Lint Cleaning Performance of a Pneumatic Fractionator: Impacts on Fiber Quality and Economic Value of Saw- and Roller-Ginned Upland Cotton

Abstract

Current saw- and pin-type lint-cleaning systems used by the ginning industry have challenges retaining lint quality. The objective of the research was to test a novel pneumatic fractionator for the lint cleaning of an Upland cotton variety that was both saw- and roller-ginned. The process variables tested were initial lint moisture content in the range of 5.5–15% w.b., line pressure in the range of 276–552 kPa, and residence time in the range of 15–45 s. Experiments were conducted based on a central composite design. Models based on response surface methodology (RSM) were developed for final lint moisture, total trash extracted during lint cleaning, and High-Volume Instrument (HVI) fiber quality. The RSM models adequately described the pneumatic fractionation process, as indicated by the coefficient of determination, predicted vs. observed plots, and residual values. The results indicated that the interactions among initial lint moisture content, residence time, and line pressure significantly affected lint quality. At the optimized pneumatic fractionator process conditions, the predicted lint quality attributes were better for both roller- and saw-ginned lint compared to lint cleaned with saw- and pin-type lint cleaners. The upper half mean length increased by 1 mm, the uniformity index was higher by 0.5–1 percentage points, the strength was 1–2 g/tex higher, and the short fiber content was reduced by more than one percentage point. Color grades were better for pneumatic fractionated lint compared to saw- and pin-type lint cleaning methods. Lint value was approximately 4 cents/kg higher for both saw- and roller-ginned pneumatic fractionated lint, compared to lint cleaned using saw- and pin-type lint cleaners. The novel pneumatic fractionator, when compared to industry-standard saw- and pin-type lint cleaners, effectively cleaned lint while retaining fiber quality and removing most of the motes and trash.

1. Introduction

The global economic impact of the cotton crop exceeds $600 billion [1]. More than 80% of the world’s cotton is grown in countries such as China, Brazil, India, Pakistan, Turkey, the USA, and Uzbekistan [2,3]. Harvesting fully mature cotton during hot, dry weather results in high-quality lint and seed [4]. After harvesting and baling into seed cotton modules, the modules are transported to gins for further processing, as illustrated in Figure 1. The process begins with seed cotton bales being cut open and dried to a lower moisture content of about 5–6% (wet basis). Next, the dried seed cotton passes through two precleaning systems to remove unwanted materials. First, stick machines and cylinder cleaners remove sticks, burrs, and leaves. The cylinder cleaner typically contains six spiked cylinders, each equipped with screens or grid bars around part of its circumference. As the seed cotton moves through the cylinders and screens, fine trash particles are removed and fall through the openings provided [5]. The stick machine, used as a second precleaning system, features three cylinders: the first two clean the cotton, where brushes remove burrs and sticks aided by centrifugal force, and the third cylinder recovers the seed cotton that was flung off by the first two cylinders [6].

After the pre-cleaning of seed cotton, the seeds are separated from the cotton through a process called ginning. In the United States, two main types of gin stands are used for this purpose: saw gins and roller gins. Saw ginning is primarily used for American Upland (Gossypium hirsutum) cotton. In contrast, extra-long staple (ELS) cotton, such as Pima (Gossypium barbadense), must be roller-ginned in the US. Some high-quality Upland (Gossypium hirsutum) cotton and Hybrid (a variety combining traits from superior species like Gossypium hirsutum and Gossypium barbadense) cotton are also roller-ginned in the USA. Figure 2a shows the saw gin stand commonly used for Upland cotton [7]. The saw-type gin stand features rotating saws located between ginning ribs. During the ginning process, the saw teeth pass through these ribs, pulling the fiber from the seed rather than cutting the lint away. Figure 2b depicts the roller gin stand [8]. This type of gin includes a rotary knife that guides seed cotton directly to the ginning point. It effectively sweeps away the seeds from the ginning area and removes any seed cotton.

After the ginning process, the lint is cleaned using saw- and pin-type mechanical lint-cleaning systems to eliminate motes, leaves, and fine trash [9]. A saw-type lint cleaner is used when seed cotton is ginned with a saw gin, while a pin-type lint cleaner is used for seed cotton ginned using a roller gin [10]. Figure 3a illustrates a diagram of a conventional controlled-batt, saw-type lint cleaner [9]. In this system, a batt of lint forms on a condenser screen, and a feed works assembly removes the batt from the condenser, directing it to a feed plate. In this process, the slowly moving batt is pinched tightly between the feed plate and feed roller, while a fast-moving saw cylinder with sharp teeth moves in the opposite direction. The saw grabs the fiber at the tip of the feed plate, potentially damaging it [9]. Research has indicated that saw-type lint cleaners generally increase the amount of neps and short fiber content. It is believed that the sharpness of the saw teeth profile on the saw cylinder contributes significantly to the damage of the fiber. Additionally, greater fiber damage is observed as the saw cylinder speed increases [11].

On the other hand, a pin-type lint cleaner is typically associated with the roller gin stand (Figure 3b) [12]. This type of lint cleaner has a round body equipped with pins. Unlike the saw-type cleaner, the pin-type cleaner does not cut fibers, helping reduce fiber damage. It is considered less aggressive than the saw-type cleaner, producing fewer neps and shorter fibers while achieving higher upper half mean lengths. The major challenge of the pin-type lint cleaner is that it leaves more trash in the lint [13]. In conclusion, the saw-type lint cleaner is effective at removing motes, leaf, and fine trash, but negatively impacts lint quality attributes, such as the upper half mean length and short fiber content, whereas pin-type lint cleaners excel at preserving lint quality parameters; however, they do not clean the lint as thoroughly as saw-type lint cleaners.

Recent discussions with cotton gin stakeholders highlighted the need for a gentler method to clean lint, preserving its properties while removing most of the trash content. Preserving lint properties such as upper half mean length and uniformity index (UI) after lint cleaning benefits spinning unit operations by producing better yarn [14]. USDA cotton ginning labs and Cotton Incorporated are working to improve and develop ginning and lint cleaning systems to enhance the quality of cotton produced in the USA. Previous studies used a lab-scale pneumatic fractionator to examine how line pressure and residence time affect the High Volume Instrument (HVI) and Advanced Fiber Information System (AFIS) properties of saw-ginned Upland cotton [15,16]. These authors showed that the pneumatic fractionator maintained fiber length and quality better than traditional saw-type lint cleaners. However, these authors did not address the impact of lint moisture on lint quality and the modeling of the process.

This work aims to explore how a wide range of pneumatic fractionator process conditions (lint moisture content, line pressure, and residence time) influence lint properties (micronaire, upper half mean length, UI, strength, short fiber content, and trash count) and pneumatic fractionator process outputs such as final lint moisture content and total trash extracted. The specific objectives of this research are to (a) develop response surface models and plots to identify and describe the interactions among pneumatic fractionation process variables on specific lint properties, (b) determine the process conditions that optimize pneumatic fractionation for both roller- and saw-ginned Upland cotton to achieve the target lint properties identified, and (c) evaluate and compare the lint properties and economic value of roller- and saw-ginned Upland cotton after ginning with no lint cleaning, after conventional lint cleaning, and after optimized pneumatic fractionation.

2. Materials and Methods

2.1. Cotton Variety

For this study, a high-fiber-quality Upland cotton variety, NG 4545, was selected to determine the impact of the conventional and pneumatic lint-cleaning systems on fiber quality. The cotton was picker-harvested and roller- and saw-ginned at the Southwestern Cotton Ginning Research Laboratory, Las Cruces, NM, USA. According to the Americot company, NG 4545 B2XF is a medium-maturing cotton variety with high yield potential and outstanding fiber quality, and tolerance to Verticillium wilt and bacterial blight [17].

2.2. Equipment Used for Processing Upland Cotton

The seed cotton NG4545 variety was obtained from a local cotton grower. The seed cotton has a low moisture content of about 5% (w.b.). As a result, drying the seed cotton in a tower dryer was not required, and it was sent directly to precleaning equipment, including cylinder cleaners and stick machines. The precleaned Upland cotton was further processed through a roller gin stand with a pin-type lint cleaner and a saw gin stand equipped with a saw-type lint cleaner.

Table 1. Specifications of the systems used for processing the NG4545 Upland cotton variety.

System	Seed Cotton Precleaning Unit Operations
	Model	Specification or Dimensions
Cylinder cleaner	Continental/Moss-Gordin (Prattville, AL, USA)	Gravity-fed six-drum inclines with 9.5 mm (3/8-in.) diameter, grids spaced 9.5 mm (3/8 in.) apart. The incline is 1300 mm (50 in.) wide and rated at 5.6 to 8.2 bales/h/m (1.7 to 2.5 bales/h/ft) of width.
Stick machine	Continental/Moss Gordin Little David (Prattville, AL, USA)	Gravity-fed, with two 349 mm (13.75-in.) diameter channel (sling off) saws and one reclaimer saw. The Little David is 1800 mm (72 in.) wide and rated at 3.3 to 6.6 bales/h/m (1.0 to 2.0 bales/h/ft) of width.
	Seed Cotton Ginning Unit Operations
Roller ginning
Roller gin stand	Hardwicke–Etter roller gin stand (Sherman, TX, USA)	Roller diameter: 380 mm (14.785 in.) and 60DO (on durometer scale) in hardness. The stationary knife is made of hardened steel. The rotary knife is a 6-bladed spiral measuring 70 mm (2.75 in.) in diameter. The gin stand is 1000 mm (40 in.) width.
Air jet cleaner	Lummus, Super-Jet Cleaner (Lummus Cotton Gin Co., Ltd.; Columbus, GA, USA)	1070 mm (42-Inch) wide. The air-jet cleaner connected to the pin-cylinder cleaner had an adjustable edge (4.77–31.75 mm) (3/16′-1 ¼″) to skim off heavier trash.
Pin-type lint cleaner	Aldrich Machine Works (Greenwood, SC, USA)	406 mm (16.0 in.) diameter pin cylinder that rotated at 1100 rpm and 16 grid bars around the pin cylinder with each leading edge spaced 19.1 mm (0.75 in.) apart. The clearance between the pins and grid bars was 23.8 mm (0.94 in.).
Saw ginning
Saw gin stand	46-Saw Continental/Murray Double Eagle saw gin stand (Prattville, Autauga County, AL, USA)	Gins saws 406.4 mm (16.0 in.) diameter spaced 1.6 mm (0.063 in.) apart, operated at 660 rpm. Motor: 22.4 kW (30 hp), with 1760 rpm.
Air jet cleaner	Air- Jet Cleaner (Northern Lucus Inc., Lubbock, TX, USA)	1220 mm (48-in.) wide, the air-jet cleaner had an adjustable edge (4.77–31.75 mm) (3/16′–1 ¼″) to skim off heavier trash
Saw-type lint cleaner	Continental/ Moss–Gordin Lodestar controlled-batt saw-type lint cleaner (Continental Gin Co., Ltd.; Prattville, AL, USA)	Spiral-wrapped lint cleaner saw 406 mm (16.0 in.) in diameter; operated at about 1000 rpm. The lint cleaner had five grid bars with 1.59 mm (0.063 in.) clearance between each bar and the saw.

outlines the specifications for the precleaning, ginning, and lint-cleaning systems. A portion of the material processed through the saw- and roller-ginning stands was subsequently used to conduct lint-cleaning studies with a pneumatic fractionator.

2.3. Pneumatic Fractionation Process

Figure 4a shows a laboratory-scale pneumatic fractionator used in the present study [16]. The pneumatic fractionator consisted of a rectangular chamber about 45.7 cm tall × 61.0 cm wide × 20.3 cm deep with rounded ends at the top and bottom that are split and hinged in the middle (Figure 4a). A pre-weighed ginned lint sample (but not cleaned) was placed in the pneumatic fractionator. When the pneumatic fractionator was started, compressed air was released from eight orifices along the back of the chamber, causing the cotton lint to tumble and flow around the chamber’s perimeter (Figure 4b). The tumbling and scrubbing action against the 4.8 mm-wide × 15.9 mm slots across the front of the chamber, and one continuous 3.2 mm (0.125-in.)- wide slot along the front wall, dislodges small trash and dust from the cotton lint and carries them out of the chamber with the airflow through the slots [15]. The trash is collected on a series of two sieves: first, the No. 6 standard sieve (3.35 mm square openings), and second, the No. 200 sieve (75 µm openings). Figure 5 shows the sample before and after lint cleaning, motes collected on a 3.35 mm screen, and leaf and fine trash collected on a 75 µm sieve.

2.4. Experimental Plan

In the present study, a laboratory-scale pneumatic fractionator was used to investigate the impact of process conditions on lint quality, extracted trash, and final lint moisture content. Initial lint moisture content varied from 5.5 to 15% (w.b.), line pressure from 276 to 552 kPa, and residence time from 15 to 45 s.

Table 2. Pneumatic fractionator process conditions tested and lint properties measured.

Process Conditions	Low	Mid	High	Lint Quality
Lint moisture content (%, w.b.) (x1)	~5.5 (−1)	9 (0)	15 (1)	Pneumatic fractionator outputs Final lint moisture content (%., w.b.) Total trash extracted (motes + leaf and fine trash) (%) HVI lint quality outputs Upper half mean length (mm) Uniformity index (%) Strength (grams/tex) Reflectance (Rd) Yellowness (+b) Short fiber content (%) Trash count (number of visible particles) Micronaire Leaf grade Spinning consistency index
Line pressure (kPa) (x2)	276 (−1)	414 (0)	552 (1)
Residence time (s) (x3)	15 (−1)	30 (0)	45 (1)

Note: Lint moisture content and quality of the roller- and saw-ginned lint were also measured (a) before lint cleaning (NO LC), and (b) after industry-standard lint cleaning using saw- and pin-type lint cleaners (1-LC).

indicates the three levels of each process condition. A central composite experimental design with three factors, three levels, and two center points, resulting in 16 total runs, was used to test saw- and roller-ginned NG 4545, picker-harvested Upland cotton. Experiments were conducted using this design, as it provides high-quality predictions across the entire design space. JMP Pro software (version 18.2.0, JMP Statistical Discovery LLC, Cary, NC, USA) was used to develop the experimental design [18]. The pneumatic fractionator outputs (lint moisture content and total trash extracted) and Uster High Volume Instrument (HVI) (Uster company (Zürich, Switzerland) properties of the lint were measured for the pneumatic fractionated lint (Table 2). Using the (HVI) properties, the spinning consistency index was calculated based on an equation developed by the Uster company (Zürich, Switzerland) [19].

$$Spinningconsistencyindex\left(SCI\right)=-414.67+2.9\times Strength-9.32\times Micronaire+49.17\times Upperhalfmeanlength+4.74\times Uniformityindex+0.65\times Relectance+0.36\times Yellowness$$

(1)

2.5. Experiment Procedure

Seventy grams of saw- and roller-ginned (but not cleaned) lint were used for each pneumatic fractionation test. Of the 70 g, 20 g were used to measure the lint’s initial moisture content before the pneumatic fractionator test. The remaining 50 g were used to conduct pneumatic fractionator tests based on a central composite experimental design. After each pneumatic fractionation test, 20 g of cleaned lint was used to measure the final lint moisture, and the remaining lint (about 25–30 g) was used to determine the lint quality properties using HVI.

2.6. Lint Moisture Adjustment Method

The initial moisture content of the lint samples was measured and used as the lower moisture content limit in the experimental design. For the other two levels, the lint was reconditioned to target moisture contents of 9% and 15% (w.b.). Based on the initial moisture content, the amount of water needed was calculated using Equation (2) [20]. In Equation (2), Ww is the weight of water (g), Ws is the weight of the lint sample (g), mf is the percent final moisture content of the lint sample (w.b.), and mi is the percent initial moisture content of the lint sample (w.b. %). The computed amount of water to be added to the 50 g of lint was sprayed onto the lint using a spray bottle. The moisture-conditioned lint was placed in zipper-closure plastic bags in the refrigerator for 2 days to allow moisture to equilibrate. Refrigerated storage of the moisture-conditioned lint samples was used to prevent color changes. Lint moisture adjustment data indicated that there is about ±0.5–1.0% (w.b.) moisture variability from the desired moisture levels of 9% (w.b.), and ±0.5–2.0% (w.b.) at higher desired moisture contents of 15% (w.b.). For the experimental data analysis, the actual lint moisture content after reconditioning was used.

$$W_w=W_s\times {\displaystyle \frac{m_f-m_i}{100-m_f}}$$

(2)

2.7. Lint Moisture Measurement (%, w.b.)

A twenty-gram lint sample (wet weight) was placed in 652 cm³ perforated stainless-steel baskets weighing approximately 91.5 g and dried at 105 °C for 2 h [21]. The lint moisture data reported after roller and saw ginning with no lint cleaning and after industry-standard saw- and pin-type lint cleaning were an average of three measurements, whereas the lint moisture content after moisture reconditioning before pneumatic fractionation and after pneumatic fractionation was based on one measurement (as the weight of the lint sample (~20 g) after reconditioning and pneumatic fractionation was sufficient for only one moisture measurement).

2.8. Total Trash Extracted (%)

After each fractionator test, trash collected on the No. 6 sieve (motes) and No. 200 sieve (leaf and fine trash) was weighed. The total trash weight was calculated by adding the material collected on both sieves (motes + leaf and fine trash). The percent total trash extracted was calculated from the initial lint weight used for pneumatic fractionation and the total trash collected on the two sieves. The reported value is of one measurement after each pneumatic fractionator experimental condition test.

2.9. HVI Properties Measured

All lint samples (a) before lint cleaning, (b) after conventional industry-standard pin- and saw-type lint cleaning, and (c) after pneumatic fractionation at various process conditions, were sent to the USDA-ARS Cotton Quality and Innovation (CQI) in New Orleans, Louisiane for High Volume Instrument analysis (HVI, Uster Technologies, Inc., Greenville, SC, USA), the same quality measurements made for every bale of cotton ginned in the U.S. by the USDA-AMS classing offices. HVI measurements include fiber upper half mean length (UHML), uniformity index (UI), strength, micronaire, color reflectance and yellowness, trash count, and area and short fiber content (short fiber content is not part of USDA-AMS classing data) [22,23]. Before cotton samples are tested, they are placed in a controlled atmosphere at 21 °C and 65% relative humidity for a set period to allow moisture to be within the range of 6.75–8.25%, before High Volume Instrument (HVI) testing. The reported values for HVI properties were averages of 5 measurements. A brief description of the HVI properties measured is given below:

• UHML is the average length of the longer half of the fibers in a sample.
• UI is the ratio of the average fiber length to the upper half mean length.
• Strength is the force needed to break a bundle of fibers.
• Micronaire is an indirect measure of fiber fineness and maturity.
• Color is the measure of cotton’s reflectance (brightness) and yellowness. These two values determine the color grade.
• Trash is the amount of non-lint material, such as leaf and bark.
• Leaf grade is calculated based on the trash meter percentage area and particle count.
• Short fiber content, which is an index calculated based on the HVI properties, is a measure of fibers less than half an inch or 12.7 mm.

2.10. Data Analysis and Modeling

The experimental data generated using the central composite design were further used to develop models (Equation (3)) using response surface methodology (RSM) to understand how the process variables impact the lint quality during pneumatic fractionation. In Equation (3), β₀, β_i, β_ii, and β_ij are regression coefficients, x_i and x_j are independent variables, k denotes the number of variables, and ε represents the error [24]. Pareto charts developed from the experimental data help identify the statistically significant process variables. The observed and predicted values were also used to see the upper and lower ranges of the residual values. The RSM model developed was used to predict responses (lint moisture content, percent total trash extracted, and lint quality properties based on HVI data) and to create surface plots that visualize the trends of process variables on maximizing and minimizing the responses. For RSM analysis, Statistica software (version 14.1.0.8, StatSoft GmbH, Hamburg, Germany) was used [25].

$${\beta }_0+{\sum }_{i=1}^k{\beta }_i{x}_i+{\sum }_{i=1}^k{\beta }_{ii}{x}_i^2+\sum {\sum }_{i<j}{\beta }_{ij}{x}_i{x}_j+\mathsf{\epsilon }$$

(3)

2.11. Pneumatic Fractionator Process Optimization

Optimization is challenging, particularly when conflicting maxima and minima occur in the model. A hybrid genetic algorithm (HGA), which integrates a genetic algorithm with a deterministic subroutine, was used for current optimization [26]. HGA input parameters, such as population size and number of iterations (set to 100), crossover and mutation rates (set to 0.8 and 0.01, respectively), and elitism (set to 0.1), were selected [8]. The RSM models developed for final lint moisture content, total trash extracted, and HVI fiber properties for the pneumatic fractionation process conditions were used in the optimization study. Instead of determining the individual optimum process conditions for each lint quality attribute, identifying the common optimum process conditions for the pneumatic fractionator that meet the desirable lint quality attributes will be more useful for process scale-up studies.

The final lint moisture content (LMC) was minimized, as the lint moisture content must be ≤7.5% for packaging. Total trash (TT) extracted during pneumatic fractionation was maximized to capture most of the total trash, including motes (immature fibers), leaf, and fine trash in the lint. For the HVI properties, the upper half mean length (UHML), uniformity index (UI), and lint strength (STR) were maximized. To achieve a better color grade, higher reflectance (Rd) and increased yellowness (+b) values were necessary, so these properties were maximized. Short fiber content (SFC), calculated from HVI properties, was minimized. Micronaire (MIC) has a premium range that spans medium values. All the experimental data were in or above that range, so MIC was minimized. Finally, the spinning consistency index (SCI), calculated from HVI data (Equation (1)), was also maximized, as the spinning industry prefers higher SCI values. Equation (4) presents the combined equation used to optimize the pneumatic fractionation process. All properties that need to be maximized were grouped together, and all properties that need to be minimized were grouped together and assigned negative signs to return a minimum value. This combined Equation (4) is further maximized using HGA. No weights were assigned to the individual lint quality equations, as all were treated equally to find the common optimum process conditions.

$$f\left(y\right)=Maximize(\left(UHML+UI+STR+Rd+\left(+b\right)+SCI+TT\right)-\left(FinalLMC+MIC+SFC+TrashCnt\right))$$

(4)

2.12. Loan Rate and Bale Value Calculation

The lint properties of each bale of cotton are used to calculate the loan value. More details on the cotton value based on HVI, including properties, are available on the USDA website [27]. For comparing the value of lint after lint cleaning treatments, the loan values were calculated based on 2021 crop loan rates of 52 cents per pound or 114 cents/kg for the saw- and roller-ginned cotton with (a) no lint cleaning, (b) lint cleaning using saw- and pin-type lint cleaners, and (c) pneumatic fractionated lint cleaned at optimized conditions based on HVI properties indicted in Table 2. In the present study, 2021 crop loan rates were used, as the Upland cotton used in the present study was harvested in 2021. Based on the total trash extracted using conventional lint cleaners (saw- and pin-type) and a pneumatic fractionator, the lint bale weight was calculated. A standardized US cotton bale typically weighs approximately 480 pounds (218 kg) of cleaned cotton lint [28], and this weight was considered for the calculation of the lint bale weight based on total trash extracted during pneumatic lint cleaning. The loan rate, calculated based on lint HVI properties for the different lint cleaning systems tested in this project, was then used to calculate the bale value in dollars for saw- and roller-ginned conventional and pneumatic fractionated lint.

3. Results and Discussion

3.1. Response Surface Models

Based on experimental data collected for the pneumatic fractionation process, response surface models were developed for lint moisture content, total trash, and HVI properties of both roller- and saw-ginned lint.

Table 3. RSM models, for saw-ginned pneumatic fractionation process variables: initial lint moisture content (x₁), line pressure (x₂), and residence time (x₃).

Lint Properties	Model	R2	Significance
Pneumatic fractionator outputs
Final lint moisture content (%, w.b.)	$y=0.187644+0.504400x_1+0.012910x_2+0.012511x_3+0.003377x_1^2-0.000013x_2^2+0.000145x_3^2-0.000358x_1{x}_2-0.004076x_1{x}_3-0.000033x_2{x}_3$	0.98	x1: p < 0.05; x1x3: p < 0.05; x1x2: p < 0.1
Total trash extracted (%)	$y=-1.82218+0.13839x_1+0.00460x_2+0.06447x_3+0.00384x_1^2+0.00001x_2^2-0.00113x_3^2-0.00057x_1{x}_2-0.00033x_1{x}_3+0.00029x_2{x}_3$	0.99	x1x2: p < 0.05; x2x3: p < 0.05
HVI properties
Upper half mean length (mm)	$y=27.88030+0.17996x_1+0.00388x_2-0.10990x_3-0.00599x_1^2-0.00001x_2^2+0.00185x_3^2+0.00024x_1{x}_2-0.00255x_1{x}_3+0.00002x_2{x}_3$	0.95	x32: p < 0.05; x3: p < 0.05; x1x3: p < 0.1; x1x2: p < 0.1
Uniformity index (%)	$y=82.41237-0.00253x_1-0.00503x_2-0.03848x_3+0.00689x_1^2+$0.0000026$x_2^2+0.00088x_3^2+0.00007x_1{x}_2-0.00196x_1{x}_3-0.00008x_2{x}_3$	0.80	ns
Trash count (number of particles)	$y=36.57372-2.88500x_1+0.03550x_2-0.41064x_3+0.16001x_1^2-0.00009x_2^2+0.00146x_3^2-0.00044x_1{x}_2-0.01077x_1{x}_3+0.00044x_2{x}_3$	0.89	x1: p < 0.1; x12: p < 0.1
Strength (grams/tex)	$y=28.37610+0.16408x_1+0.00347x_2-0.09910x_3-0.00360x_1^2-0.00001x_2^2+0.00170x_3^2+0.00039x_1{x}_2-0.00361x_1{x}_3+0.00002x_2{x}_3$	0.82	ns
Short fiber content (%)	$y=7.166637-0.356263x_1-0.000406x_2+0.173204x_3+0.005440x_1^2+0.000009x_2^2-0.003258x_3^2-0.000142x_1{x}_2+0.005464x_1{x}_3+0.000026x_2{x}_3$	0.92	x32: p < 0.1; x1x3: p < 0.1; x3: p < 0.1
Reflectance (Rd)	$y=76.60507-0.21413x_1+0.02101x_2+0.00887x_3+0.01213x_1^2-0.00002x_2^2+0.00077x_3^2-0.00002x_1{x}_2-0.00045x_1{x}_3-0.00008x_2{x}_3$	0.77	x2: p < 0.05; x22: p < 0.1
Yellowness (+b)	$y=7.185996+0.002597x_1+0.006959x_2+0.023253x_3+0.005561x_1^2-0.000007x_2^2-0.000266x_3^2-0.000127x_1{x}_2-0.000903x_1{x}_3+0.000009x_2{x}_3$	0.83	x2: p < 0.1; x22: p < 0.1; x1x2: p < 0.1
Micronaire	$y=4.327300+0.065148x_1-0.000887x_2-0.007605x_3-0.004563x_1^2-0.000001x_2^2+0.000137x_3^2+0.000105x_1{x}_2-0.000554x_1{x}_3+0.000011x_2{x}_3$	0.88	x1x2: p < 0.05; x12: p < 0.05; x1: p < 0.05; x1x3: p < 0.05
Spinning consistency index	$y=124.2766+0.0668x_1+0.0182x_2-0.5975x_3+0.0630x_1^2-0.0001x_2^2+0.0118x_3^2+0.0009x_1{x}_2-0.0201x_1{x}_3-0.0004x_2{x}_3$	0.83	ns

ns, not significant.

and

Table 4. RSM models, for roller-ginned pneumatic fractionation process variables: initial lint moisture content (x₁), line pressure (x₂), residence time (x₃).

	Model	R2	Significance
Pneumatic fractionator outputs
Final lint moisture content (%, w.b.)	$y=1.195175+0.359572x_1+0.011415x_2-0.001147x_3+0.003887x_1^2-0.000014x_2^2+0.000293x_3^2-0.000073x_1{x}_2-0.003957x_1{x}_3-0.000018x_2{x}_3$	0.98	x1x3: p < 0.05; x1: p < 0.1
Total trash extracted (%)	$y=0.605038-0.407652x_1+0.016467x_2+0.118420x_3+0.027262x_1^2-0.000005x_2^2-0.001408x_3^2-0.000494x_1{x}_2-0.002246x_1{x}_3+0.000188x_2{x}_3$	0.99	x12: p < 0.05; x2x3: p < 0.05; x1x2: p < 0.1
HVI properties
Upper half mean length (mm)	$y=30.56433+0.45672x_1-0.01905x_2-0.00010x_3-0.01628x_1^2+0.00002x_2^2-0.00007x_3^2+0.00004x_1{x}_2-0.00230x_1{x}_3+0.00007x_2{x}_3$	0.77	x1: p < 0.1; x2: p < 0.1
Uniformity index (%)	$y=88.68152+0.00076x_1-0.03418x_2+0.02505x_3+0.00876x_1^2+0.00003x_2^2-0.00116x_3^2-0.00012x_1{x}_2-0.00013x_1{x}_3+0.00009x_2{x}_3$	0.91	x2: p < 0.05$;$x22: p < 0.05
Trash count (number of particles)	$y=91.48584-2.59638x_1-0.14385x_2-1.66912x_3+0.00836x_1^2+0.00006x_2^2+0.01183x_3^2+0.00394x_1{x}_2+0.02241x_1{x}_3+0.00113x_2{x}_3$	0.95	x3: p < 0.05; x1x2: p < 0.05; x2x3: p < 0.05; x2: p < 0.05 x32: p < 0.1; x1x3: p < 0.1; x1: p < 0.1
Strength (grams/tex)	$y=39.26359-0.14563x_1-0.05744x_2+0.12637x_3-0.00212x_1^2+0.00006x_2^2-0.00172x_3^2+0.00069x_1{x}_2-0.00078x_1{x}_3+0.00005x_2{x}_3$	0.61	x2: p < 0.1
Short fiber content (%)	$y=3.295573-0.364407x_1+0.013093x_2+0.190303x_3+0.005150x_1^2-0.000010x_2^2-0.002741x_3^2+0.000123x_1{x}_2+0.002479x_1{x}_3-0.000083x_2{x}_3$	0.90	x3: p < 0.05; x32: p < 0.05
Reflectance (Rd)	$y=76.59609+0.29431x_1+0.00269x_2+0.13551x_3-0.01614x_1^2-0.0000015x_2^2-0.00202x_3^2-0.00004x_1{x}_2+0.00060x_1{x}_3-0.0000016x_2{x}_3$	0.63	ns
Yellowness (+b)	$y=\mathrm{8.6730.42}-0.023575x_1+0.003635x_2+0.000920x_3+0.001300x_1^2-0.000005x_2^2-0.000001x_3^2+0.000045x_1{x}_2-0.000032x_1{x}_3+0.000001x_2{x}_3$	0.88	x22: p < 0.1; x2: p < 0.1
Micronaire	$y=4.130914-0.055023x_1+0.002306x_2-0.009309x_3+0.002668x_1^2-0.000003x_2^2+0.000173x_3^2-0.000002x_1{x}_2+0.000116x_1{x}_3-0.000002x_2{x}_3$	0.63	x11: p < 0.10
Spinning consistency index	$y=193.1217+1.1610x_1-0.3839x_2+0.6602x_3-0.0310x_1^2+0.0004x_2^2-0.0136x_3^2+0.0015x_1{x}_2-0.0081x_1{x}_3+0.0007x_2{x}_3$	0.83	x2: p < 0.05; x22: p < 0.05

ns, not significant.

present the response surface models and coefficient of determination (R²) values for saw- and roller-ginned pneumatic fractionated lint, respectively. The saw-ginned pneumatic fractionated process models developed for lint moisture content and total trash extracted fit well, based on an R² of 0.99. Additionally, the models were statistically significant (p < 0.001). For roller-ginned pneumatic fractionated lint, the models for lint moisture content and total trash extracted had R² values of 0.95 and 0.99, respectively, and both models were statistically significant (p < 0.001). All the models of HVI properties for the saw-ginned pneumatic fractionated lint had R² values ≥ 0.80, except for reflectance (Rd), which had R² = 0.77 (Table 3). Also, except for the uniformity index and spinning consistency index models, the HVI models were statistically significant. For the roller-ginned pneumatic fractionated lint, HVI property models of uniformity index, trash count, short fiber content, yellowness (+b), and spinning consistency index had R² ≥ 0.80, and the upper half mean length model had R² = 0.77. The strength, reflectance, and micronaire models had R² values greater than 0.60 (Table 4). The uniformity index, trash count, short fiber, yellowness, and spinning consistency index models were statistically significant (p < 0.05), whereas the other models were not. The model-predicted and residual values were examined for both saw- and roller-ginned fractionated lint to assess prediction accuracy, particularly for models with lower R² values. Analysis of residual data for all models showed that, even at an R² of about 0.60, most residuals were within the precision limits of the HVI equipment [23], suggesting that the model predictions could be useful for optimizing the pneumatic fractionation process. The complete set of experimental data used for the RSM modeling of roller- and saw-ginned pneumatic fractionated Upland cotton is provided in the Supplementary Data File (Tables S1–S4).

3.1.1. Final Lint Moisture Content and Total Trash Extraction Models

Pareto charts were used to understand the statistical significance of the pneumatic fractionation process variables for saw- and roller-ginned lint properties. Table 3 and Table 4 show the statistically significant pneumatic process variables regarding final lint moisture content and total trash extracted for both saw- and roller-ginned pneumatic fractionated lint. In the case of final lint moisture content for roller- and saw-ginned lint, initial lint moisture content and the interaction between initial lint moisture content and residence time were statistically significant for both saw- and roller-ginned pneumatic fractionated lint. For saw-ginned lint, the interaction between initial lint moisture and line pressure was also statistically significant. For the total trash extracted during pneumatic fractionation, interactions among lint moisture content, line pressure, and residence time were statistically significant for pneumatic fractionated saw-ginned lint. For roller-ginned lint, line pressure interactions with residence time and initial lint moisture content interactions with line pressure, along with quadratic terms of lint moisture, were found to be statistically significant.

3.1.2. HVI Properties Models

Table 3 and Table 4 list the significant pneumatic fractionator process variables for HVI properties of saw- and roller-ginned lint, respectively. For saw-ginned and pneumatic fractionated lint, initial lint moisture content was a significant variable (p < 0.1) in the models for every lint property, except uniformity index, strength, reflectance, and spinning consistency (Table 3). Initial lint moisture was a significant variable for only a few lint properties (upper half mean length, trash count, and micronaire) of roller-ginned-pneumatic fractionated lint (Table 4). It was the only significant variable for saw-ginned lint trash count and roller-ginned lint micronaire. Line pressure was a significant variable in the models for saw-ginned lint upper half mean length, reflectance, yellowness, and micronaire. It was also a significant variable in models describing upper half mean length, uniformity index, trash count, strength, yellowness, and spinning consistency index for roller-ginned fractionated lint. Line pressure was the only significant variable for saw-ginned lint reflectance and roller-ginned lint uniformity index, strength, yellowness, and spinning consistency index. Residence time was significant only for upper half mean length, short fiber content, and micronaire from saw-ginned pneumatic fractionated lint, and for only trash count and short fiber content from roller-ginned lint. It was always paired with one of the other process variables, except for the roller-ginned lint short fiber content model, where it was the sole significant variable. The interaction between initial lint moisture content and the other two process variables (line pressure and residence time) affects the final lint moisture content, total trash, and most HVI properties. Previous work on the pneumatic fractionator indicated that line pressure and residence time at low moisture (about 5%, w.b.) for saw-ginned Upland cotton showed less interaction between these two variables, as observed in this study [15,16]. However, the initial lint moisture content interacted with the other two variables (line pressure and residence time) and affected the final moisture content, total trash extracted, and most HVI properties. Surface plots derived from these models provided a better understanding of the interactions among the three process variables and were used for subsequent optimization studies.

3.2. Response Surface Plots for Pneumatic Fractionated Saw- and Roller-Ginned Lint

The response surface models (Table 3 and Table 4) were used to develop response surface plots for the pneumatic-fractionated lint properties of saw- and roller-ginned lint. The entire model, incorporating both significant and non-significant variables, was used to predict data across various levels of process conditions, and the resulting predictions were used to generate surface plots. This was also performed to compare how the three pneumatic fractionator process variables tested impacted the lint properties for the roller- and saw-ginned lint, and especially to understand the lint moisture interaction with line pressure and residence time (even though some were statistically non-significant), since the effect of ranging initial lint moisture content on lint quality during pneumatic lint cleaning is not available in the literature. As there were three process variables, one was fixed at the central point of the experimental design, and the surface plot was generated for the other two by varying the process conditions. Lint moisture content, total trash content extracted, and HVI property variable surface plots showing the effect of the significant pneumatic fractionator process variables are shown in Figure 6, Figure 7, Figure 8 and Figure 9. The complete set of surface plots, drawn for all process conditions with respect to the studied responses, is given in the Supplementary File (Figures S1–S11).

3.2.1. Final Lint Moisture Content (%, w.b.) and Total Trash Extracted

Figure 6 presents surface plots of the final lint moisture content and total trash extracted with respect to key process variables of the pneumatic fractionation process. The surface plots indicate that the initial moisture content of the lint significantly affects its final moisture level. At lower initial moisture levels of 5.5% (w.b), there is only a marginal change in the lint moisture content (a reduction of approximately one percentage point). However, at higher initial moisture levels, the loss of moisture is considerable, decreasing from 15% to about 7.5% (Figure 6a,b) for both saw- and roller-ginned pneumatic fractionated lint. The higher moisture loss in the lint at higher moisture content may be due to most of the moisture being surface-bound and having left the lint quickly, whereas at lower moisture content, the loss is lower, and the moisture is more tightly bound to the lint. For the total amount of trash extracted during pneumatic fractionation, both line pressure and residence time had a significant effect. Increasing the line pressure from 276 to 542 kPa, along with extending the residence time from 15 to 45 s, nearly doubled the total trash extracted, increasing it from about 3-4% to about 8–10% (as illustrated in Figure 6e–f). The surface plots (Figure 6c–d) indicate that the lint moisture content had a lesser effect than residence time and line pressure.

3.2.2. HVI Properties

The upper half mean length of the pneumatic fractionated lint was influenced more by the initial lint moisture content than by line pressure. For saw-ginned pneumatic fractionated lint, higher lint moisture levels around 13–15% (w.b.) and line pressures between 320 and 480 and 460 kPa resulted in an upper half mean length of over 28 mm (Figure 7a). In contrast, for roller-ginned pneumatic fractionated lint, moisture levels of about 10–14% (w.b.), combined with higher line pressures exceeding 540 kPa or less than 300 kPa, produced the longest upper half mean lengths, which were greater than 29.8 mm (Figure 7b). The trend for the uniformity index showed that higher initial lint moisture content and lower line pressure resulted in a higher length uniformity index, exceeding 81% for saw-ginned and 83% for roller-ginned pneumatic fractionated lint (Figure 7c,d). Short fiber content was lowest (6–7%) at high initial lint moisture content and low residence time, and peaked (>9.5% for saw-ginned lint and >8% for roller-ginned lint) at about 30 s residence time and lower moisture content (Figure 7e,f). Interestingly, lower short fiber content could be achieved by extending residence time past 30 s at higher lint moistures contents. Studies on cotton fiber humidification at cotton ginneries found that higher moisture levels resulted in increased upper half mean length and uniformity index [29]. Another probable reason for the increase in the upper half mean length and uniformity index, and the reduction in short fiber content at higher moisture content, is the extraction of most of the immature fibers or motes, leaf, and fine trash, which typically contain higher short fiber content.

Lint strength was positively influenced by moisture content. At initial lint moisture contents of 14–15% (w.b.), medium to high line pressures (greater than 400 kPa for saw-ginned and greater than 500 kPa for roller-ginned pneumatic fractionated lint) resulted in higher lint strength, which exceeded 29.4 g/tex for saw-ginned and 31.5 g/tex for roller-ginned lint (Figure 8a,b). In contrast, lower lint moisture levels (5.5% w.b.) along with higher line pressures resulted in the lowest lint strength, below 27.6 g/tex for saw-ginned and below 29.2 g/tex for roller-ginned pneumatic fractionated lint. The total trash count (number of particles) decreased with increasing line pressure, at a medium to higher lint moisture contents of 9–13% (w.b.). The lowest trash particle count, 7–9, for saw-ginned pneumatic fractionated lint was observed across a wide range of lint moisture levels (8–14% w.b.) and line pressures (520–540 kPa; Figure 8c). For roller-ginned pneumatic fractionated lint, a trash count of less than 12 was observed at line pressures of 520–540 kPa and lower to medium initial lint moisture content of 5.5–8% (w.b.) (Figure 8d). The highest micronaire values for saw-ginned pneumatic fractionated lint, exceeding 4.3, were found at lint moisture levels of 8–9% (w.b.) and line pressures ranging from 280 to 400 kPa (Figure 8e). The lowest micronaire values (<4.11) were observed at both lower and higher lint moisture levels (5–5% and 14–15% w.b.), where line pressures in the range of 280–340 kPa were preferable at higher lint moistures, and 460–540 kPa at lower lint moistures. At 14–15% lint moisture content (w.b.), lower line pressures of 280–320 kPa were desired to lower the micronaire values (<4.06) (Figure 8e). In contrast, roller-ginned pneumatic fractionated lint exhibited the lowest micronaire values, below 4.08, at lower and higher pressures (<280 kPa and 520–540 kPa) and at medium lint moisture levels of 8-10% (w.b.) (Figure 8f). To produce lint with optimal micronaire values of 3.7 to 4.2 [23] for saw-ginned lint, both higher and lower moistures, and higher line pressures are desirable; for roller-ginned lint, medium and higher line pressures, and medium lint moisture levels are desirable. The surface plots indicate that at higher lint moisture contents, due to the extraction of most of the trash, including motes, leaf, and fine trash, and due to the fibrous nature of the lint, the fiber strength and length might have improved. The micronaire is an inherent property of the fiber; changes in its value can be due to variations in the amount of immature fiber, short fiber content, and trash count in the lint.

The color attributes of the lint, such as reflectance and yellowness, were measured for both pneumatic fractionated saw- and roller-ginned lint. The highest reflectance values, exceeding 81, for saw-ginned pneumatic fractionated lint were recorded at higher and lower moisture levels (>14% and <6% (w.b.)) and line pressures of 440–500 kPa (Figure 9a). For roller-ginned pneumatic fractionated lint, maximum reflectance was noted at a medium moisture level of 7–11% (w.b.) and line pressures greater than 500 kPa (Figure 9b). Yellowness values increased (>9.5) at higher moisture levels of 13–15% (w.b.) and line pressures between 380 and 480 kPa for both saw- and roller-ginned pneumatic fractionated lint (Figure 9c,d). The spinning consistency index (SCI) values were highest (exceeding 120) for saw-ginned pneumatic fractionated lint at higher moisture levels of 12–15% (w.b.) and low line pressures ranging from 280 to 340 kPa (Figure 9e). For roller-ginned pneumatic fractionated lint, higher and lower line pressures of 280–300 kPa and 520–540 kPa, along with high lint moisture contents of 14–15% (w.b.), resulted in high SCI values, exceeding 143 (Figure 9f). Studies indicated adding moisture to the lint up to 7.9% (w.b.) after ginning does not affect reflectance values for up to 6 months, but can slightly increase yellowness values [30]. The present observation regarding the yellowness and reflectance corroborates this observation, where the final moisture content of the lint after pneumatic fractionation for both roller- and saw-ginned lint, which is <7.5% (w.b.), resulted in marginal changes in reflectance and yellowness values.

3.3. Optimization of Pneumatic Fractionation Process

Figure 10 presents the HGA-based optimization output for roller-ginned, pneumatic, fractionated lint. The surface plot shown is based on Equation (3) and relates lint moisture content and residence time to the desired lint quality attributes.

Table 5. Optimized process conditions for the pneumatic fractionation process.

Optimized Process Conditions	Saw-Ginned Lint	Roller-Ginned Lint
Lint moisture content (%, w.b.)	14.64	14.83
Line pressure (kPa)	276.01	276.12
Residence time (s)	44.92	35.07

lists the optimized process conditions for saw- and roller-ginned pneumatic-fractionated lint, determined using Equation (3). Achieving optimal fiber properties in saw-ginned lint required longer residence times, whereas roller-ginned lint only needed a moderate residence time. Both lint types benefited from a lower line pressure of about 276 kPa and higher lint moisture content (14.6–14.8% w.b.) to reach desirable fiber properties.

Comparison of Lint Properties: No Lint Cleaning (NO LC), Industry-Standard Lint Cleaning (1-LC), and Pneumatic Fractionation

Table 6. Saw- and roller-ginned lint properties: (a) no lint cleaning, (b) industry-standard saw- and pin-type lint cleaner, and (c) pneumatic fractionator optimized process conditions.

		Saw-Ginned Cotton				Roller-Ginned Cotton
S. No.	Lint properties	No LC *	Saw LC *	PF *	No LC *	Pin LC *	PF *
1	Final Lint moisture content (% w.b.)	5.25	5.29	7.28	5.35	5.30	7.25
2	Total Trash Extracted (%)		4.35	4.77		3.19	5.77
3	Upper Half Mean Length (mm)	28.4	26.9	27.9	29.2	28.4	29.6
4	Uniformity Index (%)	81.8	80.0	80.7	82.6	82.5	83.2
5	Strength (grams/tex)	29.4	27.3	28.6	29.6	28.5	30.6
6	Reflectance (Rd)	78.4	80.3	81.0	77.5	79.6	80.4
7	Yellowness (+b)	8.68	9.02	9.31	8.64	9.38	9.44
8	Micronaire	4.38	4.23	4.09	4.58	4.20	4.22
9	Short Fiber Content (%)	7.0	9.4	8.2	6.1	6.6	6.2
10	Trash count (number of particles)	43	13	13	97	32	14
11	Spinning Consistency Index	126	111	119	130	130	143
	Color Grade **	31	21	11	31	11	11
	Leaf Grade ***	3.4	1.1	1.2	5.4	2.8	1.4

Note: The properties reported in the table for PF are predicted values at the optimized process conditions using the RSM models; for Saw LC, Pin LC, and No LC, the data reported are the actual experimental values. * Lint Cleaning Treatments: No LC = no lint cleaning, Saw LC = saw-type lint cleaners, and Pin LC = pin-type lint cleaner; PF = pneumatic fractionator. ** Color grade values determined based on Rd and +b values. *** Leaf grade values for no lint cleaning and saw-type lint cleaning were measured values, whereas for pneumatic fractionation, leaf grade values were taken from the experimental data close to the optimized process conditions (Table 5).

provides the lint HVI properties measured with no lint cleaning (after ginning), after lint cleaning using saw- and pin-type lint cleaners, and the predicted lint properties at the optimized pneumatic process conditions for saw- and roller-ginned pneumatic fractionated lint using response surface models. At the optimized pneumatic fractionator process conditions, there is about 50% moisture loss from the saw- and roller-ginned pneumatic fractionated lint, reducing it from 14.64% to 7.28% (w.b.) (50.27% reduction) and from 14.83% to 7.31% (w.b.) (50.7% reduction), respectively. The total trash (motes, leaf & fine trash) extracted from the lint at the optimized pneumatic fractionator process conditions was higher for roller-ginned lint (5.80%) compared to saw-ginned lint (4.77%). The conventional lint cleaners (saw- and pint-type lint cleaners) extract less total trash than the pneumatic fractionator. The saw-type lint cleaner removed 4.35% total trash compared to 4.77% for the pneumatic fractionator, and the pin-type lint cleaner for roller ginning removed 3.19% total trash compared to 5.8% for the pneumatic fractionator. Comparing the two industrial-type lint cleaners, the saw-type lint cleaner removed more total trash than the pin-type lint cleaner (4.35% versus 3.19%).

Many of the HVI properties of pneumatic fractionated lint were better than those of standard saw- and pin-type lint cleaners. The upper half mean length was about 1.0 mm longer for pneumatic fractionated lint compared to the saw- and pin-type lint cleaners. The upper half mean length values for pneumatic fractionated roller-ginned lint were closer to no lint cleaning. Uniformity index values were 0.75 percentage points higher for saw- and roller-ginned pneumatic fractionated lint compared to the standard saw-type and pin-type lint cleaning. The pneumatic fractionated lint uniformity index was about 1.0 point lower than that of lint with no lint cleaning for saw-ginned lint, and about 0.7 points higher for roller-ginned lint. For saw-ginned cotton, the pneumatic fractionator produced lint that was about 1.3 g/tex stronger than the saw-type lint cleaner, and for roller-ginned cotton, about 2 g/tex stronger than the pin-type lint cleaner. Like the uniformity index results, pneumatic fractionated lint was stronger than no lint cleaning (30.6 vs. 29.6 g/tex) after roller ginning, and it was weaker (28.6 vs. 29.4 g/tex) after saw ginning.

The reflectance (Rd) and yellowness (+b) values, which define the color grade of the lint, were higher after pneumatic fractionation lint cleaning, both saw- and roller-ginned lint, compared to the saw- and pin-type lint cleaners and no lint cleaning. Color grade calculated from the Rd and +b values improved from 31 with no lint cleaning, to 21 after saw-type lint cleaning, to 11 after pneumatic fractionation. In contrast, roller-ginned lint showed an improvement in color grade from 31 after no lint cleaning to 11 after pin-type and pneumatic fractionator lint cleaning. Micronaire values were lower after conventional lint cleaning and pneumatic fractionation. Pneumatic fractionation produced lint falling within the optimal range (3.7–4.2) for both saw and roller ginned lint, and were better than those without lint cleaning, which ranged from 4.4 to 4.6. Short fiber content of saw-ginned cotton increased by about 1.2 percentage points after pneumatic fractionation, compared to no lint cleaning, but increased by about 2.4 percentage points after saw-type lint cleaning. For roller-ginned cotton, the differences in short fiber content were marginal, increasing from 6.1% with no lint cleaning to 6.2% after fractionation to 6.6% after pin-type lint cleaning. The pneumatic fractionator produced lint with similar short fiber content as that of no lint cleaning for the roller-ginned lint. The trash count decreased by almost 70% after lint cleaning with the saw- and pin-type lint cleaners. The saw-type lint cleaner was more effective at reducing the trash count than the industry-standard pin-type lint cleaning system (Table 6). The trash count was almost similar for saw-ginned lint cleaned with the saw-type lint cleaner and pneumatic fractionator (13-14 particles). In contrast, roller-ginned pneumatic fractioned lint had about 56% lower trash count than lint cleaned with the pin-type lint cleaner (14 vs. 32 particles). The spinning consistency index calculated based on Equation 1 for HVI data decreased more with the saw-type lint cleaner compared to no lint cleaning than with the pneumatic fractionator (126 for no lint cleaning, 111 for the saw-type lint cleaner, and 119 for the pneumatic fractionator). In contrast, the spinning consistency index for roller-ginned pneumatic fractionated lint was 143, better than lint produced after roller ginning (no lint cleaning) and with an industry-standard pin-type cleaner (130). The leaf grade values after saw ginning were lower (3.4) than roller ginning (5.4), whereas saw-type lint cleaning reduced the leaf grade values to 1, and the pin-type lint cleaner resulted in a leaf grade value of 2.8. The pneumatic fractionator for both saw- and roller-ginned lint resulted in leaf grade values between 1 and 2.

3.4. Loan Value

The loan value is calculated based on the HVI properties for pneumatic fractionated lint with respect to industry-standard lint cleaning (saw- and pin-type lint cleaners), and no lint cleaning is given in Figure 11. The loan values were calculated based on the HVI properties of lint cleaned with saw- and pin-type lint cleaners, and on the predicted HVI properties of pneumatic fractionated lint at optimized process conditions, using the 2021 cotton loan premium and discounts [27]. The loan value for pneumatic fractionated saw-ginned lint was about 4 cents/kg higher compared to lint cleaned with saw- and pin-type lint cleaners (1LC) and 2 cents/kg lower compared to no lint cleaning (NO-LC) (Figure 11). In the case of roller-ginned pneumatic fractionated lint, the loan value was about 4 cents/kg higher than that of lint cleaned using the pin-type lint cleaner. The loan values were higher for roller-ginned lint than for saw-ginned lint after conventional and pneumatic fractionation lint cleaning. The loan value decreased by about 2 cents/kg after saw-type lint cleaning of saw-ginned lint compared to no lint cleaning; whereas for roller-ginned lint, pin-type lint cleaning increased the loan value by about 14 cents/kg (110–124 cents/kg) compared to no lint cleaning (Figure 11). The primary reason for this increase in loan value for roller-ginned lint with pin-type lint cleaning is the reduction in the leaf grade from 6 to 2. Further increases in loan values after pneumatic fractionation for both saw- and roller-ginned lint are primarily due to improvements in length, uniformity index, and fiber strength. The micronaire values for saw-ginned, industry-standard-cleaned lint and for pneumatically cleaned roller-ginned lint were not in the premium range (3.7–4.2), whereas the micronaire value for saw-ginned, pneumatically cleaned lint was.

The loan rate and lint bale weight, calculated based on the total trash extracted, indicated that pneumatic fractionated saw-ginned lint had a higher value (about 8 $/bale) compared to lint cleaned using a saw-type lint cleaner (259.7 $/bale for saw-type lint cleaner versus 267.1$/bale pneumatic fractionator), even though the total trash extracted was slightly higher for pneumatic fractionator (Table 6). In the case of roller-ginned lint, the change in the lint bale value for roller-ginned pin-type lint cleaner to pneumatic fractionator was marginal (271.9 versus 271.4 $/bale), even though the total trash extracted for roller-ginned, pneumatic, fractionated cotton was about 2.6% higher compared to pin-type lint cleaner (Table 6).

3.5. Discussion

In general, higher moistures are not added to the lint before lint cleaning at cotton gins, as excessive moisture reduces cleaning efficiency of saw- and pin-type lint cleaners, and also, the baled lint moisture must be below 7.5% to be eligible for the cotton loan program [28]. This work showed that even if a higher moisture content of about 15% (w.b.) is added to the lint after ginning, it can be effectively removed during lint cleaning to ≤7.5% (w.b.) and, at the same time, help retain most of the lint properties similar to those after ginning. The results indicated that the primary drivers for most of the surface moisture removal from the lint during pneumatic fractionation were the line pressure of the ambient air entering the chamber and the lint’s residence time in the cleaning chamber, with higher line pressure and residence time increasing moisture reduction, resulting in most of the surface moisture loss in the lint.

The saw-type lint cleaner combines centrifugal force, the scrubbing action of the saw cylinder and grid bars, and gravity, assisted by an air current, to clean cotton lint [6]. These combined actions can result in decreased upper-half mean length and increased short-fiber content during lint cleaning, as observed in the present study. In the case of the pin-type lint cleaner, the principle of separation is similar to that of the saw-type lint cleaner, but instead of saws, a pin cylinder removes the trash by scrubbing against grid bars, which is a gentler process [6] that has resulted in lower trash removal but retained the lint length parameters such as upper half mean length and uniformity index. Past research has shown that standard lint cleaners used by the ginning industry reduce fiber length and uniformity [12]. Studies on controlled-batt-saw lint cleaners indicated that they positively impact color and trash; negatively effect length, length uniformity, short fiber, nep content, and elongation; and have no effect on strength, micronaire, fineness, maturity, number, and size of the seed coat neps and fiber nep size. Meanwhile, on the other hand, batt-less saw lint cleaners had similar effects on fiber quality, although not as severe as batt-type lint cleaners [31]. The improvement in fiber quality during lint cleaning, especially length uniformity, can be achieved by (a) proper moisture management and (b) the use of alternative lint cleaning technologies that are more gentle. In the pneumatic lint-cleaning process, lint is tumbled across slotted screens in a closed container using air, and trash is removed through the fractionator screens. The experimental data indicated that this process is gentler and can handle wide range of lint moistures and more effective than industry-standard saw- or pin-type lint-cleaning systems. Also, the present experimental data showed that more total trash is extracted with the pneumatic fractionator than with saw- and pin-type lint cleaners, suggesting that more immature fibers and short fibers are removed during cleaning, thereby improving the lint length parameters.

The strength of pneumatic fractionated lint is greater than that of lint cleaned using saw- and pin-type lint cleaners, which can be due to the addition of moisture to the lint before cleaning. Byler [32] concluded that moisture content significantly affects fiber strength. According to Anthony and Mayfield [6] the fiber breakage rate during ginning is inversely proportional to the fiber’s moisture content at the gin stand and lint cleaners. Studies on lignocellulosic biomass drying by thermal and chemical methods have shown that thermal drying reduces pore size and volume compared to chemical drying with dimethyl ether. Many researchers have found that surface moisture removal in biomass fibers is driven by evaporation (convection/diffusion) and can affect the mechanical and morphological properties of the fibers [33,34,35]. These authors concluded that surface moisture leaves the fiber surface layers by forming liquid bridges that pull fibers together via capillary action. This process changes the fiber morphology, affecting the physical properties such as strength, and the morphological properties such as pore size and pore volume. In the present study, surface moisture removal from lint using ambient air (which is gentler than thermal methods) as a function of residence time and line pressure might have helped retain the lint’s morphological features and preserved its length and strength properties.

The improvements in the color parameters’ reflectance and yellowness after pneumatic fractionation may be due to the removal of greater amounts of trash. According to Baker and Brashears [36], reflectance values after lint cleaning increase due to a decrease in trash count, and yellowness increases at higher moisture content [37]. These observations corroborate the present results.

Improved lint properties after pneumatic fractionation resulted in a higher loan rate and, in turn, a higher bale value compared to lint cleaned using saw- and pin-type lint cleaners. For saw-ginned pneumatic fractionated lint, the bale value is approximately 8 $ higher compared to lint cleaned using a saw-type lint cleaner. In contrast, for roller-ginned pneumatic fractionated lint, even though the lint properties are superior compared to lint cleaned using a pin-type lint cleaner, the bale value was similar, mainly because the total trash extracted was higher. Our future research will focus on analyzing the total amount of lint extracted from roller-ginned, pneumatic, fractionated total trash to determine whether we are removing any good-quality lint during the cleaning process.

Another critical aspect of the pneumatic fractionator is its operational safety. According to the US Department of Labor, Occupational Safety and Health Administration (OSHA), saw-and-pin-type lint cleaners pose safety challenges [38]. In contrast, pneumatic fractionators, which primarily involve screens with no moving parts, can reduce the risk of injury.

Future work on pneumatic fractionation includes conducting tests at optimized conditions to validate the fiber properties, and developing a laboratory-scale modular pneumatic lint cleaning system. The modular system will include the following features: (a) changeable screens with different cleaning slot sizes, shapes, and surface areas, (b) high-speed video capture to better understand the mechanism of lint and motes, leaf and fine trash separation, (c) flowmeters and control valves to enable precise control of air pressure, flow rate, and distribution within the lint-cleaning chamber, and (d) newly designed length extensions to aid scale-up studies.

4. Conclusions

A pneumatic fractionator, which was tested for cleaning saw- and roller-ginned Upland cotton, resulted in better lint properties compared to saw- and pin-type lint cleaners used by the industry. The RSM models adequately described the pneumatic fractionation process, as indicated by the coefficient of determination and by comparisons between observed and predicted values. The response surface analysis and pareto charts showed that the initial lint moisture content interacted with line pressure and residence time, influencing total trash extraction, lint moisture content, and lint HVI properties. Optimization of the RSM models indicated that residence times are different—45 sec for saw-ginned lint and 35 sec for roller-ginned lint—but line pressure (about 276 kPa) and lint moisture content (14–15%, w.b.) were similar. Higher initial lint moisture and lower line pressure positively impacted most lint properties, especially the upper half mean length, uniformity index, strength, and short fiber content of saw- and roller-ginned pneumatic fractionated lint. The reflectance and yellowness of pneumatic-fractionated lint samples were higher than those achieved with industry-standard lint-cleaning methods, resulting in better color grades. The pneumatic fractionation of saw-ginned lint had lower micronaire values than the saw-type lint cleaner, whereas for roller-ginned lint, the change in the micronaire was marginal. The spinning consistency index was higher for pneumatic fractionated roller- and saw-ginned lint compared to lint cleaned using industry-standard saw- and pin-type lint cleaners. Loan values are higher for the pneumatic fractionated saw- and roller-ginned lint, approximately 4 cents/kg, compared to lint cleaned using industry-standard methods.