1. Introduction
The chili pepper is an important economic crop in Korea that is planted in early May and continuously harvested at least three times from the end of July; however, the production of the chili pepper is declining due to the aging of farm households and the slow mechanization rate. The harvest area and production of the chili pepper in Korea are both decreasing due to the increase in imports of chili peppers. The chili pepper harvest area, production, and self-supply rate have all decreased from 74,471 ha, 193,786 tons, and 89%, respectively, in 2000, to 31,146 ha, 60,076 tons, and 40%, respectively, in 2020 [
1]. On the other hand, imports of chili peppers have increased 9% each year over the same period, from 30,000 tons to 123,000 tons [
2].
The mechanization of the pepper harvester started in the 1970s, and research on the pepper harvester has been carried out in various countries [
3,
4,
5,
6,
7]. Researchers have actively studied and manufactured chili harvesters to increase pepper harvest and production, but the current pepper harvesters had a high PTR, thus requiring additional manpower to separate twigs from peppers [
8,
9,
10]. In agriculture, the cost of harvest labor accounts for up to 50% of the total cost of production. The cost of harvest labor could be reduced up to 10% with mechanization or the use of robotic harvesting [
11]. However, harvesting by robot has limitations in use depending on the cultivar and environment and is more suitable for greenhouses and small-scale farms [
12,
13,
14,
15].
Research on chili pepper varieties for use in mechanical harvesting has been conducted [
16]. Recently, to evaluate the suitability for harvest mechanization, previous research was conducted on the nutritional and physiological components of various varieties [
17,
18,
19,
20,
21] as well as plant spacing, productivity, and quality [
22,
23,
24] according to cultivation methods.
The self-propelled chili pepper harvester consists of the picking and transfer parts that remove peppers from plants, a separation sorting part that removes small branches and leaves, and a collecting part that contains the peppers [
25]. Previous studies were conducted on the separation sorting part using a star wheel and a card cleaner [
26,
27,
28,
29] as well as using an air blower [
26,
30] to remove small branches. When bigger particles, such as stemmed chili peppers, flow in from the picking part, the development of additional separating and sorting devices are required. Kim et al. [
31] obtained a PTR of 24.2–37.8% in a chili harvester comparison test, and suggested the need for a stem-separation and sorting device to reduce the input of additional manpower for sorting work after the chili harvest.
Studies on stem separation used a rotating roller method and a drum-type threshing method. A study on a rotating roller method conducted by Wolf et al. [
32] reported that, as a result of a pair of feeding rollers rotating in opposite directions putting a whole paprika plant, including roots, into the two pairs of picking elements, the number of stem-attached peppers was reduced by 50% compared to that of work without a stem-separation system. Esch et al. [
33] reduced the amount of stem-attached cherry peppers by more than 90% by placing stem-attached cherry peppers between roller beds of different sizes. Wolf et al. [
34] developed a system for separating stems using three cylindrical squeeze rollers rotating clockwise at the top and eight downstream rollers rotating counterclockwise at the bottom. Seiphepi et al. [
35] designed and simulated an automated system that detects the color of bell peppers and separates the pepper from the stem by a roller-driven belt.
A study by Gan-Mor et al. [
36] examined a drum-type threshing method that used a porous rotating drum to separate leaves from sorghum and parsley stems, and its SSE and DR were 90% and 10%, respectively. Sudajan et al. [
37] comparatively analyzed the drum type, rotational speed, and sunflower supply for sorting sunflowers. The optimal sorting drum was found to be the Rasp bar drum, and when the rotation speed was 700 rpm and the supply was 3000 kg/h, the SSE, DR, and loss were 99%, 1.39%, and 0.36%, respectively. To confirm the efficiency of a tractor-attached small tree-cutting chipper, the fuel consumption rate and productivity were compared for the drum and disk types [
38]. As a result of the test, the drum-type chipper reduced fuel consumption by 19% and improved productivity by 8% compared to the disk-type chipper. A cassava harvester for stem removal was developed by Jyoti et al. [
39]. This machine installed a blade on the outside of the drum to find the optimal angle and thickness of the blade. The best result was obtained at a shear angle of 20°, an approach angle of 30°, and a thickness of 6 mm.
Choi [
40] developed a device for separating peppers from stems by depositing harvested pepper stems into a drum-type threshing machine. Pepper SSE and DR were 82.7% and 4.4%, respectively, and the smaller the pepper, the lower the pepper SSE and DR. Funk et al. [
25] manufactured and tested a sorting device using a card cleaner and a cylindrical coil drum, and 60–75% of the debris was removed as a result, but the PTR was about 10%, which means additional work was needed to separate stems from peppers. Nam et al. [
41] attempted to determine the optimal drum rotation speed, the space between drums, and the drum rotation speed ratio for three rotating drums with steel-bar teeth. They found that the optimal conditions were a drum rotation speed of 70 rpm, a 5 mm space between drums, and a drum rotation speed ratio of 7:3:5, and the pepper SSE and DR were 60% and 15%, respectively. The method suggested in the current reference has low pepper SSE and high DR, so it was necessary to develop a drum-type pepper stem-separation device to improve SSE and DR.
The purpose of this study was to design and manufacture a pepper-stem-separation device for a pepper harvester, and to evaluate its performance. The detailed objectives were as follows.
To design and manufacture three-shaft and four-shaft drum stem-separation devices.
To find the optimal drum rotation speed.
To conduct a comparative performance test and evaluation of the three-shaft and four-shaft drum stem-separation devices.
3. Results and Discussion
Table 4 presents the performance test results for each factor level according to the speed of the three-shaft drum stem-separation device. When operating from factor level 1 to factor level 6 using the three-shaft drum, the SSE of the AR legend increased, and the PTR and DR of the AR legend decreased. At factor level 6, the average SSE was 76.8%, the average PTR was 20.9%, and the average DR was 2.3%, thus showing the highest efficiency. In this study, to select the optimal drum rotation speed, the rotation speeds of A, B, and C using the three-shaft drum with a factor level of 6 were set to 0.9, 2.7, and 1.3 m/s, respectively, as this showed the highest efficiency. In the four-shaft drum, the rotation speeds of A, B, and C were the same as that of the three-shaft drum, and the rotation speed of D was set to 1.3 m/s, which is the same as C. The conveyor speed was 0.5 m/s.
Table 5 lists the field performance test results for AR Legend and Jeokyoung. In the non-moving status during non-picking operation, the average SSEs of AR Legend and Jeokyoung with the three-shaft drum were 70.0% and 69.3%, respectively; the average PTRs were 26.9% and 27.7%, respectively; and the average DRs were 3.1% and 3.0%, respectively, which means the results were similar regardless of pepper species. The average SSEs of AR Legend and Jeokyoung with the four-shaft drum were 64.0% and 72.3%, respectively; the average PTRs were 25.5% and 17.8%, respectively; and the average DRs were 10.5% and 9.9%, respectively, which means that the average PTR of Jeokyoung was lower than that of AR Legend. The overall DR with the four-shaft drum was slightly higher compared to that with the three-shaft drum, but the average PTR decreased, thus confirming that the four-shaft drum is more suitable for stem separation.
In the non-moving status during picking operation, the average SSEs with the three-shaft drum were 86.0% and 90.1%, respectively; the average PTRs were 9.3% and 6.2%, respectively; and the average DRs were 4.7% and 3.7%, respectively. Compared to the three-shaft drum, the average SSE with the four-shaft drum increased by 4.1%, the average PTR decreased by 3.1%, and the average DR increased by 1%, which means that the four-shaft drum had higher performance than the three-shaft drum. The three-shaft and four-shaft drum-type stem-separation devices used in this study showed higher SSE (86.0%, 90.1%) than that of Nam [
41] (63.8%) and Choi [
40] (82.7%), and the average DR (4.7%, 3.7%) significantly decreased compared to Nam [
41] (12.9%), which was considered to have higher performance.
Table 5 lists the field performance test results in the moving status during picking operation for an 8 m sector using a pepper harvester equipped with the pepper stem-separation device developed in this study. The average performances of the three-shaft drum and four-shaft drum showed SSEs of 65.7% and 69.3%, respectively; average PTRs of 22.1% and 13.0%, respectively; and average DRs of 9.5% and 11.2%, respectively. SSE increased by 3.6%, PTR decreased by 9.1%, and DR increased by 1.7% in the four-shaft drum compared to the three-shaft drum, so the four-shaft drum had better stem-separation device performance than the three-shaft drum. However, further research is needed to determine how to reduce DR.
Table 6 presents the results of a two-way analysis of variance intended to check the effects on SSE, PTR, and DR according to the pepper species and the number of drums. For the statistical analysis of two-way analysis of variance, the commercial software SAS (Version 9.4, SAS Institute, Cary, NC, USA, 2019) was used, and the analysis was performed at a confidence level of 95%. As a result of the variance analysis of PTR and DR in regard to the pepper species and the number of drums, the number of drums was confirmed to affect PTR and DR with P-values of 0.0219 and 0.0001, respectively, which were both smaller than the significance level of 0.05. In addition, the interaction between the number of drums and pepper species turned out to have no impact on PTR.