air assists. It was evident that sprays without air assist lost their propulsion after being discharged, and they were pulled back between tree rows by air current generated from the forward movement of the sprayer (fig. 8a), which would lead to the droplets depositing on the ground or drifting away from the tree row. As the intensities of air assists for spray increased, the droplets tended to move forward to tree canopies; however, their fates were differed by the intensities of air assist. For example, with the air assist at 30% DC to the spray, spray droplets generally traveled forward to the tree canopies, although some spray droplets deposited on the ground (fig. 8b). However, when DC for the air assist was increased to 50% or above, most spray droplets moved forward to the tree canopies and reached tree canopies with less droplets to the ground (figs. 8c and 8d). This would increase the chances of spray droplets being deposited on the tree canopies rather than on the ground; however, these levels (> 50% DC) of air assist would promote spray being drifted further away from the application site (figs. 7d and 7e).
Table 3 shows the average CPTSCs of each tree from spray applications with various intensities of air assistance from the EAAS over the growing season. The CPTSCs generally decreased as the average LAI increased because more leaves in the tree canopies were available for spray droplets to be deposited. The CPTSCs had high CVs (up to 188%) at various canopy depths, which spray droplets needed to penetrate to reach the collectors as they were behind the canopy. The CPTSCs were varied by trees as expected since tree canopy size, shapes, and foliage densities were different. When there was no air assistance, the average CPTSCs were less than 7%, regardless of canopy densities, while its changes were erratic because wind conditions and air current from the sprayer to manipulate the majority spray droplets deposited elsewhere were the main factors in the CPTSCs being less than 7%. Spray applications with the air assist with DCs of 70% (expected air velocity of 8.5 m s-1 at the opening) and 100% (expected air velocity of 11.3 m s-1 at the opening) for the EAAS had average CPTSCs of 14.2% to 24.2% and 17.1% to 44.6%, respectively, over the growing season. This suggests that such levels of air assist for spray applications might be too intense to force spray droplets to pass through the canopy instead of being deposited, which might drift away from the application site eventually. However, air assist from the EAAS with 30% DCs (expected air velocity of 3.4 m s-1 at the opening) and 50% DC (expected air velocity of 5.9 m s-1 at the opening) generally resulted in low average CPTSCs from 0.5% to 12.0% and from 4.1% to 19.9%, respectively, throughout the growing season. Relatively low CPTSCs compared to CPTSCs from 70% and 100% air assist indicate that spray applications with the air assist of the EAAS with 30% and 50% DC of the EAAS might be sufficient when trees have medium canopy densities (LAI < 2.17).
 |
| Figure 8. Example photos of spray plume movement after discharging with various levels of air assist from the EAAS. |
| Table 3. Average canopy pass through spray coverages of each tree from 400 L ha-1–spray application with different air assist levels over a growing season. | |||||||||||||||
| Avg. LAI | No air | Air assist from the EAAS with 30% DC | Air assist from the EAAS with 50% DC | ||||||||||||
| Tree | Avg. | Tree | Avg. | Tree | Avg. | ||||||||||
| 1 | 2 | 3 | 1 | 2 | 3 | 1 | 2 | 3 | |||||||
| 0.00 | 0.0 (60.9)[a] | 0.3 (94.0) | 0.1 (187.5) | 0.1 (117.8) | 0.0 (122.1) | 0.4 (54.8) | 1.0 (76.2) | 0.5 (102.0) | 1.2 (77.7) | 6.6 (114.2) | 4.6 (58.7) | 4.1 (66.4) | |||
| 1.02 | 0.5 (101.2) | 0.4 (86.3) | 0.6 (51.6) | 0.5 (20.5) | 1.0 (69.7) | 10.1 (50.9) | 17.4 (109.2) | 12.0 (39.1) | 20.3 (65.3) | 24.7 (50.7) | 14.6 (47.2) | 19.9 (25.4) | |||
| 1.60 | 0.3 (108.2) | 0.6 (105.0) | 0.2 (85.6) | 0.4 (64.2) | 1.6 (58.0) | 6.4 (109.7) | 7.5 (32.5) | 5.1 (61.6) | 8.6 (46.4) | 17.0 (67.1) | 6.7 (53.8) | 10.7 (51.0) | |||
| 1.88 | 0.0 (136.9) | 0.7 (65.5) | 0.1 (103.0) | 0.3 (130.0) | 1.9 (93.0) | 2.6 (87.8) | 4.0 (95.7) | 2.5 (62.9) | 13.8 (43.1) | 16.9 (88.2) | 7.3 (95.7) | 12.7 (38.5) | |||
| 2.07 | 2.4 (64.1) | 6.9 (66.5) | 1.0 (84.9) | 3.4 (90.6) | 2.7 (58.4) | 3.6 (84.9) | 0.6 (94.0) | 2.3 (68.8) | 7.8 (65.1) | 13.1 (72.1) | 4.9 (90.0) | 8.6 (48.7) | |||
| 2.17 | 0.4 (78.9) | 0.7 (83.8) | 0.4 (125.5) | 0.6 (20.0) | 0.6 (70.7) | 5.3 (68.1) | 2.5 (104.6) | 2.8 (85.6) | 8.6 (123.1) | 7.2 (73.8) | 4.3 (145.6) | 6.7 (32.8) | |||
| Avg. LAI | Air assist from the EAAS with 70% DC | Air assist from the EAAS with 100% DC | |||||||||||||
| Tree | Avg. | Tree | Avg. | ||||||||||||
| 1 | 2 | 3 | 1 | 2 | 3 | ||||||||||
| 0.00 | 17.2 (70.3) | 20.5 (64.3) | 22.3 (39.8) | 20.0 (12.9) | 55.0 (46.3) | 42.2 (57.0) | 36.6 (25.3) | 44.6 (21.1) | |||||||
| 1.02 | 23.9 (63.2) | 22.6 (53.2) | 25.3 (50.2) | 23.9 (5.9) | 29.9 (36.7) | 32.1 (69.7) | 46.9 (47.7) | 36.3 (25.1) | |||||||
| 1.60 | 31.2 (58.0) | 22.8 (76.8) | 18.5 (50.0) | 24.2 (26.6) | 36.5 (39.4) | 31.9 (64.8) | 40.9 (46.9) | 36.5 (12.4) | |||||||
| 1.88 | 16.2 (90.5) | 25.9 (71.3) | 18.1 (48.0) | 20.1 (25.6) | 31.4 (78.6) | 35.2 (49.3) | 31.4 (60.8) | 32.7 (6.8) | |||||||
| 2.07 | 10.2 (55.6) | 16.8 (71.0) | 15.8 (51.7) | 14.2 (25.1) | 15.6 (62.0) | 12.0 (55.0) | 23.7 (49.3) | 17.1 (35.1) | |||||||
| 2.17 | 12.0 (143.4) | 21.4 (74.2) | 22.4 (78.5) | 18.6 (30.7) | 16.6 (81.2) | 17.9 (62.4) | 18.1 (77.1) | 17.5 (4.6) | |||||||
[a] Coefficient of variation of spray coverages (%) are presented in parentheses. | |||||||||||||||
Figure 9 shows the changes in average CPTSCs of three trees over their average LAI over the growing season with air assist from the EAAS, with DCs of 30%, 50%, and 70%, excluding the CPTSC data without leaves on the trees due to the wind direction. The average CPTSCs decreased linearly from 12.0% to 2.8%, 19.9% to 6.7%, and 23.9% to 18.6% for air assist with 30%, 50%, and 70% DCs, respectively, as the average LAI increased from 1.02 to 2.17. The data shows that the CPTSCs generally decreased by approximately 6.9%–10.3% when the average LAI increased by one. The CPTSCs decreased further for spray applications with the air assist with lower DCs as expected since lower DCs created relatively low exit air speeds (3.4–5.9 m s-1) thus, obstacles, e.g., leaves or branches, had better chances to block spray droplets from passing through the canopies.
 |
| Figure 9. Average CPTSC of three trees over the growing season with different levels of air assist from the EAAS. |
A multivariable regression model was developed with the overall average CPTSC data (with air assist from the EAAS with DCs of 30%, 50%, and 70%) and the average LAI to predict a DC for the EAAS to achieve the intended average CPTSC (fig. 9). The CPTSC data from the EAAS with DCs of 30%–70% were included as the potential minimum and maximum DCs for more dense tree canopies. The CPTSC data from the EAAS with 100% DC was excluded as its intense air assist caused too high CPTSCs. The model-predicted DCs were linearly correlated with actual DCs, with average and maximum absolute prediction errors of 3.1% and 8.7%, respectively. This suggests that determining DCs of the EAAS could be possible with the intended CPTSC and tree LAI with the maximum errors less than 10% DC, which likely caused variations in the exit airflow of less than 1.16 m s-1 (equivalent to 258.4 m3 h-1) from the EAAS if it were controlled automatically. In addition, the implications of these variations on spray depositions and coverages should be evaluated in field experiments.
These evidences showed that air assist was essential for tree crop spray applications to increase the chances of spray droplets being deposited on tree canopies, although their intensities would vary with crop conditions, e.g., LAI. The research work presented here was a preliminary step to vary intensities of air assist by developing a prototype of an electric variable air assist system and characterizing its performance and capabilities. In addition, the multivariable model developed shows that it could determine the DCs of the PWM controller to adjust the air assist intensities in real time with crop conditions, e.g., LAI, and potential spray drift risk (intended CPTSCs).
However, there were limitations with the EAAS as well. Currently, input parameters, e.g., LAI and intended CPTSCs, for the model would need to be determined based on the applicator's experience. If the system controller was connected to a sensing system, tree LAI estimations could be automated based on a sensing data processing algorithm (Mahmud et al., 2022). In addition, a process to determine an intended CPTSCs should be established because the level of tolerable intended CPTSC could vary with active ingredients, application conditions, e.g., wind, temperature, RH, and others, or application sites, e.g., the middle or edge of the field or proximities to the watershed or waterbody. Another concern was the time taken to generate 100% airflow. The current system required approximately 4 s to reach its maximum air speed when it was stopped, which might limit the system's ability to be a variable air assist system. However, based on our results from apple orchard tests, the airflow of 100% DC for the EAAS was not required to reduce CPTSCs, but the airflow of 30% to 50% DC, which would require substantially less time to attain, was sufficient for apple trees with LAIs from 1.02 to 2.17. If intended CPTSCs below 10% could be acceptable, the EAAS might require only running DCs between 30% and 50% for apple trees with LAIs from 1.02 to 2.17 based on the model developed from the CPTSC data.
Therefore, field experiments with a sprayer equipped with the EAAS must be carried out to understand the impacts of its variable air assist in spray deposition and coverage in tree canopies before proceeding to the automatic variable air assist system. In addition, further development work for a control algorithm to optimize the operation of the EAAS is necessary for automatic variable air assists, along with tests to understand its performance characteristics in airflow and spray depositions and coverages in tree canopies.
 |
| Figure 10. Accuracy of multivariable linear regression model in predicting intended duty cycles to achieve intended canopy pass through spray coverage. |
Summary
Modulating intensities of air assists for tree crop spray applications is a challenging task because tree structures change rapidly; however, typical axial fans of conventional sprayers are unable to change their speed in a timely manner. This often leads to having one or two settings for constant air assist intensities, which are often prone to spray drift. The EAAS presented herein addressed such issues by using an electric fan coupled with a PWM controller to electrically control the speeds of the fans. The EAAS required approximately 4 s to reach 100% airflow from the complete stop while drawing over 30 A of the startup current and approximately 18 A to maintain the maximum air speeds of 11.3 m s-1 and 5.3 m s-1 at the opening and 1 m away, respectively. Field tests were carried out to acquire potential off-target sprays with a range of air assist intensities from EAAS in an apple orchard. The test results showed that air assists with an exit air speed of 3.4 m s-1 (the EAAS with 30% DC) to 5.9 m s-1 (the EAAS with 50% DC) were sufficient, and those intensities resulted in relatively small CPTSCs. A multivariable regression model was developed using CPTSC and LAI data to determine DCs of the EAAS to attain intended airflow to achieve intended CPTSCs for trees with given LAIs, although further study and development for determining those parameters were needed (fig. 10).
The EAAS had notable potentials, e.g., simple adjustment of air assist intensities, and potentials of varying air assist intensities based on tree crop canopy characteristics in real time, although there were limitations, e.g., additional power source, slow response time (4 s to reach the maximum airflow from the stop) and were more feasible for tower sprayer types. Since the EAAS was a prototype and a base system aiming to provide automatic air assistance, the limitations must be addressed before the development. In addition, the evaluation of the EAAS under a field condition must be carried out for its effectiveness and benefits on spray deposition and coverage of a sprayer.
Acknowledgments
The authors express their appreciation to Adam Clark, Andy Doklovic, and Wia Jian Chia for their technical assistance. This research was supported by the USDA-ARS in-house project 5082-21620-001-00D.
References
Chen, Y., Zhu, H., & Ozkan, H. E. (2012). Development of a variable-rate sprayer with laser scanning sensor to synchronize spray outputs to tree structures. Trans. ASABE, 55(3), 773-781. https://doi.org/10.13031/2013.41509
Colaço, A. F., Molin, J. P., Rosell-Polo, J. R., & Escolà, A. (2019). Spatial variability in commercial orange groves. Part 1: Canopy volume and height. Precis. Agric., 20(4), 788-804. https://doi.org/10.1007/s11119-018-9612-3
Fox, R. D., Derksen, R. C., Zhu, H., Brazee, R. D., & Svensson, S. A. (2008). A history of air-blast sprayer development and future prospects. Trans. ASABE, 51(2), 405-410. https://doi.org/10.13031/2013.24375
Gil, E., Escolà, A., Rosell, J. R., Planas, S., & Val, L. (2007). Variable rate application of plant protection products in vineyard using ultrasonic sensors. Crop Prot., 26(8), 1287-1297. https://doi.org/10.1016/j.cropro.2006.11.003
Giles, D. K., Delwiche, M. J., & Dodd, R. B. (1988). Electronic measurement of tree canopy volume. Trans. ASAE, 31(1), 264-273. https://doi.org/10.13031/2013.30698
Gu, J., Zhu, H., Ding, W., & Wang, X. (2014). Characterization of air profiles impeded by plant canopies for a variable-rate air-assisted sprayer. Trans. ASABE, 57(5), 1307-1315. https://doi.org/10.13031/trans.57.10646
Holownicki, R., Doruchowski, G., Swiechowski, W., Godyn, A., & Konopacki, P. J. (2017). Variable air assistance system for orchard sprayers; concept, design and preliminary testing. Biosyst. Eng., 163, 134-149. https://doi.org/10.1016/j.biosystemseng.2017.09.004
Jeon, H. Y., & Zhu, H. (2012). Development of a variable-rate sprayer for nursery liner applications. Trans. ASABE, 55(1), 303-312. https://doi.org/10.13031/2013.41240
Jeon, H., & Zhu, H. (2022). Stereo vision controlled variable rate sprayer for specialty crops: Part I. Controller development. J. ASABE, 65(6), 1397-1410. https://doi.org/10.13031/ja.15227
Khot, L. R., Ehsani, R., Albrigo, G., Larbi, P. A., Landers, A., Campoy, J., & Wellington, C. (2012). Air-assisted sprayer adapted for precision horticulture: Spray patterns and deposition assessments in small-sized citrus canopies. Biosyst. Eng., 113(1), 76-85. https://doi.org/10.1016/j.biosystemseng.2012.06.008
Landers, A. J., & Muise, B. (2010). The development of an automatic canopy sprayer for fruit crops. Aspects Appl. Biol., 29-34.
Maghsoudi, H., Minaei, S., Ghobadian, B., & Masoudi, H. (2015). Ultrasonic sensing of pistachio canopy for low-volume precision spraying. Comput. Electron. Agric., 112, 149-160. https://doi.org/10.1016/j.compag.2014.12.015
Mahmud, M. S., Zahid, A., He, L., Zhu, H., Choi, D., Krawczyk, G., & Heinemann, P. (2022). Development of an automatic airflow control system for precision sprayers based on tree canopy density. J. ASABE, 65(6), 1225-1240. https://doi.org/10.13031/ja.14972
Marucco, P., Tamagnone, M., & Balsari, P. (2008). Study of air velocity adjustment to maximise spray deposition in peach orchards. Agric. Eng. Int.: CIGR J.
Miranda-Fuentes, A., Rodríguez-Lizana, A., Gil, E., Agüera-Vega, J., & Gil-Ribes, J. A. (2015). Influence of liquid-volume and airflow rates on spray application quality and homogeneity in super-intensive olive tree canopies. Sci. Total Environ., 537, 250-259. https://doi.org/10.1016/j.scitotenv.2015.08.012
Nackley, L. L., Warneke, B., Fessler, L., Pscheidt, J. W., Lockwood, D., Wright, W. C.,... Fulcher, A. (2021). Variable-rate spray technology optimizes pesticide application by adjusting for seasonal shifts in deciduous perennial crops. HortTechnology, 31(4), 479-489. https://doi.org/10.21273/horttech04794-21
Pai, N., Salyani, M., & D. Sweeb, R. (2009). Regulating airflow of orchard airblast sprayer based on tree foliage density. Trans. ASABE, 52(5), 1423-1428. https://doi.org/10.13031/2013.29122
Pascuzzi, S., Cerruto, E., & Manetto, G. (2017). Foliar spray deposition in a “tendone” vineyard as affected by airflow rate, volume rate and vegetative development. Crop Prot., 91, 34-48. https://doi.org/10.1016/j.cropro.2016.09.009
Román, C., Jeon, H., Zhu, H., Campos, J., & Ozkan, E. (2023). Stereo vision controlled variable rate sprayer for specialty crops: Part II. Sprayer development and performance evaluation. J. ASABE, 66(5), 1005-1017. https://doi.org/10.13031/ja.15578
Salas, B., Salcedo, R., Ortega, P., Grella, M., & Gil, E. (2022). Use of ultrasound anemometers to study the influence of air currents generated by a sprayer with an electronic control airflow system on foliar coverage. Effect of droplet size. Comput. Electron. Agric., 202, 107381. https://doi.org/10.1016/j.compag.2022.107381
Spal Automotive Srl. (2022). Axial motor fan catalog.
Stajnko, D., Berk, P., Lešnik, M., Jejcic, V., Lakota, M., Štrancar, A.,... Rakun, J. (2012). Programmable ultrasonic sensing system for targeted spraying in orchards. Sensors, 12(11), 15500-15519. https://doi.org/10.3390/s121115500
Syngenta Crop Protection, LLC. (2021). Aprovia Fungicide product label. EPA Reg. No. 100-1471.
Warneke, B. W., Zhu, H., Pscheidt, J. W., & Nackley, L. L. (2021). Canopy spray application technology in specialty crops: A slowly evolving landscape. Pest Manag. Sci., 77(5), 2157-2164. https://doi.org/10.1002/ps.6167
Yan, T., Zhu, H., Sun, L., Wang, X., & Ling, P. (2019). Investigation of an experimental laser sensor-guided spray control system for greenhouse variable-rate applications. Trans. ASABE, 62(4), 899-911. https://doi.org/10.13031/trans.13366
Zhu, H. (2023). Target-oriented intelligent spraying. In Q. Zhang (Ed.), Encyclopedia of smart agriculture technologies (pp. 1-14). Cham: Springer. https://doi.org/10.1007/978-3-030-89123-7_165-1
Zhu, H., Derksen, R. C., Guler, H., Krause, C. R., & Ozkan, H. E. (2006). Foliar deposition and off-target loss with different spray techniques in nursery applications. Trans. ASABE, 49(2), 325-334. https://doi.org/10.13031/2013.20400
Zhu, H., Salyani, M., & Fox, R. D. (2011). A portable scanning system for evaluation of spray deposit distribution. Comput. Electron. Agric., 76(1), 38-43. https://doi.org/10.1016/j.compag.2011.01.003