An Adaptive Spring-Loaded Yam Peeling Machine for Reducing Post-Harvest Processing Losses in Irregularly Shaped Tubers

1. Introduction

Yam is an important root and tuber crop for food supply, household consumption, and agro-processing. Its commercial value depends not only on production but also on timely and efficient post-harvest operations. Mechanized post-harvest equipment can reduce drudgery and processing losses while improving labor productivity and product consistency [1]. Peeling is particularly important because it precedes slicing, drying, milling, cooking, and flour production. However, yam peeling remains difficult to mechanize because tubers vary substantially in diameter, length, curvature, and local surface profile.

Manual peeling is flexible enough to follow irregular tuber geometries, but it is labor-intensive, time-consuming, operator-dependent, and unsuitable for high-throughput processing [2,3]. Chemical and thermal approaches can loosen the peel, but chemical use, residue management, energy demand, and partial cooking can limit their suitability for decentralized processing [4,5,6]. Mechanical peeling therefore remains attractive where fresh tissue must be preserved and processing is required at small or medium scale.

Abrasive drums and rotating brush systems are simple and productive, but incomplete contact in recessed regions and excessive contact at protruding regions can produce a trade-off between peel removal and flesh loss. A multi-tuber peeling machine demonstrated the potential of mechanized processing but also showed that performance depends on product properties and operating conditions [7]. For yam specifically, a spring-loaded knife machine reported efficiencies of 71.2–100% and losses of 3.67–14.29% [8]. A combined yam peeling and slicing machine achieved 87.86% efficiency, 12.1% peeling loss, and 114 kg h⁻¹ throughput [9]. Earlier drum-based equipment reported efficiency up to 95%, but peeling loss reached 23% [10]. These results demonstrate the continuing need to combine high peel removal, reduced edible-matter loss, and practical processing capacity.

Recent work has explored more adaptive approaches. A computer-vision-assisted cassava system used contour recognition to guide peeling and reported improved control of efficiency and flesh loss [11]. A waterjet potato-peeling study used fluid–structure interaction to investigate precision-peeling parameters [12]. Although such systems demonstrate the value of adaptation and precision, practical mechanical solutions for highly irregular yam tubers remain limited.

The engineering hypothesis of the present study is that the cutting elements should accommodate the local geometry of the tuber rather than force the tuber through a fixed global cutting envelope. Accordingly, this work develops an adaptive spring-loaded yam peeling machine that combines localized knife compliance, reciprocating cutting motion, and controlled angular indexing. The objectives were to: (i) design and fabricate the machine using accessible workshop processes; (ii) establish the force, power, and spring requirements; (iii) evaluate peeling efficiency, peeling loss, peeling rate, and capacity using ten tubers; and (iv) benchmark the prototype against previously reported yam-peeling technologies.

2. Materials and Methods

2.1. Design Requirements and Engineering Concept

The machine was developed for dry yam tubers stored for at least two months. The principal design requirements were to accommodate substantial geometric variability, use locally available materials and conventional fabrication methods, permit straightforward maintenance, and provide a processing capacity suitable for small- and medium-scale operations. The design envelope covered tuber diameters of 50–200 mm and lengths of 100–600 mm. The target capacity was 240–300 kg h⁻¹.

The developed machine consists of five principal subsystems: a reciprocating cutting head, two holding chucks, an indexing and quick-return transmission, linear guide rails, and a structural frame. The tuber is held approximately along its longitudinal axis between an adjustable non-indexing chuck and an indexing chuck. During the forward stroke, the cutting head traverses the tuber length while the spring-supported knives progressively remove the peel. At the end of each stroke, the cutting head returns to its initial position and the indexing chuck rotates the tuber to expose a new surface region. The spring-loaded knives are mounted on roller bearings and can deflect independently, enabling localized conformity to irregular yam profiles. Figure 1 presents the CAD assembly model and fabricated prototype developed in this study.

2.2. Materials Selection and Prototype Fabrication

Materials were selected by considering strength, wear resistance, corrosion resistance, machinability, cost, and local availability. High-speed steel was used for the knife edges to retain sharpness under repeated interaction with fibrous yam tissue. Mild steel was used for the knife shanks and housings because it is readily available and machinable. A36 carbon-steel sections and plates were used for the pivot arm and structural frame because of their strength, ductility, and weldability.

The prototype was fabricated using conventional workshop processes. As shown in Figure 1(b), the completed machine was constructed from standard steel sections, compression springs, bearings, and commercially available transmission components. The fabricated system closely followed the CAD configuration shown in Figure 1(a), ensuring that the intended reciprocating cutting action, adaptive knife compliance, and angular indexing mechanisms were implemented. The use of locally available materials and conventional fabrication techniques enhances maintainability and supports future adoption by small- and medium-scale yam processors.

2.3. Cutting-Force and Power-Transmission Design

The maximum peeling shear stress, $P_p$, was taken as 9.5 N mm⁻² based on reported resistance values for root and tuber crop processing [13]. The nominal peel thickness was 2 mm. The peel area engaged during one stroke was estimated as

$$A=C_{p}t,$$

(1)

where $C_p$ is the maximum circumference engaged by the tool and t is the peeling thickness. For $C_p=63$ mm and $t=2$ mm,

$$A=126{\mathrm{mm}}^2.$$

(2)

Using Merchant’s cutting relation, the shear angle was calculated as

$$\varphi ={45}^{\circ }+{\displaystyle \frac{\alpha }{2}}-{\displaystyle \frac{\beta }{2}},$$

(3)

where $\alpha ={20}^{\circ }$ and $\beta ={tan}^{-1}\left({\mu }_t\right)$. With ${\mu }_t=0.58$, $\beta =30.{11}^{\circ }$ and $\varphi =39.9^{\circ }$. The shear force was

$$F_s=P_{p}A=1197\mathrm{N}.$$

(4)

The cutting and thrust forces were estimated as

$$F_c={\displaystyle \frac{F_{s}cos(\beta -\alpha )}{cos(\varphi +\beta -\alpha )}},$$

(5)

$$F_t={\displaystyle \frac{F_{s}sin(\beta -\alpha )}{cos(\varphi +\beta -\alpha )}},$$

(6)

which gave $F_c=1835$ N and $F_t=327$ N. Friction in the reciprocating head was represented by

$$F_{rc}={\mu }_c{M}_{c}g,$$

(7)

where ${\mu }_c=0.2$, $M_c=5.8$ kg, and $g=9.81$ m s⁻². Thus, $F_{rc}=11.4$ N and the total operating force was

$$F_r=F_c+F_{rc}=1846.4\mathrm{N}.$$

(8)

For an 80 mm gear diameter, the calculated torque at gear A was

$$T_A={\displaystyle \frac{F_r{d}_A}{2}}=73.86\mathrm{N}\mathrm{m}.$$

(9)

A 2:1 gear ratio gave 36.93 N m at gear B. The calculated power requirement was 110.784 W. Applying a design factor of 1.8 produced a minimum selected specification of 200 W, up to 30 rpm, and 100 N m torque.

2.4. Spring-Loaded Compliance Design

Knife compliance is required because peel resistance and local geometry vary along the tuber surface. Controlled mechanical compliance has also been used in adaptive elastomeric mounting systems to manage force transmission under variable loading [14,15]. In the present machine, three identical compression springs support the knife system and permit local displacement of the blades.

Figure 2. Spring-loaded knife assembly providing localized compliance during peeling of irregular yam surfaces.

Figure 2. Spring-loaded knife assembly providing localized compliance during peeling of irregular yam surfaces.

The spring preload was

$$F_p=k{t}_p,$$

(10)

where k is the combined spring rate and $t_p$ is the preload deflection. The residual force under thrust loading was

$$R_p=F_t-F_p,$$

(11)

and the effective instantaneous cutting depth was

$$t_e=t-{\displaystyle \frac{R_p}{k}}.$$

(12)

A parametric study was used to identify a spring rate that required minimal adjustment across the expected range of peel resistance. The selected combined rate was 80 N mm⁻¹. Therefore, for three identical springs, the individual spring rate was 26.67 N mm⁻¹. The spring geometry was evaluated using

$$k_i={\displaystyle \frac{d^{4}G}{8D^3{N}_a}},$$

(13)

where $D=17$ mm, $N_a=6$, and $G=81,700$ N mm⁻². The resulting wire diameter was 2.96 mm.

2.5. Performance Evaluation and Data Analysis

Ten dry yam tubers with different masses, diameters, and lengths were used for the expanded prototype performance evaluation. Each tuber constituted one independent experimental observation. The tubers were evaluated using the same machine configuration and performance-measurement procedure. The initial tuber mass, diameter, peel length, peeling time, and post-peeling mass were recorded. Derived values were calculated from these measurements.

Peeling efficiency was defined as

$${\zeta }_{\mathrm{eff}}={\displaystyle \frac{A_{\mathrm{tuber}}-A_{\mathrm{unpeeled}}}{A_{\mathrm{tuber}}}}\times 100,$$

(14)

where $A_{\mathrm{tuber}}$ is total tuber surface area and $A_{\mathrm{unpeeled}}$ is the surface area remaining unpeeled. Peeling loss was

$${\beta }_L={\displaystyle \frac{m_t-m_{pt}}{m_t}}\times 100,$$

(15)

where $m_t$ and $m_{pt}$ are tuber masses before and after peeling, respectively. Peeling rate was

$$R_{pa}={\displaystyle \frac{l_t}{t_p}},$$

(16)

where $l_t$ is peeled length and $t_p$ is peeling time. Results were summarized using the range, arithmetic mean, and sample standard deviation. Linear or quadratic regression curves were fitted only as descriptive indicators of patterns within the ten-observation dataset. No treatment-based inferential analysis was performed because the experiment did not contain replicated groups defined by controlled operating settings.

3. Results

3.1. Peeling Performance

Table 1 summarizes the ten independent tuber observations. Mean peeling efficiency was $94.65\pm 3.61$ %, mean peeling loss was $3.92\pm 0.98$ %, and mean peeling rate was $44.94\pm 3.04$ mm s⁻¹. The estimated machine capacity was approximately 270 kg h⁻¹, within the target range of 240–300 kg h⁻¹.

3.2. Descriptive Relationships

Peeling efficiency ranged from 89.9% to 99.9%. Figure 3 shows a positive descriptive relationship between tuber diameter and peeling efficiency:

$${\zeta }_{\mathrm{eff}}=0.2377D_t+66.5774,R^2=0.686.$$

(17)

The relationship remained positive in the expanded dataset, although it was weaker than the relationship obtained from the preliminary five-observation subset. Diameter therefore contributed to the observed efficiency variation, but it did not fully explain machine performance.

Peeling loss ranged from 2.67% to 5.91%. The descriptive quadratic relationship was

$${\beta }_L=-0.000895D_t^2+0.17657D_t-4.32352,R^2=0.131.$$

(18)

The low coefficient of determination indicates that tuber diameter alone explained little of the observed variation in peeling loss. Local curvature, surface irregularity, peel adhesion, moisture condition, and tissue heterogeneity are likely to be influential.

Figure 4. Relationship between tuber diameter and peeling loss for the ten-tuber dataset.

Figure 4. Relationship between tuber diameter and peeling loss for the ten-tuber dataset.

Peeling rate ranged from 41.76 to 51.92 mm s⁻¹. The fitted descriptive relationship was

$$R_{pa}=-3.9321m_t+55.6616,R^2=0.350.$$

(19)

This relationship indicates a moderate tendency for peeling rate to decrease with increasing tuber mass, but additional controlled observations are required before the equation can be used predictively.

Figure 5. Relationship between tuber mass and peeling rate for the ten-tuber dataset.

Figure 5. Relationship between tuber mass and peeling rate for the ten-tuber dataset.

The relationship between peeling loss and efficiency was

$${\zeta }_{\mathrm{eff}}=-1.8137{\beta }_L+101.7541,R^2=0.244.$$

(20)

The negative slope is consistent with a general efficiency–loss trade-off, but the low coefficient of determination shows that peeling loss alone does not explain most of the efficiency variation.

Figure 6. Efficiency–loss relationship for the ten-tuber dataset.

Figure 6. Efficiency–loss relationship for the ten-tuber dataset.

4. Discussion

4.1. Comparison with Existing Yam-Peeling Technologies

Table 2 places the present results beside selected yam-peeler values reported in the literature. Direct comparison should be interpreted cautiously because the studies used different tuber varieties, maturity conditions, test protocols, and definitions of performance. Nevertheless, the present machine combined an efficiency range of 89.9–99.9%, loss of 2.67–5.91%, and estimated capacity of approximately 270 kg h⁻¹. The loss range was lower than those reported for the selected earlier yam-peeling machines, while capacity exceeded the 114 kg h⁻¹ reported for the combined peeler-slicer [9].

The principal design distinction is that each spring-supported knife can respond locally to tuber geometry, while angular indexing exposes successive surface regions. This avoids requiring the tuber to conform to a fixed circular opening or full-profile cutting envelope. The approach offers a mechanical alternative to more complex computer-vision and waterjet solutions [11,12], with potential advantages in workshop manufacture and maintainability.

4.2. Post-Harvest and Practical Implications

For small- and medium-scale yam processors, the main practical objectives are to reduce drudgery, preserve edible tissue, increase throughput, and use equipment that can be maintained locally. The estimated capacity of approximately 270 kg h ⁻¹ met the design objective, while the observed loss range indicates that the compliant knife arrangement can limit excess flesh removal in the tested tubers. If demonstrated through replicated and field-oriented trials, these features could support labor productivity and reduce avoidable material loss during primary processing.

The machine was intentionally based on conventional mechanical components rather than specialized sensing or robotic hardware. This design choice may improve accessibility, but practical deployment will also depend on operator safety, sanitation, cleaning time, knife replacement, energy use, lifecycle cost, and performance under continuous operation. The developed machine demonstrates potential for reducing manual peeling drudgery and improving labor productivity where mechanization options remain limited.

4.3. Interpretation of the Expanded Dataset

The addition of five independently recorded tuber observations broadened the tested ranges of mass, diameter, peel length, and peeling time. The overall mean values remained close to those estimated from the preliminary five-observation subset, while the sample standard deviations for efficiency, loss, and rate decreased. This consistency supports the stability of the central performance estimates. However, the expanded dataset also reduced the coefficients of determination for several regression relationships. In particular, the weak diameter–loss and efficiency–loss relationships indicate that performance cannot be explained by one geometric or outcome variable alone.

The results therefore support the engineering function of the prototype more strongly than they support any universal regression model. The machine achieved high efficiency and relatively low loss across the observed tubers, but future experiments should explicitly vary and replicate operating speed, spring preload, indexing angle, cultivar, storage duration, moisture content, and tuber-shape descriptors.

4.4. Limitations and Future Work

Although the dataset contains ten independent tuber observations, the experiment did not include replicated treatment groups involving controlled variations in machine settings or tuber conditions. The regression equations should therefore be interpreted as descriptive trends rather than predictive models. Tuber cultivar, moisture content, storage history, peel adhesion, curvature, and surface defects were not separately controlled. The surface-area method used for estimating peeling efficiency should also be documented with a repeatable measurement protocol in future studies.

Further work should evaluate larger randomized samples across cultivars and storage conditions; include replicated operating speeds, spring preloads, and indexing angles; report uncertainty and inferential statistics; quantify energy consumption, hygiene, and cleaning requirements; and conduct operator-safety and economic assessments. Continuous-duration testing would also be necessary before commercial-scale recommendations are made.

5. Conclusions

An adaptive spring-loaded yam peeling machine was developed to address the post-harvest challenge posed by irregular tuber geometry. The machine combines reciprocating cutting, localized spring compliance, and controlled angular indexing. In an evaluation involving ten dry yam tubers, peeling efficiency was 89.9–99.9% (mean $94.65\pm 3.61$ %), peeling loss was 2.67–5.91% (mean $3.92\pm 0.98$ %), peeling rate was 41.76–51.92 mm s⁻¹ (mean $44.94\pm 3.04$ mm s⁻¹), and estimated capacity was approximately 270 kg h⁻¹. The comparative results indicate that the developed concept can combine high peel removal with relatively low edible-matter loss.

The expanded dataset strengthens the prototype-validation evidence but does not replace the need for replicated, statistically designed experiments. Larger field-oriented evaluations are required to establish robustness, economics, sanitation, safety, and applicability across yam varieties and processing conditions. The adaptive peeling concept may contribute to the development of accessible post-harvest processing technologies for root and tuber crops, particularly where labor requirements, processing inefficiencies, and material losses remain important challenges.Future studies should include larger sample populations, replicated operating conditions, and techno-economic assessment to support commercial implementation.