Experimental Studies and Modelling of Innovative Peeling Processes for Tough-Skinned Vegetables
Experimental studies and modelling of innovative peeling processes for tough-skinned vegetables
Bagher Emadi
M.Sc. Mechanics of Agricultural Machinery B.Sc. Agricultural Machinery
Thesis submitted as a requirement for the degree of
Doctor of Philosophy
School of Engineering Systems Faculty of Built Environment and Engineering
Queensland University of Technology 2005
This dissertation is dedicated to my wife, Masoumeh, and my two lovely children, Roya and Amir Reza
ii Keywords peeling, mechanical peeling, abrasive peeling, mechanical properties, tough-skinned vegetables, model, mathematical model, peeling rate, peeling efficiency, peel losses, pumpkin, melon
iii Abstract:
Tough-skinned vegetables such as pumpkin and melon currently are peeled either semi-automatically or automatically. The main limitation of both methods, especially for varieties with an uneven surface, is high peeling losses.
Improvement of current mechanical peeling methods and development of new mechanical methods for tough-skinned vegetables which are close to the “ideal” peeling conditions using mechanical properties of the product were the main objectives of this research.
This research has developed four innovative mechanical peeling methods on the basis of the mechanical properties of tough-skinned vegetables. For the first time, an abrasive-cutter brush has been introduced as the best peeling method of tough-skinned vegetables. This device simultaneously applies abrasive and cutting forces to remove the peel. The same peeling efficiency at concave and convex areas in addition to high productivity are the main advantages of the developed method. The developed peeling method is environmentally friendly, as it minimises water consumption and peeling wastes.
The peeling process using this method has been simulated in a mathematical model and the significant influencing parameters have been determined. The parameters are related to either the product or peeler. Those parameters appeared as the coefficients of a linear regression model. The coefficients have been determined for Jap and Jarrahdale as two varieties of pumpkin. The mathematical model has been verified by experimental results.
The successful implementation of this research has provided essential information for the design and manufacture of a commercial peeler for tough-skinned vegetables. It is anticipated that the abrasive-cutting method and the mathematical model will be put into practical use in the food processing industry, enabling peeling of tough-skinned
iv vegetables to be optimised and potentially saving the food industry millions of dollars in tough-skinned vegetable peeling processes.
v Contents
Keywords iii Abstract iv List of Symbols and Abbreviations xvi Authorship xviii Acknowledgement xix 1 Introduction 1.1 Significance and motivation of the research 1 1.2 Objectives of the research 4 1.3 Originality and major contribution of the thesis 4 1.3.1 Definition of tough-skinned vegetables 4 1.3.2 Determination of mechanical properties of tough-skinned vegetables 5 1.3.3 Developing new mechanical peeling methods for tough-skinned vegetables 5 1.3.4 Developing current peeling methods 5 1.3.5 Introducing the best mechanical peeling method applicable in peeling industry for tough-skinned vegetables 6 1.3.6 Mathematical modelling of mechanical peeling 6 1.3.7 Determination of design parameters of tough-skinned vegetable peeler 6 1.4 Thesis organization 7 1.5 Publications (refereed) of the author arising from the PhD Research 8 2 Literature review 2.1 Mechanical properties of fruits and vegetables and methods of testing 11 2.1.1 Introduction 11 2.1.2 States of product to be tested 11 2.1.3 Compression test 12 2.1.4 Cutting test 13
vi 2.1.5 Shear strength test 13 2.1.6 Coefficient friction test 14 2.2 Peeling methods of fruits and vegetables 15 2.2.1 Introduction 15 2.2.2 Mechanical peeling 15 2.2.2.1 Abrasive devices 16 2.2.2.2 Devices using drums 17 2.2.2.3 Devices using rollers 17 2.2.2.4 Knives or blades 17 2.2.2.5 Milling cutter 20 2.2.3 Thermal peeling 22 2.2.3.1 Flame (dry heat peeling) 23 2.2.3.2 Steam (wet heat peeling) 23 2.2.3.3 Thermal blast peeling 25 2.2.3.4 Freeze-thaw 25 2.2.3.5 Vapour explosion (Vacuum peeling) 26 2.2.4 Chemical peeling 26 2.2.4.1 Caustic (lye) peeling 26 2.2.4.2 Enzymic peeling 28 2.3 The current situation of peeling tough-skinned vegetables 30 2.4 Mathematical modelling of peeling process 30 2.5 Conclusions and discussion 32 2.6 Summary 33 3 Testing of mechanical properties of tough-skinned vegetables 3.1 Introduction 35 3.2 Design and construction of instrumentation for testing vegetables properties 37 3.2.1 Cutter 37 3.2.2 Holder of unpeeled sample 37 3.2.3 Holder of skin sample 38 3.2.4 Indentor 38 3.2.4.1 Spherical end indentor 38 3.2.4.2 Flat end indentor 38 3.2.4.3 Cutting indentor 39
vii 3.2.5 Curvature meter 39 3.2.6 Friction coefficient tester 39 3.3 Testing methodology 39 3.3.1 Force-deformation test 41 3.3.2 Shear strength test 41 3.3.3 Cutting force test 42 3.3.4 Friction coefficient test 42 3.3.5 The relative contribution of skin to the unpeeled mechanical properties 43 3.4 Results and discussion 43 3.4.1 Force-deformation relationship 44 3.4.2 Toughness 47 3.4.3 Cutting force 47 3.4.4 Maximum force of shear strength 48 3.4.5 Shear strength 49 3.4.6 Static coefficient of friction 50 3.4.7 The relative contribution of skin to unpeeled mechanical properties 50 3.4.8 Application of investigated mechanical properties 53 3.5 Summary 55 4 Testing equipment for investigation of mechanical peeling methods 4.1 Introduction 56 4.2 Objectives of the design 57 4.2.1 Adaptability for investigation of different mechanical peeling tools 57 4.2.2 Possibility of accommodation of different product size 57 4.2.3 Possibility of peeler head position adjustment 57 4.2.4 Possibility of peeler tool position adjustment 57 4.2.5 Possibility of rotation of peeler tool at different angular velocities 58 4.2.6 Possibility of rotation of vegetable holder at different angular velocities 58 4.2.7 Simplicity and low cost of manufacturing 58 4.3 Enforcement of the objectives 58
viii 4.3.1 Chassis and chamber 58 4.3.2 Vegetable holder 58 4.3.3 Peeler head 61 4.3.4 Attachments 63 4.4 Performance of the test rig 64 4.5 Summary 65 5 Preliminary trials of different mechanical peeling methods 5.1 Introduction 66 5.2 Trials of different tools 67 5.2.1 Wire brush 67 5.2.1.1 Rotary wire brush 67 5.2.1.2 Twisted wire brush 68 5.2.2 Ball chain tool 71 5.2.3 Milling cutter 72 5.2.4 Mower trimming lines 73 5.2.5 Abrasive ropes 74 5.2.6 Abrasive pads 74 5.2.7 Abrasive foams 75 5.2.8 Rope covered by spiral blade 75 5.2.9 Sandpaper belt 77 5.2.10 Abrasive plates 78 5.2.11 Abrasive-cutter brush 79 5.2.12 Abrasive bristle products 80 5.3 Conclusions 81 5.4 Summary 82 6 Experimental investigation of mechanical peeling methods 6.1 Introduction 83 6.2 The criteria of experiments 84 6.2.1 Peel losses 84 6.2.2 Peeling efficiency 84 6.2.3 Estimated responses 85 6.2.4 Data analysis 86 6.3 Peeling by using milling cutter 86 6.3.1 Introduction 86
ix 6.3.2 Material of experiments 86 6.3.3 Results and discussion 88 6.3.4 Optimization and estimation of the responses 90 6.4 Peeling by using abrasive pads 92 6.4.1 Introduction 92 6.4.2 Material of experiments 92 6.4.3 Results and discussion 94 6.4.4 Optimization and estimation of the responses 98 6.5 Peeling by using abrasive foams 98 6.5.1 Introduction 98 6.5.2 Material of experiments 99 6.5.3 Results and discussions 100 6.5.4 Optimization and estimation of the responses 104 6.6 Peeling by using abrasive-cutter brush 105 6.6.1 Introduction 105 6.6.2 Material of experiments 106 6.6.3 Results and discussion 108 6.6.4 Optimization and estimation of the responses 111 6.7 The comparison of the four innovative peeling methods 112 6.8 Potential industrial application of abrasive-cutter brush 113 6.9 Conclusions 113 6.10 Summary 115 7 Abrasive-cutter brush, full factorial experiments, and ANOVA 7.1 Introduction 116 7.2 Material of experiments 117 7.3 Peeling rate 119 7.4 Data analysis 119 7.5 Results and discussion 119 7.5.1 The effect of p. speed on LnP. rate 122 7.5.1.1 The effect of p. speed on LnP.rate for different levels of coarseness 123 7.5.1.2 The effect of p. speed on LnP.rate in different locations of product 125 7.5.2 The effect of coarseness on LnP. rate 126
x 7.5.2.1 The effect of coarseness on LnP.rate at different p. speed 127 7.5.2.2 The effect of coarseness on LnP.rate at different locations of pumpkin 128 7.5.3 The effect of location of product’s surface on LnP.rate 129 7.5.3.1 The effect of location on LnP.rate in different coarseness of brush 130 7.5.3.2 The effect of location of product’s surface on LnP.rate at different p. speed 131 7.6 Conclusions and discussion 132 7.7 Summary 134 8 Modelling of mechanical peeling as sum of consumed energy in peeling process 8.1 Introduction 135 8.2 Theory of the model 136 8.2.1 The assumptions 136 8.2.2 Development of the model 136 8.2.3 Determination of the model coefficients 145 8.2.4 Model validation 145 8.3 Results and discussion 146 8.3.1 Model coefficients 146 8.3.2 Model validation 149 8.3.3 Applicability of the model 150 8.4 Conclusions 151 8.5 Summary 151 9 Conclusions and perspectives 9.1 Thesis summary and conclusions 153 9.2 Directions for future research 155 Appendices 158 1.1 Multiple comparisons of the mean of the mechanical properties 158 1.2 Multiple Comparisons of contribution of skin to the mechanical properties 171 1.3 Mechanical properties of varieties of melon and pumpkin in three different states including skin, unpeeled, and flesh 176
xi 1.4 Relative contribution (%) of skin to different mechanical properties for three pumpkin varieties including Jarrahdale, Jap, and Butternut 177 1.5 Drawings of instrumentations 178 2.1 Test rig 184 3.1 Experimental results of using milling cutter 196 3.2 Experimental results of using abrasive pads 196 3.3 Experimental results of using abrasive foams 197 3.4 Experimental results of using abrasive-cutter brush 198 4.1 Normality assessment of peeling rate (g/min) of Jap and Jarrahdale varieties 199 4.2 Multi comparisons of the mean of LnP.rate among different levels of independent variables 205 Bibliography 208
xii List of Figures
1.1 The top view of pumpkin 3
2.1 Force-deformation curve 13
2.2 An industrial application of an abrasive roller peeler for tuberal
products such as potato 18
2.3 General feature of milling cutter in use 21
2.4 Enzymic peeled (right side) and manual (left side) peeled grapefruit 29
3.1 The instrumentations of testing mechanical properties of vegetables 40
3.2 Effects of force-deformation test (a-d) and relationship between
force (N) and deformation (mm) for melon and pumpkin in two cases
of skin and unpeeled (e-f) 45
3.3 Rupture force of skin and unpeeled states for different varieties
of pumpkin and melon 46
3.4 Toughness of skin and unpeeled states for different varieties of
pumpkin and melon 47
3.5 Cutting force of skin, flesh and unpeeled states for different
varieties of pumpkin and melon 48
3.6 The maximum shear strength force of skin, flesh and unpeeled
states for different varieties of pumpkin and melon 49
3.7 The shear strength of skin, flesh and unpeeled states for different
varieties of pumpkin and melon 50
3.8 The relative contribution (%) of skin to different mechanical
xiii properties of Pumpkin and Melon 52
4.1 Test rig 59
4.2 Product holder and two available positions 60
4.3 The two D.C. sources for vegetable holder and peeler head 60
4.4 Product holder 61
4.5 Details of the peeler head 62
4.6 Peeler head 62
4.7 Flap with holes in spiral pattern 63
4.8 The auxiliary peeler head as attachment 64
5.1 Rotary wire brush and its peeling effect on pumpkin 67
5.2 Twisted wire brush before and after loosening strands 68
5.3 Affected areas of pumpkin after using twisted wire brush 69
5.4 Improved design of artificial twisted brush in second stage
and its effects 70
5.5 The twisted wire brush in third stage and its peeling effect 71
5.6 Ball chain and its peeling effect after application 71
5.7 Different investigated lathe tools (a) and cutter in cylindrical
shape with triangular side section (b) 72
5.8 The effect of peeling by wedgy side cutter 73
5.9 Abrasive rope and its peeling effect on pumpkin 74
5.10 Different shapes of abrasive foam and their peeling effect
on pumpkin 76
5.11 Rope covered by spiral blade (a), and its peeling effect on pumpkin (b) 77
5.12 Sandpaper belt installed on test rig with and without pumpkin 77
5.13 Grater plate peeling unit and its effect on pumpkin 78
xiv 5.14 Abrasive cutter brush and its effects on two stages 80
5.15 Abrasive bristle products 81
6.1 Disk shape milling cutter with triangular contour 88
6.2 The contribution of independent variables to responses resulted
from using milling cutter 89
6.3 The effects of independent variables on responses resulted from
using milling cutter 91
6.4 Abrasive peeler pads and accessories 93
6.5 The contribution of independent variables to responses resulted
from using abrasive-pads 95
6.6 The effects of independent variables on responses resulted from
using abrasive pads 96
6.7 Abrasive foam and accessories 99
6.8 The contribution of independent variables to responses resulted
from using abrasive foams 102
6.9 The effects of independent variables on responses resulted from
using abrasive foams 103
6.10 Abrasive-cutter brush 107
6.11 The contribution of independent variables to responses resulted
from using abrasive-cutter brush 109
6.12 The effects of independent variables on responses resulted
from using abrasive-cutter brush 110
7.1 The different type of stripes of coarseness used for fabrication
of abrasive-cutter brush 118
7.2 Different parts of product as levels of location variable 118
xv 7.3 The effect of mean p. speed on LnP.rate 124
7.4 The effect of p. speed on LnP.rate at different levels of coarseness 124
7.5 The effect of p. speed on LnP.rate at different in different
locations of pumpkin 125
7.6 The effect of mean coarseness on LnP.rate 127
7.7 The effect of coarseness on LnP.rate at different speed of
abrasive-cutter brush 128
7.8 The effect of coarseness on LnP.rate at different location
of product 129
7.9 The effect of mean location on LnP.rate 130
7.10 The effect of location on LnP.rate in different coarseness 131
7.11 The effect of location on LnP.rate at different p. speeds 132
8.1 The view of abrasive-cutter brush after penetration to the peel 137
8.2 The cross-sectional view of one protrusion
(two out of four teeth are shown) 138
8.3 Experimental versus predicted values of p. rate (gr/min) 149
xvi List of Tables
3.1 Static coefficient of friction of three varieties of pumpkin in the flesh,
unpeeled, and without periderm state on three different materials
including stainless steel, Teflon, and wood………………………………....51
6.1 Taguchi experimental design for independent variables and levels………....88
6.2 Taguchi experimental design for independent variables and levels…………94
6.3 Taguchi experimental design for independent variables and levels………...101
6.4 Taguchi experimental design for independent variables and levels………...108
7.1 The results of frequencies analysis on peeling rate…………………………120
7.2 The results of frequencies analysis on LnP.rate…………………………….121
7.3 The results of Levene’s test for homogeneity of variance of LnP.rate……..121
7.4 ANOVA of the mean of LnP. rate among different levels of
independent variables ………………………………………………………122
8.1 The results of multiple regression analysis for coefficients of two
mechanical peeling models……………………………………………...... 147
xvii List of Symbols and Abbreviations
Pt total expenditure power, N. mm/min P1 expenditure power at fracture part of cutting, N. mm/min P2 expenditure power at forming part of cutting, N. mm/min ηc total peeling efficiency n the number of installed brushes on peeler head ωp the angular velocity of abrasive-cutter brush, rpm Ep penetration energy of abrasive-cutter brush, N. mm Ed the deflection energy of abrasive-cutter brush, N. mm K1 average shearing resistance per unit length of stroke, N/mm Vip linear penetration velocity of brush’s teeth inside peel, mm/s t1 time of stroke, s δ1 deflection of product, mm δ2 the depth of average penetration, mm γ the ratio of toughness of product (Tp) to toughness of tool (Tt) Tp the toughness of product, N. mm Tt the toughness of abrasive-cutter brush, N. mm α the density of protrusions on a brush, number/mm2 l1 the effective length (covered by abrasive strip) of brush, mm d1 the diameter of brush, mm τ the shear strength of product, N/mm2 d2 the diameter of protrusion’s hole, mm l2 the length of each tooth on protrusion, mm θ1 the angle of teeth in protrusion, degree E the modulus of elasticity of the brush, N.mm-2 I the geometrical moment of inertia of the brush, mm4 δ3 the average deflection of the brush at fracture stage, mm L the whole length of brush, mm δ3max the maximum deflection of brush in fracture stage, mm Vop the linear velocity of brush’s teeth in scratching stage, mm/s Fc total cutting force, N Ff friction force, N Fd disintegration force on the structure of product, N Fe the spent force for elastic and plastic deformation, N Ef the expended friction energy, N. mm h the length of removed peel, mm K2 the friction coefficient α the density of protrusion, number/mm2 φ the degree of unevenness of product’s surface μd the dynamic coefficient of friction between the brush’s tooth and product Rv the total normal reaction, N Fde deflection force of brush, N N the normal reaction force to the weight of brush, N W1 the weight of one brush, g
xviii θ2 the angle between direction of the weight and direction of the line passes through the gravity centre of brush and is perpendicular to the surface of product in contact point, degree δ4 the average deflection of brush in second stage of cutting, mm l3 the total projected lengths of protrusion’s teeth engaged in cutting, mm K3 the coefficient of elastic and plastic force E2 the total required energy of peeling in second stage, N. mm K4 the coefficient of disintegration force K5 scratching coefficient in second stage, number/min ωv angular velocity of vegetable holder, rpm β the number of scratches, number/min P. rate peeling rate, g/min LnP.rate the logarithmic transform of P. rate, g/min K6 transform coefficient of Pt to p. losses, g/N. mm v. speed the angular velocity of vegetable holder, rpm p. speed the angular velocity of peeler head, rpm peeling losses the substantial amount of usable vegetable flesh that is being discarded because of peeling, % of weight of whole produce before peeling peel losses the ratio of the weight of removed peel to the weight of whole produce before peeling divided by time of peeling, %/min peeling efficiency the percentage peel that is removed from the initial skin per unit time, %/min peeling rate the weight of removed peel divided by peeling time, g/min
xix Authorship
The work contained in this thesis has not been previously submitted for a degree or diploma at this or any other educational institute. To the best of my knowledge and belief, the thesis contains no material previously published or written by another person except where due reference is made.
Signature: …………………… Date: …………………....
xx
Acknowledgement
The foremost gratitude to my God, Allah, who offered me this opportunity to learn and progress. I appreciate Him for delighting me with kind parents who supported and encouraged me. I am thankful to Him for blessing me with my wife, Masoumeh, who showed patience and support during this research.
I would like to express my appreciation to my principal supervisor, Dr. Vladis Kosse, for his time and valuable suggestions in supervising this thesis. I also thank my associate supervisor, Prof. Prasad KDV Yarlagadda, for his support and suggestions.
The author wishes to thank Dr. Mahalinga Iyer, Dr. Prasad Gudimetla, and Dr. Kunle Oloyede, School of Engineering Systems, QUT, for reading thesis and their valuable comments.
I wish to express my gratitude to the government of Islamic Republic of Iran for financial support through a PhD scholarship award.
I also wish to thank all former and present members of School of Engineering Systems. Special thanks to Mr. Terry Beach, Mr. Mark Hayne, and Mr. Abdul Sharif for their technical support.
Finally, I would like to thank my fellow students and colleagues for their company and interesting discussions over the past three years.
xxi Chapter 1
Introduction
1.1 Significance and motivation of the research
The food and beverage sector was the largest sector of Australia’s manufacturing industry in 2002-03, providing about 20% of total sales and services income. The total income from sales and services for the Australian food processing industry was estimated at $65.9 billion in the same year. The industry value added by the food processing industry was recorded as $16.6 billion in 2002-03 (Department of Agriculture, Fisheries and Forestry, NSW).
Peeling is an important preliminary stage of fruit and vegetable processing. The quality and the final price of the processed product is highly dependant on this stage. Manual peeling is possible for any kind of product but high losses and considerable consumption of time and labour have encouraged the peeling industry to use other methods. Mechanical, thermal, and chemical peelings are conventional methods (Luh, and Woodroof., 1988), each of which has its own benefits and limitations. Those methods apply mechanical tools, heat or cold, and lye respectively to peel off the fruit and vegetable skins. As none of the current methods can satisfy all requirements of producers and consumers, other kinds of peeling methods such as enzymatic peeling have been developed. Changing the peeling technique has changed the kind of merits and demerits but the problem is still unsolved. Getting closer to the “ideal” peeling method is the aim of every researcher in this field. The “ideal” peeling method possesses the following features (Radhakrishnaiah settee et al., 1993): -minimizes product losses -peels to the extent dictated by the products
1 -minimizes energy and chemical usage -minimizes the pollution loads -minimizes heat ring formation
Among the current peeling methods, mechanical methods can attract the satisfaction of consumers as these methods possess some important benefits of the “ideal” method such as the freshness of the peeled product. As the view of consumers is important to the food processing industry researchers are encouraged to continue the search for mechanical peeling methods that are closer to the “ideal” peeling, in spite of the high losses so far experienced with these methods.
Approaching the “ideal” peeling method through trial and error is difficult and time consuming. The design of new methods and improvement of current peeling methods to achieve the “ideal” peeling conditions using physical and mechanical properties of the product is one of the objectives of researchers (Ohwovoriole et al., 1988). Depending on the proposed technique, different properties of products have been applied to improve the efficiency of purposed peeling methods or devices. However, despite all attempts, some fruits and vegetables, such as mangoes, are commonly peeled manually (Radhakrishnaiah settee et al., 1993) and methods for others such as pumpkin are far from the “ideal” peeling conditions.
Tough-skinned vegetables such as pumpkin and melon currently are peeled either semi-automatically (i.e. Comet Food Ltd.) or automatically (i.e. Dornow Ltd.). Comet Food Ltd. is a Brisbane-based organization visited by this researcher which utilises semi-automatic peeling methods of pumpkin and melon. Circular shapes of rotating graters are applied in the semi-automatic method. Segments of the product are brought into contact with the grater by an operator. This process is tedious and time consuming. In the latter method, whole pumpkins are passed through automatic machines where the floor is covered by many rotator disks (carborundum or blade). The main limitation of both methods especially for varieties with an uneven surface is high peeling losses. For pumpkins as a case study, the minimum peeling losses in an optimistic situation is about 35% (Department of State and Regional Development, NSW). As Figure1.1.a shows, penetration to the inside of
2 concave areas to peel off through grooves is accompanied by high removal of flesh from convex areas in the current peeling methods.
Concave area Layer to be Convex area removed
a. Current peeling methods
Concave area Layer to be Convex area removed
b. The “ideal” peeling method
Fig.1.1.The top view of pumpkin (a) Current peeling methods (b) “ideal” peeling method
The investigation of concepts for new mechanical peeling methods and development of design recommendations for peeling of tough-skinned vegetables is a challenging task in this research. The main objective is to achieve even peeling
3 efficiency from different areas of uneven surfaced products with minimum peeling losses (Figure1.1.b).
1.2 Objectives of the research
This research develops new mechanical methods and tools for peeling of tough- skinned vegetables and recommends important design parameters for a peeler on the basis of the mechanical properties of those products. The main objectives of the research are as follows: • Measurement of the mechanical properties of tough-skinned vegetables, with pumpkin and melon as the case studies (three varieties each) (addressed in Chapter 3). • Development of new innovative mechanical peeling methods and adaptation of existing methods for tough-skinned vegetables (addressed in Chapter 5). • Selection of the best peeling method close to the “ideal” peeling method (addressed in Chapter 6). • Simulation and mathematical modelling of the mechanical peeling process (addressed in Chapter 8). • Development of recommendations for design parameters of a mechanical peeler for tough-skinned vegetables (addressed in Chapters 3 and 8).
1.3 Originality and major contribution of the thesis
The originality and major contributions of the thesis are summarized in the following sections.
1.3.1 Definition of tough-skinned vegetables
Despite common use of this term, the review of literature shows that there is no clear definition for the term “tough-skinned” vegetable. As a result of this thesis, for the first time, the tough-skinned vegetables could be scientifically defined. This definition will be the first effort for classification of fruits and vegetables on the basis of their mechanical properties.
4 1.3.2 Determination of mechanical properties of tough-skinned vegetables
For the first time, some mechanical properties of pumpkin and melon - as two kinds of tough-skinned vegetables - will be determined. The three investigated varieties of pumpkin are Jarrahdale, Jap and Butternut and the three investigated varieties of melon are Rockmelon, Watermelon, and Honeydew. Rupture force, cutting force, shear strength force, and shear strength will be measured. All properties will be investigated in three states - skin, flesh, and unpeeled product - except rupture force and toughness that will be studied in two states - skin and unpeeled product. The static coefficient of friction also will be measured for three states of product - unpeeled, without periderm and flesh. This property will be determined for three different materials - stainless steel, Teflon and wood.
1.3.3 Developing new mechanical peeling methods for tough- skinned vegetables
Inflexibility of current mechanical peeling methods that causing uneven peeling and high peeling losses is one of the main current problems for the peeling industry especially for tough-skinned vegetables. This thesis develops new mechanical peeling devices suitable for tough-skinned vegetables and close to the conditions of the “ideal” peeling method. The highlighted feature of these designed devices is that they do not share the limitations of current peeling tools.
1.3.4 Developing current peeling methods
Peeling by the use of a milling cutter has been successfully introduced by some researchers (Cailliot and Serge, 1988; Gardiner et al., 1963; Boyce, Jose and Calif, 1961). Clogging as the main limitation prevented the industrial application of the milling cutter as a peeling device. A new shape of milling cutter will be developed by this study which can be used without the limitation of previous works. The criteria of this research (evenness of peeling in different areas of product and higher peeling efficiency) also will be considered in the developed method.
5 1.3.5 Introducing the best mechanical peeling method applicable in the peeling industry for tough-skinned vegetables
Developing and introducing an innovative mechanical peeling method for tough- skinned vegetables is the main contribution of this study. The development of method leads to the provision of necessary and industrially applicable knowledge for the design and manufacture of the proposed peeler machine. The design knowledge includes information about the peeling tool and product.
1.3.6 Mathematical modeling of mechanical peeling
No mathematical model for mechanical peeling has been known so far. For the first time a mathematical model of mechanical peeling will be developed in this thesis. The model simulates the cutting process by abrasive-cutter brush (developed in this research). The applicability of the model for the peeling industry will be emphasized. The variables of input and output of the model will be chosen from important effective parameters related to the product and peeling tool. Those variables should be easily measurable and empirically adjustable.
1.3.7 Determination of design parameters of tough-skinned vegetable peeler
Some necessary design parameters of tough-skinned vegetable peelers are recommended by the obtained results of this study. These recommendations can be used by the peeling industry to improve the performance of peeler equipment. The research attempts to explain the role of these determined parameters and those which require determination in the mechanical peeling processing.
6 1.4 Thesis organization
The thesis consists of three main parts: investigation of mechanical properties of tough-skinned vegetables, development of peeling methods and mathematical modeling of the results. The content is organized in nine chapters as follows:
Chapter 1 is the introductory chapter that explains the motivation of the research, and outlines the main objectives and major contributions of the thesis.
Chapter 2 reviews critically the existing peeling methods and background of the research. This chapter also reviews work done on properties (especially mechanical) of fruits and vegetables and the developed models in the area of fruit and vegetable peeling.
Chapter 3 focuses on the mechanical properties of tough-skinned vegetables including pumpkin and melon (three varieties each). It explains developed and fabricated instrumentations, measuring methods, results and statistical comparisons of the results. The mechanical properties, including rupture force, toughness, cutting force, shear strength force, shear strength, and static coefficient of friction are investigated in different specimen states.
Chapter 4 describes the test rig. Design and fabrication procedure are described, specifications and advantages of the test rig that was used to investigate different peeling methods and devices are also discussed.
Chapter 5 outlines all the attempts undertaken to explore suitable peeling methods and devices. Although the devices might look simple, they were preliminary prototypes that were made with regard to the specifications of the products and the main defined objectives of the research.
Chapter 6 presents the experiments carried out for investigation of four different peeling methods and devices for the Jap variety of pumpkin. The four different investigated methods were abrasive pads, abrasive foams, milling cutter, and
7 abrasive-cutter brush. Results and analysis of the results are comprehensively explained in this chapter.
Chapter 7 reports the results and analysis of full factorial experiments conducted on two varieties of pumpkin (Jap and Jarrahdale) by abrasive-cutter brush as the best selected method. The influencing parameters that are related to the product and abrasive-cutter brush have been comprehensively studied.
Chapter 8 describes mathematical modelling of mechanical peeling. The procedure of mechanical peeling method using abrasive-cutter brush is simulated and the influence of different parameters related to the product and peeler are modeled. The value of the coefficients of the model is determined for Jap and Jarrahdale varieties of pumpkin. The validation of the model is carried out using experimental data and the results are discussed using the results of the study of mechanical properties of tough-skinned vegetables.
Chapter 9 summarizes and concludes the thesis. It also proposes directions for future studies related to this area.
1.5 Publications (refereed) of the author arising from the PhD research
1) Emadi, B., Kosse, V., Yarlagadda, P. K. D. V., Mechanical properties of pumpkin, International Journal of Food Properties, 8 (2), 277-287.
2) Emadi, B., Kosse, V., Yarlagadda, P. K. D. V., Design and manufacturing of test rig for investigation of improved mechanical peeling methods of fruits and vegetables, International conference on manufacturing and management: GCMM- 2004, pp.574-579, Vellore, India, Dec. 2004.
3) Emadi, B., Kosse, V., Yarlagadda, P. K. D. V., Mechanical properties of three varieties of melon, Journal of Texture Studies, Submitted, 20/11/2005.
8 4) Emadi, B., Kosse, V., Yarlagadda, P. K. D. V., Experimental investigation of abrasive peeling of pumpkin, 4th International congress on food technology, pp.118- 117, Athens, Greece,Feb.2005.
5) Emadi, B., Kosse, V., Yarlagadda, P. K. D. V., Relationship between mechanical properties of pumpkin and skin thickness, 4th International congress on food technology, pp.111-123, Athens, Greece, Feb. 2005.
6) Emadi, B., Kosse, V., Yarlagadda, P. K. D. V., Abrasive peeling of pumpkin, part 1: using abrasive pads. Journal of Food Engineering, Submitted, 23/02/2005.
7) Emadi, B., Kosse, V., Yarlagadda, P. K. D. V., Abrasive peeling of pumpkin, part 2: using abrasive foams. Journal of Food Engineering, Submitted, 22/02/2005.
8) Emadi, B., Kosse, V., Yarlagadda, P. K. D. V., Using abrasive disk as innovative peeling method for vegetables with uneven surfaces, BEE Postgraduate Research Conference, Brisbane, Australia, Dec. 2005.
9 Chapter 2
Literature review
Peeling of fruits and vegetables in theory and practice has been studied by many researchers. These studies are very broad and cover various methods of peeling and products to be peeled. Some products (such as potato) and methods of peeling (such as chemical methods) are particular matters of interest and have attracted more attention. The study of those products and methods has been extended to investigate the physico-chemical behaviour of the product during the peeling process. There is very little literature pertaining to the peeling of tough-skinned vegetables while there has been little effort to quantify the tissue properties or any process modelling of products such as pumpkin and melon.
From a mechanical peeling standpoint, there is evidence that more research is needed to develop modelling approaches for many peeling operations, for example the peeling of tough-skinned vegetables such as pumpkin.
This chapter sets out to identify and critically analyse all the previously published literature with regard to the mechanical properties of fruits and vegetables, peeling methods, and modelling of the peeling process.
10 2.1 Mechanical properties of fruits and vegetables and methods of testing
2.1.1 Introduction
The study of the physical and mechanical properties of fruits and vegetables is very important. It can improve the efficiency of processing equipment, especially peelers. The agricultural products are subjected to either wanted or unwanted mechanical loads after harvesting. During the process of mechanical peeling, the products are loaded purposefully with wanted mechanical loads that are always accompanied by unwanted loads. The unwanted mechanical loading (compression, impact, and vibration) is the main reason for bruising of fruits and vegetables during post harvesting operations (Brusewitz et al., 1991). Reducing the harmful effects of unwanted loads and improving the effectiveness of wanted loads can be achieved by knowledge of the mechanical properties (i.e. toughness, cutting force and shear strength) of the products. Mechanical properties of products can also be used in design of their mechanical peeler. Researchers have used various techniques to investigate the mechanical properties of different produce. The following sections address these issues.
2.1.2 States of product to be tested
The properties of products can be studied for different states including skin, flesh and unpeeled (overall) state. Among different states of product to be tested, determination of the mechanical properties of skin always poses a challenge. It can be done by carrying out experiments on skin directly or indirectly. Several researchers such as Grotte et al. (2001), Jackman and Stanley (1994), and Voisey et al. (1970) have used the difference between unpeeled (overall) product and that of product measured without skin (flesh) to indirectly obtain the result of the force- deformation test of the skin. This experimental procedure has not been accepted by others due to the likelihood of some errors in the result. For example, Thompson et al. (1992) stated that to identify the contribution of the skin to the external puncture force (in the case of cucumbers) by making measurements before and after skin
11 removal is accompanied by error. In addition, Jackman and Stanley (1992) concluded that the difference of puncture load displacement between unpeeled (overall) and without skin (flesh) product cannot provide the load displacement of the skin itself. Jackman and Stanley’s point was proved by an increase in effective area of compression during puncture of the skin.
2.1.3 Compression test
The compression (force-deformation) test is the basic and one of the most important tests in the study of the mechanical properties of fruits and vegetables. The force- deformation test shows the behaviour of the product under different levels of compression forces. Some mechanical properties of different states of products including skin, flesh, and unpeeled states can be determined by using this test (Mohsenin, 1970). The test can be used for determination of an extended range of mechanical properties such as modulus of elasticity, rupture force, rupture stress, toughness, and firmness (Jackman and Stanley, 1994; Voisey and Lyall, 1965a; Voisey and Lyall, 1965b; Grotte et al., 2001; Holt 1970; Voisey et al., 1970; Behnasawy et al., 2004; Rybczynski and Dobrzanski, 1994). The same purposes can be achieved by using the tensile test (Jackman and Stanley, 1994) but, in comparison with the compression test, implementation of the tensile test is difficult because of some limitations, such as difficulty in holding the skin specimens during the test (Su and Humphries, 1972) and creation of premature tensile failure during specimen preparation (Clevenger and Hamann, 1968; Thompson et al., 1992).
Two important properties resulting from the force-deformation test are rupture point and toughness (Figure 2.1). Rupture point is a point on the force-deformation curve at which the axially loaded specimen ruptures under a load (Mohsenin, 1970). The work required to cause rupture in the product is known as the toughness (Mohsenin, 1970). Finney (1969) also defined the toughness as the area under the force- deformation curve, recorded through the point of tissue rupture or failure (Figure 2.1). He explained the intercellular adhesion or cementing substances and cell wall strength are factors quite likely to influence toughness of fruits and vegetables.
12 2.1.4 Cutting test
The cutting test is not as common as the compression test and is used to determine the resistance of tissue to loading cutting force. Ohwovoriole et al. (1988) applied this test to identify the necessary cutting force of unpeeled and peeled cassava tuber. They used the resulting data to design a cassava peeler. A cutting edge in the form of a sharpened 1.5 mm thick piece of sheet metal was placed between the plungers of the Universal Testing Machine for that purpose.
Fig.2.1. Force-deformation curve (Finney, 1969)
2.1.5 Shear strength test
Shearing strength of the product is determined by shearing a plug from a slice of the product. The shearing strength of the product indicates the degree to which the cells are held together. Knowing shearing force (F), the diameter of the solid cylindrical die with flat end (d), and the thickness of the slice (t); shearing strength (S) can be determined (Mohsenin, 1961):
F S = (2.1) π × d × t
Ohwovoriole et al. (1988) reported using the shear strength test to measure shear stress of peeled and unpeeled cassava tuber. An 8.7 mm diameter rod was used on
13 an Instron machine to reveal the necessary data for cutting cassava tubers through the peel. Their report does not mention the shape of the applied indentor.
2.1.6 Coefficient friction test
The coefficient friction test is applied to identify the coefficient of friction for different states of product on various surface materials. The effective parameters on these properties are the moisture content of the specimen and the kind of surface material. Chung and Verma (1989) concluded that the surface material is more effective on the dynamic than on the static coefficients of friction while Carman (1996) reported a higher influence of moisture compared with surface material on the static coefficient of friction. Chung and Verma (1989) concluded the ratio of static and dynamic coefficient of friction remained almost invariant irrespective of the moisture content of the sample and the type of surface material. The coefficient of friction, either static or dynamic, has been measured using different methods. Those coefficients can be measured by analog or digital systems. Data obtained by a digital system are more accurate than those taken by an earlier analog system (Chung and Verma, 1989). In the analog system, specimens are placed on a table which is manually and slowly tilted until movement of the specimen and the static coefficient of friction would be the tangent of the slope angle of the table measured with a protractor. This method has been used by some researchers (Oje and Ugbor, 1991; Bahnasawy et al., 2004; Helmy, 1995; Saif and Bahnasaway, 2002; Ohwovoriole, 1988). The main benefit of using analog systems is simplicity of the method.
The digital measuring system works based on a friction device (disk) modified by Tsang-Mui-Chung et al. (1984) and improved by Chung and Verma (1989). The latter authors used a personal computer for data acquisition. They applied the following equations to calculate the static and dynamic coefficient of frictions:
μs = Ta /Wf .q (2-2)
μd = Tm /W f .q (2-3)
14 where, μs is static coefficient of friction, Ta is beginning value of torque, μd is the dynamic coefficient of friction, Tm is the average value of the torque, q is the length of torque arm, and Wf is the weight of fruits to calculate the dynamic and static coefficients of friction. The average value of the torque during the rotation of the disk and the maximum value of torque obtained as the disc started to rotate were used.
This method has been used by other researchers such as Marakogˇlu et al. (2005), Çalı ır et al. (2005) and Gupta and Das (1998). They applied this method to measure the friction coefficients of fresh blackthorn fruits, wild plum fruits, and sunflower seed and kernel respectively.
2.2 Peeling methods of fruits and vegetables
2.2.1 Introduction
For some kinds of fruits and vegetables, such as mango, manual peeling is commonly in use. The requirement to develop new methods and tools for peeling that can be mechanised and automated has led to the versatile current peeling methods, machinery and equipments. Peeling methods fall into three main groups: mechanical, thermal and chemical peeling. Much research has been published related to different methods of peeling and the range of publication is considerably extensive regarding the variety of products and peeling methods. A literature review, arranged on the basis of the technique used, along with examples of the latest works of interest is given here.
2.2.2 Mechanical peeling
There is a variety of mechanical peelers designed to suit the peeling of either a particular product or a group of products. In general, mechanical peelers are classified on the basis of the type of mechanism that is incorporated into the peeling system. Commercial mechanical peelers include abrasive devices, devices with
15 drums, rollers, knifes or blades, and milling cutters. Generally, the quantity of losses in this kind of peeling is high, but the quality of the final peeled vegetables in terms of features such as freshness is good. Those devices are briefly described using related works of interest as follows:
2.2.2.1 Abrasive devices
The abrasive method can be implemented in a very simple way by using gloves with an abrasive outside layer. Vegetables are rubbed by these gloves to remove the skin. That is a common method for the peeling of potatoes in small amounts. Somsen et al. (2004) have proved that manually peeling of potatoes using sandpaper results in the lowest possible peel losses. They proved that these losses were normally expected losses (wanted losses). Although some researchers realised that the abrasive method has a lower quality compared to hand peeling (Barry-Ryan, 2000), it is still used commonly for root vegetables. These researchers criticized this method as it bruises the underlying tissue to varying degrees and leaves the new outer layer of cells damaged which leads to leakage of cellular fluids and encourages microbial growth.
Jasper et al. (2001) patented a peeler equipped with a rough exterior surface. The peeler abrades the outside surface of fruits and vegetables when it comes into contact with the outside surface of the product. The surface roughness of the peeler can be adjusted depending on the skin of the vegetable to be peeled. One of the clear disadvantages of this device is that it can only be used in a domestic kitchen environment.
Agrawal et al. (1982) discuss the development of an abrasive brush type ginger peeling machine. Two continuous and abrasive vertical brush belts are the main parts of the machine. Ginger, as a product with an irregular shape, passes between these two belts while it moves in the opposite direction with a downward relative velocity. The opposite direction of movement of the two belts causes an abrasive action while the downward relative velocities provide the downward movement of the product. Agrawal et al. have reported a peeling efficiency of about 75-85%.
16 2.2.2.2 Devices using drums
Singh (1995) describes a power-operated batch type potato peeler that includes a peeling drum (670 mm in length and 450 mm in diameter) and a water-spraying unit. The peeling drum with protrusions (2.5-3.0 mm) on the inside surface rotates and removes skin from potatoes by means of abrasion. The space between two centres of protrusions in rows and columns were 14 and 7 mm, respectively. A speed of 30 rev/min batch, load of 20 kg and time of 8 min were found to be the best combination providing higher peeling efficiency and lower peel losses. As a result of this combination, the peeling efficiency and peel losses were 78% and 6%, respectively. A large amount of wastewater is a shortcoming of this method.
2.2.2.3 Devices using rollers
Suter (2002) developed a peeling machine for efficient and effective control of the peeling operation. It applies a set of abrasive rollers (Figure 2.2). The rollers come together in a longitudinal direction and the distance between them is adjustable. The feeder feeds the rollers controllably on the basis of the sensed load inside the rollers by a related sensor. Also, Zittel (1991) used a plurality of rotating abrasive rollers. The machine had a frame with a pair of end plates that rotatively carried the longitudinal rollers. Each roller is powered by an individual motor, which is coupled to that roller. However, those inventors neither achieved higher efficiency nor could reduce peeling losses because their patents focused mainly on control and drive mechanisms.
2.2.2.4 Knives or blades
Tardif and He (1999) released a machine equipped with blades to peel vegetables. The vegetable, which is located in the hollow base of the machine, can be rotated by a threaded rod on the top. The rod is rotated manually by a handle. A blade, which is coupled to the supporting rod and urged by a spring, moves towards the vegetable to be peeled. While the vegetable is rotated, the blade removes the peel.
17 Harding (2001) described an apparatus for peeling a convex surface of a section of a fruit or vegetable. The machine is attached to a U-shaped peeling blade and a feeder. The feeder grips the fruit or vegetable at a position about opposite the apex of the peeling blade. Fruits or vegetables have to pass in front of a peeling blade being guided by at least one guide. He also patented a melon peeler in 2000. The semi- manual peeler includes a curved peeling blade which presents a curved surface to the convex surface of the section of produce. It facilitates peeling of curved sections of products but can not follow the unevennesses of surface.
Fig.2.2. An industrial application of an abrasive roller peeler for tuberal products such as potato (Dornow Food Technology GmbH, 2004)
Gingras (2001) presented equipment for the peeling of vegetables of a round, oval or elongated shape such as cucumbers. The machine is equipped with a frame including an adjustable hole to receive and let pass the vegetable to be peeled. The frame also carries several knives that can be slid toward the corner of a hole.
Ridler (2000) described a peeling apparatus including a traversed blade, which continuously and intermittently rotates in the opposite direction to a rotating vegetable. The apparatus is controlled and powered manually. The user rotates the
18 vegetable that is mounted on a slender arbour by one hand and controls the peeling blade by another hand at the same time.
Martin (2000) reported a peeling machine equipped with a lower and upper holding assembly connected to a frame securing and rotating the vegetable to be peeled. A carriage assembly including a cutting assembly is engaged with the end of a second air cylinder. The extension of the second air cylinder pushes the cutting assembly against the vegetable as the carriage assembly moves upwards; as a result peeling is done.
Protte (1999) described a peeling machine for stalk-like vegetables, including a plurality of knife stations that are successively arranged along the vegetables moving inside the machine. There are pluralities of pairs of feed rollers, and every pair is supported between successive knife stations.
Sommer (1997) discussed a device for the peeling of elongated vegetables especially asparagus. The device comprises housing equipped with a passage designed to permit a stick of asparagus to be inserted. There are several peeling blades inside the housing, which are orientated in various directions of the passage and act on the stick of asparagus. One blade as a minimum can move crosswise to the elongated direction of the passage and presses flexibly towards the stick of asparagus.
Rauschning (2001) expounded a device for the peeling of root vegetables. The device includes a container equipped with at least two rotating discs at its bottom. These discs have a grating or cutting surface on their upper side.
Odigboh (1976) revealed the development, design and construction of a continuous process mechanical cassava peeler. The machine includes two cylinders which are located parallel with 20 mm space and inclined at 15° to the horizontal plane. The surface of the driver cylinder is covered by knives and rotates clockwise at 200 rev/min. The driven cylinder which has a roughened surface, also rotates clockwise at 88 rev/min. The cassava pieces with 100 mm length are fed lengthwise to the spaces between cylinders. Products are being peeled off while they rotate anti-
19 clockwise and move down. Low efficiency of peeling especially for small sizes of roots (40 mm and less) and inability to be set up for roots of specific sizes are reported as disadvantages. Although the continuous option of the apparatus is considered to be one advantage, the necessity to cut roots to pieces of about 100 mm length is a limitation.
Srivastava et al. (1997) reported the design and development of an onion-peeling machine which uses four scoring blades assisted by compressed air jets to slit the outer layers of the onion skin. The skin will be loosened and dislodged from the bulb by the compressed air under the peel. The compressed air penetrates under the skin through the created slits by scoring blades. A pair of high-speed saw blades is used to cut the ends of the onion. Srivastava et al. (1997) reported peeling losses as 17%.
2.2.2.5 Milling cutter
Cailliot and Serge (1988) claimed that peeling using the milling cutter is one of the known methods for spherical shape products (Figure 2.3). In this method, one or more fixed or rotary peeling tools (i.e. knife) with at least one cutting edge take the peel off the product in a similar way to manual peeling. In the first stage of this method, a fixed knife or blade was used to peel a rotary spherical product. As the knife had no flexibility, it could not follow the irregular shape of the product exactly and in particular it could not penetrate to the inside of thin grooves. The history of using a milling cutter goes back to Boyce et al. (1961) who used a milling cutter in the form of a very flat milling cutter, having a large number of cutting teeth distributed over a considerably large diameter, in order to produce small chips of peel, the discharge of which is left to chance. The big diameter of the cutter and the shape of the teeth (like spoons) were two reasons that did not allow the cutter to properly follow the shape of the product.
To remedy the production of a continuous peel and resulting clogging, Gardiner et al. (1963) tested a milling cutter with a cylindrical cutting edge, combined with a disc which supports the cylindrical cutting edge and which was provided with apertures sharpened in the plane of the disc, so as to cut the ribbon of peel
20 transversely into smaller portions making it easier to discharge it. Although the problem of clogging was solved, the peeling production was not sufficient. Similar limitations were experienced by Polk (1972) who used a large rotary milled edge and rotary vegetable holder.
Couture and Allard (1979) invented a cutting head comprising a blade strip bent longitudinally into a generally cylindrical shape. The peeler head was pivotally connected to the body to follow the irregular shape of the product. The machine included means for moving the cutting head along the supported vegetable in contact therewith as the vegetable is rotated with the cutting edge and a continuous strip of peel is cut around the vegetable.
Fig.2.3. General feature of milling cutter in use (Boyce, San Jose and Calif, 1961)
As the cutting head had no rotation itself and had to follow the shape of the rotating vegetable, the chance of it getting stuck especially for sharp irregular shapes was high. Cailliot and Serge (1988) noted the disadvantages of this appliance such as
21 producing a continuos peel and clogging the peeling tool in an automatic appliance. Cailliot and Serge (1988) claimed one vertical cutter type in their patent. The diameter of the rotary cutter is small and equipped with two teeth to give balance. 20000 rpm for the rotary cutter gives a speed that is equal to tangential speed of at least 20 m/s and needs low torque. The depth of penetration is limited by the choice of a small diameter cutter for example 20 mm. The extra length of each tooth from outside the diameter of the cutter defines the depth of penetration in this plan. Cailliot et al. claimed to solve the common problem (clogging) but it seems the low number of teeth and the shape of the teeth that come out from the surface of the plate cause clogging for irregular shapes of product such as pumpkin with irregular, thin and deep grooves. Tardif and He (1999) in another trial used a similar method to Couture and Allard (1979) with a simple knife (non-rotating). It was found that low flexibility will definitely lead to clogging as well. All attempts which have been carried out so far were unsuccessful in solving the clogging problem especially for products with an uneven surface. It is believed that the high number of teeth in special shapes without any convex section may reduce the chance of clogging to zero.
2.2.3 Thermal peeling
Thermal peeling as well as chemical peeling is used for thick-skinned vegetables. This method can be performed by wet heat (steam) or dry heat (flame, infra red, hot gases). Floros and Chinnan (1988a) reported that the widespread application of steam peeling is due to its high level of automation, precise control of time, temperature and pressure by electronic devices to minimize peeling losses, and due to the reduced environmental pollution as compared to chemical peeling. This method of peeling - especially dry heat - causes a cauterizing of the surface, wound areas, and small pieces of charred skin, which if not removed, give a poor appearance to vegetables, especially canned ones (Weaver et al., 1980). Different types of thermal peeling are described below with reference to related works of interest.
22 2.2.3.1 Flame (dry heat) peeling
Some vegetables such as pepper can be peeled by dry heat (flame). In this method, vegetables are exposed to direct flame (for about 1 min at 1000ºC) or hot gases in rotary tube flame peelers. Heat causes steam to develop under skins and this puffs the skins up so that they can be washed away with water. Each heat treatment should be immediately followed by cooling in water.
Weaver et al. (1980) report a flame application for the peeling of tomatoes. Dual Maxon gas-fired burners are mounted at the top of a live-roller conveyor at the height of 6 in. The affecting time of the flame is controlled by adjusting the speed of the conveyor. Tomatoes are affected by the flame or infrared irradiation alone or this is accompanied with boiling water or steam at an atmospheric pressure at 100ºC. Each heat treatment was immediately followed with a cooling stage by exposure of the product to water. A rapid splitting of the skin with very little charring is achieved in thirty seconds of infrared radiation, but it cannot improve the peeling of green-shoulder tissue on many cultivars. Weaver et al. state that flame peeling can efficiently remove skin over green shoulders and immature green or yellow areas of the fruit.
Davies (1996) discusses a vegetable peeling apparatus that has a heating station. The heating at this station can be carried out by infrared radiation and that is sufficient to at least partially lift the skin from the flesh.
2.2.3.2 Steam (wet heat) peeling
To eliminate charring, but to keep the effects of the high-temperature of the infrared or flame, superheated steam is used. The steam pressure that is used in wet heat is about 10 atm and it leads to the softening of skins and underlying tissues. When the pressure is suddenly released, steam under the skin expands and causes the skin to puff and crack. Then the skin is washed away with jets of water at high pressure (up to 12 atm).
23 Kunz (1978) patented a method and device for peeling pumpkin by using wet steam. While the pumpkins are shifting on endless conveyors, they are cut into halves and are placed with the pulp facing downwards. Then they are exposed to pressurised wet steam for a short time followed by a water cleaning step. Kunz’s device uses water sprays from the top to remove the skin and water sprays from the bottom to remove pips and pip pulp. Weaver et al. (1980) applied the superheated steam method for peeling of tomatoes as well. Fruits were affected directly by the flow of steam in an open-mesh basket. Tomatoes were exposed two or three times to superheated steam at three different levels of steam pressure and temperature. Each heat treatment was instantaneously followed by cooling with water at about 22ºC. Most efficient peel removal was achieved by using steam at temperatures and flow rates of 425 to 480ºC and 12-15 lb steam per ft2 min respectively.
Smith (1984) developed a method of superheated steam peeling of apples by refining conventional caustic and steam peeling methods. He used a batch-type laboratory pilot-model steam peeler of ¼ bu (8.8 litres) capacity. His pilot-model accepted either saturated steam at 100 psig (7 kg/cm2) or superheated steam at 100 psig (7kg/cm2) at mean inlet temperatures of 371ºC. He found that steam peeling with saturated steam followed by flash cooling by injection of water increased yields, saved labour, eliminated the need for expensive caustic solutions and caustic-solution disposal, and finally, resulted in high quality apples for further processing. Peeled yields in excess of 95% were attained in peeling treatment using superheated steam with or without water injection. Furthermore, as the thermal conductivity of superheated steam is considerably lower than that of saturated steam at the same pressure, the amount of heat penetration into the flesh of fruits should be controlled more easily.
Floros and Chinnan (1988a) explained that ‘in single stage steam peeling, the mesocarp cells would separate from the rest of the fruit. A large portion of edible fruit will be washed away during the pressurised cold water treatment, which follows the steam treatment. Many attempts have been made to date to improve the efficiency of steam peeling for several commodities’ and therefore, they developed a multi-stage process for steam peeling of pimiento peppers. Each process included several repeated cycles at a constant temperature of 215º C and steam pressure of
24 480 KPa. Each cycle was 10 seconds long (except the last which was 5 seconds). Steam was supplied and pressure built_up for the first 5 seconds. For the next 5 seconds, the steam was disconnected and the door of the chamber was opened, the temperature was considerably reduced and allowed the pressure to drop immediately to atmospheric. They observed the effect of various treatments on the fruit surface by scanning electron microscopy. They also observed considerable reduction of losses compared to single stage peeling. The reason was the short consecutive heat treatments that supply sufficient heat to break down the outer layers of the mesocarp cells with minimum effect on the next layers.
2.2.3.3 Thermal blast peeling
In a patented process developed by Harris and Smith (1986), vegetables and fruits are placed in a closed and elevated pressure vessel. The products are affected by infrared heat from the vessel wall and conductive heat from the superheated steam atmosphere. The heat treatment leads to an increased plasticity of the skin tissues caused by drying. The plasticized tissues will increase the resistance of peel to rupture so steam can spread laterally to build the peeling area under the skin. This stage is too short for heat to penetrate to the edible portion. After heat treatment, pressure is reduced to atmospheric pressure by instantly opening the vessel. An explosion leads to blowing up the product from the vessel and blasting the peel up simultaneously as a result of the instantly and highly energized moisture under skin. They applied this peeling method to many fruits and vegetables and observed better results than lye and saturated steam peeling. For example, they tested this method for Alagold pumpkin under 343.33°C within 45 minutes and got 89.4 percent yield by weight. They could reduce peeling losses for this variety of pumpkin from 28% to about 11% for saturated steam and thermal blast peeling respectively.
2.2.3.4 Freeze-thaw
Brown et al. (1970), Thomas et al. (1976), Goud (1983), and Woodroof and Luh (1988) attempted to eliminate the use of caustic solutions in the peeling of tomatoes by the use of the freeze-thaw method. In this method tomatoes are immersed in liquid nitrogen for 5-15 seconds, and then thawed in warm water at 66ºC for 30
25 seconds to loosen the peel. The loss was about 5-7% but this method was not effective on immature yellow and green shoulder tissues. It was mentioned that the method is applicable for peaches as well.
2.2.3.5 Vapour explosion (vacuum peeling)
Drooge et al. (1999) tested the vapour explosion method for removing the skins of fruits and vegetables by explosive vaporization of the moisture under the skin of fruits and vegetables. They placed the vegetable in a peeling vessel, and the pressure in the vessel was rapidly reduced (below atmospheric pressure), leading to explosive vaporization of the moisture. Drooge et al. suggested that it is possible to reduce the air pressure and to cool the vegetable before the vapour explosion. Kliamow et al. (1977) called this method vacuum peeling. They applied vacuum at 600-700 mm Hg to tear the peel off tomatoes. They reported high peeling efficiency, retention of high fruit quality and low energy consumption as well as cost for this method.
2.2.4 Chemical peeling
To reduce the losses during mechanical and thermal peeling methods, chemical peeling has been considered. In this method, skins can be softened from the underlying tissues by submerging vegetables in hot alkali solution. The quantity of solution and the period of time are different for different kinds and varieties of vegetables. Generally, lye may be used at a concentration of about 0.5-3%, at about 93ºC (2000 F), for a short period of time (0.5-3 min). The loosened skins are washed away by high velocity jets of water or compressed air. This method of peeling reduces the losses but it has harmful effects on the flesh of vegetables and also is not environmentally friendly. Different kinds of chemical peeling are briefly described with reference to related works of interest in the following section.
2.2.4.1 Caustic (lye) peeling
26 Floros et al. (1987) evaluated the effect of lye concentration (4 to 12% NaOH), process temperature (80 to 100ºC) and time (1.5 to 6.5 min) on the yield, peeling loss and unpeeled skin, in a lye peeling process of pimiento peppers. They optimized the process to achieve maximum removal of the skin and minimum loss of edible fruit. They revealed that the high lye concentration (12% NaOH) accompanied with short processing time (1.6 to 2 min) at a moderate temperature of around 90ºC should yield an optimum process with removal of all of the skin and peeling losses as low as 20%. Floros and Chinnan (1988b) found that the double-stage process was more effective than the conventional single-stage operation. They tested a double-stage lye
(NaOH) peeling process involving pre-treatment (concentration, c1; temperature, T1; time, t1), holding time (th), and post-treatment (c2, T2, t2). The effect of seven factors on four responses (unpeeled skin, peeling loss, product yield, and texture) was studied. Processing times and lye concentrations were the most important factors, while processing temperatures and holding times had no significant effect on the peeling operation. A mild pre-treatment with 3.2% NaOH for 130 seconds combined with a relatively strong post-treatment of 8% NaOH for 60 seconds at 84ºC and the holding time of 45 seconds were found to result in an optimum process.
Garrote et al. (1993) surveyed the effect of NaOH concentration (4-20%), process temperature (55-95ºC) and time (1-7 min) on the yield, peeling quality, unpeeled skin and total usage of NaOH. They also evaluated titratable NaOH in the potato tissue, NaOH penetration and “heat ring” depth. The best peeling quality, maximum yield and minimum total usage of NaOH resulted with the following conditions: concentration, 11-13%; time, 5-5.70 minutes and temperature, 90-95ºC. The maximum temperature for which the “heat ring” and NaOH penetration depth were equal was 72ºC where, at 20% NaOH and 7 minutes, peeling quality was very good and the “heat ring” was eliminated.
Walter et al. (1982) investigated the effects of heat penetration on sweet potato tissue under three lye-peeling treatments. Heat-mediated, starch gelatinization, cell wall separation, chromoplast disruption, and enzymatic discoloration were evaluated in different conditions according to the peeling treatment. Starch
27 gelatinization, cell wall separation, and chromoplast disruption reduced in the order: 15 minute peel; 30 minute pre-soak (water 78-83ºC); followed by a 6 minute peel. Discoloration occurred in significant amounts only in the 6 minute peel because heat penetration was sufficient to disrupt lacticifer organization but insufficient to inactivate the polyphenol oxidising (PPO) enzyme. The 30 minute pre-soak and 6 minute peel treatment resulted in the best product.
Floros et al. (1987) observed microstructural changes of pimiento peppers, which were treated with varying degrees of NaOH (lye) solutions (1, 4, and 9%), maintained at 80ºC, for different times (1, 2, and 3 minutes). They found that NaOH removes the epicuticular and cuticular waxes, diffuses uniformly into the fruit where it breaks down epidermal and hypodermal cell walls, and solubilizes the middle lamella causing separation of the skin. In severe treatments the lye also dissolves the parenchyma cells of the mesocarp resulting in considerable loss during processing.
Walter et al. (1982) used the different treatments of peeling sweet potatoes in a boiling, NaOH solution. The different treatments were: 6 minute peel (6p), 20 minute pre-soak in water (55ºC) followed by a 6 minute peel (20S), 30 minute pre- soak in water (80ºC), followed by a 6 minute peel (30S), 15 minute peel (15p). The area of tissue which was affected by heat was excited and analysed for o- dihydroxyphenols (DP) and carotene destruction and sugar formation. The data showed that roots peeled by 6P or 20S treatments could discolour as a result of the PPO-DP reaction. 15P and 30S did not show discoloration because both treatments are vigorous enough to inactivate the PPO system. All treatments except 6P caused the inactivation of amylolytic enzymes. Carotenoid destruction was not detected.
2.2.4.2 Enzymic peeling
The adherence of peel to the fruits is done by pectin, cellulose and hemicellulose as the polysaccharides (Toker and Bayindirli, 2003). Therefore, using corresponding glycohydrolases to treat the product will lead to enzymic peeling. Janser (1996) has claimed better texture and appearance for product after enzymatic peeling because of fewer amounts of broken segments and juice losses. Researchers (Ben-Shalom et
28 al., 1986; Rouhana and Mannheim, 1994; Soffer and Mannheim, 1996; Pretel et al., 1997) have proved suitability of this method for peeling citrus fruits. Prakash et al. (2001) studied enzymic peeling of ‘Indian tough nature grapefruit’ by vacuum infusion (Figure 2.4). They used two commercial peeling enzymes coded as “Brand A” and “Brand B”. “Brand A” produced by Aspergillus Niger, involves a mixture of pectinases and cellulase, while “Brand B” produced by Niger and Trichoderma Reesi, contains pectinases, cellulase and hemi cellulase. Their conclusion was that:
a scalding time of 2 min in a boiling water bath, scoring the peel with four radial lines, immersion in an mentioned enzyme bath containing enzymes at 1ml⋅l-1, vacuum infusion at 760 mmHg for 1 min and incubating the fruit in the enzyme bath for 12 min at ambient temperature (30±2ºC), followed by hand-peeling under running tap water, were found to be necessary for easy peeling of Indian tough nature grapefruit.
In continuing the research for peeling of citrus fruits, some trials have been carried out to assess the feasibility of this method for stone fruits such as apricot, nectarines, and peaches (Toker and Bayindirli, 2003; Janser, 1996).
Fig.2.4. Enzymic peeled (right side) and manual (left side) peeled grapefruit (Prakash et al., 2001)
29 2.3 The current situation of peeling tough-skinned vegetables
The current situation of tough-skinned vegetable peeling was assessed by this researcher during an industrial visit to Comet Food Pty. Ltd., Brisbane. This company carries out fruits and vegetables processing involving peeling. During this visit technologies that are used in peeling of different vegetables were observed. In particular, pumpkin is chopped into segments by pivoted hand operated knife, and then, each segment is rubbed against rotating grater drum. Because of the unevenness of surface of some pumpkin varieties such as Jap and Jarrahdale (Figure 1.1), this method of peeling has high peeling losses. This limitation is common for the peeling of tough-skinned vegetables. For example, one of the famous companies which is active in design and manufacturing of food processing machinery is Dornow Food Technology GmbH, Germany. Dornow has introduced automated peelers as the latest peeler for the Hokkaido pumpkin variety which has an even surface. Whole pumpkins are passed continuously through the machine. The floor of the machine is equipped with many rotating disks. These disks could be carborundum or blade. The main limitation of those peelers is also that they have no flexibility to follow the uneven surface of some varieties of pumpkin such as Jarrahdale and Jap (Figure 1.1). This limitation causes high peeling losses. The value of peeling losses for pumpkin is not stated by the company but using similar machines for peeling potatoes can produce peeling losses from 2.4% to 24% (Dornow Food Technology GmbH, Year). The rate of peeling losses depends on the degree of desired peeling. In this case, complete peeling to remove all eyes from potatoes leads to 24% peeling losses.
2.4 Mathematical modelling of peeling processes
The mathematical modelling of peeling processes has been largely limited to the chemical peeling process and, only rarely, to the thermal peeling process. Chemical peeling is a complex phenomenon which involves mass diffusion and chemical reactions. The general approach to the problem is the assumption that the rate of peeling is a function of the lye concentration, temperature and treatment time, as
30 well as other variables intrinsic to the product such as form and geometry, ripeness, peel thickness, and type or variety (Barreiro et al. 1995). The relationship among the above parameters must be identified for each product to attain the maximum efficiency and minimum losses during the peeling process.
There are numerous research efforts that have established relationship amongst lye concentration, temperature and time, and practical results are available for the chemical peeling of various fruits and vegetables, including products such as peaches (Olsen, 1941; Lankler and Morgan, 1944), pimiento peppers (Floros and Chinnan, 1987,1988b), and so on. Most of the previous by mentioned investigations have been conducted empirically to determine the peeling conditions. For example, Athanasopoulos and Vagias (1987) adjusted the results for peeling mandarin segments to a zero and first order reaction for peeling temperature and lye concentration respectively. Also Floros and Chinnan (1987, 1988b) used a response surface methodology to optimize the peeling process of pimiento peppers based on empirical peeling data in one and two stage processes.
Barreiro et al. (1995) developed a mathematical model for the chemical peeling process of foods with spherical shapes. They used the concept of the unreacted core model for this purpose. They established some equations for the prediction of peeling times, weight losses and texture changes during the peeling of guava as a function of the variables involved in the peeling process. Chavez et al. (1996) applied a mathematical model to describe the chemical peeling process. They intended to identify the minimum processing losses of vegetables, energy and NaOH solution consumption. They used the shrinking core model and the second Fick`s law to formulate the mathematical model on the basis of mechanisms included in the peeling process of potatoes. They compared their results to experimental data and observed good agreement.
Few publications could be found on mathematical modelling of thermal peeling methods. Somsen et al. (2004) developed two models on the basis of experimental data of steam peeling for three varieties of potato. They successfully predicted the heat ring and peel losses as the function of some independent variables such as size, variety, conditioning temperature, steam pressure, and steam exposure time.
31
No publication on mathematical modelling of mechanical peeling methods or processes has been found by the date of writing the thesis.
2.5 Conclusions and discussion
The foregoing sections have elaborated on the current state-of-the-art technology in peeling processes, highlighting some affective mechanical properties of products in mechanical peeling, the various methods applicable to different food produce especially tough-skinned vegetables such as pumpkin.
It is evident that each peeling method has its merits and limitations. Some fruits and vegetables are not well adapted to thermal peeling because this method may cause a cauterizing of the surface and wound areas. In the dry heat method, small pieces of charred skin, which are not removed, give a poor appearance to the canned product. A large amount of wastewater and considerable loss of flesh are other important disadvantages of this method. It is necessary to find other methods of thermal peeling that reduce the time required to expose vegetables to heat and subsequently reduce flesh damage. Floros and Chinnan (1988a) reported that the widespread application of steam peeling is due to its high level of automation, precise control of time, temperature and pressure by electronic devices to minimize peeling losses, and due to the reduced environmental pollution as compared to chemical peeling.
Chemical peeling methods similar to thermal methods also have limitations. Those methods should be used in very restricted conditions otherwise high peeling losses will result. For example, Smith et al. (1986) pointed out that if caustic chemicals such as lye are used for a long time or at high temperature, it will lead to the softening of the product with a high degree of losses in edible tissues, expensive caustic solutions and caustic solution disposal. Otherwise, as they stated, this method cannot be used for the peeling of some vegetables and fruits such as pumpkin. Also, this researcher believes that there is no guarantee of absence of harmful effects resulting from the use of chemical materials.
32 The comparison of different peeling methods indicates that mechanical peeling is preferred as it maintains the freshness of vegetables and protects the flesh either from the effects of chemical materials or from heat in the thermal methods. However, mechanical peeling methods have some limitations. For example, every machine is designed for a specific vegetable and shape. Further, peeling losses and the amount of wastewater are considerable. Low efficiency is another concern.
To date no recorded work has been conducted to investigate the limitation of current mechanical peeling methods for tough-skinned vegetables such as pumpkin. Also there is no knowledge about the tissue properties of this kind of product. High peeling losses resulting from the current peeling methods of products such as pumpkin necessitate the development of new automated peeling methods for tough- skinned vegetables. The selection and development of peeling methods that would be commercially viable requires identification of tissue properties. These properties should be determined in relation to the particular peeling methods. For example, identifying some mechanical properties (i.e. toughness, cutting force and shear strength) for different states of product will help to clarify the best technique to mechanically separate skin from under layers. The kind and rate of necessary forces to apply for the separation of skin from flesh without damaging flesh can be determined by the study of mechanical properties of product to be peeled.
The study of mechanical properties is also necessary for any further mathematical modelling of peeling process. Modelling the peeling process enables identifying effective parameters related to the peeler and product to be peeled. Knowing those parameters helps machine manufacturers meet the needs of their food industry customers to be able to fully control the peeling process, reduce peeling losses and increase the peeling rate. At this point in time, no publications on the mathematical modelling of mechanical peeling of fruits and vegetables have been found.
2.6 Summary
The common peeling methods of fruits and vegetables were critically reviewed in this chapter. The benefits and limitations of mechanical, thermal, and chemical
33 peeling methods along with examples of some interesting works were highlighted. It was concluded that in spite of low flexibility and high peeling losses, mechanical peelers are preferred because of their ability to preserve the freshness and quality of the peeled product. The current situation of tough-skinned vegetable peeling, with pumpkin as a case study, was considered. Semi-manual peeling of these vegetables leads to high peeling losses especially on pumpkin varieties that have an uneven surface. In addition, available automated peelers of pumpkin are not able to follow uneven surfaces of produce. It is necessary to develop innovative mechanical peeling methods for these kinds of products. It was concluded that the selection and development of the mechanical peeling method should be directly related to the mechanical properties of products. Some mechanical properties such as toughness, cutting force and shear strength were considered in this review. The comparison of these properties among different states of products, especially skin and unpeeled product, will help designers to select the best kind of separating forces and in proper rates. Modelling of the peeling process was discussed in the last part of review. The literature research has shown that to date there are no published materials on mathematical modelling of mechanical peeling.
Therefore, the demand for a commercially applicable mechanical peeling method and the lack of knowledge about mechanical properties of tough-skinned vegetables, with pumpkin and melon as case studies, encouraged this research. The results of this research could be commercially viable both in terms of the peeling methods developed and the mathematical model. The model will be able to explain the behaviour of the proposed method mathematically on the basis of properties related to the product and peeling tool.
It is identified thus, that the first step towards developing mechanical peeling methods followed by a mathematical modelling for tough-skinned vegetables is the correct and accurate identification of the mechanical properties of pumpkin and melon (case studies). This effort is elaborated in Chapter 3.
34 Chapter 3
Testing of mechanical properties of tough-skinned vegetables
3.1 Introduction
One of the reasons for studying the physical and mechanical properties of fruits and vegetables is the improvement of efficiency of processing equipment, especially mechanical peelers. The tissues of most fruit and vegetable products are subjected to different wanted and unwanted mechanical forces and strains during the post harvesting stage. The wanted mechanical loading takes place basically in food processing equipment such as slicers and peelers and is always accompanied by unwanted loads. Further, unwanted mechanical loading (compression, impact, and vibration) is the main cause of bruising of products during post harvesting operations (Brusewitz et al., 1991). Knowledge of the physical and mechanical properties of products will be useful for the purpose of increasing the effects of wanted, and decreasing the effects of unwanted, mechanical loading.
As discussed in Chapter 2, compression testing is one of the basic and most important tests in the study of mechanical properties of fruits and vegetables. Mechanical properties of skin, flesh, and unpeeled products can be identified by using this test (Jackman and Stanley, 1992; Voisey and Lyall, 1965a; Voisey and Lyall, 1965b; Grotte et al., 2001; Holt, 1970; Voisey et al. 1970; Behnasawy et al., 2004). In addition to the compression test, a tensile test is carried out to determine skin properties such as skin strength (Jackman and Stanley, 1994). The latter method is difficult to implement because of shortcomings such as difficulty in the
35 holding of skin specimens during the test (Su and Humphries, 1972) and the creation of premature tensile failure during specimen preparation (Clevenger and Hamann, 1968; Thompson et al., 1992).
Several researchers for example, Grotte et al.( 2001), Jackman & Stanley ( 1994) and Voisey et al. (1970) have used the difference between unpeeled (overall) compression test and peeled compression test (flesh only) to obtain indirectly the result of the force-deformation test of the skin. This experimental procedure has been rejected by some researchers because of the likelihood of some errors in the result. For example, Thompson et al. (1992) concluded that to determine the contribution of the skin to the external puncture force (in the case of cucumber) by making measurements before and after skin removal is subject to error. Further Jackman and Stanley (1992) concluded that the difference in puncture load displacement between unpeeled (overall) and without skin (flesh) product cannot provide the load displacement of the skin itself. The reason they determined for this is the increase in the effective area of compression during puncture of the skin.
Attempts to find any useful published data on physical and mechanical properties of pumpkin and melon varieties have been unsuccessful. This part of the study has been done in response to the two main queries explained as follows.
Pumpkin and melon belong to the Cucurbitaceous family and are idiomatically called tough-skinned vegetables. This term is commonly used but no scientific (including botanical) definition of this term was found in the literature review. This chapter explains the degree of similarity between these two products of the Cucurbitaceous family and defines a new classification with the name of ‘tough- skinned vegetables’ on the basis of mechanical properties.
Then, recommendations are developed with regard to the design and optimization of processing equipment, especially peelers, for pumpkin and melon. The current study was conducted on three common varieties each of pumpkin and melon with the aim of investigating some mechanical properties. Rockmelon, Honeydew, and Watermelon were the three chosen varieties of melon and the three chosen varieties of pumpkin were Jarrahdale, Jap and Butternut. Finding the similarities
36 and differences between properties will enable specification of the range of application of different peelers. The investigation was carried out by means of using compression testing on different states of the product including skin, flesh, and unpeeled. The investigation of the skin properties was carried out directly on separated skin as well as flesh. Force-deformation behaviour, and rupture force for the two states - skin and unpeeled product - were investigated and toughness was calculated for both states. Shear strength, cutting force, and maximum force of shear strength were evaluated for three states - flesh, skin, and unpeeled product. In addition, the relative contribution of skin to unpeeled product for each property was calculated. The mean arithmetic value of each property was statistically compared for different varieties.
3.2 Design and construction of instrumentation for testing vegetables properties
A number of different devices were designed and built for testing the mechanical properties of vegetables. As product juices caused the test environment to be acidic, stainless steel was used as the material for the instrumentations. They are described below.
3.2.1 Cutter
A drum with a diameter of 80 mm and the edge sharpened at 30° (Figure 3.1.a) was fabricated from stainless steel. It was used to cut and take samples from vegetables. The external diameter of the cutter was fitted to the internal diameter of the unpeeled specimen holder. The specimen taken by the cutter fitted into the sample holder.
3.2.2 Holder of unpeeled sample
The holder was designed and made to hold the unpeeled specimen. It was made from stainless steel with internal diameter of 80 mm which is ten times the diameter of the indentor as shown in Figure 3.1.b. The diameter was chosen to be
37 at least ten times that of the indentor to ensure that the sample reflects the properties of the whole vegetable. The holder was designed to be installed on the Universal Testing Machine (UTM) directly. It was used in those experiments to test the resistance of vegetables to shearing, puncturing and cutting forces.
3.2.3 Holder of skin sample
The holder was designed and made to keep the specimens of skin. It was used in those experiments to test the resistance of the skin to shearing, puncturing and cutting forces. It was made from stainless steel with a diameter of 54 mm as shown in Figure 3.1.c.
3.2.4 Indentor
Three different shapes of indentors were used as the main devices for the shearing, puncturing and cutting tests. The results of the experiments were different depending on the shape of the indentor’s end. The following indentors were designed and made for different experiments.
3.2.4.1 Spherical end indentor
The spherical end indentor was made from stainless steel to be used in the force- deformation test. On the basis of American Society of Agricultural Engineers (ASAE), the cylindrical indentor of 8 mm in diameter was made with spherical end (ASAE S368.4) and a curvature diameter of 25 mm as shown in Figure 3.1.d.
3.2.4.2 Flat end indentor
A flat end indentor was made from stainless steel to be used in the shear strength test (ASAE S368.4). The end of the indentor was flat and had a diameter of 8 mm (Figure 3.1.e). 3.2.4.3 Cutting indentor
38 A sharpened (30° included angle of the edges) indentor, which was made from a 1.5 mm thick piece of stainless steel (Figure 3.1.f) was to be used for the cutting force test (Ohwovoriole et al., 1988).
3.2.5 Curvature meter
The ASAE standard (ASAE S368.4) requires the measurement of the curvature radius on a sample before force-deformation tests. A special device was designed and built (Figure 3.1.g) for that purpose. It included a cross bar with four pairs of holes to install measuring pins at four different positions and a dial gage indicator with an accuracy of 0.01 mm. The holes on the cross bar provided four different positions of measurement adapted to the size and the shape of convex specimen. These positions were 40, 60, 80, and 100 mm. It enabled the measurement of the maximum and minimum radii of curvature at the point of indentation.
3.2.6 Friction coefficient tester
The apparatus was designed and built to measure the frictional coefficient of vegetable samples by means of sliding on a stainless steel chute with adjustable tilt angle. The chute could also be plated with wood or Teflon as shown in Fig.3.1.h.
3.3 Testing methodology
Experiments were carried out to investigate mechanical properties of different varieties of melon and pumpkin (Cucurbitaceae family) by using instrumentations for testing vegetable properties. Toughness, rupture force, shear strength, and cutting force were determined for the Jarrahdale, Jap, and Butternut varieties of pumpkin and the Rockmelon, Honeydew, and Watermelon varieties of melon. The investigation was carried out in three states - flesh, skin and unpeeled product - ignoring the toughness and rupture force of flesh. Relative contribution of skin -
39 i i ffi i
a. Vegetable cutter b. Sample holder
c. Skin sample holder d. Spherical end indentor
e. Flat end indentor f. Cutting indentor
g. Curvature meter h. Friction coefficient tester
Figure 3.1.The instrumentations of testing mechanical properties of vegetables
40 to unpeeled state of each property was estimated. The obtained results were statistically analysed.
3.3.1 Force-deformation test
A study of the force-deformation relationship of agricultural products will give valuable data for engineering analysis and design. Rupture point, stiffness or rigidity, toughness, bioyield point, plasticity and degree of elasticity could be considered as important parameters. The behaviour of samples under the force- deformation test and determination of rupture force and toughness for two different states involving skin and unpeeled samples were the objectives of this test. Tests were carried out according to ASAE standard (ASAE S368.4). On the basis of the standard, the maximum and minimum radii of curvature at the point of indentation on the product were measured and recorded by means of the curvature meter. Because conducting the test on the whole vegetable was impossible, the unpeeled specimens with a circular shape of 80 mm in diameter and 10 mm thick (flesh depth) were prepared from whole pumpkin by using a special cutter device. Skin specimens were prepared with a diameter of about 30 mm and flesh was removed by means of scraping to avoid any ambient effects on the test results. Experiments were carried out using a Universal Testing Machine (Hounsfield Test Equipment-H5000M). The machine subjected the specimens to compression at the speed of 20 mm/min (ASAE S368.4). The specimens were kept in their special holders during tests. The increasing compression force led to loosening the resistance of the sample and finally rupturing of the sample at rupture point. Applied forces (N) and the resulting deformations (mm) could be read off the machine and recorded on the connected computer. Every experiment gave one force-deformation curve. The test was repeated at least 20 times.
3.3.2 Shear strength test
The maximum shear stress that a material is capable of sustaining is called shear strength (Mohsenin, 1970). It is calculated from the maximum load during a shear or torsion test and on the basis of original dimensions (the cross section) of the
41 specimen. The shear strength test was carried out on three different states of the products: skin, flesh, and unpeeled sample. Initially the test was carried out on skin, which is not supported by the flesh. The holder of skin sample and flat end indentor were used. Skin samples had a diameter of at least three times that of the indentor to facilitate them being held by the skin holder. Tests for the different varieties of pumpkin and melon (three varieties each) were carried out. The samples in different states were selected from whole product of different sizes and from different parts of the vegetable (from the top, bottom, and middle parts of all varieties and from convex and concave areas for pumpkin). The size and other specifications of the skin and unpeeled specimens were similar to that described in the previous section. Middle layers of flesh close to skin were chosen to be picked up as circular shaped flesh specimens of 30 mm diameter and 5 mm thickness. The UTM machine with the speed of 20 mm/min was used to apply shearing force. The test was repeated at least 20 times for every state.
3.3.3 Cutting force test
This test was carried out to determine the necessary cutting force of a product in three states involving unpeeled, skin and flesh of different varieties of pumpkin and melon. Samples were prepared from different parts of the product and kept in the holder. The cutting indentor and UTM were applied. The speed of the loading cutting force was 20 mm/min. The test was repeated ten times and the amount of applied force was read directly from the machine and from the force-displacement plot.
3.3.4 Friction Coefficient test
The test was carried out to measure the static coefficient of friction of each variety of melon and pumpkin. Products were tested in three different states: unpeeled, without periderm, and flesh. They were also tested on three different surface materials: stainless steel, Teflon, and wood by using the friction coefficient tester. There was no necessity to do statistical comparisons for the obtained results because the purpose of the test was to develop design recommendations for
42 different possible materials applied to the peeling machine. The test was carried out on samples of different sizes of product and repeated ten times for each. The moisture content of samples was measured simultaneously with the testing of properties. The sample which was used for the moisture content test was removed from the original piece of product for the property test. Each sample was a cube with 10 mm in each dimension. The experiments were conducted under the same conditions of temperature and humidity as in the previous experiments on mechanical properties.
The sample was placed on the chute with adjustable tilt angle. The chute was moved up until the sample started to slide. The tangent of angle of chute in this position was the static coefficient of friction. The chute could also be plated with wood or Teflon to measure the static coefficient of friction on this materials.
3.3.5 The relative contribution of skin to the unpeeled mechanical properties
The percentage contribution of skin to unpeeled mechanical properties was calculated by dividing the values of the mechanical properties of skin by the same property of unpeeled products and multiplying by 100. In addition to this determination of the skin’s contribution to the unpeeled mechanical properties, this property can be used in comparisons of investigated varieties with other fruits and vegetables. Although direct comparison of the same properties obtained by different researchers for different vegetables is extremely difficult because of different conditions of experiments, it is possible to assess the relative contribution (%) of skin. The standard experimental methodology suggested by ASAE was used in this research to simplify comparisons of the same properties for different products in the future.
3.4 Results and discussion
The results of the mechanical properties investigation for the three varieties of both pumpkin and melon are shown in Figures 3.2 to 3.8 and Table 3.1. The
43 obtained results for each property except friction coefficient and also the results of statistical comparison among different varieties are shown in Appendix 1. One- way analysis of variance (ANOVA) with post-hoc comparisons by using SPSS (version 12 for windows) software was used for statistical comparisons of the results. Variations in the mean values of each property for different varieties of products were determined by Least Significant Difference (LSD). LSD values were calculated at the 5% level of probability (p).
3.4.1 Force-deformation relationship
The sectional view of a pumpkin’s skin layers and flesh before and after the force- deformation test is shown in Figure 3.2.a-d. High cohesion of the cells leads to a high need of force and deformation to cause the rupture of the unpeeled product. The high rate of rupture force causes bruising of the flesh to a depth of twice the depth of penetration immediately after rupture. The necessary rupture force of skin was 41 N for Jarrahdale and it reached 189 N for Butternut which also showed a strong cohesion of skin cells itself. Typical examples of force- deformation curves for skin and unpeeled products are shown in Figures 3.2.e and 3.2.f. The peak point of the curve (rupture point) is important because the area under the curve from the origin to this point represents the toughness of the sample. The high rupture force and low deformation of Jarrahdale compared to Rockmelon can be easily recognised but this does not necessarily mean a high difference of toughness. The linear compression representation of the skin and unpeeled states for both vegetables along a clear peak point (rupture point) can be easily seen in the curvatures. The rupture point also shows the maximum point in the range of flesh elasticity such that flesh collapses at this point. Both melon curves, especially for the skin case, showed a round shape at the peak points. Harker et al. (1997) found similar behaviour for the shearing of Watermelon flesh. They attributed this to either a change in the mechanical properties of the cell walls just before tissue failure or a gradual progression of breakage of individual cells until rupture point. After reaching the rupture point, the force falls to a much lower level for Rockmelon than it does for Jarrahdale. This shows that Jarrahdale has tougher flesh than Rockmelon. The other minor peak points after reaching the
44 a. Pumpkin sample before test b. Tissue deformation before rupture (Indenter ∅8 mm, Tip radius R = 12.5mm Depth of indenter penetration h= 4.20mm)
c. Tissue deformation after rupture d. Pumpkin sample after indentation (Indenter ∅8mm, Tip radius R = 12.5 test mm Tissue is affected to the depth of Depth of indenter penetration h=8.44 twice the depth of penetration mm) 120 Unpeeled 300 Unpeeled Skin 250 Skin 100 200 80 150 60
100 Force, N 40 Force, N 50 20 0 0 0 5 10 15 -50 0246810 Deformation, mm Deformation, mm e. Force-deformation curve of f. Force-deformation curve of Jarrahdale variety of pumpkin (skin Rockmelon variety of melon (skin and unpeeled) and unpeeled)
Figure 3.2. Effects of force-deformation test (a-d) and relationship between force (N) and deformation (mm) for melon and pumpkin in two states (skin and unpeeled) (e-f)
45 rupture point for unpeeled Jarrahdale may be considered the result of friction on the sides of the indentor (Grotte et al., 2001).
Pumpkin varieties showed higher unpeeled rupture force compared to the varieties of melon (Figure 3.3). There was no significant difference among unpeeled varieties of pumpkin (Appendix 1). The rupture force of unpeeled Rockmelon was significantly (p < 0.05) different than the two other varieties of melon.
While the varieties of pumpkin were significantly different than melon in unpeeled rupture force, some similarities were found in the rupture force of the skin (Appendix 1). Jap pumpkin and Rockmelon had similar rupture force of skin as well as Butternut with Honeydew and Watermelon. Rupture force of skin and unpeeled product were in the range of 40.68-189.37 N and 100.01-265.49 N respectively.
300 Unpeeled 250 skin 200 150 100 50 Rupture force (N) Rupture 0
t n n le o w o Jap el e l da yd h m e rme Butternu ck te Jarra o Hon a R W Vegetable
Figure 3.3. Rupture force of skin and unpeeled states for different varieties of pumpkin and melon
46 3.4.2 Toughness
Unpeeled toughness of the varieties of pumpkin and melon were statistically alike and also no significant difference was found between Honeydew and Watermelon (Appendix 1). The range of difference of unpeeled toughness was between 601.26 and 1079.66 N. mm. The comparison of skin toughness among all products revealed no significant difference between Butternut and Rockmelon as well as between Rockmelon and Honeydew. The skin toughness of Jap and Jarrahdale was also statistically similar. The skin toughness varied from 13.87 to 436.36 N. mm for Jarrahdale and Watermelon respectively (Figure 3.4).
) 1200 Unpeeled 1000 skin 800 600 400
Toughness (N.mm Toughness 200 0 t e n w l lo e lon Jap da e h ternu m yd me ra ut k e r r B te Ja oc Hon a R W Vegetable
Figure 3.4. Toughness of skin and unpeeled states for different varieties of pumpkin and melon
3.4.3 Cutting Force
Except for the Butternut and Jarrahdale, the products did show close values of unpeeled cutting force. There was no significant difference between Jap and the other three varieties of melon. While Honeydew and Watermelon were like the Jap they were significantly different to unpeeled Rockmelon. The cutting force of Jarrahdale with 5.15 N and Butternut with 20.48 N were the lowest and the
47 highest unpeeled cutting force respectively (Figure 3.5). The same order was also seen in the cutting force of skin. The difference varied from 2.82 to 17.31 N. The skin cutting force of Jap was like that of Honeydew and Watermelon. Honeydew and Rockmelon also showed similar properties. The cutting forces of flesh for different varieties of melon were statistically alike and significantly (P < 0.05) lower than pumpkin varieties. It ranged from 0.27 to 5.43 N for Honeydew and Butternut respectively.
25 Unpeeled 20 Skin Flesh 15 10 5 Cutting force (N) force Cutting 0 t e w n l lon e lo Jap da e h m yd me utternu e r B ck te Jarra Hon a Ro W Vegetable
Figure 3.5. Cutting force of skin, flesh and unpeeled states for different varieties of pumpkin and melon
3.4.4 Maximum Force of Shear Strength
The maximum shear strength force of unpeeled product varied from 99.69 N for Rockmelon to 250 N for Butternut. There was no significant difference among pumpkin varieties in this case as Figure 3.6 clearly shows. Although Watermelon was like the other varieties of melon, Honeydew and Rockmelon did show significant difference (P < 0.05). Similarities were found between Jap and Rockmelon as well as Butternut with Honeydew and Watermelon in the case of skin. The maximum shear strength force varied between 57 and 168 N for Jarrahdale and Butternut, respectively. The investigated property in the flesh state did not show significant difference among melon
48 varieties and also between Jap and Butternut. The lowest and highest values of this property belong to Rockmelon and Butternut (8.82-64.15 N).
300 Unpeeled 250 Skin Flesh 200 150 100 50
Max. shear force (N) shear Max. 0 t e n w n l lo e lo Jap da e h tternu m yd me ra u k e r r B te Ja oc Hon a R W Vegetable
Fig.3.6.The maximum shear strength force of skin, flesh and unpeeled states for different varieties of pumpkin and melon
3.4.5 Shear Strength
There was no significant difference among varieties of melon in the case of unpeeled shear strength. Jap and Butternut were alike in this property. The difference ranged between 0.51 and 2.42 N.mm-2 for Rockmelon and Jap, respectively (Figure 3.7). The same order of vegetables was seen for the shear strength of skin and it ranged from 0.71 to 3.29 N.mm-2. The skin of Jarrahdale, Butternut, and Honeydew had similar shear strength. There was also no significant difference between Rockmelon and Watermelon in this case. In the statistical comparison of shear strength of flesh only varieties of melon were alike. The lowest and highest values of shear strength of flesh belong to Honeydew (0.09 N. mm-2) and Butternut (0.55 N.mm-2) respectively.
49 )
-2 3.5 Unpeeled 3 Skin 2.5 Flesh 2 1.5 1 0.5 0
Shear strength (N.mm t e w n l nu e lo Jap da r e h tte melon yd ra u k e r B Ja Hon Roc Waterm Vegetable
Figure 3.7. The shear strength of skin, flesh and unpeeled states for different varieties of pumpkin and melon
3.4.6 Static coefficient of friction
Wood, stainless steel, and Teflon generally showed increasing static coefficients of friction, respectively (Table 3.1). The static coefficient of friction of unpeeled products ranged from 0.15 to 1.70 for Watermelon and Jap respectively on Teflon and wood. That was varied for flesh from 0.27 (Watermelon on stainless steel) to 2.40 (Butternut on wood). The varieties of melon did not show any sliding on wood in the case of flesh. The products in the state without periderm in total had a higher rate of this property. While Watermelon without periderm just showed slide on Teflon (0.88 gradient), the rest were found in the range of 0.34 (Jap on Teflon) and 2.42 (Rockmelon on stainless steel).
3.4.7 The relative contribution of skin to unpeeled mechanical properties
The relative contribution percentage of skin to shear strength of unpeeled product was significantly high for all products (Figure 3.8). It ranged from 102 to 336% for Butternut and Honeydew, respectively. The higher percentage of contribution
50 revealed higher strength of skin to shear compared to the unpeeled state, or in other words, the tough-skinned product.
Table.3.1. Static coefficient of friction of three varieties of pumpkin in the states of flesh, unpeeled, and without periderm on three different materials including stainless steel, Teflon, and wood.
Jarrahdale Jap Butternut Rockmelon Honeydew Water- melon Flesh St. Steel 0.46 0.45 0.56 1.04 0.78 0.27 Teflon 0.43 0.33 0.60 0.83 0.65 0.40 Wood 1.05 0.63 2.40 No Slide No Slide No Slide Unpeeled St. Steel 0.30 0.62 0.44 0.60 0.41 0.46 Teflon 0.19 0.63 0.16 0.19 0.17 0.15 Wood 0.45 1.70 0.50 0.77 0.40 0.47 Without St. Steel 0.61 0.76 0.99 2.42 -* No Slide periderm Teflon 0.40 0.34 0.69 0.53 -* 0.88 Wood 0.92 1.01 0.97 2.10 -* No Slide
*Honeydew had no periderm
The relative contribution of the skin of Honeydew to shear strength was significantly (P < 0.05) different to that of other vegetables. Butternut was also statistically different to Jap, Watermelon and Jarrahdale. The lowest relative contribution (%) of skin to the investigated properties of unpeeled products belongs to toughness. Jarrahdale and Jap were significantly similar and had the lowest contribution (%) of skin to toughness. Butternut statistically had no difference with the Rockmelon and Honeydew varieties of melon. The relative contribution (%) of skin to the toughness of unpeeled investigated products ranged between 1 and 50 N. mm. Grotte et al. (2001) reported 45% of the same property for the Golden Delicious apple at harvest. This value increased to 78% after 210-day storage at 2º C. They did the measurement indirectly by calculating the difference of toughness with and without skin.
51 The comparison of contributions in cutting force showed that Jap and Butternut were similar to the varieties of melon except Rockmelon. This property ranged from 53 to 102 N for all varieties. Relative contribution of skin to cutting force of unpeeled Cassava tuber was calculated indirectly by using reported data (Ohwovoriole et al., 1988) between peeled and unpeeled product. It was 76% and is very close to the obtained values for Jap and Butternut varieties.
400 Rupture force 350 Toughness 300 Cutting force 250 Max. shear strength force 200 Shear strength 150 100 50
Contribution (%) 0 t u on Jap n l ew ahdale tter me eyd rr u n Ja B Rock Ho Watermelon Ve ge table
Figure 3.8. The relative contribution (%) of skin to different mechanical properties of pumpkin and melon
The Butternut skin in rupture force was similar to that of Honeydew. That was close to the reported values for other products. Thompson et al. (1992) reported 58 to 88% of the same property by doing a puncture test on different fruits including avocado, Bartlett pear, McIntosh apple and also vegetables such as eggplant, green bell pepper, slicing-type cucumber, zucchini and squash. Also Grotte et al. (2001) obtained 65 to 70% of relative contribution of skin to different unpeeled varieties of apples. Jap with Jarrahdale and Rockmelon with two other varieties of melon were found to be similar in the case of rupture force. The varieties of pumpkin were significantly different to the melon varieties in skin contribution to maximum shear strength force. No difference was found among the varieties of melon in this case. The skin of Honeydew and Watermelon had
52 higher and lower contribution (%) to shear strength of unpeeled product respectively. The skin contribution (%) of Honeydew was significantly (p<0.05) different to all other vegetables and except for the significant existing difference of Jap with Jarrahdale and Watermelon, the remaining were alike in this case.
Table3.3. Relative contribution (%) of skin to different mechanical properties
(Mean ±Standard Deviation)
Properties Rupture Toughness Cutting force Max. shear Shear Varieties force N. mm N Strength force, strength N N N.mm-2 Jarrahdale 16±9 2±1 53±12 28±7 153±55 Jap 22±7 1±0.7 85±23 42±27 145±41 Butternut 72±6 22±9 84±8 67±8 102±16 Rockmelon 89±6 28±14 102±17 95±8 141±30 Honeydew 82±16 21±18 102±25 89±17 336±86 Watermelon 97±2 50±15 100±14 97±8 178±43
3.4.8 Application of investigated mechanical properties
The value of investigated properties could be used in the development of design specifications for a peeler and the application range for different vegetables. The “ideal” peeling target is removal of the skin without wasting the underlying good tissue. Maximum permitted rupture force to break the skin without rupturing the whole pumpkin with regard to standard deviations can be considered about 200 N. Values higher than 200 N will lead to rupturing and wasting of the whole Jarrahdale and less than that, probably would not give results in rupturing of the Butternut skin. The above range can not be used for melon varieties and determination of a specific limit for them is not possible. The rupture forces of different varieties of melon are different from each other and it is close for the
53 skin and unpeeled states of each variety. It is not applicable in the case of Watermelon for which the values of rupture force of skin and unpeeled states are the same. The suggested applicable rupture force of skin to protect the whole product and save energy is 95 and 160 N for Rockmelon and Honeydew respectively. Values of toughness showed different work necessary to perform rupture of skin for different varieties. To save energy, it should not be considered more than 22, 62, and 164 N. mm for Jarrahdale, Jap and Butternut varieties of pumpkin respectively. The suggested values of melon varieties are 266, 397, and 527 N. mm for Rockmelon, Honeydew, and Watermelon respectively. Cutting force and maximum shear strength force of skin showed the maximum force needed to cut through the peel without injuring the whole product. Although 188 N as the maximum force of shear strength can be applied effectively without any harm to the whole pumpkin except the Jarrahdale, determination of any value for the melon variety is not possible because of the overlap of skin and unpeeled melon in this case. Determining one value of cutting force for all varieties is also impossible. Injuring the variety with lower thickness as well as wasting energy are the reasons to consider different cutting force of skin to that obtained. There is also overlap in the skin and unpeeled states for cutting force of Jap and all three varieties of melon. Butternut showed the same value for shear strength of skin and unpeeled specimens due to higher skin thickness. This property for skin is considerably higher than the unpeeled state for all varieties of investigated vegetables. Higher value of shear strength for skin when compared with unpeeled products of different varieties can be a possible reason for skin toughness of melon and pumpkin, which should be considered as an important point for peeler design. The scientific definition of tough-skinned vegetables in spite of common use could not be found in literature review and can be suggested by this writer as follows:
Vegetables could be defined as tough-skinned vegetables when the shear strength of skin is equal to or higher than the shear strength of unpeeled product in the same condition of experiment.
The significance, therefore, of this research is that no work has been done to classify the vegetables and fruits on the basis of mechanical properties and the
54 above definition can introduce the first classification of the vegetables on the basis of mechanical properties.
The obtained results of static coefficients of friction recommend using wood, Teflon, and stainless steel as the peeler material in the ranges of 0.15-1.70, 0.27- 2.4, and 0.34-2.42 for unpeeled, flesh, and without periderm states respectively.
3.5 Summary
Some mechanical properties and their applications for three different varieties of melon (Rockmelon, Honeydew, and Watermelon) and pumpkin (Jarrahdale, Jap, and Butternut) were investigated. The investigation was conducted on different states of product, namely, skin, flesh and unpeeled. The measurement of the skin properties was carried out directly on separated skin and also flesh. Rupture force and toughness for two states of unpeeled and skin in addition to cutting force, shear strength and maximum force of shear strength for skin, flesh, and unpeeled states were determined by means of a direct compression test for each case. The existence of similarity among those six varieties of tough-skinned vegetables was investigated statistically. The results were applied firstly to define tough-skinned vegetables on the basis of mechanical properties for the first time. Secondly, they were used to find out the best potential peeling methods in the preliminary stage. Those results, thirdly, were suggested for consideration as the design parameters for tough-skinned vegetable peelers.
55 Chapter 4
Testing equipment for investigation of mechanical peeling methods
4.1 Introduction
The simulation of real peeling circumstances was made possible by the design and fabrication of a new test rig. The purpose of the design was investigation of different mechanical peeling methods. Indeed, design was conducted for the purpose of investigation of the effects of different peeling tools on the product; peeling of the whole product (design of an industrially applicable peeler machine) was not an objective of this research. The effect of peeling tools on different areas of skin was assessed for the circular band area formed on a product. The width of the band on skin depended on the kind of peeling tool. At the stage of design of the test rig, the prediction of what methods or tools will be used was difficult. Therefore the process of design and manufacture of the test rig was completed during testing, and additional components are described in section 4.3.4 (“Attachment”) of this chapter. The important objectives and the methods of their achievement are explained below and relevant drawings are shown in Appendix 2.
56 4.2 Objectives of the design
4.2.1 Adaptability for investigation of different mechanical peeling tools
The test rig was designed to be as versatile as possible, to enable testing of different mechanical peeling tools: blades, knives, and abrasive devices. The flexibility of the test rig for investigation of improved features of those existing and of new mechanical peeling tools was considered. Milling cutter, wire brush and abrasive ropes were some examples of peeling tools of interest to be investigated in the test rig.
4.2.2 Possibility of accommodation of different product size
As the variation in product size is considerable, it was attempted to design a test rig in which it would be possible to use different sizes of products. The range of product size variation was taken into account in designing the peeler head.
4.2.3 Possibility of peeler head position adjustment
To cover the whole surface of products of different sizes, it was necessary to enable the peeler head to adjust its position. It was desirable to adjust its position in three main directions: axial, lateral and vertical. As the number of runs for different experiments could be large, it was essential to provide the means for simple and quick adjustment.
4.2.4 Possibility of peeler tool position adjustment
To enable investigation of different angles of acting forces on the product by peeler tools, it was necessary to make possible positioning of the peeler head in both the vertical and horizontal planes.
57 4.2.5 Possibility of rotation of peeler tool at different angular velocities
In some methods, the rotation of the peeler tool at different angular velocities is needed. Rotary blades and some abrasive tools require rotational movement to accomplish the task. The role of impact force is fundamentally important in the efficiency of peeling for those methods. Changing angular velocities was the easiest way to adjust the impact force in the test rig.
4.2.6 Possibility of rotation of vegetable holder at different angular velocities
Peeling the whole circumference of one product was necessary for evaluation of the peeling results. The rotation of the product was assumed to be easier than rotating of the peeler head around the product. Further, as the different angular velocities of the product during peeling lead to different results, the table with a product holder should be spun to achieve a large range of speed variation.
4.2.7 Simplicity and low cost of manufacturing
Low manufacturing cost is one of the objectives of every design. Attempts were made to reduce the number of components of the test rig. Simple spring and screw mechanisms were used to provide necessary adjustments.
4.3 Enforcement of the objectives
4.3.1 Chassis and Chamber
The chassis was designed as a portable body equipped with one chamber at the top and expandable to two separate chambers. Stainless steel was used as the material in order to provide corrosion resistance. The spacious chamber was designed to accommodate large size products and the peeler head. The product holder was
58 mounted at the base of the chamber and the peeler head was installed at the front side of the chamber (Figure 4.1). There were two possible positions of the product holder; on the centre line of the peeler head and offset (50 mm width and 90 mm depth) in the lateral direction (Figure 4.2). Such a solution was selected for two reasons: firstly, to enable handling of different product sizes (120 to 240 mm diameter), and secondly, to enable peeling by both or just one side of the peeler head.
Peeler head Vegetable holder
DC Motor
Fig.4.1. Test rig
4.3.2 Vegetable holder
The product holder was designed as a rotating table that can carry the product (Figure 4.2). The product could be fixed on the disc by three sharp blades that form a pyramid to provide access to the sides and the top. The pyramid with sharp blades was made in two different sizes to enable handling of small and large vegetables (120 to 240 mm in diameter). Although the blades make cuts on the vegetable, from the standpoint of simplicity of machine and access to the most areas of product, the use of blades was preferred for the test rig in this stage.
59
Fig.4.2. Product holder and two available positions
A 24 V DC motor coupled to the worm gearbox was selected to provide up to 270 rpm depending on the voltage supplied (Figure 4.3). The adjustment of angular velocity of the product holder was carried out by changing the supplied voltage through the voltage volume in the DC supplier. The speed was measured by an optical tachometer (G4958, Smiths). The DC motor was installed outside under the base of the chamber and transferred the torque directly to the shaft of the vegetable holder. This assembly (details shown in Figure 4.4) can be easily repositioned. To reduce the friction between the vegetable holder and the table of the test rig, a Teflon washer (N.6 in Figure 4.4) was used. The diameter of the plate (position 3 in Figure 4.4) was chosen to meet two limitations: firstly, the plate should be able to carry heavy products mostly with large diameters and secondly, it should not interfere with the peeling tools during peeling of small sized products.
Fig.4.3.The two DC sources for vegetable holder and peeler head
60
Fig.4.4. Product holder 1. Shaft 2.Tube 3.Plate 4. Blade 5, 7.Bush 6.Teflon
4.3.3 Peeler head
The mechanism of the peeler head was designed to enable adjustment in three different directions. Two vertical rods enable movement in the vertical direction (Z axis) along the front wall of the chamber (Figure 4.1). The adjustment range varied from 0 to 300 mm in this direction. Position adjustment in the longitudinal direction (X axis) is provided by a spring (position 7 in Figure 4.5) and screw thread of main shaft of the peeler head. The adjustment range varied from 0 to 90 mm. The adjustment of the peeler head in the lateral direction (Y axis) can be achieved by shifting the vegetable holder in two fixed positions (80 mm distance). Resilient ability of the holder of the peeling tools was required to enable tools to follow the irregular shape of different products. The spring mechanism enabled this adjustment. Peeler tools can be installed on rotary flaps (position 9 in Figure 4.5). There are six flaps with an adjustable angular position with the plane of rotation parallel to the product.
61
Fig.4.5. Details of the peeler head 1.Shaft 2.Lock nut 3.Nut 4.Motor 5.Frame 6, 13.Washer 7, 12.Spring 8.Bush 9.Flap 10.Nut Screw 11.Grip screw
Fig.4.6. Peeler head
Ten holes were made in each flap to facilitate the installation of different peeling tools. The holes were placed in a spiral pattern to improve the yield of peeling production (Figure 4.7). The angular position of flaps could be adjusted from 0 to 30o. A diameter of 251 mm was selected for the size of the peeler head (flap’s angle is 0°) to accommodate tough-skinned vegetables of diameters from 120 to 240 mm. Flaps were adjusted by means of a screw mechanism that contains a spring and a lock screw (positions 10 and 12 in Figure 4.5). The angular adjustment enables flaps to accommodate different shapes and sizes of product. The main shaft is
62 driven by a DC motor that can provide angular velocities up to 300 rpm. The adjustment of angular velocity of peeler head was carried out by changing the supplied voltage through voltage volume in DC supplier. The DC motor was installed at the end of the shaft. The rpm could be changed by changing the supply voltage of the DC motor and was measured by using an optical tachometer (G4958, Smiths). Different peeler tools can be installed on the flaps using holes and fixtures.
Fig.4.7. Flap with holes in spiral pattern
4.3.4 Attachments
Preliminary experiments (see Chapter 5) revealed that adjustment of flaps of the peeler head parallel to the product can not provide enough penetration for the peeling tools. For example, abrasive brushes have to reach the inside of concave areas of the product for efficient peeling. It was noticed that the possibility of motion in the vertical plane and perpendicular to the surface of the product will lead to higher peeling efficiency especially in concave areas because centrifugal force will help the abrasive brushes to stand straight. To achieve this objective, one auxiliary peeler head as an attachment was fabricated. The attachment was basically a pivoted frame equipped with two solid plates that can carry peeling tools in the plane perpendicular to the product surface. The shaft was coupled to a DC motor (Figure 4.8). The DC motor was 24 V DC that could provide up to 2000 rpm. The rpm was controlled by the supply voltage and was measured by an optical tachometer.
63 Pivoted point
DC Motor Solid plates
Fig.4.8.The auxiliary peeler head as attachment
The attachment was installed on the frame of the main peeler head. The position in the vertical direction was adjustable by means of two main rods similar to the previous peeler head. The position in the other two directions could be adjusted by using two bars with a set of holes. The peeler head was installed on a pivoted bracket and connected to the frame by a spring to enable following the shape of product. Therefore the abrasive tool which was installed on one end of the peeler head was able to move in the radial direction (related to the product). Constant force of interaction of the peeling tool with the product was maintained by means of dead weight attached to the bracket through pulleys.
4.4 Performance of the test rig
As the test environment is acidic because of product juices, stainless steel was used as the material of the test rig. In application, the test rig showed good performance and versatility enabling the use of different peeling tools and handling tough- skinned vegetables of different size. Flexibility of the test rig and the ease of adjustment and installation of different peeling tools including abrasive, knife and blade tools were excellent. The test rig enabled access to the whole surface of product except the area engaged with the mounting table.
64 The test rig has shown the ability to extend the range of application for investigation of different new mechanical peeling tools. In addition, the peeling of some other fruits and vegetables can be investigated in the future by the use of this test rig.
4.5 Summary
The test rig for investigation of new concepts of mechanical peeling methods was designed and manufactured. Versatility and the ability to use different peeling tools on different sized products and different prospective peeling tools were the criteria in design of the test rig. High flexibility and possibility of peeler head adjustments as well as simplicity and low cost of manufacturing enabled experimental verification of a wide range of mechanical peeling devices.
The test rig proved to be reliable and easy to use. Those capabilities allow the extension of the range of test rig applications for more and different kinds of products in the future. It is also believed that investigation of new concepts of peeling tools is easily possible on the available test rig.
65 Chapter 5
Preliminary trials of different mechanical peeling methods
5.1 Introduction
Mechanical peeling methods can be classified depending on the type of peelers used. Mechanical peeling methods include the use of different types of abrasive tools, knifes, disks etc., and the choice of a peeler is dictated by the type of product that needs to be peeled. The efficiency of all mechanical peeling tools depends to a large degree on their shape. The shapes of peeling tools have been developed to accommodate the different shapes of fruits and vegetables. In addition to the shape, the specifications of the product to be peeled such as its physical and mechanical properties, and the reasons for peeling it, are also important. Trial and error as a simple approach to the development of new shapes of peeling tools was used in this research. The target was to find a tool of an appropriate shape that could follow the surface contours of tough-skinned vegetables such as pumpkin. The shape of the tool had to be able to impart necessary forces to remove the tough skin. A review of some important trials of different shaped of peeling tools on the Jap variety of pumpkin form this chapter. The criteria of effectiveness of peeling in experiments were equal peeling from convex and concave areas with high peeling rate. In each experiment different parameters were selected that would have a significant influence on peeling process. The preliminary results and conclusions could help the author to choose the best tools and methods for further investigations.
66 5.2 Trials of different tools
5.2.1 Wire brushes
Different kinds of wire brushes were tested because of their flexibility and ability to accommodate the pumpkin shape. The product could be peeled by the end tip of a wire brush. Two main kinds of wire brushes are available in the market - rotary and twisted wire brushes. In this case study the effects of both kinds of wire brushes were investigated.
5.2.1.1 Rotary wire brush
This abrasive tool is designed to strip paint and clean objects (Figure 5.1.a). These brushes are available in different diameters and thicknesses. They have a trapezoidal contour in cross-section. For this investigation, it was necessary to reshape the side section of the brushes for two reasons: firstly, edges with an acute angle will cause uneven tissue removal and secondly, right angled edges couldn’t access grooves effectively. The side section of the brush was reshaped to a triangular contour with a fillet radius of 3 mm. Preliminary experiments were carried out. The effective parameters were found to be the pushing force of the brush, the angular velocity of the vegetable holder (v. speed) and angular velocity of the peeler head (p. speed).
(a) (b)
Fig.5.1.Rotary wire brush and its peeling effect on a pumpkin
67 The brush was brought into contact with the pumpkin with different pushing forces (0.2-1.4 N), angular velocities of the peeler head (300-1000 rpm) and the vegetable holder (5-20 rpm). The uncovered ranges of values were unlikely to generate effective results. The pressure of the pushing force on the peeling tool was applied through a spring and a pulley. The results showed that the brush did not completely follow the shape of pumpkin and caused excessive flesh removal in convex areas (Figure 5.1.b). Indeed these convex areas completely disappeared. This high loss of flesh in the peeling process was the main limitation.
5.2.1.2 Twisted wire brush
A multi-strand stainless steel wire brush was used (Figure 5.2.a). Firstly, one end of the wire was loosened to form the shape of the brush and it was fixed to a holder bush (Figure 5.2.b) at the other end. Several brushes were installed on the flaps of the peeler head (Figure 5.2.b). Preliminary experiments were conducted. The significant parameters were found to be the v. speed, p. speed and the angle of flaps. Experiments were carried out under different angular velocities of the peeler head (300-1000 rpm) and pumpkin (5-20 rpm). The flaps were also tilted at different angles (0-30°). The uncovered ranges of values were unlikely to generate effective results. The values produced from investigated ranges did not generate satisfactory results. There was excessive flesh removal at convex areas (Figure 5.3) but the concave areas left unpeeled.
Holder bush
(a) (b)
Fig.5.2. Twisted wire brush before and after loosening the strands
68
Fig.5.3. Affected areas of pumpkin after using the twisted wire brush
In the second stage, an attempt was made to improve the shape and the flexibility of the brush. For the stainless steel brush, first, the strands were loosened and then wires in the secondary strands were also loosened (Figure 5.4.a). Four brushes in a new configuration were clamped between the two disks of the peeler head attachment at 90 degrees to each other with each pair having a different length (Figure 5.4.b). Experiments were conducted under the same conditions as in the previous stage. Better results were obtained compared to the previous stage. Peeling was initiated at the grooves and then worked towards the convex areas. For one adjusted distance, peeling at the grooves by the brush tips was quite effective but in convex areas the side strands of the wires were just touching the pumpkin. More flexibility of the wires at the side areas caused low peeling productivity compared to at the grooves. So in order to peel in the convex areas, a large amount of flesh needed to be removed from the concave areas. Figures 5.4.c and d show the pumpkin after 30 minutes of continuous peeling.
At the third stage, to reduce the flexibility of the side areas of the brush, each secondary strand was restricted by an aluminium clamp (Figure 5.5.a). Experiments were conducted under the same conditions like as in the previous stages.
The results were not as encouraging as at the second stage because there was much lower flexibility at the side areas of the brush. Also, because the width of the brush tip was bigger than the width of grooves, more flesh was needed to be removed from the concave areas in order to reach to the floor of grooves.
69
(a) (b)
(c) (d)
Fig.5.4.Improved design of the artificial twisted brush in the second stage and its effects
At the fourth stage, each secondary strand was cut at a different length and then held by a clamp before being loosened. The brushes were installed on the attachment of the peeler head and experiments were conducted under the same conditions as in the previous stages. Peeling at the grooves was effective but only 70% of the convex areas were peeled. This could be attributed to the flexibility of the both the wire brush and the peeler head.
70 Aluminium clamp
(a) (b)
Fig.5.5.The twisted wire brush in the third stage and its peeling effect
5.2.2 Ball chain tool
In this method, stainless steel ball chains (Figure 5.6.a) were used. It was believed that the heat produced from the energy released from impact would enable the pumpkin to be easily peeled by softening the affected areas. Several ball chains were prepared and clamped between two discs and installed on the peeler head attachment.
(a) (b)
Fig.5.6.Ball chain and its peeling effect after application
71 The preliminary experiments were conducted at different levels of significant parameters such as the angular velocity of the peeler head (300-1000 rpm) and pumpkin (5-20 rpm). The uncovered ranges of values were unlikely to generate effective results. The values produced from investigated ranges did not generate satisfactory results. (Figure 5.6.b). Peeling was irregular, and the tool did not follow the entire surface.
5.2.3 Milling cutters
Rotary milling cutters were also tested. Different shaped of cutters are available in the market. For peeling purposes, five different shaped of cutters were selected: cylindrical (in two different diameters), semi-elliptical, spherical, tapered (Figure 5.7.a), and cutters of a cylindrical shape with triangular side sections (Figure 5.7.b) were the different shapes to be investigated.
(a) (b)
Fig.5.7.Different milling cutter tools investigated (a) and cutter of a cylindrical shape with triangular side section (b)
The preliminary experiments were carried out. The effective parameters were found to be the shape of milling cutter, v. speed and p. speed. The results were compared for different levels of v. speed (5-20 rpm), p. speed (300-1000 rpm) and each shape of the milling cutter. The uncovered ranges of values were unlikely to generate effective results. The first four shapes quickly became clogged during the peeling
72 process. Also the removal rate of the peel was very low and the decision was taken to discontinue experiments with these tools. The cylindrical shaped cutter with the triangular side section showed significantly better results in preliminary experiments. High removal rates of peel with no clogging were important outcomes of these experiments. The tool followed the uneven surface of the pumpkin very well, and its action was self cleaning. Figures 5.8.a and 5.8b show how the pumpkin was peeled by the cylindrical shape cutter with triangular side section. Figure 5.8.a shows the comparison between the effects of peeling by the milling cutter (bottom side) and the effects of peeling by the twisted wire brush (top side). The results were encouraging.
(a) (b)
Fig.5.8.The effect of peeling by milling cutter of a cylindrical shape with triangular side sections
5.2.4 Mower trimming lines
The trimmer line that is used in lawn movers for gardening was used in this experiment. The one available in black colour with “star” cross-section shapes of 3 mm thickness was chosen. One brush was made from several lines (stands) and installed between two discs of peeler head attachment. The effectiveness of peeling at different angular velocities of the peeler head (300-1000 rpm) and the product (5- 20 rpm) as two significant parameters were examined. The uncovered ranges of values were unlikely to generate effective results. The result was unsatisfactory, and
73 this could be attributed to the light weight and speed of lines and consecutively low kinetic energy. The available weight of the trimmer line needed to be of a much higher rpm, and this was impossible to achieve in the current test rig.
5.2.5 Abrasive ropes
Abrasive particles of different abrasion grades and sizes were glued to the end of syntetic ropes (4 mm diameter) and installed between two discs on the peeler head attachment (Figure 5.9.a). The angular velocity of the abrasive ropes (p. speed) and pumpkin (v. speed) were significant parameters. Experiments for different angular velocities of abrasive ropes (300-1000 rpm) and of pumpkins (5-20 rpm) were conducted. The uncovered ranges of values were unlikely to generate effective results. The results were unsatisfactory (Figure 5.9. b). Despite having a high degree of flexibility they could only scratch the convex areas. The unsatisfactory result could be attributed to light weight of the tool and the low speed that resulted in a low energy of impact.
(a) (b)
Fig.5.9. Abrasive rope and its peeling effect on a pumpkin
5.2.6 Abrasive Pads
Abrasive pads that are used for removing paints were tested. Rectangular shaped abrasive pads were prepared from a standard pad (Grit: 60/60, S/carbide, Size:
74 120+98+ 13 mm). They were preliminarily tested by hand and the removal rate of peel proved satisfactory. As the result was promising, they were chosen for a future experiment. Abrasive tools using of these pads were made and tested and the results are described in the next chapter.
5.2.7 Abrasive foams
As the results of applying abrasive pads were good, an attempt was made to improve this method, especially in respect to the shape of the peeler tool. To maintain flexibility during the peeling process and also test the ease of using of different shapes of peelers, foam was chosen as the material. Foams of the same material (HR 30-60 density) but of different shapes were used. (Figure 5.10.a-d).
Different shapes of abrasive foams including horny shape (a), spiral paraboloid (b), disk with flat face and 60˚ included angle (c), and disk with mirrored edges and 75˚ included angle (d) were tested. Abrasion grade, the angular velocity of abrasive foam and pumpkin were realised as effective parameters on peeling experiments. Abrasive particles with different abrasion grades were glued to the foams and the experiments were conducted at different angular velocities of peeler head (300-1000 rpm) and pumpkin (5-20 rpm). ). The uncovered ranges of values were unlikely to generate effective results. Shapes (a) and (b) showed unsatisfactory results. Experiments with shapes (c) and (d) were promising (see Figure 5.10.e).
5.2.8 Rope covered by spiral blade
Spiral blades made from thin stainless steel strips were used. One edge of the blade was sharpened in a chisel pattern. The blade was twisted around a rope and clamped in a peeler head (Figure 5.11.a). The preliminary experiments with different levels of independent variables including angular velocities of ropes (300-1000 rpm) and product (5-20 rpm) as two significant parameters were conducted. The values out of those ranges were unlikely to be effective in the results. Results were unsatisfactory (Figure 5.11.b); blades were easily bent and peeling was uneven.
75
(a) (b)
(c) (d)
(e)
Fig.5.10. Different shapes of abrasive foam and their peeling effect on pumpkin (a) horny shape; (b) spiral paraboloid; (c) disk with flat face and 60˚ included angle; and (d) disk with mirrored edges and 75˚ included angle; (e) abrasive foam in use as peeler tool
76
(a) (b)
Fig.5.11. Rope covered by spiral blade (a), and its peeling effect on pumpkin (b)
5.2.9 Sandpaper belt
In this experiment narrow strips of sandpaper were used (Figure 5.12). During the preliminary test it was proven that the strip can not follow the irregular shape of the pumpkin. The strip was easily twisted in places of changing curvature.
(a) (b)
Fig.5.12. Sandpaper belt installed on test rig with and without pumpkin
77 5.2.10 Abrasive plates
Grater plates were shaped and applied to investigate another peeling tool. Grater plates were cut to form a triangular contour to produce better access to grooves (Figure 5.13.a). They were made in different lengths from coarse grade of grater to access the groove floor. The grater plates were fixed on the flaps of the peeler head and the experiments were conducted at different tilt angles (0-30°). The peeling efficiency in concave and convex areas was also examined at different angular velocities of the peeler head (300-1000 rpm) and pumpkin (5-20 rpm). These two parameters were found to be significant. The values out of those ranges were unlikely to be effective in the results. The results did show high removal rate of skin from convex areas but the peeling units were not able to penetrate inside the grooves. Peeling off the grooves required high peel removal from convex areas (Figure 5.13.b) and caused high peeling losses. Two circumstances caused unsatisfactory results: firstly, flexibility of abrasive plates was not sufficient to allow penetration into the grooves and secondly, the shape and design of the peeler head was not fully suitable for that purpose. The abrasive plate along the flaps of the peeler head was rotating in the plane parallel to the surface of the pumpkin. When one plate was in the position to penetrate inside the groove, other plates which were in contact with convex areas lifted flaps, reducing the effectiveness of the peeling.
(a) (b)
Fig.5.13. Grater plate peeling unit and its effect on pumpkin
78 5.2.11 Abrasive-cutter brush
Abrasive-cutter brushes were designed and applied to eliminate the limitation of the abrasive plate tool. The stainless steel ropes were covered from one end with twisted strips of coarse grater. At the first trial, the strip of coarse grade was simply wrapped around the thin rope (2.5 mm diameter) of stainless steel (Figure 5.14.a). This design was more flexible than the previous one described in section 5.2.10. The two ropes, 120 mm long, were installed between the two disks of peeler head attachment (Figure 5.14.b). The attachment did provide the possibility of peeling the pumpkin in the plane perpendicular to the surface of the product to solve the fundamental problem (penetration into grooves) of the peeler head. Experiments were conducted at different angular velocities of abrasive-cutter brush (300-1000 rpm) and pumpkin (5-20 rpm) as two significant parameters. The values out of those ranges were unlikely to be effective in the results. The results of peeling especially for higher angular velocities of brushes (1000 rpm) and lower speed of the vegetable holder (5 rpm) were satisfactory (Figure 5.14.b). The second variant of design was developed to improve flexibility and productivity of the abrasive- cutter brush. The strips of coarse grade were twisted around of thicker rope in 4 mm diameter (Figure 5.14.c). They were installed on the peeler attachment in the same way as before. Preliminary experiments were conducted under the same conditions as in previous ones.
The performance of this more flexible brush in the experimental environment was investigated and good results were obtained (Figure 5.14.d). Peel at affected areas - either concave or convex - was removed evenly. Since this tool appeared to be one of the most promising more experiments were planned to be carried out and the results are described in the next chapter.
79 (a) (b)
(c) (d)
Fig.5.14. Abrasive-cutter brush and its effects on two stages First design of abrasive-cutter brush (a) and its effect (b), the second design (c) and its effect (d)
5.2.12 Abrasive bristle products
These products are ready to use and available in the market. They are made from rubber-like substance with abrasive particles embedded into it. They are used in the polishing industry and are available from different suppliers in Australia (i.e. 3M Australia). The supplier advertises the benefits of bristle products such as higher safety, productivity, versatility, and lower cost compared to normal abrasive wires. They are available in two different shapes of wheels and disks in different sizes and abrasion grades. The brown, yellow and green colour radial disks (75 mm) with abrasion grade of 220, 80, and 50 (Figure 5.15) respectively were applied to investigate the peeling effects. The abrasion grade of disks, angular velocity of
80 disks and pumpkin were found to be significant parameters. The investigation was carried out at different angular velocities of disks (300-1000 rpm), pumpkin (5-20 rpm) and different pushing forces (1-5 N). The values out of those ranges were unlikely to be effective in the results.
(a) Abrasive bristle disk (b) Variety of shapes and sizes
Fig.5.15. Abrasive bristle products
The green disks at high value of speed and low value of angular velocity of pumpkin gave positive results only if the bristle disks are brought into contact with the pumpkin with the application of high pushing forces (bigger than 4 N). Application of high pushing forces (compression) to the tool decreased the flexibility of peeler and produced an uneven peeled surface.
5.3 Conclusions
The comparison of the results of preliminary experiments proved that the use of abrasive-cutter brush, abrasive foam, abrasive pads, and milling cutters could be considered for further investigation. They could perform peeling of an uneven surface of product equally with higher productivity. Flexibility and the capability of imparting necessary peeling forces were the main advantages of these tools.
81 5.4 Summary
Preliminary experiments were carried out to identify some peeling tools suitable for tough-skinned vegetables with an uneven surface such as pumpkin. High peeling rate with equal flesh removal from convex and concave areas was the main criteria of comparison. Different shapes of tools were tested. Ease of manufacture of these tools and possibility of their installation on the test rig were considered. The selected tools mostly work on the basis of the abrasive peeling method and possess the required flexibility. Abrasive-cutter brush, abrasive foam, abrasive pads, and milling cutters were selected from the tested tools for further investigation, which will be addressed in Chapter 6.
82 Chapter 6
Experimental investigation of mechanical peeling methods
6.1 Introduction
New approaches to mechanical peeling methods of tough-skinned vegetables were tested. The four selected peeling tools from preliminary experiments (Chapter 5) were used for further investigation. The chosen methods were the milling cutter, abrasive pads, abrasive foams, and abrasive-cutter brush. The significant factors that could mainly reflect the results of applying each method were selected for experiments. The effects of either five factors (abrasive foam method) or four factors (the three other methods) that have significant impact on peeling were considered for analysis. Factors were investigated at three levels each. As carrying out full factorial peeling experiments (at least 81 runs each method) is costly and time consuming, the Taguchi method was used. This fractional factorial design allowed the running of a minimum number of necessary experiments for determination of criteria for each peeling method. The criteria involving the evenness of efficiency (%/min) in concave and convex areas of the product and the peel losses (%/min) were optimized by using the recommended Taguchi formula. The optimized results were compared to identify the best peeling method of tough- skinned vegetables for further investigation.
83 6.2 The criteria of experiments
The following two criteria were used to identify the optimum combination of independent variables in each method and also comparison of different methods: • A minimum possible difference between peeling efficiencies in concave and convex areas • High efficiency of peeling in concave and convex areas along with low peel losses
6.2.1 Peel losses
Peel losses in percentage can be calculated by using the weight of product before and after peeling (Willard, 1971), by applying the following formula:
( W1 −W2 y1 = ×100 (6.1) W1 × t
where, Ў1 is peel losses in %/min; t is time of peeling in minute; W1 and W2 are weight of unpeeled and peeled pumpkin respectively; and W2 is not equal either to zero or to W1.
Pumpkins were weighed before and immediately after peeling by analogue scale with ± 0.1 gram accuracy.
6.2.2 Peeling efficiency
Peeling efficiency is the percentage peel that is removed from the initial skin per unit time (min). Three places (120° including angle) at the circular affected area on the pumpkin for each convex and concave area were considered for the measurement of peeling efficiency. The peeling efficiency (%/min) after each time interval (t1 to t4) of peeling was measured at the same place, and the mean value was calculated for further discussion. An indicator with internal diameter of 15 mm was used for the measurement of effect. Optical judgement was made. Judgement
84 was made by three observers and the average value was reported. Remaining peel inside the indicator with notice of the different colours of skin layers, thickness and area were the main criteria for assessment. The suggested formula by Singh and Shukla (1995) was used and developed for calculation of peeling efficiency as follows:
( A1 − A2 y2 = ×100 (6.2) A1 × t
where, Ў2 is peeling efficiency in %/min; t is peeling time in minute; A1 is the
fraction of peel inside the indicator before peeling (assumed 100); and A2 is the fraction of remaining peel inside the indicator after peeling. In this chapter, the terms of “concave efficiency” and “convex efficiency” refer to the peeling efficiency at concave and convex areas respectively.
6.2.3 Estimated responses
The estimation of the mean response in optimum combination of variables was carried out on the basis of the Taguchi ANOVA by applying the following equation (Ranjit, 1990):
( (6.3) μ = T + (LS x1 − T )+ (LS x2 − T )+L+ (LS xn − T )
where: μ( = estimate of the mean response
T = mean of all experimental data
LSxn = optimal level sum response for the significant factor at the level of interest.
85 6.2.4 Data analysis
Analysis of variance (ANOVA) was carried out on the basis of the Taguchi method. It was used to calculate the percentage contribution of independent variables and their effect on the response variables. Optimization was also carried out using the suggested Taguchi method.
6.3 Peeling by using milling cutter
6.3.1 Introduction
Five different shapes of milling cutters were used in the preliminary experiments of peeling (Chapter 5). A semi-ovate, elongated cylindrical shape in small and large diameter and a disk shape with triangular contour were investigated (Figure 5.7). It was revealed that except for the disk shape with triangular contour, other shapes of milling cutters became clogged very quickly. This could be attributed to arrangement of the teeth in a perpendicular direction that did not allow the clogged cutter to self clean. The disk shape with a triangular contour did produce good results without any clogging in preliminary experiments. The teeth were tilted in two directions so that they could easily discharge the removed peel. This shape of milling cutter was considered for further tests.
6.3.2 Material of experiments
The Jap variety of pumpkin (Cucurbitaceous family) from different local farms around Brisbane (Queensland, Australia) was used for the experiments. The products were randomly selected from ripe and defect-free and quite similar sized (18-23 cm diameter) pumpkins. They were kept under conditions of controlled temperature and humidity for at least 24 hours before testing. The environment temperature was limited to 20-25 ˚C and 50-55% relative humidity.
86 The milling cutter in a disk shape with triangular contour was used (Figure 6.1). The large and small diameters of head of the cutter were 25 and 21 mm respectively. The teeth in the serrated area were tilted to one side and this facilitated discharge of particles from the cutter during the peeling. The milling cutter was installed on the peeler head attachment (Figure 5.8).
Fig.6.1. Disk shape milling cutter with triangular contour
The Taguchi method was used for the design of the experiments (DOE). Among different variables that potentially could affect the results four more significant independent variables were chosen. The array of L9 was applied for those independent variables including the angular velocity of the peeler head (p. speed), the angular velocity of the vegetable holder (v. speed), the location of the peeling over on the vegetable (location: M for middle; T for top; and B for bottom areas), and applied force for pushing the milling cutter towards the pumpkin (force). Independent variables were investigated in three levels each (Table 6.1).
Experiments were carried out in four time intervals (t1 to t4). The measurements were taken after each time interval (5 min). Peel losses, and peeling efficiency in two cases including convex and concave areas were calculated and measured respectively as dependent variables. The average values of the results (t1 to t4) were calculated and used for discussion.
87 6.3.3 Results and discussion
The experimental results for three dependent variables are shown in Appendix 3. The contribution of four independent variables involving force, v. speed, p. speed, and location to three dependent variables while neglecting the interactions is shown in Figure 6.2. There was no likely interaction among variables to be considered.
Table 6.1. Taguchi experimental design for independent variables and levels
Exp.no.a Variable levelsb
χ1 χ2 χ3 χ4 1 0.66 M 5 1000 2 0.66 T 10 800 3 0.66 B 15 600 4 1 M 10 600 5 1 T 15 1000 6 1 B 5 800 7 0.33 M 15 800 8 0.33 T 5 600 9 0.33 B 10 1000
aExperiments were randomly performed. b χ1=force (N), χ2=location (M: middle; T: top; and B: bottom areas), χ3=v. speed
(rpm), χ4=p. speed (rpm)
As the aim of this study was to investigate the method and not the peeler, the manufactured test rig was designed to enable peeling on a circumferential band (40 mm average width) around the whole pumpkin and experimental data relates to that area. Experimental results for three dependent variables involving peel losses (%/min), peeling efficiency (%/min) at concave and convex areas were obtained
88 and the main effects of independent variables on them are illustrated in Figure 6.3. In this figure logarithmic scale for y axes was used to enable comparison between peel loses and efficiencies.
The comparison of the results shows that concave efficiency was considerably influenced by location and force variables (Figure 6.2). The location also made a high contribution to the peel losses. While the rate of convex efficiency remained constant for different areas of pumpkin (Figure 6.3), the concave efficiency at the top of product was smaller than at other areas. As the concave efficiency at the bottom of pumpkin was smaller than at the middle as well, the direction of application of the force on the milling cutter was also important. The line of action was perpendicular for middle areas while the included angle between that line and tangential line to the surface of pumpkin for the other areas was smaller than 90˚ depending on the curvature of the whole pumpkin in the vertical direction. Peel losses showed a similar trend as concave efficiency except at the bottom area. Although the concave efficiency at the bottom was higher than at the top areas, the peel losses were smaller.
70 60 Peel losses Concave efficiency 50 Convex efficiency 40 30 20
Contribution (%) 10 0 Force Location v.speed p.speed Independent variables
Fig.6.2. The contribution of independent variables to responses resulted from using milling cutter
89
One likely reason of that might be the thinner thickness of skin at the bottom of the pumpkin that is less exposed to the sun. Force as the second important contributor, had greater effect on convex than concave efficiency (20%/min). While convex efficiency remained constant for the last two levels, concave efficiency did show an increasing trend up to the last level of force which obtained almost the same value as convex efficiency. Although the contribution of the force to the peel losses was not considerable, applied force at mid-level (0.66 N) caused lower peel losses. The contribution of angular velocity of the milling cutter and vegetable on peeling of concave areas was negligible. p. speed also did show negligible contribution to peel losses and the same contribution to convex efficiency as v. speed. The effect of different values of v. speed and p. speed on peeling efficiencies in general, remained constant when considering a little deviation for mid-levels. In spite of that constancy, the peel losses were considerably higher for mid level of the p. speed and the last high level of the v. speed (15 rpm).
6.3.4 Optimization and estimation of the responses
High peeling efficiency in concave and convex areas with low peel losses as well as a small difference between concave and convex efficiency were the criteria for the best combination of parameter levels. The performance at optimum combination of variables was estimated only from the significant factors. Variables were considered significant when their contribution percentage to the dependent variables was at least 10%. Therefore, the counting of the v. speed in estimation of the concave efficiency and p. speed in concave efficiency and peel losses was neglected because of inconsiderable contribution. Also the effect of the force on calculation of response estimation of peel losses was neglected for the same reason. High level of force (1 N) that showed high efficiencies and small difference for concave and convex areas was chosen for optimization. Although peeling in the middle areas of the pumpkin gave high peel losses, because of equality of peeling efficiencies, it was chosen as the criteria of optimization. A lower level of the v. speed (5 rpm) because of lower peel losses and difference of peeling efficiencies was selected for optimization.
90 100 100
10 10 Peel losses (%) Peel losses (%) Concave efficiency (%) Convex efficiency (%) Concave efficiency (%) Convex efficiency (%) 1 1 0.33 0.66 1 MTB Peel losses & efficiencies (%/min) (%/min) efficiencies & losses Peel 0.1 (%/min) efficiencies & losses Peel 0.1 Force (N) Location (M, T, and B)
(a) (b)
100 100
10 Peel losses (%) 10 Peel losses (%) Concave efficiency (%) Convex efficiency (%) Concave efficiency (%) Convex efficiency (%)
1 1 51015 600 800 1000 Peel losses & efficiencies (%/min) (%/min) & efficiencies losses Peel Peel losses & efficiencies (%/min) efficiencies & losses Peel 0.1 0.1 V. speed (rpm) P. speed (rpm)
(c) (d)
Fig.6.3. The effects of independent variables on responses resulted from using milling cutter
The p. speed in the mid level (800 rpm) led to higher and more equal peeling efficiencies and was used for calculating optimized results. Estimated mean responses for concave and convex efficiency were obtained at 27.22 and 25.85%/min respectively at 0.59%/min peel losses per minute.
91 6.4 Peeling by using abrasive pads
6.4.1 Introduction
Preliminary experiments using abrasive pads in reciprocating motion under pressure contact with the products showed that this method can remove the peel satisfactorily. The abrasive peeling pads were designed and manufactured for the first time. Nothing similar has been found in the research literature. The innovative design was carried out to reflect the contradictory requirements for the new peeling tool. Tool flexibility and the ability to peel evenly in concave and convex areas simultaneously with contact under necessary force were considered as the main criteria of design.
6.4.2 Material of experiments
The Jap variety of pumpkin (Cucurbitaceous family) from different local farms around Brisbane (Queensland, Australia) was used for the experiments. The products were randomly selected from ripe and defect-free and quite similar sized (18-23 cm diameter) pumpkins. They were kept under conditions of controlled temperature and humidity for at least 24 hours before testing. The environment temperature was maintained in the range of 20-25 ˚C as well as 50-55% humidity.
Abrasive pads were shaped and installed on the flaps of the peeler head. Rectangular abrasive pads were selected among the pads available in the market (standard pad, Grit: 60/60, S/carbide, Size: 120×98×13 mm). They were cut in two different shapes: triangular and rectangular (Figure 6.4.a). Those shapes were prepared in short and long lengths. Pads were placed in metal holders and glued to them. To provide flexibility to each abrasive unit, foams in a wedge shape were used at the peripheral side of the units (Figure 6.4.b). The wedge shape was considered to give easy access to the grooves of the product during peeling. As triangular shape units are supposed to penetrate and work inside grooves, they were positioned higher on the flaps of the peeler head. Then, two different thicknesses of foams were considered: a thicker one (35 mm) and a thinner one (25 mm) for triangular and rectangular units respectively.
92
The peeler head was equipped with six flaps. On each flap, two different shapes of abrasive pad units were installed. The pads were installed on the peeler head at two different circles (Figure 6.4.c). The short units were placed in the inner circle due to space limitation. Units were installed on flaps located on flexible spring suspension to enable following of the product shape. The peeler head was installed in the test rig offset (90 mm) to the centre of pumpkin. Just one side of the peeler head was in contact with the pumpkin during peeling to increase efficiency.
(a) Abrasive units in different shapes (b) Foams in wage shape under each abrasive unit
(e) Assembled peeler head
Fig.6.4. Abrasive peeler pads and accessories Experiments were planned on the basis of the Taguchi method. L9 array was used. The experimental design with uncoded and coded levels is shown in Table 6.2. It
93 enabled experiments for four factors in three levels each. Factors were the angular velocity of the peeler head (p. speed), angular velocity of the vegetable holder (v. speed), angle of the flaps (angle) and overlap between radius lines of abrasive unit and pumpkin (overlap). Experiments were carried out in four time intervals (t1 to t4) at 5 minutes increments. The dependent variables were measured after each five minutes and the mean in percentage per unit time (minute) was used for assessment.
Table 6.2. Taguchi experimental design for independent variables and levels
Exp.no.a Variable levelsb
χ1 χ2 χ3 χ4 1 0 140 10 21.5 2 0 200 20 26.5 3 0 160 5 16.5 4 5 140 20 16.5 5 5 200 5 21.5 6 5 160 10 26.5 7 10 140 5 26.5 8 10 200 10 16.5 9 10 160 20 21.5 aExperiments were randomly performed. b χ1=angle (degree), χ2=peeler speed (rpm), χ3=vegetable speed (rpm), χ4=overlap (mm)
6.4.3 Results and discussion
The experimental results for three dependent variables are shown in Appendix 3. The contribution of four independent variables involving overlaps, v. speed, p. speed, and angle of flaps on three dependent variables while neglecting the interactions is shown in Figure 6.5. There was no likely interaction among variables
94 to be considered. As the aim of this study was to investigate the method and not the peeler, the available test rig was modified to enable peeling on a circumferential band (40 mm average width) around the whole pumpkin and experimental data relate to that area. Experimental results for three dependent variables involving peel losses (%/min), and peeling efficiency (%/min) at concave and convex areas were measured and the main effects of independent variables on them are illustrated in Figure 6.6. Very close values of removal rate of peeling in convex and concave areas could be seen as an important point. As the difference was not significant, equal peeling in different areas can easily be achieved by improved design. Some indication of foam clogging was noticed during experiments. Although clogging of some foam was observed during experiments because of the fine size of particles (Grade: 60), there were no remaining separated abrasive particles within the flesh after the experiments. The significant difference of contribution (%) among independent variables was noted in concave efficiency. While the overlap made a considerable contribution (78 %), v. speed and angles with 2 and 4% were not effective contributors to concave efficiency.
90 80 Peel losses 70 Concave efficiency 60 Convex efficiency 50 40 30
Contribution (%) 20 10 0 angle p.speed v.speed overlap Independent variables
Fig.6.5. The contribution of independent variables to responses resulted from using abrasive pads
Significantly higher contribution of overlap to the concave efficiency compared with the two other independent variables revealed that different overlap levels can
95 significantly change the depth of penetration of abrasive pads through the grooves in concave areas. Figure 6.6.a shows that increasing the overlap can considerably increase the concave efficiency.
10 10
1 1 16.5 21.5 26.5 51020 P.losses P.losses Concave efficiency Concave efficiency Convex efficiency Convex efficiency 0.1 0.1 P.losses & efficiencies (%/min) P.losses efficiencies & P.losses & efficiencies (%/min) 0.01 0.01 Ove rlap (mm) V.speed (rpm)
(a) (b)
10 1000
100 P.losses Concave efficiency 1 Convex efficiency 140 160 200 10 P.losses Concave efficiency Convex efficiency 1 0.1 0510 0.1 P.losses & eficiencies (%/min) P.losses & efficiencies (%/min) (%/min) efficiencies & P.losses 0.01 0.01 P.speed (rpm) Angle (degree)
(c) (d)
Fig.6.6. The effects of independent variables on responses resulted from using abrasive pads
The concave efficiency also is decreasing in response to higher angular velocities of pumpkin (Figure 6.6.b) although its contribution was the lowest. A possible reason
96 for the reduction could be a limited engagement time of abrasive pads with grooves for higher velocities of vegetable.
The sinusoidal function of efficiency and peel losses (%/min) for different angles of the peeler flaps specified 0 degree as the best angle to provide the best access to the inside of grooves. The p. speed as the second important contributor to the concave efficiency caused reduction of peeling efficiency in concave areas at 160 rpm velocity but it showed higher removal at 140 compared to 200 rpm. Similar to concave efficiency, angle also made a low contribution to the convex efficiency with similar effect. The increasing of pumpkin velocity (p. speed) led to decreasing convex efficiency (Figure 6.6.b) at a similar rate compared to concave efficiency but made much higher contribution. It was revealed that higher v. speed caused a decrease of imposing time of grooves and lower contact with abrasive pads while it led to a higher rate of contact for convex areas.
The p. speed showed the largest contribution (38%) to the convex efficiency compared to the other variables. While the increase of the p. speed was expected to increase the convex efficiency because of increasing the number of contacts, the decrease was observed at 160 rpm of p. speed. Convex efficiency (%/min) did not change for the first two levels of overlap but it increased for the higher level of overlap (26.5 mm). For the highest amounts of overlap, the efficiency of peeling in convex and concave areas was approaching the same value. This means that higher overlap could provide the same access to the grooves as to the other areas of the pumpkin for abrasive pads. Generally the independent variables except the angle and v. speed did show similar effects on peel losses as peeling efficiency especially in concave areas. While increasing the overlap led to the higher level of peel losses, the mid-level of the p. speed did result in lower losses of peel. Angle and v. speed, despite low contribution, significantly affected peel losses at higher levels (Figures.6.6.b and d). Increasing the angle may decrease the covered area of the pumpkin for peeling and then decrease the peel losses.
6.4.4 Optimization and estimation of the responses
97
High peeling efficiency in concave and convex areas with low peel losses as well as minimum difference between concave and convex efficiency were the criteria for the best combination of variable levels. The performance at optimum condition was estimated only from the significant factors related to their contribution percentage to the dependent variables. Variables were considered significant when their contribution percentage to the dependent variables was at least 10%. Then, the effect of v. speed and angle in estimation of concave efficiency was neglected because of low contribution. Angles of 0 and 5 degrees did show higher efficiency and lower peel losses. The difference between concave and convex efficiencies was smaller at 0 degree, which could be considered as the optimum value of angle. The overlap of 26.5 mm did also result in higher efficiencies with smaller difference. The peeling productivity at 140 and 200 rpm of p. speed was higher than the same parameter at 160 rpm but because of the small difference between efficiencies at 140 rpm, it was chosen as the best level of peeler head speed. The best level of v. speed was chosen as 10 rpm because of higher peeling productivity and lower difference between efficiencies in concave and convex areas compared to other levels. Estimated mean responses for concave and convex efficiency were obtained at 5.31 and 6.24%/min respectively at 0.12%/min peel losses per minute.
6.5 Peeling by using abrasive foams
6.5.1 Introduction
The attempt to increase the flexibility of the abrasive tool by using abrasive pads was successful. This method was developed using an improved tool. Abrasive foam as a new approach to the development of the abrasive peeling tool was different to abrasive pads in shape. The foam works as a unit like the milling cutter and removes the peel by abrasive particles which cover the active surface of the foam. In addition to the shape, different sizes of abrasive particles (abrasion grades) were investigated.
6.5.2 Material of experiments
98
The Jap variety of pumpkin (Cucurbitaceous family) from different local farms around Brisbane (Queensland, Australia) was used for the experiments. The products were randomly selected from ripe, defect-free and similar sized (18-23 cm diameter) pumpkins. They were kept under conditions of controlled temperature and humidity for at least 24 hours before testing. The surrounding environment temperature was maintained between 20-25 ˚C as well as 50-55% relative humidity.
The shaped foam disk installed on the peeler head of the test rig was used for experiments (Figure 6.7.a). Abrasive particles in three different abrasion grades (Grades of 24, 46, and 60 representing coarse, medium, and fine abrasive particles respectively) were glued (using grit glue) to the shaped foam disks.
(a) Abrasive peeler disk works (b) Abrasive disk, shape A perpendicular to pumpkin surface
(c) Abrasive disk, shape B (d) Abrasive disk , shape C
Fig.6.7 Abrasive foam and accessories
99 Foam disks were shaped in three different shapes of sharpened edge disks marked as A, B, and C. The foam disks had the same material (HR 30-60 density) and size (50 mm thickness and 165 mm external diameter sharpened at 75˚). Shape A (Figure 6.7.b) was continuous without any slots. Shape B (Figure 6.7.c) had six deep and wide slots (8 mm wide and 20 mm deep) in the outer circumference of the sharpened edge foam with 60˚ including angle.
The slots were perpendicular to the outer surface of the foam. Shape C (Figure 6.7.d) had narrow and shallow slots (4 mm wide and 5 mm deep). The slots were made at 45˚ to the radius in the direction of rotation.
Experiments were planned on the basis of the Taguchi method. L27 array was applied. The first five columns of the L27 table were used for five independent variables in three levels each. The experimental designs with uncoded and coded levels are given in Table 6.3.
Factors were the shape (A, B, and C), abrasion grade (Grade: 24, 46, and 60), angular velocity of the pumpkin (v. speed: 5, 10, and 20 rpm), angular velocity of the peeler head (p. speed: 600, 800, and 1000 rpm), and force (1, 1.65, and 2.30 N).
Experiments were carried out in four time intervals (t1 to t4) each 1 minute. The dependent variables were measured after each minute and the mean in percentage per unit time (minute) was used for assessment.
6.5.3 Results and discussion
The experimental results for three dependent variables are shown in Appendix 3. The contribution of five independent variables involving shape, abrasion grade, v. speed, p. speed, and force to three dependent variables while ignoring the interactions is shown in Figure 6.8. There was no likely interaction among variables to be considered because in this stage of investigation only the main effects of independent variables were important.
100 Table6.3. Taguchi experimental design for independent variables and levels
Exp. No.a variable levelsb
χ1 χ2 χ3 χ4 χ 5 1 A 60 5 1000 1.65 2 A 60 5 1000 2.30 3 A 60 5 1000 1.00 4 A 46 10 800 1.65 5 A 46 10 800 2.30 6 A 46 10 800 1.00 7 A 24 15 600 1.65 8 A 24 15 600 2.30 9 A 24 15 600 1.00 10 B 60 10 600 1.65 11 B 60 10 600 2.30 12 B 60 10 600 1.00 13 B 46 15 1000 1.65 14 B 46 15 1000 2.30 15 B 46 15 1000 1.00 16 B 24 5 800 1.65 17 B 24 5 800 2.30 18 B 24 5 800 1.00 19 C 60 15 800 1.65 20 C 60 15 800 2.30 21 C 60 15 800 1.00 22 C 46 5 600 1.65 23 C 46 5 600 2.30 24 C 46 5 600 1.00 25 C 24 10 1000 1.65 26 C 24 10 1000 2.30 27 C 24 10 1000 1.00
aExperiments were randomly performed. b χ1=shape, χ2=abrasion grade, χ3=v. speed (rpm), χ4=p. speed (rpm), χ5=force (N)
As the aim of this study was to investigate the method and not the peeler, the available test rig facilitated the peeling trial on a circumferential band (28 mm
101 average width) around the whole pumpkin and experimental data relates to that area. Experimental results for three dependent variables involving peel losses (%/min), peeling efficiency (%/min) in concave and convex areas were measured and the effects of independent variables on them are illustrated in Figure 6.9.
40 35 30 peel losses Concave efficiency 25 Convex efficiency 20 15
Contribution (%) 10 5 0 Shape Grade V.speed P.speed Force Independent variables
Fig.6.8. The contribution of independent variables to responses resulted from using abrasive foams
There was no significant difference between the removal rate of peeling in convex and concave areas. Despite concerns about embedment of abrasive particles into the flesh of the peeled pumpkin no sign of that was observed. The fine grade (Grade: 60) of abrasive foam did show some clogging. This was also reported in previous experiments (peeling using abrasive pads). As expected, abrasion grade made a higher contribution to the all responses compared to the other independent variables.
The convex and concave efficiencies, and peel losses (%/min) respectively were more affected by the grade of abrasive particles. The coarse grade (Grade: 24) did show significantly higher skin removal from all areas of product while the two other grades had almost identical removal rates (Figure 6.9.a). The peel removal was even in convex and concave areas for different grades of particles except the mid-grade.
102 30 25 25 20 20 15 15 P.losses Peel losses 10 Concave efficiency (%/min) 10 Concave efficiency (%/min) Convex efficiency 5 Convex efficiency 5 0 Peel losses & efficiencies efficiencies & Peel losses 0 Peel losses & efficiencies efficiencies & Peel losses 24 46 60 ABC Grade (abrasion) Shape (foam)
(a) (b)
25 25 20 20 15 15 Peel losses Peel losses 10 10 Concave efficiency (%/min) Concave efficiency (%/min) Convex efficiency 5 Convex efficiency 5 0 0 Peel losses & efficiencies efficiencies & Peel losses efficiencies & Peel losses 51015 600 800 1000 V.speed (rpm) P.speed (rpm)
(c) (d)
25 20 15 Peel losses 10 Concave efficiency (%/min) Convex efficiency 5 0 Peel losses & efficiencies efficiencies & losses Peel 1 1.65 2.3 Force (N)
(e)
Fig.6.9. The effects of independent variables on responses resulted from using abrasive foams
103 The even peeling of different areas of pumpkin also was observed for the shape variable (Figure 6.9.b). The shape of the abrasive unit had the same order of contribution as abrasion grade. The skin removal from concave and convex areas was very close for shape A although it was expected to be much closer for other shapes. The existence of slots in shapes B and C was expected to increase the continuous contact time between peeler disk and product and finally higher and more even peeling of different areas. But shape was not a significant contributor to the peel losses (%/min) and v. speed did show higher contribution to that. The mid- level of v. speed (10 rpm) caused higher losses of peel (Figure 6.9.c) and efficiency especially in convex areas, decreasing with an increase of v. speed. The better peel removal was observed for the first two levels of v. speed. The development of efficiency for higher levels of p. speed was in the opposite direction of v. speed (Figure 6.9.d). Peeling at convex areas for different levels of p. speed, except mid- level, was higher than at concave areas although the difference was small. Increase in peel losses was the same as efficiency with increasing p. speed although the contribution of p. speed to peel losses was lower compared to efficiency. The higher contribution of force to concave efficiency revealed that applying an appropriate amount of force can lead to even peeling at different areas of product. High level of force caused higher peel removal through grooves at concave areas (Figure 6.9.e). Mid-level of force (1.65 N) showed higher efficiency of even peeling in different areas.
6.5.4 Optimization and estimation of the responses
High peeling efficiency in concave and convex areas with low peel losses as well as a small difference between concave and convex efficiency were the criteria for the best combination of variables. The performance at optimum condition was estimated only from the significant factors which had high percentage of contribution to the response variables. Variables were considered significant when their contribution percentage to the dependent variables was at least 5%. This limit was chosen because of the highest contribution of variables (35% for grade to concave efficiency). Therefore, the effect of force, p. speed, and shape in estimation of peel losses and also the effect of v. speed in estimation of efficiencies was
104 neglected because of insignificant contribution. The abrasion grade of 24 did show significantly higher efficiency of even peeling and was selected as the best abrasion grade. The shape A also caused higher peeling efficiency compared to the other levels. As the small difference between efficiency of peeling at concave and convex areas was observed at all levels, shape A was chosen as the best shape. The lower level of v. speed (5 rpm) led to the higher efficiency and lower amount of peel losses. This level was selected as the best level of v. speed because a small difference between efficiencies was achieved.
The difference between concave and convex efficiencies was almost similar for all levels of p. speed and because of higher efficiency of peeling at 1000 rpm it was chosen for the calculation of optimized results. Mid-level of the force variable (1.65 N) did show higher peeling efficiency with lower difference between concave and convex efficiency. It was selected as the best level compared to the upper level with lower efficiency and higher peel losses.
Estimated mean responses for concave and convex efficiency were obtained at 30.96 and 31.66%/min respectively at 0.97%/min peel losses per minute. The obtained results compared to the results of abrasive pads revealed higher productivity of this device per unit time. Relative comparison of the results between the two works also proved that average efficiency has increased 5.42 times; this increase for peel losses was 8.08 times. The difference possibly shows either the higher peeling losses of using abrasive foams compared to abrasive pads, or higher density of inner layers of skin.
6.6 Peeling by using abrasive-cutter brush
6.6.1 Introduction
The mechanical peeling method is done mostly by using either abrasive tools or knife and blades. Combining these two tools led to the development of another new innovative tool named the abrasive-cutter brush. The new tool can utilise the benefits of the two mentioned peeling tools. The tool is basically twisted stainless
105 steel wires with grater strips wrapped around. The flexibility of wires could provide easy access of the grater protrusions to different areas of the product. Each protrusion acted as a small cutting unit cut and removed the peel pieces. The cutting action caused effective peeling with higher production compared with the other investigated tools.
6.6.2 Material of experiments
The Jap variety of pumpkin (Cucurbitaceous family) from different local farms around Brisbane (Queensland, Australia) was used for the experiments. The products were randomly selected from similar sized (18-23 cm diameter) ripe and defect-free pumpkins. They were kept under controlled temperature and humidity conditions for at least 24 hours before testing. The environment temperature was maintained in the range of 20-25 ˚C as well as 50-55% relative humidity.
The attachment of peeler head was used for conducting the experiments. Two abrasive-cutter brushes (Figure 6.10.a) were installed between solid discs of the peeler head attachment for each trial. They could be fine, coarse, or a combination of the two depending on the status of the planned experiment.
Each abrasive-cutter brush was made from stainless steel wire covered by a twisted strip (Figure 6.10.b) of grater of different grades (coarse or fine). The stainless steel wires were already double twisted wires of the same materials. The strips of the grater were cut from common kitchen graters available in the market. The total length of each brush was 165 mm and the weight 14.75 grams. The materials and methods used to fabricate the brush provided high flexibility of the brush.
106 (a) (b)
(c) (d)
(e) (f)
Fig. 6.10.Abrasive-cutter brush a. abrasive-cutter brush; b. strip; c. affected peeled area (after 5 minutes); d. affected peeled area (after 2 minutes); e and f. other views after peeling (5 minutes)
Experiments were planned on the basis of the Taguchi method. L9 array was used. The experimental design with uncoded and coded levels is shown in Table 6.4. It
107 enabled experiments for four factors in three levels each. Factors were the angular velocity of the abrasive-cutter brush (p. speed), angular velocity of the vegetable holder (v. speed), vertical position of the brush (position) and the coarseness of the brush (coarseness: C for coarse; F for fine; and Com. for combined types of brush).
Experiments were carried out in four time intervals (t1 to t4) each 1 minute. The dependent variables were measured after each time interval and the mean in percentage per unit time (minute) was used for assessment.
Table 6.4. Taguchi experimental design for independent variables and levels
Exp.no.a Variable levelsb C 5 700 -20 1 C 10 850 20 2 C 15 550 0 3 F 5 850 0 4 F 10 550 -20 5 F 15 700 20 6 Com. 5 550 20 7 Com. 10 700 0 8 Com. 15 850 -20 9 C 5 700 -20 aExperiments were randomly performed. b χ1=coarseness (course, fine, combined), χ2=vegetable speed (rpm), χ3=brush speed
(rpm), χ4=vertical position (mm)
6.6.3 Results and discussion
The experimental results for the three dependent variables are shown in Appendix 3. The contribution of four independent variables involving v. speed, p. speed, coarseness, and vertical position to three dependent variables, while ignoring the
108 interactions, is shown in Figure 6.11. There was no likely interaction among variables to be considered. As the aim of this study was to investigate the method and not the peeler, the available test rig facilitated the peeling trial on a circumferential band (40.66 mm average width) around the whole pumpkin and experimental data relates to that area. Experimental results for three dependent variables involving peel losses (%/min), and peeling efficiency (%/min) in concave and convex areas were measured and the main effects of independent variables on them are illustrated in Figure 6.12.
60 Peel losses Concave efficiency 50 Convex efficiency 40 30 20
Contribution (%) 10 0 V. speed Roughness P. speed Position Independent variables
Fig.6.11. The contribution of independent variables to responses resulted from using the abrasive-cutter brush
There was no significant difference between removal rate of peeling in convex and concave areas except in the third level of position (20 mm). The higher contribution of coarseness to concave efficiency can be seen as the first important point in Figure 6.11. This independent variable also had higher percentage of contribution to peel losses and convex efficiency compared with the other independent variables.
109 100 100
10 P.losses Concave efficiency 10 Peel losses Convex efficiency Concave efficiency (%/min)
Convex efficiency (%/min) 1 FCM
Peel losses and efficiencies andPeel losses efficiencies 1 Peel losses and efficiencies andPeel losses efficiencies 51015 0.1 V. speed (rpm) Roughness
(a) (b)
100 100
Peel losses 10 Concave efficiency Peel losses Convex efficiency 10 Concave efficiency
(%/min) Convex efficiency
(%/min) 1 550 700 850 Peel losses & efficiencies efficiencies & losses Peel 1 Peel losses & efficiencies efficiencies & Peel losses 0.1 -20 0 20 P. speed (rpm) Position (mm)
(c) (d)
Fig.6.12. The effects of independent variables on responses resulted from using abrasive-cutter brush
The contribution of coarseness to the efficiencies at concave and convex areas was almost similar. The fine type of abrasive-cutter brush had higher impact on peel losses and efficiencies among the four investigated variables (Figure 6.12.b). The combined and coarse abrasive-cutter brushes were located at the next orders respectively in the statistical comparison. p. speed was the second higher contributor after coarseness (Figure 6.11). p. speed had lower contribution to
110 response variables than the other two independent variables. This variable, as seen in Figure 6.12.c, highly affected peel losses and efficiencies at the third level (850 rpm). It did show medium effect on response variables with lower peel losses at the first level (550 rpm). Position and v. speed showed insignificant and similar contribution to all response variables. The position’s contribution to convex efficiency was less than 5% and could be neglected. It means the peel removal in concave areas can be controllable in micro and micro levels by coarseness and position parameters respectively. Both v. speed and position variables affected more peel losses at the mid-level.
6.6.4 Optimization and estimation of the responses
High peeling efficiency in concave and convex areas with low peel losses as well as small difference between concave and convex efficiency were the criteria for the best combination of variables. The performance at optimum condition was estimated only from the significant factors which had high percentage of contribution to the response variables. Variables were considered significant when their contribution percentage to the dependent variables was at least 10%. Then the position variable was not considered in the calculation of optimized result of efficiency in convex areas because of insignificant contribution in that scenario.
The fine type of abrasive-cutter brush did show significant even peeling in different areas of the pumpkin. High productivity of peel removal with low difference between concave and convex areas caused its selection as the optimized level of coarseness. The first and the third level of v. speed did show similar productivity with lower peel losses. Although the peel losses were higher for 5 rpm it was chosen as the optimized level for v. speed because of the lower difference between efficiencies in different areas of the pumpkin. Except for the third level of p. speed that caused high peel losses, the other two levels had close efficiencies. The first level of p. speed (550 rpm) was chosen as the best level because of significantly lower peel losses. Analysis of the results for the vertical position of the abrasive- cutter brush showed significant difference at the third level (20 mm) and higher peels losses at the mid-level (0 mm). The first level of position (-20 mm) was
111 recognised as the best level for optimization because of the low difference of efficiencies and lower peel losses. Estimated mean responses for concave and convex efficiency were obtained as 68.74 and 68.75%/min respectively at 1.1%/min peel losses per minute.
6.7 The comparison of the four innovative peeling methods
The abrasive-cutter brush can be identified as the best peeling tool among the four investigated peeling methods of tough-skinned vegetables, with the Jap variety of pumpkin as a case study. Comparison of the optimized results of efficiencies in concave and convex areas showed that the abrasive-cutter brush has significantly higher peeling efficiencies (68.74-68.75%/min). The lower efficiencies with significant difference belonged to abrasive pads (5.31- 6.24 %/min). The peeling efficiencies of the other two methods including abrasive foams (30.96- 31.66%/min) and milling cutter (27.22-25.85%/min) were close and located in- between. In the case of efficiency difference between concave and convex areas, generally all four methods showed good results.
The results revealed small difference for the abrasive-cutter brush (0.01) and higher difference for the milling cutter (1.37). Lower difference of efficiencies can be considered as the main sign of lower peeling losses and therefore what peel losses (%/min) show could be considered as another peeling production parameter. The lowest and highest peel losses (%/min) belonged to the abrasive pads (0.12%/min) and abrasive cutter brush (1.1%/min) respectively. Comparison of the proportion of peel losses and efficiencies for different tools showed peel losses of the abrasive- cutter brush could be considered as wanted peeling losses that are expected from “ideal” peeling. This proportion for the other three methods showed higher peel losses in proportion to obtained efficiencies. For example, abrasive foam with 0.97 %/min of peel losses which is very close to the abrasive-cutter brush (1.1%/min) could only produce efficiencies (30.96- 31.66%/min) less than half of abrasive-cutter brush (68.74-68.75%/min).
112 6.8 Potential industrial application of abrasive-cutter brush
In addition to meeting the conditions of the “ideal” peeling method, the abrasive- cutter brush has high potential for industrial application. Peeling machines for tough- skinned vegetables can be designed and manufactured on the basis of this method. A peeler can be equipped with one or several peeling units. Each unit would be able to continuously peel the product. At least six peeler heads should be installed at different positions to carry out peeling along the six circumferential bands. The product could be moved on a linear conveyor or a rotary conveyor and spin around its axis to enable peeling. The linear speed of product, angular velocity of product and peeler heads should be adjustable. Holding the product during peeling can be planned either by with using the method introduced in this research or using pneumatic systems. Using blades to hold the product makes some damage to the product at the top and bottom centres but it is not a matter of concern especially if peeling is followed with dicing. The products after the peeling stage should be cut for further processing (i.e. removing the seeds and cutting into the small pieces). The product could be also clamped using small disks supported by pneumatic actuators.
The peeler ability can be improved using available technologies. It is obvious that the abrasive-cutter brush itself needs to be improved for industrial application. In addition, the peeling process can be controlled using image processing to identify when to stop peeling and thereby minimise peeling losses.
6.9 Conclusions
Four selected innovative peeling methods including abrasive pads, abrasive foams, milling cutter and abrasive-cutter brush were investigated and compared. Significant independent variables for each method and peeling tool were investigated. The independent variables of v. speed and p. speed were chosen as significant parameters for each peeling trial method. Results generally showed that, in spite of the significant contribution by the v. speed to response variables, its
113 contribution was lower compared to the other independent variables for each experiment. The v. speed showed greater contribution to the peel losses than efficiencies for all methods. Its highest contribution was to the peel losses of the milling cutter (29.89%/min). The best level of v. speed for all methods was found to be 5 rpm which produces optimum results for defined criteria. It was impossible to specify the best level of p. speed because the range of variation of the optimized p. speed for different methods was broad. While the optimized level of p. speed for the abrasive foam (1000 rpm) and the milling cutter (800 rpm) were close and higher compared with other levels of the same variety, the optimized p. speed of the abrasive-cutter brush (550 rpm) and abrasive pads (140 rpm) were at a lower level compared with other levels of the same variety for each method. It is clear that the main reason for the variation is because of the technique used for each method and determination of more accurate optimized level of p. speed needs further investigation with more levels of p. speed.
Other effective independent variables for all four methods were overlap and the pushing force of the peeling tool on the product. Higher overlap significantly increased the efficiencies and peel losses. Optimization of the results showed high overlap of the abrasive pads (26.5 mm) and the abrasive-cutter brush (-20 mm position) leads to the best results. The overlap indirectly affected the milling cutting as maximum pushing force (1N) and led to optimized results and also abrasive foams at the middle level of force (1.65 N). Although higher level of force 2.3 N caused higher peel losses (%/min) for abrasive foam it was not chosen as the best level because of larger difference of efficiencies and insignificant lower efficiencies compared to the middle level of force.
The comparison of results at optimum conditions revealed that the abrasive-cutter brush can be selected as the best peeling method for tough-skinned vegetables. Higher and almost equal peeling efficiencies at concave (68.74%/min) and convex (68.75%/min) areas along with accepted peel losses (1.1%/min) compared to the other three methods were the main reasons. This method also has high potential for industrial application.
114 6.9 Summary
Four peeling tools from preliminary experiments (reported in Chapter 5) were selected and tested. The chosen tools including the milling cutter, abrasive pads, abrasive foam, and abrasive-cutter brush were compared in the peeling of a Jap variety pumpkin by using the Taguchi method. The criteria of comparison were high efficiency with low difference of peeling efficiency in concave and convex areas and low peel losses. The results showed that the abrasive-cutter brush has significantly higher productivity at lower peel losses. Estimated mean responses for concave and convex efficiency were obtained at 68.74 and 68.75 %/min respectively, at 1.1 %/min peel losses per minute by using the abrasive-cutter brush. Also the results showed that v. speed at 5 rpm for each method including the abrasive-cutter brush method can provide the best result. The comparison of the results also showed that overlap (for abrasive pads and abrasive-cutter brush) or pushing force (for abrasive foam and milling cutter) to the pumpkin leads to closer results for those criteria.
The selected innovative abrasive-cutter brush has high potential for industrial application. Meeting all conditions of the “ideal” peeling method of fruits and vegetables makes this method the most suitable for design and development of industrial peeling machinery. For this purpose more information about significant factors and their influences is needed and this can be obtained with full factorial experiments which will be explained in Chapter 7.
115 Chapter 7
Abrasive-cutter brush, full factorial experiments, and ANOVA
7.1 Introduction
The abrasive-cutter brush was selected as the best method after conducting a series of designed experiments by preliminary chosen mechanical peeling tools on the Jap variety of pumpkin (Chapter 6). The analysis of the results showed almost uniform peeling in convex and concave areas along with higher values of peeling efficiencies for this method compared with other methods including abrasive pads, abrasive foam, and milling cutter. Statistical analysis (ANOVA) suggested the optimum results can be obtained at 5 rpm angular velocity of pumpkin, 550 rpm of angular velocity of brushes, -20 mm overlap between brushes and pumpkin, and with application of fine type of abrasive-cutter brush. Further experiments are required to investigate the influence of different parameters related to the product and the abrasive-cutter brush on the peeling rate. Two full factorial design tests, each 64 runs, were conducted on two different varieties of pumpkin including Jarrahdale and Jap. The abrasive-cutter brushes were applied in four different coarseness levels - very coarse, coarse, mild, and fine. The peeling rate (g/min) as a dependent variable was assessed as it was affected by other independent variables including angular velocity of brushes (4 levels), and location of peeling area on the product (4 locations). The selection criteria for parameters and their levels were the importance and commercial applicability of them. This chapter outlines the conducted experiments and the results of ANOVA.
116 7.2 Material of experiments
Jap and Jarrahdale as two different varieties of pumpkin were used for experiments. Defect-free and mature products in similar sizes (18-23 cm diameter) were chosen from the farms around Brisbane, Queensland. The pumpkins were kept in a controlled environment at least 24 hours before the test. The storage temperature and relative humidity were controlled in the range of 20-25 ˚C and 50-55% respectively.
The test rig with attachment, as described in Chapter 4, was used. Abrasive-cutter brushes were manufactured and installed on the peeler head attachment by using the same procedure as described in Chapter 6. In addition to fine and coarse brushes, two other brushes labelled “mild” and “very coarse” were made and applied (Figure 7.1). The mild and very coarse brushes were fabricated by opening further the protrusions of the strip of the fine and coarse grater respectively. This change increased the teeth angle by about 20° for each type of strip. It was increased from about 70° to 90° for both types of strips. This change caused a bigger size of protrusions and upright standing of the teeth compared with the initial shape (fine or coarse). The toughness of the four different strips of grater was measured using a force-deformation test, described in Chapter 3. Each protrusion of those strips was located under the compression force of the spherical end indentor till it reached rupture point (loosening the resistance). The test was repeated 15 times for each type of coarseness. The calculated toughness of the strips was 90, 160, 110, and 170 N. mm for fine, mild, coarse, and very coarse respectively.
Full factorial design was used for the design of experiments (DOE). The total number of runs was 128 which were divided equally for the two varieties of products, Jap and Jarrahdale. High number of runs on two different vegetables and long time of each experiment (half an hour) were reasons to ignore replications. The experiments were conducted on three independent variables including coarseness of abrasive-cutter brush (coarseness: coarse, very coarse, fine, and mild), the angular velocity of the peeler head (p. speed: 400, 550, 700, and 850 rpm), and the peeling
117 location on the product (location: top, top-side, bottom-side, bottom) as shown in Figure 7.2.
Fig7.1. The strips with different type of coarseness used for fabrication of the abrasive-cutter brush (from left: very coarse, coarse, mild, fine)
Top
Top-side
Bottom-side
Bottom
Fig.7.2. Different parts of product as levels of location variable
The tests were carried out on four levels of each independent variable. Two brushes were installed on the peeler head attachment for each run of experiment. The
118 angular velocity of the vegetable holder (or product) was not considered as a variable because the results of previous experiments (Chapter 6) showed that lower angular velocity of the product (5 rpm) leads to the optimum results. Therefore, the angular velocity of the pumpkin was kept fixed at 5 rpm. Also the overlap of the abrasive-cutter brush and product was considered fixed and equal to 10 mm for all experiments. This overlap was an independent variable in the previous experiment (Chapter 6) and results showed that up to an overlap of up to 20 mm can lead to optimum peel losses. A new overlap distance was chosen as 10 mm to facilitate adjustment of the peeler head for different locations of product. The running time of the experiments was 5 minutes and this was long enough to cover the necessary peeling time. Experiments were carried out continuously in each run until enough effective peeling (preferably 100%) could be seen. In some combinations complete peeling (100%) was achieved within less than the determined time (5 min). As continuation of peeling into flesh caused reduction of accuracy of results, the run was discontinued after completed skin removal and actual elapsed time was considered in calculation of the mean of peeling rate. It was measured by a analogue scale (±0.1 gr accuracy) and weighting of the pumpkin before and after testing. Mean value per unit time (minute) was calculated for further assessment.
7.3 Peeling rate
Peeling rate is equal to the weight difference of the pumpkin before and after peeling divided by peeling time in g/min.
7.4 Data analysis
The software package SPSS (version 12.0.1) was used for data analysis of the results.
7.5 Results and discussion
The results of frequencies analysis on peeling rate are shown in Table 7.1. The peeling rates were distributed between 0.3 and 8.4 g/min for the Jarrahdale and also
119 ranged from 0.3 to 8 g/min for the Jap. There was a considerable difference between the mean and median of both products. It was 0.385 and 0.362 g/min for the Jarrahdale and the Jap respectively. The difference placed the normality of distribution of data under question. Therefore, further assessment of normality was needed.
Table 7.1.The results of frequencies analysis on peeling rate
Variety Mean Median Mode Std. Variance Range Min Max. deviation Jarrahdale 2.735 2.35 1.00 1.889 3.570 8.10 0.3 8.40 Jap 2.062 1.70 1.20 1.538 2.366 7.70 0.3 8.00
Normality as a prerequisite for many inferential statistical techniques was explored graphically in different ways including histogram, stem and leaf plot, boxplot, normal probability plot, and detrended normal plot (Appendix 4). The tests were carried out separately for both products. The results of all mentioned techniques suggested that the peeling rate (g/min) as the dependent variable is not normally distributed but is significantly positively skewed for each variety of pumpkin. The boxplots indicated that there are four and two outliers (illustrated by circles) for the Jarrahdale and Jap varieties respectively. The outlying values of the Jarrahdale were 7.33, 7.6, 7.9, and 8.4 g/min. The outliers of the Jap variety were 5.66 and 5.30 g/min. The boxplot of the Jap also showed two extreme cases as 7.62 and 8 g/min. Therefore, a natural logarithmic transformation was appropriate to transform the distribution to normal. The results of assessing normality after logarithmic transformation are shown in Appendix 4. The preceding statistics and graphs showed that the natural logarithmic transformation was appropriate because the distribution of transformed peeling rate, called the LnP.rate, was normal. All the diagnostic data were satisfactory after transformation and the results of frequencies analysis for the transformed peeling rate are shown in Table 7.2.
120 Table 7.2. The results of frequencies analysis on LnP.rate
Variety Mean Median Mode Std. Variance Range Min. Max. deviation Jarrahdale 0.770 0.854 0.00 0.725 0.526 3.33 -1.20 2.13 Jap 0.496 0.530 0.18 0.681 0.465 3.28 -1.20 2.08
One-way ANOVA with post-hoc analysis was carried out to investigate the influence of three different independent variables including coarseness, p. speed, and location on peeling rate (g/min) as the only dependent variable. LSD at 0.05 was applied in mean comparison. Population normality and homogeneity of variance were assumed as two basic conditions of conducting ANOVA. Levene’s test showed the homogeneity assumption has been violated for the peeling rate and also non-normality of data population for that has already been confirmed. The homogeneity of variance for the LnP.rate was significantly high (Table 7.3). As two conditions were met by the LnP.rate, therefore ANOVA focused on this variable.
Table 7.3. The results of Levene’s test for homogeneity of variance of LnP.rate
Independent Variety Levene statisticdf1 df2 Sig. variable LnP.rate Product 0.358 1 126 0.551 Coarseness Jarrahdale 0.100 3 60 0.960 Jap 0.424 3 60 0.737 P. speed Jarrahdale 1.301 3 60 0.282 Jap 1.003 3 60 0.398 Location Jarrahdale 0.172 3 60 0.915 Jap 1.124 3 60 0.346
The results of multiple comparisons (Table 7.4) showed that the mean difference of LnP.rate between two groups of product (Jarrahdale and Jap) was significant (p <
121 0.05). The comparison of means of the coarseness effect on dependent variable showed significant difference (p < 0.05) of the F test among different levels of coarseness for two varieties. The mean comparison of the p. speed for both products also illustrated significant difference (p < 0.01) among different levels of angular velocities of peeler head. Although the mean of the LnP.rate in different places of the Jap variety was significantly different (p < 0.05) the Jarrahdale generally did not show any difference.
7.5.1 The effect of p. speed on LnP.rate
The high significant effect of the mean of different angular velocities of peeler head on the mean of LnP.rate (g/min) as the transformed type of peeling rate is shown in Figure 7.3. The increase of the p. speed led to an increase of the LnP.rate (g/min). It increased from -0.13 and -0.24 g/min at 400 rpm to 1.51 and 1.22 g/min at 850 rpm for Jarrahdale and Jap varieties respectively. The pattern of increase was linear with R square of 0.97 and 0.99 for the Jarrahdale and Jap respectively. This result was expected because increasing p. speed leads to higher energy and rate of impacts of brushes on the surface of product.
Table 7.4. ANOVA of the mean of LnP. rate among different levels of independent variables
Variable Variety Sum of df Mean F Sig. squares square Product 2.400 1 2.400 4.844 0.030
Coarseness Jarrahdale 4.170 3 1.390 2.881 0.043 Jap 4.393 3 1.464 3.527 0.020 P. speed Jarrahdale 23.363 3 7.788 47.883 0.000 Jap 18.716 3 6.239 35.369 0.000 Location Jarrahdale 3.798 3 1.266 2.590 0.061 Jap 4.091 3 1.364 3.246 0.028
122 The results of multi comparison of the LnP.rate among different levels of p. speed (Appendix 4) showed high significant difference (p < 0.05) between each of the two levels of p. speed for both varieties of pumpkin. General comparison of the two products showed a higher LnP.rate of Jarrahdale. This could be attributed to lower effective cutting parameters including cutting force, rupture force, and shear force of the Jarrahdale’s skin. Also it may be because of higher skin toughness of Jap, as discussed in Chapter 3. The difference of the LnP.rate at 850 rpm (0.29) was greater than at 400 rpm (0.11) which confirms that the difference of the peeling rate (LnP.rate) between the two varieties was greater for higher levels of p. speed. It means the slope of increasing trend for the Jarrahdale was higher than for the Jap. The increase of p. speed provides the same amount of impact force for both varieties but the Jarrahdale did show higher response when increasing the slope of trend in higher levels of p. speed. It means the resistance of skin tissues against elastic and plastic deformation for the Jarrahdale decreases more at higher levels of p. speed compared to the Jap.
7.5.1.1 The effect of p. speed on LnP.rate for different levels of coarseness
The general effect of mean p. speed on the mean LnP.rate for different mean coarseness of brushes was similar for both varieties (Figure 7.4). The removal rate of peel in the form of LnP.rate increased in order of the following coarseness: very coarse, mild, coarse, and fine (Figure 7.4) for both products. The pattern of increase was almost linear for mild and coarse brushes.
The higher removal rate of fine and coarse brushes means the oblique angle of teeth helps to remove more peel than the other two brushes. The existence of trends for different types of coarseness of abrasive-cutter brush from 550 to 700 rpm could be seen clearly. The output difference out of this range of speed increased especially for lower speeds (400 rpm). In addition to increasing trend for growing speed all coarseness levels except mild did show an extra increase of LnP.rate while illustrating a convex point at 550 rpm for Jarrahdale. The fine brush did show a subsidence at 700 rpm for both products. The comparison of the very coarse type of brush showed that it can produce a close output to the other types of brushes
123 between 550 and 700 rpm for both varieties and it considerably subsided at 400 and 850 rpm.
2
Jap
) 1.5 Jarrahdale 1
0.5
LnP. rateLnP. (gr/min 0 400 550 700 850 -0.5 P. speed (rpm)
Fig.7.3. The effect of mean p. speed on LnP.rate
2.00 Coarseness 2.00 Coarseness Very coarse Very coarse Mild Mild 1.50 Coarse 1.50 Coarse Fine Fine
1.00 1.00
0.50 0.50 Mean LnP.rate Mean LnP.rate 0.00 0.00
-0.50 -0.50
-1.00 -1.00
400.00 550.00 700.00 850.00 400.00 550.00 700.00 850.00 P. speed Pspeed Jarrahdale Jap
Fig. 7.4. The effect of p. speed on LnP.rate at different levels of coarseness
124 7.5.1.2 The effect of p. speed on LnP.rate in different locations of product
Both varieties were affected similarly by p. speed at different parts of product (Figure 7.5). While a general growing trend of peeling rate for increasing p. speed was seen at all parts of the two varieties of pumpkin, except at the top areas that illustrated a peak point at 550 rpm, which reveals high peel removal at this speed. This result confirms selection of 550 rpm as the optimal angular velocity of the peeler head equipped with the abrasive-cutter brush in the previous chapter (Chapter 6).
2.00 Location Location Bottom Bottom Top Top Bottom-side 1.50 Bottom-side 1.50 Top-side Top-side
1.00 1.00
0.50 0.50 Mean LnP.rate Mean LnP.rate 0.00 0.00
-0.50
-0.50 -1.00 400.00 550.00 700.00 850.00 400.00 550.00 700.00 850.00 Pspeed Pspeed Jarrahdale Jap
Fig. 7.5. The effect of p. speed on LnP.rate at different locations of pumpkin
The skin removal at the top and top-side areas varied at different levels of p. speed for both varieties. While the peel removal at these areas was approaching each other at 400 and about 700 rpm, the difference increased at 550 and 850 rpm. Lower removal at the top areas for 550 rpm was the cause of difference because the LnP.rate for 550 rpm of p. speed at top areas did not follow the shape of other locations at this point. The peel removal at top side areas became smaller than at top points at higher levels of p. speed for both products. This reduction happened for
125 Jap and Jarrahdale before and after 700 rpm respectively. Although the depth and width of grooves at the top side areas are considerably bigger than at other areas of both products, lesser peel removal in this area than at top points for higher speeds could be explained in different ways. The most likely reason is reduction in coverage area of the concave points at top side areas for higher p. speed. The higher radius of rotation for those points compared with smaller radius at the top points caused higher circumferential speed and decreased the access to the grooves by the abrasive-cutter brushes.
7.5.2 The effect of coarseness on LnP.rate
The comparison mean of LnP.rate as one transformed type of peeling rate for different coarseness levels of brush revealed an almost linear relationship (Figure 7.6). The increasing trend of LnP.rate was found for both products in the same order for very coarse, mild, coarse, and fine respectively. The Jarrahdale variety showed significantly higher LnP.rate than the Jap in all types of coarseness of brush. The lower rate of peeling rate for the Jap variety again emphasized the greater toughness of skin that demands higher cutting, shearing, and rupture forces compared with the Jarrahdale. Although the results showed that all types of brushes with different coarseness levels have toughness greater than skin toughness of the two products, the production of peel losses per unit time becomes higher for the finest skin of Jarrahdale. In addition fine and coarse brushes with lower toughness showed a higher LnP.rate. This was mainly because of the oblique position of the teeth in these types of brushes. More comparison of the results also revealed higher output of the fine than the mild brush and also of the coarse than the very coarse brush. It means the size of protrusions is effective on the LnP.rate. The smaller size could increase the LnP.rate for both varieties. The mean of LnP.rate for the fine type of brush was significantly (p < 0.05) bigger than for the coarse type (Appendix 4) and this was in agreement with the previous obtained results (Chapter 6). The reasons for this result could be explained as more flexibility and capability of penetration into concave areas by smaller sizes such as the fine brush. Despite those acceptable results the coarse brush type revealed significantly (p < 0.05) higher LnP.rate than the mild type of brush. The results of applying the mild brush showed expected non significant higher peel removal compared to the very coarse type of brush. Results
126 generally showed the increase of the peeling rate when reducing the coarseness of brush, irrespective of the position of the coarse type of brush.
1.2 Jap 1 Jarrahdale 0.8
0.6
0.4
LnP.rate (gr/min) 0.2
0 Very coarse Mild Coarse Fine Coarseness
Fig.7.6. The effect of mean coarseness on LnP.rate
7.5.2.1 The effect of coarseness on LnP.rate at different p. speed
The general form of an increasing trend of different p. speed levels for both varieties of pumpkin except at 550 rpm for the Jarrahdale variety was observed (Figure 7.7). The mild brush at 550 rpm did show the opposite trend. The curvature change of increasing trend at this point to concave for the Jarrahdale revealed lesser peel removal than the expected LnP.rate. Higher mean value of the LnP.rate than expected normal growth of trend at 550 rpm was easily revealed except for the mild brush on the Jarrahdale. The mean of LnP.rate at 400 rpm was similar for different brushes for both varieties of pumpkin. That means this speed is not sufficient to impose necessary cutting force for notable penetration and peel removal on different product. The idea was proved by the revelation of a big difference between the results at 400 rpm and the other higher angular velocities of every type of brush. The increase of the p. speed above 400 rpm led to higher peel removal in the Jarrahdale compared to the Jap. Both products showed a sharp increase of peel removal from course to mild at 400 and 850 rpm. It means considerable reduction
127 of coarseness and increasing p. speed can significantly affect the peeling rate in the form of the LnP.rate.
2.00 Pspeed 2.00 Pspeed 400.00 400.00 550.00 550.00 700.00 1.50 1.50 700.00 850.00 850.00
1.00 1.00
0.50 0.50 Mean LnP.rate Mean Mean LnP.rate 0.00 0.00
-0.50 -0.50
-1.00 -1.00
Very coarse Mild Coarse Fine Very coarse Mild Coarse Fine Coarseness Coarseness Jarrahdale Jap
Fig.7.7. The effect of coarseness on LnP.rate at different speeds of abrasive-cutter brush
7.5.2.2 The effect of coarseness on LnP.rate at different locations of pumpkin
The reduction of the coarseness of brush generally caused an increase of the LnP.rate as the peeling rate at different areas of both products. The form of increasing trend was different between the two varieties at different areas except the top. All types of brushes revealed higher peel removal at the top areas compared to the bottom except the very coarse brush for the Jap variety (Figure 7.8). This is because of the existence of grooves (deep and thin in some products) at top areas compared with a flat surface at the bottom. Therefore, the interaction area for each removal impact of brush and the peel removal in total was higher at top areas. For the same reason the removal rate of coarse and very coarse grades at top areas was lower. This reduction clearly can be seen for the Jap variety because of the tougher skin of the Jap compared with the Jarrahdale. Higher mean of LnP.rate at the top- side areas than the bottom-side can be explained by considering the larger depth of grooves at the top side areas. It means the possibility of more effective peeling for
128 each impact of all types of brushes at top side areas. The mean LnP.rate at the bottom side areas of the Jap was higher than at the top side for coarse and fine types of brush.
1.40 Location 1.25 Location Bottom Bottom Top 1.20 Top Bottom-side 1.00 Bottom-side Top-side Top-side 1.00 0.75 0.80
0.50 0.60 Mean LnP.rate Mean Mean LnP.rate 0.40 0.25
0.20 0.00
0.00 -0.25 Very coarse Mild Coarse Fine Coarseness Very coarse Mild Coarse Fine Coarseness Jarrahdale Jap
Fig.7.8. The effect of coarseness on LnP.rate at different locations of product
7.5.3 The effect of location of product’s surface on LnP.rate
The mean of LnP.rate of the Jarrahdale variety was higher than that of the Jap variety at different locations on the product. The difference was significantly bigger (more than two times) at top and bottom areas (Figure 7.9). The results of the post- hoc test (Appendix 4) showed significant difference (p < 0.05) between the bottom and side areas. The results revealed no difference between the top and other areas of product. Study of the range of change between the maximum and minimum mean of LnP.rate at all areas for each variety showed a small difference. The difference for the Jarrahdale (0.63 g/min) was higher than for the Jap (0.58 g/min). This can be attributed to generally lesser skin toughness of the Jarrahdale variety that results in more peel removal per unit time compared with the Jap. Comparison of two products did show also two other important differences.
129 1.2
) Jap 1 Jarrahdale 0.8 0.6 0.4
LnP. rateLnP. (gr/min 0.2 0 Bottom Top Bottom- Top-side side Location
Fig.7.9. The effect of mean location on LnP.rate
The existence of a concave area at the top of the Jap variety and also the small difference of mean LnP.rate between the top and bottom side areas of Jap could be identified in Figuere 7.9. It means the Jap variety of pumpkin has more similarity in curvature at side areas and also between top and bottom areas.
7.5.3.1 The effect of location on LnP.rate in different coarsenesses of brush
Higher peel removal rate as LnP.rate for the Jarrahdale variety compared with the Jap can be seen in Figure 7.10. While both products showed the largest difference at the top area for different coarseness of brush, the smallest difference of the Jarrahdale and Jap was observed at the bottom-side and top-side areas respectively. The effect of the mild brush could only produce similar form of increasing trend for both products. In other words, the variation of the effect of different types of brushes on peel removal at different locations of both varieties was considerable. While the very coarse brush, as a less effective brush, produced more LnP.rate at the top area of the Jarrahdale than bottom, the results for the Jap variety were opposite with small difference. The fine type of brush, as a more effective brush, produced higher peeling rate at side areas. The LnP.rate was high at the top side area for the Jarrahdale and at the bottom side area for the Jap. Although there was no significant difference between sides areas in both varieties (Appendix 4), side
130 areas of products were found to be similar especially for fine and mild brushes on the Jarrahdale and mild and coarse on the Jap.
1.40 Coarseness 1.25 Coarseness Very coarse Very coarse Mild Mild 1.20 Coarse 1.00 Coarse Fine Fine 1.00 0.75
0.80 0.50 0.60 Mean LnP.rate Mean Mean LnP.rate Mean 0.25 0.40
0.00 0.20
0.00 -0.25
Bottom Top Bottom-side Top-side Bottom Top Bottom-side Top-side Location Location
Jarrahdale Jap
Fig.7.10. The effect of location on LnP.rate in different coarseness
7.5.3.2 The effect of location of product’s surface on LnP.rate at different p. speed
The increase of p. speed showed a higher LnP.rate for both varieties (Figure 7.11). The peel losses as LnP.rate appeared to be higher for the Jarrahdale compared with the Jap. Peeling at different speeds of abrasive-cutter brush revealed that higher and lower values of mean LnP.rate difference take place at the bottom-side and top areas for both products respectively. All p. speed levels showed a lower LnP.rate at the bottom area and a maximum in one of the side areas. The biggest LnP.rate at the bottom side area was for the Jap variety at 850 rpm.
131 2.00 Pspeed Pspeed 400.00 400.00 550.00 550.00 1.50 1.50 700.00 700.00 850.00 850.00
1.00 1.00
0.50 0.50 Mean LnP.rate Mean LnP.rate 0.00
0.00 -0.50
-0.50 -1.00
Bottom Top Bottom-side Top-side Bottom Top Bottom-side Top-side Location Location
Jarrahdale Jap
Fig.7.11. The effect of location on LnP.rate at different p. speeds
7.6 Conclusions and discussion
Mechanical peeling of two varieties of pumpkin including Jarrahdale and Jap was experimentally investigated by using an abrasive-cutter brush. The tests were carried out using full factorial design. The effect of three independent variables, namely, the angular velocity of brushes (four levels), coarseness of brushes (four levels), and location of peeling on product (four levels) on LnP.rate was investigated as the transformed dependent variable of peeling rate (g/min).
The increase of mean p. speed led to a linear increase of the mean LnP.rate (g/min). This result was expected because increasing the p. speed leads to higher energy and frequency of the impacts of brushes on the surface of product. The results of multi comparisons of the LnP.rate among different levels of p. speed showed high significant difference (p < 0.05) between each two levels of p. speed for both products. The comparison of the two products showed higher LnP.rate of the Jarrahdale compared with the Jap. This could be attributed to lower effective cutting parameters including cutting force, rupture force, and shear force of the Jarrahdale
132 skin. It might also be because of higher skin toughness of the Jap and these aspects were already discussed in Chapter 3. The increase of p. speed supplies the same amount of impact force for both varieties but the Jarrahdale did show more response with increases of the curve slope at higher levels of p. speed. It means the resistance of skin tissues against elastic and plastic deformation for the Jarrahdale decreases more at higher levels of p. speed compared to the Jap.
The mean comparison of LnP.rate as one transformed type of peeling rate for different coarseness of brush revealed an almost linear relationship. The increasing trend of LnP.rate was observed for both products in the same order including very coarse, mild, coarse, and fine brushes respectively. The Jarrahdale variety showed significantly higher LnP.rate than the Jap in all types of coarseness of brush. The lower rate of peeling rate of the Jap variety again emphasized the greater toughness of skin that requires higher cutting, shearing, and rupture forces compared with the Jarrahdale. Fine and coarse brushes with lower toughness showed a higher LnP.rate. This was mostly because of the oblique position of teeth in these types of brushes. More comparison of the results also revealed higher output of the fine than mild brushes and also of the coarse than the very coarse brush. It means the size of protrusions impacts on the LnP.rate. The smaller size could increase the LnP.rate for both varieties. The mean of LnP.rate for the fine type of brush was significantly (p < 0.05) bigger than the coarse type and this was in agreement with the previously obtained results. This could be attributed to higher flexibility and capability of penetration into concave areas by smaller sizes of protrusions such as in the fine brush. Results generally showed that with the increase of peeling rate the coarseness of brush is reduced, ignoring the position of the coarse type of brush.
The mean of LnP.rate of the Jarrahdale was higher than the Jap at different locations on the product. The difference was significant (more than two times) at the top and bottom areas. The results of the post-hoc test showed significant difference (p < 0.05) between bottom and side areas. The results revealed no difference between the top and other areas of product. Study of the range of change between the maximum and minimum mean of LnP.rate in all areas for each variety showed a small difference. The difference for Jarrahdale (0.63 g/min) was higher than for the Jap (0.58 g/min). It can be explained by the generally less tough skin of
133 the Jarrahdale producing more peel removal per unit time compared with the Jap. The existence of a concave zone at the top area of the curve of the Jap variety and also small difference of mean LnP.rates between the top and bottom-side areas of the Jap showed that this variety of pumpkin has more similarity in curvature at side areas and also between top and bottom areas.
7.7 Summary
Abrasive-cutter brush as the best peeling method of tough-skinned vegetables was comprehensively investigated. Full factorial experiment design was applied in peeling experiments of two varieties of pumpkin (Jap and Jarrahdale). The experiments were carried out in 128 runs (64 runs for each variety of pumpkin). Three effective independent variables related to either the peeling tool (p. speed and the roughness of brush) or product (the location of peeling on product) were selected for investigation. The influence of these significant parameters on LnP.rate as logarithmic transform of peeling rate was studied. The results showed that LnP.rate is increasing continuously with the increase of p. speed. The effect of the brush roughness as a categorical variable on peeling rate was investigated in four levels. The response variable, LnP.rate, increased in order of coarse, mild, coarse and fine. Another categorical variable was location. The surface of the whole pumpkin was divided into four sections named “Top”, “Top-side”, “Bottom-side” and “Bottom”. These areas were selected as four levels of location of peeling. It was revealed that the LnP.rate increases in order of bottom, top, bottom-side and top- side. The difference was attributed to the number and size of grooves in these areas.
134 Chapter 8
Modelling of mechanical peeling as sum of consumed energy in the peeling process
8.1 Introduction
Modelling of the results can be illustrated in different ways. Mechanical peeling is carried out mostly on the basis of cutting forces. The simulation of mechanical peeling on the basis of the cutting process helps to better understand the effective forces of peeling and the role of different influence parameters of product and peeling tool. Mechanical peeling using an abrasive-cutter brush was simulated on the basis of the cutting process. Choosing the input and output variables which would be industrially applicable was attempted. Three variables, namely, angular velocity of abrasive-cutter brush (ωp), the degree of unevenness of product surface (φ), and the shape of the abrasive-cutter brush (λ), were chosen as independent variables and the peel losses per unit time were chosen as the output of the model. The developed model was verified using the experimental results of peeling for two varieties of pumpkin including Jarrahdale and Jap. In this chapter the theory of a mathematical model based on the cutting procedure of the fibrous material is discussed. This model is developed and investigated for the first time.
135
8.2 Theory of the model
8.2.1 The assumptions
The following assumptions were applied:
1. Removing peel is assumed to occur in layers and in the form of chips. 2. Peeling rate is in linear proportion to peeling energy. 3. The angular velocity of product is assumed to be zero. 4. The size and the weight of products are assumed to be the same and constant for products of each variety.
8.2.2 Development of the model
The cutting procedure and removing the peel can be split into two main stages including fracture of the skin and scratching along removing the peel as formed chips. The total energy spent on cutting and removing the skin can be written as given below:
1 Pt = (P1 + P2 ) , (8-1) ηc
where, Pt is the total required power of peeling in N. mm/min; P1 and P2 are the required power for cutting and forming removed skin, respectively, in N. mm/min; and ηc is total peeling efficiency.
The energy consumed at the fracture stage itself is spent to penetrate the abrasive- cutter brush (teeth) inside the skin (Figure 8.1). The penetration depth depends on the stroke force developed by the rotational kinetic energy of the brush and neglecting the air resistance. The total penetration energy can be calculated as given below:
P1 = nω p (E p + Ed ) , (8-2)
136 where, P1 is the total fractural power per unit time in N. mm/min; n is the installed number of brushes on the peeler head; ωp is the angular velocity of the brush in rpm;
Ep and Ed are the necessary penetration and deflection energy of one brush in N. mm respectively.
δ2 δ3
Without deflection
Rotation centre of brush Product After deflection
Fig.8.1.The view of abrasive-cutter brush after penetration into the skin
The energy required for penetration of one brush through the skin, Ep, is given below:
E p = K1(Vip .t1 − δ1) ×δ 2 , (8-3)
-1 where, K1 is the average shearing resistance per unit length of stroke in N.mm ; Vip is the linear penetration velocity of brush teeth inside the skin in mm/s; t1 is the time of stroke in s; δ1 is the deflection of product in mm; and δ2 is the average penetration depth of teeth inside the skin in mm. The deflection of the product in reaction of stroke, δ1, is neglected because of rigid support of the vegetable holder which allows the simplifying of equations. The penetration depth of the teeth of the brush inside the skin also depends on the linear velocity of the teeth through skin and the time of stroke as given below:
137
Vip ⋅t1 = δ 2 , (8-4)
The coefficient K1 is considerably affected by the ratio of the material’s toughness and other stated parameters according to the following equation:
γαl1πd1τd2 4l2 sinθ1 K1 = , (8-5) δ 2
where, γ is the ratio of product toughness (Tp) to the toughness of tool (Tt) as follows:
T γ = p (8-6) Tt
2 α is the density of protrusions on a brush in number/mm ; l1 is the effective length
(covered by abrasive strip) of the brush in mm; d1 is the diameter of the brush in mm; 2 τ is the shear strength of the product in N/mm ; d2 is the diameter of the protrusion’s hole in mm; l2 is the length of each tooth on protrusion in mm; and θ1 is the angle of the teeth in protrusion in degrees (Figure 8.2).
d2
θ1
l2
Fig.8.2. The cross-sectional view of one protrusion (two out of four teeth are shown)
Each brush stroke is accompanied by its deflection. In ideal conditions, the end of the brush will show a deflection of δ3 because of reaction to the stroke. The average expenditure energy for this deflection when considering a brush as a cantilever beam can be written as follows:
138
3EIδ E = 3 δ , (8-7) d L3 3
where, Ed is the expenditure energy of one brush due to deflection in N. mm; E is the modulus of elasticity of the brush in N.mm-2; I is the geometric moment of inertia of 4 the brush in mm ; L is the whole length of the brush in mm; and δ3 is the average deflection of the brush at the fracture stage in mm. As the deflection of the brush varies from zero at the root to the maximum at the end of brush and the resistance of the brush against bending is linearly proportional to deflection, the average deflection is considered as estimated below:
0 + δ δ = 3max , (8-8) 3 2
where, δ3max is the maximum deflection of the brush in the fracture stage in mm. Therefore, replacing coefficients in 8-2 leads to the final form of required power in the fractural stage of peeling as given below:
3EIδ 2 P = nω (4πγαl l d d δ τSinθ + 3 ) , (8-9) 1 p 1 2 1 2 2 1 L3
The second stage of energy is spent after fracture of the peel. This energy is required to scratch and remove the skin in chip form. The total power expenditure at this stage can be written as follows:
P2 = Fc ⋅Vop , (8-10)
where, P2 is the total power required at the second stage of peeling in N. mm/min; Vop is the linear velocity of scratching teeth inside the skin in mm/s; and Fc is the cutting force in N. The cutting force of fibrous material such as pumpkin is comprised of three effective forces (Dowgiallo, 2005) as given below:
Fc = Ff + Fe + Fd , (8-11)
139 where, Fc is the total cutting force in N; Ff is friction force in N; Fe is force spent for elastic and plastic deformation in N; and Fd is disintegration force exerted by brush teeth on the product structure in N. Ff and Fe as two important effective forces will be included in detail in the model. The expenditure energy due to Fd is released mostly as heat and depends significantly on some parameters such as the geometrical dimensions of teeth, the cutting speed, and resistance of product to cutting. Fd will be included in the model as part of the efficiency of cutting in this stage. The energy spent by Ff for one brush can be written using the law of friction as given below:
E f = K2 × Ff × h , (8-12)
where, Ef is the energy spent on friction in N. mm; Ff is the friction force in N; h is the length of removed peel in mm; and K2 is the friction coefficient related to the properties of the product and geometrical parameters of brush. This coefficient can be written on the basis of the most significant parameters as follows:
πl1d1αδ 2 K2 = ϕ , (8-13) d2
2 where, K2 is the friction coefficient; α is the density of protrusion in number/mm ; and φ is the degree of unevenness of the product surface. The degree of unevenness increases in order of bottom, top, bottom-side, and top-side for both products. The force required to overcome friction is estimated using the following relationship:
Ff = μd ⋅ Rv , (8-14)
where, Ff is the friction force in N; μd is the dynamic coefficient of friction between the brush’s tooth and product; and Rv is the total normal reaction in N, that is complimented from two main important forces as follows:
Rv = N + Fde , (8-15)
140 where, Fde is the deflection force of brush in N; and N is the normal reaction force to the weight of the brush in N, that can be calculated as:
N = W1 cosθ2 , (8-16)
where, W1 is the weight of one brush in gr; and θ2 is the angle between direction of the weight and direction of the line that passes through the gravity centre of brush and is perpendicular to the surface of product at the contact point. With the replacing of 8-15 and 8-16 into 8-14 the total friction force can be obtained as follows: 3EIδ F = μ (W cosθ + 4 ) , (8-17) f d 1 2 L3
where, δ4 is the average deflection of brush at the second stage of cutting in mm. This deflection should be bigger than the deflection at the previous stage, δ3, due to passing of the product by the brush.
The total energy expenditure of one brush by friction force can be written as:
3EIδ E = hK μ (W cosθ + 4 ) (8-18) f 2 d 1 2 L3
The force spent on elastic and plastic deformation is another effective force at the second stage of cutting. As fibrous materials are not less stiff than friable or crystalline materials (Dowgiallo, 2005), this force forms a considerable part of cutting force at the second stage. The elastic and plastic deformation force can be determined as the equation given below:
Fe = K3 ×τ × h × l3 , (8-19)
where, Fe is the total elastic and plastic deformation forces of one brush in N; τ is the 2 shear strength of product in N/mm ; h is the length of removed peel in mm; and l3 is the total projected lengths of the protrusion’s teeth engaged in cutting in mm. Figure 8.2 shows the top view of protrusion (in direction of cutting). As the direction of movement and cutting are not the same (because of angle θ1) then the projected
141 length (l3) should be considered in the above equation to calculate the work done by this force. With consideration of Figure 8.2 the length of l3 can be replaced with:
l3 = 2l2 cosθ1 (8-20)
The coefficient of K3 depends on some geometrical parameters of the brush as stated below:
K3 = πl1d1α (8-21)
Therefore, the total elastic and plastic deformation energy can be given as follows:
2 Ee = 2π ⋅l1 ⋅ d1 ⋅α ⋅τ ⋅ h ⋅l2 cosθ1 (8-22)
With consideration of the disintegration force as a coefficient for both Ef and Ee and adding up 8-18 and 8-22, the total energy expenditure at the second stage of cutting could be represented as given below:
⎡δ2μdϕ ⎛ 3EIδ4 ⎞ ⎤ E2 = K4hπl1d1α ⎢ ⎜W1 cosθ2 + 3 ⎟ + 2.τ.hl2 cosθ1⎥ , (8-23) ⎣ d2 ⎝ L ⎠ ⎦
where, E2 is the total required energy of peeling in the second stage in N. mm; and K4 is the coefficient of disintegration force of the product structure. The equation 8-10 of required power for scratching and removing peel at the second stage of cutting can be rewritten as given below:
P2 = K5E2 , (8-24)
where, P2 is the total required power for one brush in the second stage of peeling in N. mm/min; and K5 as the scratching coefficient at the second stage of peeling could be expressed as:
142
ωv K5 = nβ , (8-25) ω p
where, K5 is the scratching coefficient in the second stage in number/min; ωv is the angular velocity of the vegetable holder in rpm; and β is the number of scratches in number/min. The length of scratching or efficiency is dependent on ωv/ωp. Also, β is considered due to the reason that a stroke of brush does not necessarily lead to scratching. Therefore, the total required power of peeling at the second stage of cutting will be given as follows:
ωv ⎡δ 2μdϕ ⎛ 3EIδ 4 ⎞ ⎤ P2 = K4 nβhπl1d1α ⎢ ⎜W1 cosθ2 + 3 ⎟ + 2⋅τ ⋅hl2 cosθ1⎥ (8-26) ω p ⎣ d2 ⎝ L ⎠ ⎦
The energy required for the abrasive-cutter brush could be obtained by integrating the component models stated by the above equations. The integration gives the final equation as follows:
2 1 3EIδ3 Pt = nω p (4πγαl1l2d1d2δ 2τSinθ1 + 3 ) + ηc L
K4 ωv ⎡δ 2μdϕ ⎛ 3EIδ 4 ⎞ ⎤ nβhπl1d1α ⎢ ⎜W1 cosθ2 + 3 ⎟ + 2⋅τ ⋅hl2 cosθ1⎥ , (8-27) ηc ω p ⎣ d2 ⎝ L ⎠ ⎦
where, Pt is the total required power of peeling in N. mm/min.
As the cutting force of fibrous material is directly in relation to the resulted deformation during cutting (Dowgiallo, 2005), it can be assumed that the peeling rate also should be in direct relation to the required power of cutting. Assuming a linear relationship between the peeling rate and the required power of peeling leads to the following equation:
P.rate = K6 ⋅ Pt , (8.28)
143 where, P. rate is the peeling rate during peeling in gr/min; and K6 is the transform coefficient of Pt to p. rate in g/N. mm. The integration of 8.27 and 8.28 will show the final equation of peeling rate during peeling as given below:
2 1 3EIδ3 P.rate = K6nω p (4πγαl1l2d1d2δ 2τSinθ1 + 3 ) + ηc L
K6K4 ωv ⎡δ 2μdϕ ⎛ 3EIδ 4 ⎞ ⎤ nβhπl1d1α ⎢ ⎜W1 cosθ2 + 3 ⎟ + 2⋅τ ⋅hl2 cosθ1⎥ (8.29) ηc ω p ⎣ d2 ⎝ L ⎠ ⎦
The input of the obtained model includes many parameters related to the product and the abrasive-cutting brush. As determination of all effective parameters is impossible at this stage, it was attempted to rewrite and arrange the above model using industrially applicable input and output variables. The review of effective parameters regarding the results of the previous chapter showed three likely independent variables. They are the angular velocity of brush (ωp), the unevenness of product surface (φ), and cosθ1 that represents the shape of the brush and actual protrusion of the brush and it is denoted as (λ). The output of model is kept as p. rate and rewriting equation 8.29 by using factorial technique on the basis of these three independent variables will show the general format of model as given below:
P.rate = C0 + C1ω p + C2ϕ + C3λ , (8.30)
where, the model coefficients C0 to C3 are as follows:
1 ωv 3EIδ4 C0 = K6K4nβhπl1d1α 3 ; (8.31) ηc ω p L
2 1 3EIδ3 C1 = K6 (4nπγαl1l2d1d2τδ2 sinθ1 + 3 ) ; (8.32) ηc L
1 ωv nβhπl1d1αδ 2μd 3EIδ 4 C2 = K6 K4 (W1 cosθ2 + 3 ) ; (8.33) ηc ω p d2 L
144
1 ωv 2 C3 = 2K6K4 nβhπl1d1ατh l2 . (8.34) ηc ω p
8.2.3 Determination of the model coefficients
The accuracy of prediction of the model depends significantly on the accuracy of model’s parameters. Some of the parameters are known (e.g. l1, l2, d1, d2, etc.) and some of them could be determined from specially designed experiments. A few mechanical properties from the latter part (e.g. Tp, τ, etc.) and some related parameters that did not appear directly in the model have been determined earlier by this research and the others need to be determined using special experiments (e.g. β,
δ3 ). Direct measurement of all appeared parameters in the model would result in accurate estimation of the abrasive-cutter brush operation. Due to the impossibility of carrying out separate direct measurements, coefficients of the model were determined indirectly using experimental data and based on the multiple regression analysis technique. The obtained data from full factorial experiments on the peeling of two varieties of pumpkin, Jarrahdale and Jap, by abrasive-cutter brush were used (Chapter 7). The dependent variable as p. rate was selected to be a function of angular velocity of brushes (ωp), type of brush (λ), and unevenness of product surface (φ) as three independent variables.
Regarding the general linear function of the model, a multiple regression analysis was carried out for both varieties of pumpkin to determine four coefficients. The SPSS software package (version 13) was used for the analysis.
8.2.4 Model validation
A portion of the data for each experiment (12.7% for Jarrahdale and 12.9% for Jap) was not used for determination of the model’s coefficients. Those randomly selected data were used for model validation. The experimental data of p. rate were compared with the corresponding predicted p. rate for each variety of pumpkin using a scattered plot. The models were considered validated if the following criteria were satisfied:
145
1. The intercept of the linear regression analysis between the modelled and experimental values should be close to zero. 2. The coefficient of linear regression analysis between the modelled and experimental values should be close to unity. 3. The correlation coefficient between the modelled and experimental values should be statistically significant.
8.3 Results and discussion
The proposed mathematical model for mechanical peeling of tough-skinned vegetables, using abrasive-cutter brushes, is a linear type and has four coefficients.
There are three independent variables, including, ωp, unevenness of product surface (φ), and the type of brush (λ). The coefficients of those variables include parameters related to the properties of the product and the brush as a peeler. The coefficients were determined indirectly by using experimental data on peeling two different varieties of pumpkin namely Jarrahdale and Jap. The experiments were carried out strictly under the same conditions for both products. Multiple regression analysis was used for determination of coefficients and the results are presented below.
8.3.1 Model coefficients
The estimated values of model coefficients along with other results of multiple regression analysis are shown in Table 8-1 and the complete results are reported in Appendix 8-1. The results revealed that those three coefficients of the model for both varieties of pumpkin are highly significant (p < 0.0001). Despite the highly significant effect of these independent variables still they could explain 88% (R2 = 0.881) of the variation in dependent variables for the Jarrahdale and about 89% (R2 = 0.894) for the Jap. Regarding the relatively small number of variables (three independent variables), the R square is sufficiently satisfied. But the difference between the full value of R square showed there are still some other effective variables on peeling rate resulting from peeling by an abrasive-cutter brush. Their significance varies with their number. If remaining variables are more than one, then their effectiveness would be smaller. However, the obtained value (almost 0.90) of
146 coefficient of determination (R2) can provide a realistic estimation of effective parameters on the mechanical peeling of both products.
Table8-1.The results of multiple regression analysis for coefficients of two mechanical peeling models Model coefficients 2 Product C0 C1 C2 C3 R F Sig. Jarrahdale -4.239 0.007 0.487 0.506 0.881 106.559 0.000 Jap -3.088 0.005 0.405 0.410 0.894 123.867 0.000
The results also showed the significance of each independent variable in the model for both products. As all variables are highly significant (p < 0.0001), it means that all inserted parameters as coefficients to the model can significantly affect the peeling rate. The general comparison of all coefficients revealed that all coefficients (except the coefficients C0) are higher for the Jarrahdale than for the Jap. The existence of a positive linear function between each independent variable and output showed that the increase of ωp, φ, and λ leads to higher values of peeling rate.
The value of C1 was higher for the Jarrahdale (0.007) than for the Jap (0.005). This indicates that the energy requirement for the penetration of the brush’s teeth into the skin of the Jarrahdale was greater than that for Jap at the first stage of the cutting process. Due to identical conditions of experiments, the reason could be related to the two important mechanical properties of pumpkin, γ, which appeared in this coefficient. Considering the same toughness of the abrasive-cutter brush for both varieties, the cause of difference should be found at the numerator of the ratio of toughness (γ). Although the skin toughness of the Jarrahdale (13.87 N.mm) is less than the skin toughness of the Jap (33.10 N.mm), the higher unpeeled toughness of the Jarrahdale (719.72 N.mm) than that of the Jap (702.73 N.mm) could affect the required power of peeling at that stage.
The comparison of the coefficient C2 showed that this coefficient for the Jap (0.405) is smaller than that for the Jarrahdale (0.487). Again, considering the same conditions of peeling for both varieties, the cause of difference should be found in properties of
147 the product. The most significant parameters in this coefficient are μd, h, and β.
Although the value of dynamic coefficient of friction (μd) has not been reported for those varieties, the values of static coefficient of friction have been measured in this research. Although the static coefficient of friction of the Jap on stainless steel in the unpeeled state (0.63) and in the state of being without periderm (0.76) are higher than that of the Jarrahdale in the unpeeled state (0.30) and without periderm state (0.61), the dynamic coefficient of friction has not been measured. The length of removed peel for each scratch (h) appeared in coefficient C2. The direct effect of h on peeling rate means producing bigger h needs more required energy of peeling or the tissue structure of product is looser. The comparison of mechanical properties supports this relationship. Lower value of properties of the Jarrahdale such as skin thickness (13.87 N.mm), shear strength of skin (2.72 N/mm2) and unpeeled (1.78 N/mm2), cutting force of skin (2.82 N) and unpeeled (5.15 N) and also rupture force of skin (40.68 N) and unpeeled (248 N) compared with the Jap with skin toughness (33.10 N. mm), shear strength of skin (3.29 N/mm2) and unpeeled (2.42 N/mm2), cutting force of skin (2.41 N) and unpeeled (10.99 N) and rupture force of skin (98.30 N) and unpeeled (249 N) are already revealed (Chapter 3). As φ appeared at the second stage of peeling, the skin properties make a heavier impact than unpeeled properties. One of the effective forces at the second stage that can affect peeling production by scratching was rupture force. Lower resistance of the Jarrahdale skin to rupture force allowed for production of longer h. The loose tissue structure of the Jarrahdale also could be significantly effective in increasing the value of β. The loose tissue has higher ability to absorb the strokes of brushes. The lower value of skin toughness of the Jarrahdale was an obvious reason for the increased ability to transform more strokes to scratches (β).
The type of brush that appeared in the model as λ had coefficient C3. The value of coefficient C3 was also revealed to be smaller for the Jap (0.410) than that value for the Jarrahdale (0.506). The related properties of product that can explain the difference in this coefficient were shear strength (τ) and the cubic value of the length of removed peel (h). The shear strength of the Jap in the states of skin (3.29 N/mm2) and unpeeled (2.42 N/mm2) are significantly bigger than the shear strength of the Jarrahdale in skin (2.72 N/mm2) and unpeeled (1.78 N/mm2) states. The effect of h on coefficient C2 is another cause of higher value of C3 for the Jap. This influence
148 because of cubic value of h can significantly affect the peeling rate more than the shear strength can.
The comparison of coefficients C0 showed the value for the Jarrahdale (-4.239) is significantly smaller than that coefficient value for the Jap (-3.088). The related parameters to the product that appeared in this coefficient were β and h. They could not explain this behaviour. This difference could be attributed to some parameters that are not represented in the model.
8.3.2 Model validation
The validation of the model was assessed using scattered plots between experimental and predicted values (Figure 8.3). The values of the regression coefficients showed about ±0.14 difference with unity (not parallel trade lines). It was 0.85 for the Jap and 1.14 for the Jarrahdale. The intercepts of the regression lines were nearly close to zero. The values of -0.05 and 0.01 were obtained in this case for the Jap and Jarrahdale respectively. Also the correlation coefficients between predicted and experimental values of the Jap (0.98) and Jarrahdale (0.96) were revealed to be statistically significant. Therefore, all assumed criteria for meeting validity were satisfied.
8 Jarrahdale
e Jap 7 Jarrahdale y = 1.1411x + 0.0121 6 R2 = 0.9372, R=0.96 5
4
3 Jap y = 0.8589x - 0.0581 2 R2 = 0.9669, R=0.98 1 Experimental values of p. rat of values Experimental 0 0246 Predicted values of p. rate
Fig.8.3. Experimental versus predicted values of p. rate (gr/min)
149
8.3.3 Applicability of the model
For the first time a generalised model was developed for mechanical peeling using abrasive-cutter brushes. The model was developed for the peeling of tough-skinned vegetables and its validity was assessed for two varieties of pumpkin, namely, Jap and Jarrahdale. For the first time the model was developed on the basis of the cutting process of mechanical peeling. The determined coefficients of models are valid only for the same conditions of experiments that were carried out in this research. The general effects of different parameters of the product and peeler on power requirement of peeling or peel losses can be extended to other tough-skinned vegetables especially those with the same shape. The behaviour of different parameters that were shown in the model as part of the cutting process could be applied in the design of peelers in different aspects. The range of application of angular velocity of brushes is from 400 to 850 rpm. The application range for the shape of brush is in four groups, namely, fine, mild, coarse, and very coarse with regard to the given toughness (90-170 N. mm) and geometrical parameters described in the experiment conditions. The model can show the effect and predict the value of peel losses at different places of pumpkin. Those places as described already are top, bottom, top-side, and bottom-sides of the pumpkin.
Some mechanical properties of product such as toughness, shear stress, and dynamic friction coefficient were considered in the model. As these properties vary even for products of one variety due to different growing conditions (e.g. soil, water, sunshine) and as this variation is not necessarily linear, the mean values of those parameters were considered in the model. However, the suggested values of coefficients are valid for the Jap and Jarrahdale varieties that are grown in Australia. The relationship among different parameters of the product and abrasive-cutter brush and their effect on the peeling rate could be used in industrial design of this innovative peeler and development of other peelers with similar operation.
150
8.4 Conclusions
Mechanical peeling of two varieties of pumpkin (Jap and Jarrahdale) was simulated and results were shown as two mathematical models. The output of modelling was peeling rate (p. rate) and the input arranged with three main independent variables, namely, the angular velocity of brushes (ωp), the degree of unevenness of product’s surface (φ), and the shape of brush (λ). The results showed all three independent variables can significantly affect the peeling rate. The relationship was revealed as linear. The results revealed that the lower value of mechanical properties of the Jarrahdale such as shear strength, cutting force, rupture force and skin toughness, caused higher values of model coefficients involving C1, C2, and C3 for the same conditions of experiments. The constant C0 was unexpectedly higher for the Jap than for the Jarrahdale.
8.5 Summary
Mechanical peeling using an abrasive-cutter brush was simulated for the first time by analysing the cutting process. The energy requirement of peeling was integrated for two stages including penetration of the brush’s teeth into the peel and scratching along the skin removing skin. Assuming a linear relationship of peeling rate with total power required for peeling led to the final shape of the output for the model. A large number of related parameters to the product and abrasive-cutter brush that appeared in the model were arranged on the basis of three independent variables using factorial technique. Those three industrially applicable factors were the angular velocity of brushes (ωp), the degree of unevenness of product surface (φ), and the shape of the brush (λ).The other variables that appeared in coefficients were assessed indirectly for two different varieties of pumpkin (the Jap and Jarrahdale). The results showed all considered parameters significantly affect the peeling rate. It was revealed that for the same conditions of experiments, mechanical properties of the product can significantly influence the results. Those effective parameters of the model related to the product were shear strength (τ), the ratio of toughness for product and brush (γ), the length of removed chip (h), and the dynamic coefficient of friction (μd). The results were in good agreement with the measured mechanical properties such as
151 toughness, shear strength, cutting force, and rupture force in different states of the two varieties.
Two derived mathematical models for both products were validated with experimental results. Those models could be used for the same conditions of experiments for these two varieties of pumpkin grown in Australia. The relationship among different parameters of the product and abrasive-cutter brush and their effect on the peeling rate can be used in industrial design of this innovative peeler and development of other peelers with similar operation.
152
Chapter 9
Conclusions and perspectives
In this chapter, the thesis is concluded by providing a summary of the thesis and the major findings of the research. Research related to this project which would require more investigation is identified.
9.1 Thesis summary and conclusions
The main purpose of this thesis was to develop innovative mechanical peeling methods for tough-skinned vegetables, using pumpkin as a case study, and modelling the influence of parameters related to the product and peeling tool. Through the course of this study the objectives set in section 1.2 have all been achieved.
An innovative mechanical peeling method, named as the abrasive-cutter brush, was developed and the role of significant related parameters on the operation of this method was shown in a mathematical model. The model has been developed on the basis of the cutting process for the first time and has industrial application in the design of peeling equipment.
Peeling is one of the most important preliminary stages of fruit and vegetable processing. Tough-skinned vegetables such as pumpkin and melon currently are peeled either semi-automatically or automatically. Circular shapes of rotating graters are applied in semi-automatic method. Segments of the product are brought into contact with the grater by an operator. This process is tedious and time consuming. In the latter method, whole pumpkins are passed through automatic machines where the floor is covered by many rotator disks (carborundum or blade). The main limitation of both methods especially for varieties with an uneven surface is high peeling losses.
153 A detailed literature review has been conducted and showed that current peeling methods include mechanical, thermal, and chemical methods and among these, the mechanical peeling methods are preferred as they possess some important features of the “ideal” method such as the maintenance of the freshness and integrity of the peeled product. The literature review revealed that no systematic research has been undertaken on mechanical peeling of tough-skinned vegetables. Generally the main cause of unsuccessful attempts to investigate peeling, including mechanical methods, was the low amount of attention paid to the properties of the product to be peeled. Therefore, to introduce a successful mechanical peeling method, the development of the peeling method in this research was carried out in four steps which were the study of mechanical properties of tough-skinned vegetables, trial of possible mechanical peeling tools, development of an innovative mechanical peeling method, and modelling of the influence of parameters of the product and peeling tool in the proposed method.
In the first step, some mechanical properties of tough-skinned vegetables were studied. There was no scientific definition of tough-skinned vegetables. Pumpkin and melon (three varieties each) were chosen as the case studies. The selected varieties were Jarrahdale, Jap, and Butternut for pumpkin; and Watermelon, Rockmelon, and Honeydew for melon. The values of toughness, rupture force, cutting force, shear strength force, shear strength, and static coefficient of friction of each variety were determined for the first time in different states of product including skin, flesh and unpeeled. Further, the contribution (%) of skin to different unpeeled properties was determined. The results were statistically compared to find the similarities and differences. The necessary range of different applied forces to peel vegetables was specified. It was also found that tough-skinned vegetables can be defined by the ratio of the skin’s shear strength to the same property of the unpeeled state.
In the second step, preliminary trials of different mechanical peeling methods (15 methods) were conducted. The results of the study of mechanical properties did help to formulate better ideas for different tools for the trials. The test rig that enabled testing of different mechanical peeling methods for different sized vegetables was designed and manufactured for the first time. Four methods involving abrasive pads, abrasive foams, a milling cutter, and an abrasive-cutter brush were developed for the first time and found to have potential industrial application as mechanical peeling methods.
154 Those four methods were compared for peeling the Jap variety of pumpkin using a fractural factorial design (Taguchi method). The criteria of comparison were higher and more even peeling efficiencies in different areas of vegetable with lower peel losses. The abrasive-cutter brush was found to be the best method compared to the other methods. Estimated results for the optimum combination of the variables were determined.
In the third step, further investigation was conducted to reveal the effects of different parameters on peeling by the abrasive-cutter brush. The experiments were carried out on two varieties of pumpkin, namely the Jap and Jarrahdale, using a full factorial design. Results showed that the peeling rate (gr/min) as the dependent variable is significantly affected by three independent variables which are the angular velocity of brushes, the coarseness of brushes, and the location of peeling on the pumpkin. The effects were discussed in detail in Chapter 7.
In the final step, mechanical peeling of the pumpkin using the abrasive-cutter brush was simulated. This was the first simulation that has been carried out on mechanical peeling of fruits and vegetables. The simulation was conducted on the basis of power required in different steps of cutting and removing peel. The results were modelled mathematically. The type and effect of different parameters related to the product and abrasive-cutter brush on the peeling rate were determined. The model was verified by experimental data of pumpkin varieties including Jap and Jarrahdale. The parameters were estimated indirectly for each variety and their effect was discussed in Chapter 8. The measured coefficients of the model proved the results of mechanical properties of the Jap and Jarrahdale that were obtained in step one. The model showed it has the ability to be used on the design of an abrasive-cutter brush for industrial application.
9.2 Directions for future research
Despite achieving the defined objectives in this thesis, there are still topics related to this research that could be further investigated. Among them, the following areas are emphasised:
155 • Determination of some other related mechanical properties of tough- skinned vegetables
Further research is needed to determine values of other mechanical properties that were not studied such as the dynamic coefficient of friction. Those properties may be applied directly in the model or may indirectly affect the peeling process. Furthermore, that investigation would be applied to develop the classification of products on the basis of mechanical properties. It could also be useful for determination of the range of applications for the proposed peeler and lead to other new ideas for mechanical peeling.
• Investigation of the effects of environmental parameters on mechanical properties of tough-skinned vegetables
In this research the mean values of mechanical properties in defined environmental conditions (temperature and humidity) were determined and used. As maintenance of these conditions is not always easy, it would be helpful to investigate the relationship between environmental conditions and the mechanical properties of tough-skinned vegetables.
• Investigation of the relationship between peeling rate and energy required for mechanical peeling
This relationship was assumed to be linear in the developed model. It is necessary to investigate this relationship and to determine the coefficient of this function. This research can be conducted on different vegetables and on different varieties. It would be useful to investigate whether there is any relationship between the proposed coefficient and the mechanical properties for different products.
• Development of the proposed mathematical model
156 The model can be extended in different ways. As discussed, there are still other influencing parameters in the proposed model. The type and their effect should be investigated to increase the accuracy and application ability of the model. Further investigation is also needed to reveal any interaction between the current determined variables. Some parameters, such as angular velocity of vegetable holder, were assumed to be fixed. Increasing the accuracy of the extended model can be achieved by considering such variables.
• Design and manufacture of an industrial peeler of tough-skinned vegetables on the basis of the abrasive-cutter brush method
This research focused on revealing the best mechanical peeling method and determination of the type and effects of the related parameters on the function of the proposed method. As the ability of the method was proved in different aspects, the manufactured peeler on the basis of the abrasive-cutter brush has potential for wide application in the food processing industry. This is emphasised because of the current lack of any satisfactory peeler for tough- skinned vegetables especially those with uneven surface.
157 Appendices
Appendix 1 1.1 Multiple comparisons of the mean of the mechanical properties
1.1.1 Multiple Comparisons of the mean of the unpeeled rupture force among varieties of pumpkin and melon
Dependent Variable: Rupture force (N) of unpeeled Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale .83571 14.54 .954 -28.36 30.04 Butternut -15.99429 18.01 .379 -52.17 20.18 Rockmelon 149.48238(*) 16.31 .000 116.70 182.25 Honeydew 66.30196(*) 16.75 .000 32.64 99.95 Watermelon 77.18349(*) 16.31 .000 44.41 109.95 Jarrahdale Jap -.83571 14.54 .954 -30.04 28.36 Butternut -16.83000 15.37 .279 -47.71 14.05 Rockmelon 148.64667(*) 13.34 .000 121.83 175.45 Honeydew 65.46625(*) 13.88 .000 37.58 93.34 Watermelon 76.34778(*) 13.34 .000 49.53 103.15 Butternut Jap 15.99429 18.01 .379 -20.18 52.17 Jarrahdale 16.83000 15.37 .279 -14.05 47.71 Rockmelon 165.47667(*) 17.06 .000 131.20 199.75 Honeydew 82.29625(*) 17.48 .000 47.17 117.41 Watermelon 93.17778(*) 17.06 .000 58.90 127.45 Rockmelon Jap -149.48238(*) 16.31 .000 -182.25 -116.70 Jarrahdale -148.64667(*) 13.34 .000 -175.45 -121.83 Butternut -165.47667(*) 17.06 .000 -199.75 -131.20 Honeydew -83.18042(*) 15.73 .000 -114.78 -51.58 Watermelon -72.29889(*) 15.26 .000 -102.95 -41.64 Honeydew Jap -66.30196(*) 16.75 .000 -99.95 -32.64 Jarrahdale -65.46625(*) 13.88 .000 -93.34 -37.58 Butternut -82.29625(*) 17.48 .000 -117.41 -47.17 Rockmelon 83.18042(*) 15.73 .000 51.58 114.78 Watermelon 10.88153 15.73 .492 -20.71 42.48 Watermelon Jap -77.18349(*) 16.31 .000 -109.95 -44.41 Jarrahdale -76.34778(*) 13.34 .000 -103.15 -49.53 Butternut -93.17778(*) 17.06 .000 -127.45 -58.90 Rockmelon 72.29889(*) 15.26 .000 41.64 102.95 Honeydew -10.88153 15.73 .492 -42.48 20.71
* The mean difference is significant at the .05 level.
158 1.1.2 Multiple Comparisons of the mean of the skin rupture force among varieties of pumpkin and melon
Dependent Variable: Rupture force (N) of skin
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale 57.62015(*) 10.60 .000 36.41 78.82 Butternut -91.07467(*) 15.72 .000 -122.52 -59.62 Rockmelon 6.99242 12.33 .573 -17.67 31.65 Honeydew -56.72667(*) 15.72 .001 -88.17 -25.27 Watermelon -77.33167(*) 12.06 .000 -101.45 -53.20 Jarrahdale Jap -57.62015(*) 10.60 .000 -78.82 -36.41 Butternut -148.69482(*) 14.64 .000 -177.96 -119.42 Rockmelon -50.62773(*) 10.91 .000 -72.44 -28.80 Honeydew -114.34682(*) 14.64 .000 -143.62 -85.07 Watermelon -134.95182(*) 10.60 .000 -156.15 -113.74 Butternut Jap 91.07467(*) 15.72 .000 59.62 122.52 Jarrahdale 148.69482(*) 14.64 .000 119.42 177.96 Rockmelon 98.06709(*) 15.93 .000 66.19 129.93 Honeydew 34.34800 18.68 .071 -3.02 71.71 Watermelon 13.74300 15.72 .386 -17.70 45.19 Rockmelon Jap -6.99242 12.33 .573 -31.65 17.67 Jarrahdale 50.62773(*) 10.91 .000 28.80 72.44 Butternut -98.06709(*) 15.93 .000 -129.93 -66.19 Honeydew -63.71909(*) 15.93 .000 -95.58 -31.84 Watermelon -84.32409(*) 12.33 .000 -108.98 -59.65 Honeydew Jap 56.72667(*) 15.72 .001 25.27 88.17 Jarrahdale 114.34682(*) 14.64 .000 85.07 143.62 Butternut -34.34800 18.68 .071 -71.71 3.02 Rockmelon 63.71909(*) 15.93 .000 31.84 95.58 Watermelon -20.60500 15.72 .195 -52.05 10.84 Watermelon Jap 77.33167(*) 12.06 .000 53.20 101.45 Jarrahdale 134.95182(*) 10.60 .000 113.74 156.15 Butternut -13.74300 15.72 .386 -45.19 17.70 Rockmelon 84.32409(*) 12.33 .000 59.65 108.98 Honeydew 20.60500 15.72 .195 -10.84 52.05
* The mean difference is significant at the .05 level.
159 1.1.3 Multiple Comparisons of the mean of the unpeeled toughness among varieties of pumpkin and melon
Dependent Variable: toughness of unpeeled product
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale -16.988 108.69 .876 -235.31 201.33 Butternut 101.470 134.65 .455 -168.99 371.93 Rockmelon 99.482 121.97 .419 -145.51 344.47 Honeydew -376.92(*) 125.26 .004 -628.53 -125.31 Watermelon -300.56(*) 121.97 .017 -545.56 -55.57 Jarrahdale Jap 16.988 108.69 .876 -201.33 235.31 Butternut 118.45 114.93 .308 -112.39 349.31 Rockmelon 116.47 99.77 .249 -83.93 316.87 Honeydew -359.93(*) 103.77 .001 -568.37 -151.50 Watermelon -283.58(*) 99.77 .006 -483.98 -83.17 Butternut Jap -101.47 134.65 .455 -371.93 168.99 Jarrahdale -118.45 114.93 .308 -349.31 112.39 Rockmelon -1.98 127.56 .988 -258.21 254.23 Honeydew -478.39(*) 130.71 .001 -740.94 -215.84 Watermelon -402.04(*) 127.56 .003 -658.26 -145.81 Rockmelon Jap -99.48 121.97 .419 -344.47 145.51 Jarrahdale -116.47 99.77 .249 -316.87 83.93 Butternut 1.98 127.56 .988 -254.23 258.21 Honeydew -476.40(*) 117.61 .000 -712.63 -240.18 Watermelon -400.05(*) 114.09 .001 -629.22 -170.87 Honeydew Jap 376.92(*) 125.26 .004 125.31 628.53 Jarrahdale 359.93(*) 103.77 .001 151.50 568.37 Butternut 478.39(*) 130.71 .001 215.84 740.94 Rockmelon 476.40(*) 117.61 .000 240.18 712.63 Watermelon 76.35 117.61 .519 -159.82 312.58 Watermelon Jap 300.56(*) 121.97 .017 55.57 545.56 Jarrahdale 283.58(*) 99.77 .006 83.17 483.98 Butternut 402.04(*) 127.56 .003 145.81 658.26 Rockmelon 400.05(*) 114.09 .001 170.87 629.22 Honeydew -76.35 117.61 .519 -312.58 159.87
* The mean difference is significant at the .05 level.
160 1.1.4 Multiple Comparisons of the mean of the skin toughness among varieties of pumpkin and melon
Dependent Variable: Toughness of skin
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale 19.23 24.88 .442 -30.50 68.97 Butternut -96.05(*) 37.19 .012 -170.39 -21.70 Rockmelon -146.84(*) 29.16 .000 -205.14 -88.54 Honeydew -186.80(*) 37.19 .000 -261.15 -112.45 Watermelon -403.25(*) 28.52 .000 -460.27 -346.23 Jarrahdale Jap -19.237 24.88 .442 -68.97 30.50 Butternut -115.28(*) 34.47 .001 -184.20 -46.36 Rockmelon -166.08(*) 25.61 .000 -217.28 -114.87 Honeydew -206.04(*) 34.47 .000 -274.96 -137.12 Watermelon -422.49(*) 24.88 .000 -472.23 -372.75 Butternut Jap 96.05(*) 37.19 .012 21.70 170.39 Jarrahdale 115.28(*) 34.47 .001 46.36 184.20 Rockmelon -50.79 37.68 .183 -126.12 24.54 Honeydew -90.75(*) 44.19 .044 -179.09 -2.41 Watermelon -307.20(*) 37.19 .000 -381.55 -232.85 Rockmelon Jap 146.84(*) 29.16 .000 88.54 205.14 Jarrahdale 166.08(*) 25.61 .000 114.87 217.28 Butternut 50.79 37.68 .183 -24.54 126.12 Honeydew -39.95 37.68 .293 -115.29 35.37 Watermelon -256.41(*) 29.16 .000 -314.71 -198.10 Honeydew Jap 186.80(*) 37.19 .000 112.45 261.15 Jarrahdale 206.04(*) 34.47 .000 137.12 274.96 Butternut 90.75(*) 44.19 .044 2.41 179.09 Rockmelon 39.95 37.68 .293 -35.37 115.29 Watermelon -216.45(*) 37.19 .000 -290.79 -142.16 Watermelon Jap 403.25(*) 28.52 .000 346.23 460.27 Jarrahdale 422.49(*) 24.88 .000 372.75 472.23 Butternut 307.20(*) 37.19 .000 232.85 381.55 Rockmelon 256.41(*) 29.16 .000 198.10 314.71 Honeydew 216.45(*) 37.19 .000 142.10 290.79
* The mean difference is significant at the .05 level.
161 1.1.5 Multiple Comparisons of the mean of the unpeeled cutting force among varieties of pumpkin and melon
Dependent Variable: Cutting force (N) of unpeeled product
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale 5.83600(*) .85222 .000 4.11 7.55 Butternut -9.49200(*) 1.03455 .000 -11.57 -7.40 Rockmelon -1.20000 .85222 .166 -2.91 .51 Honeydew 1.44000 .97756 .148 -.53 3.41 Watermelon .82500 .85222 .338 -.89 2.54 Jarrahdale Jap -5.83600(*) .85222 .000 -7.55 -4.11 Butternut -15.32800(*) 1.01591 .000 -17.37 -13.28 Rockmelon -7.03600(*) .82949 .000 -8.70 -5.36 Honeydew -4.39600(*) .95781 .000 -6.32 -2.46 Watermelon -5.01100(*) .82949 .000 -6.68 -3.33 Butternut Jap 9.49200(*) 1.03455 .000 7.40 11.57 Jarrahdale 15.32800(*) 1.01591 .000 13.28 17.37 Rockmelon 8.29200(*) 1.01591 .000 6.24 10.33 Honeydew 10.93200(*) 1.12313 .000 8.66 13.19 Watermelon 10.31700(*) 1.01591 .000 8.26 12.36 Rockmelon Jap 1.20000 .85222 .166 -.51 2.91 Jarrahdale 7.03600(*) .82949 .000 5.36 8.70 Butternut -8.29200(*) 1.01591 .000 -10.33 -6.24 Honeydew 2.64000(*) .95781 .008 .70 4.57 Watermelon 2.02500(*) .82949 .019 .35 3.69 Honeydew Jap -1.44000 .97756 .148 -3.41 .53 Jarrahdale 4.39600(*) .95781 .000 2.46 6.32 Butternut -10.93200(*) 1.12313 .000 -13.19 -8.66 Rockmelon -2.64000(*) .95781 .008 -4.57 -.70 Watermelon -.61500 .95781 .524 -2.54 1.31 Watermelon Jap -.82500 .85222 .338 -2.54 .89 Jarrahdale 5.01100(*) .82949 .000 3.33 6.68 Butternut -10.31700(*) 1.01591 .000 -12.36 -8.26 Rockmelon -2.02500(*) .82949 .019 -3.69 -.35 Honeydew .61500 .95781 .524 -1.31 2.54
* The mean difference is significant at the .05 level.
162 1.1.6 Multiple Comparisons of the mean of the skin cutting force among varieties of pumpkin and melon
Dependent Variable: Cutting force (N) of skin
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale 6.59111(*) 1.09249 .000 4.39 8.78 Butternut -7.90111(*) 1.09249 .000 -10.09 -5.70 Rockmelon -3.23778(*) 1.06483 .004 -5.37 -1.09 Honeydew -.55528 1.39266 .692 -3.35 2.24 Watermelon -.71944 1.02193 .485 -2.77 1.33 Jarrahdale Jap -6.59111(*) 1.09249 .000 -8.78 -4.39 Butternut -14.49222(*) 1.09249 .000 -16.69 -12.29 Rockmelon -9.82889(*) 1.06483 .000 -11.97 -7.68 Honeydew -7.14639(*) 1.39266 .000 -9.94 -4.34 Watermelon -7.31056(*) 1.02193 .000 -9.36 -5.25 Butternut Jap 7.90111(*) 1.09249 .000 5.70 10.09 Jarrahdale 14.49222(*) 1.09249 .000 12.29 16.69 Rockmelon 4.66333(*) 1.06483 .000 2.52 6.80 Honeydew 7.34583(*) 1.39266 .000 4.54 10.14 Watermelon 7.18167(*) 1.02193 .000 5.12 9.23 Rockmelon Jap 3.23778(*) 1.06483 .004 1.09 5.37 Jarrahdale 9.82889(*) 1.06483 .000 7.68 11.97 Butternut -4.66333(*) 1.06483 .000 -6.80 -2.52 Honeydew 2.68250 1.37107 .056 -.07 5.44 Watermelon 2.51833(*) .99231 .015 .52 4.51 Honeydew Jap .55528 1.39266 .692 -2.24 3.35 Jarrahdale 7.14639(*) 1.39266 .000 4.34 9.94 Butternut -7.34583(*) 1.39266 .000 -10.14 -4.54 Rockmelon -2.68250 1.37107 .056 -5.44 .075 Watermelon -.16417 1.33802 .903 -2.85 2.52 Watermelon Jap .71944 1.02193 .485 -1.33 2.77 Jarrahdale 7.31056(*) 1.02193 .000 5.25 9.36 Butternut -7.18167(*) 1.02193 .000 -9.23 -5.15 Rockmelon -2.51833(*) .99231 .015 -4.51 -.52 Honeydew .16417 1.33802 .903 -2.52 2.85
* The mean difference is significant at the .05 level.
163
1.1.6 Multiple Comparisons of the mean of the flesh cutting force among varieties of pumpkin and melon
Dependent Variable: Cutting force (N) of flesh
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale .94643(*) .20648 .000 .5277 1.3652 Butternut -3.06500(*) .22949 .000 -3.5304 -2.5996 Rockmelon 1.85214(*) .20648 .000 1.4334 2.2709 Honeydew 2.08643(*) .20648 .000 1.6677 2.5052 Watermelon 1.95833(*) .21637 .000 1.5195 2.3971 Jarrahdale Jap -.94643(*) .20648 .000 -1.3652 -.5277 Butternut -4.01143(*) .24534 .000 -4.5090 -3.5139 Rockmelon .90571(*) .22396 .000 .4515 1.3599 Honeydew 1.14000(*) .22396 .000 .6858 1.5942 Watermelon 1.01190(*) .23311 .000 .5391 1.4847 Butternut Jap 3.06500(*) .22949 .000 2.5996 3.5304 Jarrahdale 4.01143(*) .24534 .000 3.5139 4.5090 Rockmelon 4.91714(*) .24534 .000 4.4196 5.4147 Honeydew 5.15143(*) .24534 .000 4.6539 5.6490 Watermelon 5.02333(*) .25372 .000 4.5088 5.5379 Rockmelon Jap -1.85214(*) .20648 .000 -2.2709 -1.4334 Jarrahdale -.90571(*) .22396 .000 -1.3599 -.4515 Butternut -4.91714(*) .24534 .000 -5.4147 -4.4196 Honeydew .23429 .22396 .302 -.2199 .6885 Watermelon .10619 .23311 .651 -.3666 .5790 Honeydew Jap -2.08643(*) .20648 .000 -2.5052 -1.6677 Jarrahdale -1.14000(*) .22396 .000 -1.5942 -.6858 Butternut -5.15143(*) .24534 .000 -5.6490 -4.6539 Rockmelon -.23429 .22396 .302 -.6885 .2199 Watermelon -.12810 .23311 .586 -.6009 .3447 Watermelon Jap -1.95833(*) .21637 .000 -2.3971 -1.5195 Jarrahdale -1.01190(*) .23311 .000 -1.4847 -.5391 Butternut -5.02333(*) .25372 .000 -5.5379 -4.5088 Rockmelon -.10619 .23311 .651 -.5790 .3666 Honeydew .12810 .23311 .586 -.3447 .6009
* The mean difference is significant at the .05 level.
164 1.1.7 Multiple Comparisons of the mean of the unpeeled shear strength force among varieties of pumpkin and melon
Dependent Variable: Shear strength force of unpeeled product
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale 29.526 15.51 .063 -1.61 60.66 Butternut -2.831 22.40 .900 -47.78 42.12 Rockmelon 148.016(*) 20.01 .000 107.84 188.18 Honeydew 99.233(*) 20.01 .000 59.06 139.40 Watermelon 109.162(*) 19.21 .000 70.61 147.71 Jarrahdale Jap -29.526 15.51 .063 -60.65 1.61 Butternut -32.357 21.15 .132 -74.80 10.09 Rockmelon 118.489(*) 18.60 .000 81.14 155.83 Honeydew 69.707(*) 18.60 .000 32.36 107.04 Watermelon 79.635(*) 17.73 .000 44.03 115.23 Butternut Jap 2.831 22.40 .900 -42.12 47.78 Jarrahdale 32.357 21.15 .132 -10.09 74.80 Rockmelon 150.847(*) 24.64 .000 101.39 200.30 Honeydew 102.064(*) 24.64 .000 52.61 151.51 Watermelon 111.993(*) 23.99 .000 63.84 160.14 Rockmelon Jap -148.016(*) 20.01 .000 -188.18 -107.84 Jarrahdale -118.489(*) 18.60 .000 -155.83 -81.14 Butternut -150.847(*) 24.64 .000 -200.30 -101.39 Honeydew -48.782(*) 22.49 .035 -93.92 -3.63 Watermelon -38.854 21.78 .080 -82.56 4.85 Honeydew Jap -99.233(*) 20.01 .000 -139.40 -59.06 Jarrahdale -69.707(*) 18.60 .000 -107.04 -32.36 Butternut -102.064(*) 24.64 .000 -151.51 -52.61 Rockmelon 48.782(*) 22.49 .035 3.63 93.92 Watermelon 9.928 21.78 .650 -33.78 53.64 Watermelon Jap -109.162(*) 19.21 .000 -147.71 -70.61 Jarrahdale -79.635(*) 17.73 .000 -115.23 -44.03 Butternut -111.993(*) 23.99 .000 -160.14 -63.84 Rockmelon 38.854 21.78 .080 -4.85 82.56 Honeydew -9.928 21.78 .650 -53.64 33.78
* The mean difference is significant at the .05 level.
165 1.1.8 Multiple Comparisons of the mean of the skin shear strength force among varieties of pumpkin and melon
Dependent Variable: Shear strength force of skin
(I) (J) Mean Std. 95% Confidence Vegetable Vegetable Difference (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale 33.82658(*) 10.60 .002 12.64 55.01 Butternut -77.01300(*) 16.30 .000 -109.57 -44.45 Rockmelon -2.51667 12.01 .835 -26.52 21.48 Honeydew -38.92700(*) 16.30 .020 -71.48 -6.36 Watermelon -43.04000(*) 12.01 .001 -67.04 -19.03 Jarrahdale Jap -33.82658(*) 10.60 .002 -55.01 -12.64 Butternut -110.83958(*) 16.21 .000 -143.21 -78.46 Rockmelon -36.34325(*) 11.89 .003 -60.09 -12.59 Honeydew -72.75358(*) 16.21 .000 -105.12 -40.37 Watermelon -76.86658(*) 11.89 .000 -100.61 -53.11 Butternut Jap 77.01300(*) 16.30 .000 44.45 109.57 Jarrahdale 110.83958(*) 16.21 .000 78.46 143.21 Rockmelon 74.49633(*) 17.16 .000 40.21 108.78 Honeydew 38.08600 20.39 .066 -2.65 78.82 Watermelon 33.97300 17.16 .052 -.31 68.25 Rockmelon Jap 2.51667 12.01 .835 -21.48 26.52 Jarrahdale 36.34325(*) 11.89 .003 12.59 60.09 Butternut -74.49633(*) 17.16 .000 -108.78 -40.21 Honeydew -36.41033(*) 17.16 .038 -70.69 -2.12 Watermelon -40.52333(*) 13.16 .003 -66.81 -14.22 Honeydew Jap 38.92700(*) 16.30 .020 6.36 71.48 Jarrahdale 72.75358(*) 16.21 .000 40.37 105.12 Butternut -38.08600 20.39 .066 -78.82 2.65 Rockmelon 36.41033(*) 17.16 .038 2.12 70.69 Watermelon -4.11300 17.16 .811 -38.39 30.17 Watermelon Jap 43.04000(*) 12.01 .001 19.03 67.04 Jarrahdale 76.86658(*) 11.89 .000 53.11 100.61 Butternut -33.97300 17.16 .052 -68.25 .31 Rockmelon 40.52333(*) 13.16 .003 14.22 66.81 Honeydew 4.11300 17.16 .811 -30.17 38.39
* The mean difference is significant at the .05 level.
166 1.1.9 Multiple Comparisons of the mean of the flesh shear strength force among varieties of pumpkin and melon
Dependent Variable: Shear strength force of flesh
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale 20.15977(*) 4.51 .000 11.07 29.24 Butternut -.13150 5.53 .981 -11.27 11.01 Rockmelon 55.20167(*) 4.43 .000 46.27 64.12 Honeydew 48.07650(*) 5.53 .000 36.93 59.22 Watermelon 54.14705(*) 4.51 .000 45.06 63.23 Jarrahdale Jap -20.15977(*) 4.51 .000 -29.24 -11.07 Butternut -20.29127(*) 5.23 .000 -30.83 -9.74 Rockmelon 35.04189(*) 4.05 .000 26.88 43.20 Honeydew 27.91673(*) 5.23 .000 17.37 38.46 Watermelon 33.98727(*) 4.14 .000 25.65 42.32 Butternut Jap .13150 5.53 .981 -11.01 11.27 Jarrahdale 20.29127(*) 5.23 .000 9.74 30.83 Rockmelon 55.33317(*) 5.16 .000 44.92 65.73 Honeydew 48.20800(*) 6.14 .000 35.84 60.57 Watermelon 54.27855(*) 5.23 .000 43.73 64.82 Rockmelon Jap -55.20167(*) 4.43 .000 -64.12 -46.27 Jarrahdale -35.04189(*) 4.05 .000 -43.20 -26.88 Butternut -55.33317(*) 5.16 .000 -65.73 -44.92 Honeydew -7.12517 5.16 .175 -17.53 3.28 Watermelon -1.05462 4.05 .796 -9.21 7.10 Honeydew Jap -48.07650(*) 5.53 .000 -59.22 -36.93 Jarrahdale -27.91673(*) 5.23 .000 -38.46 -17.37 Butternut -48.20800(*) 6.14 .000 -60.57 -35.84 Rockmelon 7.12517 5.16 .175 -3.28 17.53 Watermelon 6.07055 5.23 .252 -4.47 16.61 Watermelon Jap -54.14705(*) 4.51 .000 -63.23 -45.06 Jarrahdale -33.98727(*) 4.14 .000 -42.32 -25.65 Butternut -54.27855(*) 5.23 .000 -64.82 -43.73 Rockmelon 1.05462 4.05 .796 -7.10 9.21 Honeydew -6.07055 5.23 .252 -16.61 4.47
* The mean difference is significant at the .05 level.
167
1.1.10 Multiple Comparisons of the mean of the unpeeled shear strength among varieties of pumpkin and melon
Dependent Variable: Shear strength of unpeeled product
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale .64250(*) .11424 .000 .4135 .8715 Butternut .13450 .16653 .423 -.1994 .4684 Rockmelon 1.91250(*) .14880 .000 1.6142 2.2108 Honeydew 1.77875(*) .14280 .000 1.4924 2.0651 Watermelon 1.88750(*) .14280 .000 1.6012 2.1738 Jarrahdale Jap -.64250(*) .11424 .000 -.8715 -.4135 Butternut -.50800(*) .15643 .002 -.8216 -.1944 Rockmelon 1.27000(*) .13740 .000 .9945 1.5455 Honeydew 1.13625(*) .13088 .000 .8739 1.3986 Watermelon 1.24500(*) .13088 .000 .9826 1.5074 Butternut Jap -.13450 .16653 .423 -.4684 .1994 Jarrahdale .50800(*) .15643 .002 .1944 .8216 Rockmelon 1.77800(*) .18319 .000 1.4107 2.1453 Honeydew 1.64425(*) .17836 .000 1.2867 2.0018 Watermelon 1.75300(*) .17836 .000 1.3954 2.1106 Rockmelon Jap -1.91250(*) .14880 .000 -2.2108 -1.6142 Jarrahdale -1.27000(*) .13740 .000 -1.5455 -.9945 Butternut -1.77800(*) .18319 .000 -2.1453 -1.4107 Honeydew -.13375 .16192 .412 -.4584 .1909 Watermelon -.02500 .16192 .878 -.3496 .2996 Honeydew Jap -1.77875(*) .14280 .000 -2.0651 -1.4924 Jarrahdale -1.13625(*) .13088 .000 -1.3986 -.8739 Butternut -1.64425(*) .17836 .000 -2.0018 -1.2867 Rockmelon .13375 .16192 .412 -.1909 .4584 Watermelon .10875 .15643 .490 -.2049 .4224 Watermelon Jap -1.88750(*) .14280 .000 -2.1738 -1.6012 Jarrahdale -1.24500(*) .13088 .000 -1.5074 -.9826 Butternut -1.75300(*) .17836 .000 -2.1106 -1.3954 Rockmelon .02500 .16192 .878 -.2996 .3496 Honeydew -.10875 .15643 .490 -.4224 .2049
* The mean difference is significant at the .05 level.
168 1.1.11 Multiple Comparisons of the mean of the skin shear strength among varieties of pumpkin and melon
Dependent Variable: Shear strength of skin
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale .57389(*) .25 .029 .05 1.08 Butternut .98389(*) .39 .016 .19 1.77 Rockmelon 2.58306(*) .29 .000 2.00 3.16 Honeydew 1.04389(*) .39 .010 .25 1.83 Watermelon 2.28556(*) .29 .000 1.70 2.86 Jarrahdale Jap -.57389(*) .25 .029 -1.08 -.05 Butternut .41000 .39 .301 -.37 1.19 Rockmelon 2.00917(*) .28 .000 1.43 2.58 Honeydew .47000 .39 .237 -.31 1.25 Watermelon 1.71167(*) .28 .000 1.13 2.28 Butternut Jap -.98389(*) .39 .016 -1.77 -.19 Jarrahdale -.41000 .39 .301 -1.19 .37 Rockmelon 1.59917(*) .41 .000 .76 2.43 Honeydew .06000 .49 .904 -.92 1.04 Watermelon 1.30167(*) .41 .003 .46 2.13 Rockmelon Jap -2.58306(*) .29 .000 -3.16 -2.00 Jarrahdale -2.00917(*) .28 .000 -2.58 -1.43 Butternut -1.59917(*) .41 .000 -2.43 -.76 Honeydew -1.53917(*) .41 .000 -2.37 -.70 Watermelon -.29750 .31 .356 -.93 .34 Honeydew Jap -1.04389(*) .39 .010 -1.83 -.25 Jarrahdale -.47000 .39 .237 -1.25 .31 Butternut -.06000 .49 .904 -1.04 .92 Rockmelon 1.53917(*) .41 .000 .70 2.37 Watermelon 1.24167(*) .41 .004 .40 2.07 Watermelon Jap -2.28556(*) .29 .000 -2.86 -1.70 Jarrahdale -1.71167(*) .28 .000 -2.28 -1.13 Butternut -1.30167(*) .41 .003 -2.13 -.46 Rockmelon .29750 .31 .356 -.34 .93 Honeydew -1.24167(*) .41 .004 -2.07 -.40
* The mean difference is significant at the .05 level.
169 1.1.12 Multiple Comparisons of the mean of the flesh shear strength among varieties of pumpkin and melon
Dependent Variable: Shear strength of flesh
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale -.09920(*) .04333 .027 -.1864 -.0120 Butternut -.23575(*) .05316 .000 -.3428 -.1287 Rockmelon .21125(*) .04256 .000 .1256 .2969 Honeydew .22425(*) .05316 .000 .1172 .3313 Watermelon .20261(*) .04333 .000 .1154 .2898 Jarrahdale Jap .09920(*) .04333 .027 .0120 .1864 Butternut -.13655(*) .05029 .009 -.2378 -.0353 Rockmelon .31045(*) .03892 .000 .2321 .3888 Honeydew .32345(*) .05029 .000 .2222 .4247 Watermelon .30182(*) .03976 .000 .2218 .3819 Butternut Jap .23575(*) .05316 .000 .1287 .3428 Jarrahdale .13655(*) .05029 .009 .0353 .2378 Rockmelon .44700(*) .04963 .000 .3471 .5469 Honeydew .46000(*) .05897 .000 .3413 .5787 Watermelon .43836(*) .05029 .000 .3371 .5396 Rockmelon Jap -.21125(*) .04256 .000 -.2969 -.1256 Jarrahdale -.31045(*) .03892 .000 -.3888 -.2321 Butternut -.44700(*) .04963 .000 -.5469 -.3471 Honeydew .01300 .04963 .795 -.0869 .1129 Watermelon -.00864 .03892 .825 -.0870 .0697 Honeydew Jap -.22425(*) .05316 .000 -.3313 -.1172 Jarrahdale -.32345(*) .05029 .000 -.4247 -.2222 Butternut -.46000(*) .05897 .000 -.5787 -.3413 Rockmelon -.01300 .04963 .795 -.1129 .0869 Watermelon -.02164 .05029 .669 -.1229 .0796 Watermelon Jap -.20261(*) .04333 .000 -.2898 -.1154 Jarrahdale -.30182(*) .03976 .000 -.3819 -.2218 Butternut -.43836(*) .05029 .000 -.5396 -.3371 Rockmelon .00864 .03892 .825 -.0697 .0870 Honeydew .02164 .05029 .669 -.0796 .1229
* The mean difference is significant at the .05 level.
170
1.2 Multiple Comparisons of contribution of skin to the mechanical properties
1.2.1. Multiple Comparisons of contribution of skin to the unpeeled rupture force among varieties of pumpkin and melon
Dependent Variable: Rupture force contribution (%)
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale 6.66034 3.819 .088 -1.0321 14.3528 Butternut -50.02543(*) 4.979 .000 -60.0552 -39.9957 Rockmelon -66.76518(*) 4.401 .000 -75.6303 -57.9000 Honeydew -59.17743(*) 4.979 .000 -69.2072 -49.1477 Watermelon -74.42587(*) 4.285 .000 -83.0581 -65.7936 Jarrahdale Jap -6.66034 3.819 .088 -14.3528 1.0321 Butternut -56.68576(*) 4.326 .000 -65.4001 -47.9714 Rockmelon -73.42551(*) 3.646 .000 -80.7695 -66.0815 Honeydew -65.83776(*) 4.326 .000 -74.5521 -57.1234 Watermelon -81.08621(*) 3.505 .000 -88.1473 -74.0251 Butternut Jap 50.02543(*) 4.979 .000 39.9957 60.0552 Jarrahdale 56.68576(*) 4.326 .000 47.9714 65.4001 Rockmelon -16.73975(*) 4.848 .001 -26.5048 -6.9747 Honeydew -9.15200 5.378 .096 -19.9854 1.6814 Watermelon -24.40044(*) 4.743 .000 -33.9546 -14.8463 Rockmelon Jap 66.76518(*) 4.401 .000 57.9000 75.6303 Jarrahdale 73.42551(*) 3.646 .000 66.0815 80.7695 Butternut 16.73975(*) 4.848 .001 6.9747 26.5048 Honeydew 7.58775 4.848 .125 -2.1773 17.3528 Watermelon -7.66069 4.132 .070 -15.9839 .6625 Honeydew Jap 59.17743(*) 4.979 .000 49.1477 69.2072 Jarrahdale 65.83776(*) 4.326 .000 57.1234 74.5521 Butternut 9.15200 5.378 .096 -1.6814 19.9854 Rockmelon -7.58775 4.848 .125 -17.3528 2.1773 Watermelon -15.24844(*) 4.743 .002 -24.8026 -5.6943 Watermelon Jap 74.42587(*) 4.285 .000 65.7936 83.0581 Jarrahdale 81.08621(*) 3.505 .000 74.0251 88.1473 Butternut 24.40044(*) 4.743 .000 14.8463 33.9546 Rockmelon 7.66069 4.132 .070 -.6625 15.9839 Honeydew 15.24844(*) 4.743 .002 5.6943 24.8026
* The mean difference is significant at the .05 level.
171 1.2.2 Multiple Comparisons of contribution of skin to the unpeeled toughness among varieties of pumpkin and melon
Dependent Variable: Toughness contribution (%)
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale -.3214 4.738 .946 -9.8602 9.2174 Butternut -20.2054(*) 6.178 .002 -32.6425 -7.7684 Rockmelon -26.8392(*) 5.317 .000 -37.5433 -16.1351 Honeydew -19.5534(*) 6.178 .003 -31.9905 -7.1164 Watermelon -48.2414(*) 5.317 .000 -58.9456 -37.5373 Jarrahdale Jap .3214 4.738 .946 -9.2174 9.8602 Butternut -19.8840(*) 5.368 .001 -30.6900 -9.0780 Rockmelon -26.5177(*) 4.349 .000 -35.2737 -17.7618 Honeydew -19.2320(*) 5.368 .001 -30.0380 -8.4260 Watermelon -47.9200(*) 4.349 .000 -56.6759 -39.1641 Butternut Jap 20.2054(*) 6.178 .002 7.7684 32.6425 Jarrahdale 19.8840(*) 5.368 .001 9.0780 30.6900 Rockmelon -6.6337 5.885 .266 -18.4811 5.2135 Honeydew .6520 6.673 .923 -12.7816 14.0856 Watermelon -28.0360(*) 5.885 .000 -39.8833 -16.1887 Rockmelon Jap 26.8392(*) 5.317 .000 16.1351 37.5433 Jarrahdale 26.5177(*) 4.349 .000 17.7618 35.2737 Butternut 6.6337 5.885 .266 -5.2135 18.4811 Honeydew 7.2857 5.885 .222 -4.5615 19.1331 Watermelon -21.4022(*) 4.974 .000 -31.4150 -11.3894 Honeydew Jap 19.5534(*) 6.178 .003 7.1164 31.9905 Jarrahdale 19.2320(*) 5.368 .001 8.4260 30.0380 Butternut -.6520 6.673 .923 -14.0856 12.7816 Rockmelon -7.2857 5.885 .222 -19.1331 4.5615 Watermelon -28.6880(*) 5.885 .000 -40.5353 -16.8407 Watermelon Jap 48.2414(*) 5.317 .000 37.5373 58.9456 Jarrahdale 47.9200(*) 4.349 .000 39.1641 56.6759 Butternut 28.0360(*) 5.885 .000 16.1887 39.8833 Rockmelon 21.4022(*) 4.974 .000 11.3894 31.4150 Honeydew 28.6880(*) 5.885 .000 16.8407 40.5353
* The mean difference is significant at the .05 level.
172 1.2.3 Multiple Comparisons of contribution of skin to the unpeeled cutting force among varieties of pumpkin and melon
Dependent Variable: Cutting force contribution (%)
Mean Difference Std. 95% Confidence (I) vegetable (J) vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale 31.3144(*) 9.436 .002 12.2268 50.4021 Butternut .2444 9.986 .981 -19.9561 20.4450 Rockmelon -17.6205(*) 8.226 .039 -34.2608 -.9803 Honeydew -17.0935 9.986 .095 -37.2941 3.1070 Watermelon -14.8145 8.226 .079 -31.4548 1.8257 Jarrahdale Jap -31.3144(*) 9.436 .002 -50.4021 -12.2268 Butternut -31.0700(*) 10.842 .007 -53.0001 -9.1399 Rockmelon -48.9350(*) 9.246 .000 -67.6370 -30.2330 Honeydew -48.4080(*) 10.842 .000 -70.3381 -26.4779 Watermelon -46.1290(*) 9.246 .000 -64.8310 -27.4270 Butternut Jap -.2444 9.986 .981 -20.4450 19.9561 Jarrahdale 31.0700(*) 10.842 .007 9.1399 53.0001 Rockmelon -17.8650 9.806 .076 -37.7015 1.9715 Honeydew -17.3380 11.324 .134 -40.2432 5.5672 Watermelon -15.0590 9.806 .133 -34.8955 4.7775 Rockmelon Jap 17.6205(*) 8.226 .039 .9803 34.2608 Jarrahdale 48.9350(*) 9.246 .000 30.2330 67.6370 Butternut 17.8650 9.806 .076 -1.9715 37.7015 Honeydew .5270 9.806 .957 -19.3095 20.3635 Watermelon 2.8060 8.007 .728 -13.3904 19.0024 Honeydew Jap 17.0935 9.986 .095 -3.1070 37.2941 Jarrahdale 48.4080(*) 10.842 .000 26.4779 70.3381 Butternut 17.3380 11.324 .134 -5.5672 40.2432 Rockmelon -.5270 9.806 .957 -20.3635 19.3095 Watermelon 2.2790 9.806 .817 -17.5575 22.1155 Watermelon Jap 14.8145 8.226 .079 -1.8257 31.4548 Jarrahdale 46.1290(*) 9.246 .000 27.4270 64.8310 Butternut 15.0590 9.806 .133 -4.7775 34.8955 Rockmelon -2.8060 8.007 .728 -19.0024 13.3904 Honeydew -2.2790 9.806 .817 -22.1155 17.5575
* The mean difference is significant at the .05 level.
173 1.2.4 Multiple Comparisons of contribution of skin to the unpeeled maximum shearing strength force among varieties of pumpkin and melon
Dependent Variable: Maximum shearing strength force contribution (%)
Mean (I) (J) Difference Std. 95% Confidence Vegetable Vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale 14.72430(*) 5.743 .013 3.1873 26.2613 butternut -24.87433(*) 8.291 .004 -41.5287 -8.2200 Rockmelon -52.72976(*) 7.408 .000 -67.6102 -37.8493 Honeydew -46.91033(*) 8.291 .000 -63.5647 -30.2560 Watermelon -54.83583(*) 7.110 .000 -69.1168 -40.5549 Jarrahdale Jap -14.72430(*) 5.743 .013 -26.2613 -3.1873 butternut -39.59863(*) 7.829 .000 -55.3248 -23.8725 Rockmelon -67.45406(*) 6.887 .000 -81.2878 -53.6203 Honeydew -61.63463(*) 7.829 .000 -77.3608 -45.9085 Watermelon -69.56013(*) 6.565 .000 -82.7469 -56.3734 butternut Jap 24.87433(*) 8.291 .004 8.2200 41.5287 Jarrahdale 39.59863(*) 7.829 .000 23.8725 55.3248 Rockmelon -27.85543(*) 9.121 .004 -46.1758 -9.5350 Honeydew -22.03600(*) 9.851 .030 -41.8243 -2.2477 Watermelon -29.96150(*) 8.880 .001 -47.7984 -12.1246 Rockmelon Jap 52.72976(*) 7.408 .000 37.8493 67.6102 Jarrahdale 67.45406(*) 6.887 .000 53.6203 81.2878 butternut 27.85543(*) 9.121 .004 9.5350 46.1758 Honeydew 5.81943 9.121 .526 -12.5010 24.1398 Watermelon -2.10607 8.062 .795 -18.2992 14.0870 Honeydew Jap 46.91033(*) 8.291 .000 30.2560 63.5647 Jarrahdale 61.63463(*) 7.829 .000 45.9085 77.3608 butternut 22.03600(*) 9.851 .030 2.2477 41.8243 Rockmelon -5.81943 9.121 .526 -24.1398 12.5010 Watermelon -7.92550 8.880 .376 -25.7624 9.9114 Watermelon jap 54.83583(*) 7.110 .000 40.5549 69.1168 Jarrahdale 69.56013(*) 6.565 .000 56.3734 82.7469 butternut 29.96150(*) 8.880 .001 12.1246 47.7984 Rockmelon 2.10607 8.062 .795 -14.0870 18.2992 Honeydew 7.92550 8.880 .376 -9.9114 25.7624
* The mean difference is significant at the .05 level.
174 1.2.5 Multiple Comparisons of contribution of skin to the unpeeled shearing strength among varieties of pumpkin and melon
Dependent Variable: Shear strength contribution (%)
Mean (J) Difference Std. 95% Confidence (I) vegetable vegetable (I-J) Error Sig. Interval Lower Upper Bound Bound Jap Jarrahdale -7.9733 18.440 .667 -45.012 29.065 Butternut 43.0586 26.620 .112 -10.409 96.526 Rockmelon 3.9481 23.784 .869 -43.824 51.721 Honeydew -191.1873(*) 26.620 .000 -244.655 -137.719 Watermelon -33.5558 22.826 .148 -79.404 12.292 Jarrahdale Jap 7.9733 18.440 .667 -29.065 45.012 Butternut 51.0320(*) 25.136 .048 .544 101.520 Rockmelon 11.9214 22.111 .592 -32.491 56.334 Honeydew -183.2140(*) 25.136 .000 -233.702 -132.726 Watermelon -25.5825 21.077 .231 -67.918 16.753 Butternut Jap -43.0586 26.620 .112 -96.526 10.409 Jarrahdale -51.0320(*) 25.136 .048 -101.520 -.544 Rockmelon -39.1105 29.283 .188 -97.927 19.706 Honeydew -234.2460(*) 31.629 .000 -297.775 -170.716 Watermelon -76.6145(*) 28.510 .010 -133.879 -19.349 Rockmelon Jap -3.9481 23.784 .869 -51.721 43.824 Jarrahdale -11.9214 22.111 .592 -56.334 32.491 Butternut 39.1105 29.283 .188 -19.706 97.927 Honeydew -195.1354(*) 29.283 .000 -253.952 -136.318 Watermelon -37.5039 25.882 .154 -89.491 14.483 Honeydew Jap 191.1873(*) 26.620 .000 137.719 244.655 Jarrahdale 183.2140(*) 25.136 .000 132.726 233.702 Butternut 234.2460(*) 31.629 .000 170.716 297.775 Rockmelon 195.1354(*) 29.283 .000 136.318 253.952 Watermelon 157.6315(*) 28.510 .000 100.366 214.896 Watermelon Jap 33.5558 22.826 .148 -12.292 79.404 Jarrahdale 25.5825 21.077 .231 -16.753 67.918 Butternut 76.6145(*) 28.510 .010 19.349 133.879 Rockmelon 37.5039 25.882 .154 -14.483 89.491 Honeydew -157.6315(*) 28.510 .000 -214.896 -100.366
* The mean difference is significant at the .05 level.
175 1.3 Mechanical properties of varieties of melon and pumpkin in three different states including skin, unpeeled, and flesh (Mean ± Standard Deviation)
Properties Varieties Cases Rupture forc Toughness Cutting forc Max. Shear Shear (N) (N. mm) (N) Strength force strength (N) (N.mm-2) Jarrahdale Skin 40.68±19 13.87±7 2.82±0.34 57.84±16 2.72±1.06 Flesh 1.41±0.51 43.86±13 0.41±0.08 Unpeeled 248.66±45 719.72±293 5.15±0.84 218.18±61 1.78±0.40 Jap Skin 98.30±54 33.10±29 9.41±3 91.67±55 3.29 ±0.99 Flesh 2.36±0.51 64.02±5 0.31±0.07 Unpeeled 249.49±27 702.73±146 10.99±2 247.71±34 2.42±0.31 Butternut Skin 189.37±14 129.16±34 17.31±0.62 168.68±19 2.31±0.22 Flesh 5.43±0.64 64.15±22 0.55±0.24 Unpeeled 265.49±18 601.26±141 20.48±1 250.54±10 2.28±0.21 Rockmelon Skin 91.31±16 179.95±85 12.65±3 94.19±7 0.71±0.13 Flesh 0.51±0.29 8.82±2 0.10±0.05 Unpeeled 100.01±14 603.25±168 12.19±1 99.69±16 0.51±0.16 Honeydew Skin 155.03±33 219.91±176 9.96±4 130.60±24 2.25±0.63 Flesh 0.27±0.09 15.94±1 0.09±0.03 Unpeeled 183.19±36 1079.66±137 9.55±2 148.47±27 0.64±0.28 Watermelon Skin 175.63±20 436.36±89 10.13±1 134.71±23 1.00±0.21 Flesh 0.40±0.12 9.87±3 0.11±0.03 Unpeeled 172.31±15 1003.306±34 10.16±1 138.54±25 0.53±0.13
176 1.4 Relative contribution (%) of skin to different mechanical properties for three pumpkin varieties including Jarrahdale, Jap, and Butternut (Mean ± Standard Deviation)
Properties Rupture Toughness Cutting Max. shear Shear Varieties force N. mm force Strength force strength N N N N.mm-2 Jarrahdale 16.17±9 2.13±1 53.88±12 27.83±7 153.03±55 Jap 22.83±7 1.80±0.74 85.19±23 42.56±27 145.05±41 Butternut 72.86±6 22.01±9 84.95±8 67.43±8 101.99±16 Melon 89.60±6 28.64±14 102.81±17 95.29±8 141.10±30 Honeydew 82.01±16 21.36±18 102.28±25 89.47±17 336.24±86 Watermelon 97.26±2 50.05±15 100.00±14 97.39±8 178.61±43
177 1.5 Drawings of instrumentations 1.5.1 The cutting indentor
178 1.5.2 The skin holder
179 1.5.3 The skin holder (details)
180 1.5.4 The skin holder (details)
181 1.5.5 The cutter of unpeeled sample
182 1.5.6 Spherical end indentor
183
Appendix 2
2.1 Test rig
184 2.1.1 Table of test rig
185 2.1.2 The chassis of test rig
186 2.1.3 The plate of test rig (N.10 in appendix 4.2)
187 2.1.4 The chamber of test rig
188 2.1.5 The vegetable holder
189 2.1.6 The vegetable holder (shaft)
190 2.1.7 The pyramid of knifes
191 2.1.8 The peeler head
192 2.1.9 The peeler head (details)
2.1.10 The peeler head (shaft)
193 2.1.11 The peeler head (flap)
194 2.1.12.1 The peeler head (frame)
195 Appendix 3
3.1 Experimental results of using milling cutter
Exp.no. Peel Peeling Peeling losses %/min efficiency %/min efficiency %/min Concave area Convex area 1 0.4 25 25 2 0.1675 16.25 25 3 0.26 17.5 25 4 0.48 25 25 5 0.4425 17.5 25 6 0.2875 23.75 25 7 0.9725 22.5 25 8 0.1725 11.25 20 9 0.1725 12.5 22.5
3.2 Experimental results of using abrasive pads
Exp.no. Peel Peeling Peeling losses %/min efficiency %/min efficiency %/min Concave area Convex area 1 0.0615 3.25 5 2 0.042 4.5 4.75 3 0.015 0.5 3.25 4 0.0195 1.5 3.5 5 0.049 2.25 4.75 6 0.0355 4 4.25 7 0.2345 5 5 8 0.027 0.75 3.5 9 0.006 0.25 0.5
196 3.3 Experimental results of using abrasive foams
Exp.no. Peel Peeling Peeling losses %/min efficiency %/min efficiency %/min Concave area Convex area 1 0.26025 22.5 25 2 0.244 22.5 22.5 3 0.37575 25 25 4 0.3405 22.5 22.5 5 0.30925 22.5 22.5 6 1.09575 12.5 17.5 7 0.70325 25 25 8 0.58075 23.75 23.75 9 0.395 25 25 10 0.1555 15 15 11 1.24 15 17.5 12 0.10625 12.5 13.75 13 0.22925 15 17.5 14 0.38425 21.25 21.25 15 0.19725 13.75 16.25 16 0.66125 25 25 17 0.805 25 25 18 0.752 25 25 19 0.41225 20 17.5 20 0.199 25 17.5 21 0.081 3.75 5 22 0.61475 25 25 23 0.13975 12.5 13.75 24 0.058 6.25 8.75 25 1.89125 25 25 26 1.9975 25 25 27 0.5085 25 25
197 3.4 Experimental results of using abrasive-cutter brush
Exp.no. Peel Peeling Peeling losses %/min efficiency %/min efficiency %/min Concave area Convex area 1 0.455 42.5 42.5 2 1.4975 57.5 80 3 0.6075 48.75 48.75 4 2.3625 82.5 82.5 5 1.4675 78.75 78.75 6 1.2975 57.5 70 7 0.6075 50 65 8 1.6025 68.75 68.75 9 1.17 71.25 75
198 Appendix 4
4.1 Normality assessment of peeling rate (g/min) of Jap and Jarrahdale varieties 4.1.1 Jarrahdale variety before transformation
Kolmogorov-Smirnov(a) Shapiro-Wilk Statistic df Sig. Statistic df Sig. Peeling .117 64 .029 .881 64 .000 rate a Lilliefors Significance Correction
Plosses Stem-and-Leaf Plot
Frequency Stem & Leaf
6.00 0 . 345777 22.00 1 . 0000122333444556667899 13.00 2 . 0123444788899 12.00 3 . 022355556899 4.00 4 . 2247 1.00 5 . 2 2.00 6 . 23 4.00 Extremes (>=7.3)
Stem width: 1.00 Each leaf: 1 case(s)
Normal Q-Q Plot of Plosses Detrended Normal Q-Q Plot of Plosses
4 1.0
0.8
2 0.6
0.4 0 0.2
Dev from Normal from Dev 0.0 Expected Normal -2
-0.2
-4 -0.4
-20246810 0246810 Observed Value Observed Value
199 10.00 Histogram
13 14 8.00 32 15 48 12
10 6.00
8
6 4.00 Frequency
4
2 2.00 Mean = 2.7356 Std. Dev. = 0 1.88944 0.00 2.00 4.00 6.00 8.00 10.00N = 64 Plosses 0.00
Plosses
200 4.1.2 Jap variety before transformation
Kolmogorov-Smirnov(a) Shapiro-Wilk Statistic df Sig. Statistic df Sig. Peeling .169 64 .000 .807 64 .000 rate a Lilliefors Significance Correction
Plosses Stem-and-Leaf Plot
Frequency Stem & Leaf
2.00 0 . 34 12.00 0 . 556777889999 14.00 1 . 01112222233344 10.00 1 . 6667788899 8.00 2 . 00134444 8.00 2 . 55556778 2.00 3 . 24 2.00 3 . 67 2.00 4 . 24 4.00 Extremes (>=5.3)
Stem width: 1.00 Each leaf: 1 case(s)
Normal Q-Q Plot of Plosses Detrended Normal Q-Q Plot of Plosses
4 2.0
1.5 2
1.0
0 0.5 Dev from Normal Dev
Expected Normal -2 0.0
-0.5 -4 02468 -2 0 2 4 6 8 Observed Value Observed Value
201 8.00 16 Histogram 31
25 6.00 48 13 20
4.00 15
Frequency 10 2.00
5
Mean = 2.062 Std. Dev. = 0.00 0 1.53827 0.00 2.00 4.00 6.00 8.00 N = 64 Plosses Plosses
202 4.1.3 Jarrahdale variety after transformation
LnP.rate Stem-and-Leaf Plot
Frequency Stem & Leaf
1.00 Extremes (=<-1.2) 2.00 -0 . 69 3.00 -0 . 333 18.00 0 . 000001122233344444 12.00 0 . 556667788889 20.00 1 . 00000011122222333444 5.00 1 . 56889 3.00 2 . 001
Stem width: 1.00 Each leaf: 1 case(s)
Detrended Normal Q-Q Plot of LnPlosses Normal Q-Q Plot of LnPlosses
0.2 4
0.0 2
0 -0.2 Expected Normal
Dev from Normal -2 -0.4
-4 -0.6 -1 0 1 2 -1 0 1 2 Observed Value Observed Value
Histogram 2.00
20
1.00 15
0.00 10 Frequency
5 -1.00 51 Mean = 0.7703 Std. Dev. = 0 0.72508 -2.00 -1.00 0.00 1.00 2.00 3.00 N = 64 LnPlosses LnPlosses
203 4.1.4 Jap variety after transformation
LnP.rate Stem-and-Leaf Plot
Frequency Stem & Leaf
1.00 Extremes (=<-1.2) 4.00 -0 . 5669 9.00 -0 . 111122333 17.00 0 . 00001111122233444 22.00 0 . 5555666667888889999999 7.00 1 . 0122344 2.00 1 . 67 2.00 2 . 00
Stem width: 1.00 Each leaf: 1 case(s)
Normal Q-Q Plot of LnPlosses Detrended Normal Q-Q Plot of LnPlosses
0.4 4
0.2 2
0.0
0
Dev fromNormal -0.2
Expected Normal Expected -2
-0.4
-1 0 1 2 -4 Observed Value -1 0 1 2 Observed Value
Histogram
2.00 25
20 1.00
15
0.00
Frequency 10
5
-1.00 Mean = 0.4965 50 Std. Dev. = 0 0.68195 -2.00 -1.00 0.00 1.00 2.00 3.00 N = 64 LnPlosses LnPlosses
204 4.2 Multi comparisons of the mean of LnP.rate among different levels of independent variables
4.2.1 P. speed a. Jarrahdale
Dependent Variable: LnP.rate
Mean Difference 95% Confidence (I) Pspeed (J) Pspeed (I-J) Std. Error Sig. Interval Lower Upper Bound Bound 400.00 550.00 -.78455(*) .14258 .000 -1.0698 -.4993 700.00 -1.17661(*) .14258 .000 -1.4618 -.8914 850.00 -1.64859(*) .14258 .000 -1.9338 -1.3634 550.00 400.00 .78455(*) .14258 .000 .4993 1.0698 700.00 -.39206(*) .14258 .008 -.6773 -.1069 850.00 -.86404(*) .14258 .000 -1.1492 -.5788 700.00 400.00 1.17661(*) .14258 .000 .8914 1.4618 550.00 .39206(*) .14258 .008 .1069 .6773 850.00 -.47197(*) .14258 .002 -.7572 -.1868 850.00 400.00 1.64859(*) .14258 .000 1.3634 1.9338 550.00 .86404(*) .14258 .000 .5788 1.1492 700.00 .47197(*) .14258 .002 .1868 .7572
* The mean difference is significant at the .05 level.
b. Jap
Dependent Variable: LnP.rate
Mean Difference 95% Confidence (I) Pspeed (J) Pspeed (I-J) Std. Error Sig. Interval Lower Upper Bound Bound 400.00 550.00 -.52862(*) .14849 .001 -.8256 -.2316 700.00 -.96187(*) .14849 .000 -1.2589 -.6649 850.00 -1.46678(*) .14849 .000 -1.7638 -1.1698 550.00 400.00 .52862(*) .14849 .001 .2316 .8256 700.00 -.43325(*) .14849 .005 -.7303 -.1362 850.00 -.93816(*) .14849 .000 -1.2352 -.6411 700.00 400.00 .96187(*) .14849 .000 .6649 1.2589 550.00 .43325(*) .14849 .005 .1362 .7303 850.00 -.50491(*) .14849 .001 -.8019 -.2079 850.00 400.00 1.46678(*) .14849 .000 1.1698 1.7638 550.00 .93816(*) .14849 .000 .6411 1.2352 700.00 .50491(*) .14849 .001 .2079 .8019
* The mean difference is significant at the .05 level.
205 4.2.2 Coarseness a. Jarrahdale
Dependent Variable: LnP.rate
Mean Difference 95% Confidence (I) Coarseness (J) Coarseness (I-J) Std. Error Sig. Interval Lower Upper Bound Bound Very coarse Mild -.31821 .24559 .200 -.8095 .1731 Coarse -.50041(*) .24559 .046 -.9917 -.0092 Fine -.69295(*) .24559 .006 -1.1842 -.2017 Mild Very coarse .31821 .24559 .200 -.1731 .8095 Coarse -.18221 .24559 .461 -.6735 .3090 Fine -.37475 .24559 .132 -.8660 .1165 Coarse Very coarse .50041(*) .24559 .046 .0092 .9917 Mild .18221 .24559 .461 -.3090 .6735 Fine -.19254 .24559 .436 -.6838 .2987 Fine Very coarse .69295(*) .24559 .006 .2017 1.1842 Mild .37475 .24559 .132 -.1165 .8660 Coarse .19254 .24559 .436 -.2987 .6838
* The mean difference is significant at the .05 level.
b. Jap
Dependent Variable: LnP.rate
Mean Difference 95% Confidence (I) Coarseness (J) Coarseness (I-J) Std. Error Sig. Interval Lower Upper Bound Bound Very coarse Mild -.31671 .22779 .170 -.7724 .1389 Coarse -.48892(*) .22779 .036 -.9446 -.0333 Fine -.71804(*) .22779 .003 -1.1737 -.2624 Mild Very coarse .31671 .22779 .170 -.1389 .7724 Coarse -.17221 .22779 .453 -.6279 .2834 Fine -.40133 .22779 .083 -.8570 .0543 Coarse Very coarse .48892(*) .22779 .036 .0333 .9446 Mild .17221 .22779 .453 -.2834 .6279 Fine -.22912 .22779 .319 -.6848 .2265 Fine Very coarse .71804(*) .22779 .003 .2624 1.1737 Mild .40133 .22779 .083 -.0543 .8570 Coarse .22912 .22779 .319 -.2265 .6848
* The mean difference is significant at the .05 level.
206 4.2.3 Location a. Jarrahdale
Dependent Variable: LnP.rate
Mean Difference (I) Location (J) Location (I-J) Std. Error Sig. 95% Confidence Interval Lower Upper Bound Bound Bottom Top -.29541 .24717 .237 -.7898 .1990 Bottom side -.52613(*) .24717 .037 -1.0205 -.0317 Top side -.63579(*) .24717 .013 -1.1302 -.1414 Top Bottom .29541 .24717 .237 -.1990 .7898 Bottom side -.23072 .24717 .354 -.7251 .2637 Top side -.34038 .24717 .174 -.8348 .1540 Bottom side Bottom .52613(*) .24717 .037 .0317 1.0205 Top .23072 .24717 .354 -.2637 .7251 Top side -.10966 .24717 .659 -.6041 .3847 Top side Bottom .63579(*) .24717 .013 .1414 1.1302 Top .34038 .24717 .174 -.1540 .8348 Bottom side .10966 .24717 .659 -.3847 .6041
* The mean difference is significant at the .05 level.
b. Jap
Dependent Variable: LnP.rate
Mean Difference (I) Location (J) Location (I-J) Std. Error Sig. 95% Confidence Interval Lower Upper Bound Bound Bottom Top -.18371 .22916 .426 -.6421 .2747 Bottom side -.57259(*) .22916 .015 -1.0310 -.1142 Top side -.58830(*) .22916 .013 -1.0467 -.1299 Top Bottom .18371 .22916 .426 -.2747 .6421 Bottom side -.38888 .22916 .095 -.8473 .0695 Top side -.40459 .22916 .083 -.8630 .0538 Bottom side Bottom .57259(*) .22916 .015 .1142 1.0310 Top .38888 .22916 .095 -.0695 .8473 Top side -.01570 .22916 .946 -.4741 .4427 Top side Bottom .58830(*) .22916 .013 .1299 1.0467 Top .40459 .22916 .083 -.0538 .8630 Bottom side .01570 .22916 .946 -.4427 .4741
* The mean difference is significant at the .05 level.
207 Bibliography
ASAE Standard. 2001. ASAE S368.4. Compression test of food materials of convex shape. American Society of Agric. Eng. 2950 Niles Road, St. Joseph, MI 49085-9659.
Athanasopoulos, P. E. and Vagias, G. A. (1987) “Lye peeling of mandarins,” J Food Process Engineering, 9: 277-285.
Department of agriculture, Fisheries and Forestry. (2005) “Australian food: statistics 2004[Online].Available:http://www.affa.gov.au/corporate_docs/publications/pdf/food/aus _food_stats_interim04.pdf. [Accessed 01 Aug. 2005].
Barreiro, J. A., Caraballo, V., and Sandoval, A. J. (1995) “Mathematical model for the chemical peeling of spherical foods,” Journal of food engineering, 25, 483-496.
Barry-Ryan, C. and O’Beirne, D. (2000) “Effects of peeling methods on the quality of ready-to-use carrot slices,” International Journal of Food Science and Technology, 35, 243-254.
Behnasawy, A.H., El-Haddad, Z.A., El-Ansary, M.Y., and Sorour, H.M. (2004) “Physical and mechanical properties of some Egyptian onion Cultivars,” Journal of Food Engineering, 62, 255-261.
Ben-Shalom, N., Levi, A., and Pinto, R. (1986) “Pectolytic enzyme studies for peeling of grapefruit segment memebrane,” Journal of Food Science, 51, 421-423.
Boyce, J., Jose, S., and Calif (1961), “Apparatus for processing fruit,” US patent: 3013595.
Brown, H. E., Meredith, F. I., Saldana, G. S., and Stephens, T.S. (1970) “Freeze peeling improves quality of tomatoes,” Journal of Food Science, 35, 485-488.
Brusewitz, G.H., McCollum, T.G., and Zhang, X. (1991) “Impact bruise resistance of peaches,” Transactions of the ASAE, 34 (3), 962-965.
Cailliot and Serge (1988) “Fruit and vegetable peeler,” US Patent: 4765234.
Çalı ır, S., Hacısefero ulları, H., Özcan, M., and Arslan, D. (2005) “Some nutritional and technological properties of wild plum (Prunus spp.) fruits in Turkey” Journal of Food Engineering, V.66 (2), pp. 233-237.
Carrote, R. L., Coutaz, V. R., Luna, J. A., Silva, E. R., and Bertone, R. A. (1993) “Optimising processing conditions for chemical peeling of potatoes using response surface methodology,” Journal of Food Science, 58 (4), 821-826.
Carman, K. (1996) “Some physical properties of lentil seeds,” Journal of Agricultural Engineering Research, 63, 87-92.
208
Chavez, M. S., Luna, J. A., and Garrote, R. L. (1996) “A mathematical model to describe potato chemical (NaOH) peeling: Energy and mass transfer model resolution,” Journal of food engineering, 32, 209-223.
Chung, J. H. and Verma, L.R. (1989) “Determination of friction coefficients of beans and peanuts,” Transaction of the ASAE, 32(30), 745–750.
Clevenger, J. T. and Hamann, D. D. (1968) “The behaviour of apple skin under tensile loading,” Transactions of the ASAE, 11, 34-37.
Couture, F. and Allard, R. (1979) “Vegetable peeling apparatus,” US patent: 4137839.
Davies, P. O., Dromgool, N. C., Collinson, G. L., and Fish, A. P. (1996) “Peeling apparatus and method,” U. S. Patent No. 5545422.
Dornow Food Technology GmbH. [Homepage of Dornow] [Online]. 01 Aug, - last update. Available: http://www.dornow.de/Web02/englisch/index_de.html.
Drooge, V. and Lodewijk, B. (1999) “Method for the removal of skins from fruits or vegetables by vapour explosion,” U. S. Patent No. 5942271.
Dowgiallo, A. (2005) “Cutting force of fibrous materials,” Journal of Food engineering, 66, 57-61.
Finney, E. E. (1969) “To define texture in fruits and vegetables,” Agricultural Engineering (United States), 50, 462-465.
Floros, D. J., Wetzstein, H. Y., and Chinnan, M. S. (1987) “Chemical (NaOH) peeling as viewed by scanning electron microscopy: pimiento peppers as a case study,” 52 (5), 1312-1320.
Floros, J. D. and Chinnan, M. S. (1988a) “Microstructural changes during steam peeling of fruits and vegetables,” Journal of Food Science, 53, 849-853.
Floros, J. and Chinnan, M. S. (1988b) “Seven factors response surface optimisation of a double-stage lye (NaOH) peeling process for pimiento peppers,” Journal of Food Science, 53 (2), 631.
Gardiner, R. G., san Jose, and Calif (1963) “Rotary cutter for peeling fruit,” US patent: 3113603.
Gingras, M. (2001) “Apparatus for peeling and optionally cutting vegetables,” U. S. Patent No. 6253670B1.
Grotte, M., Duprat, F., and Loonis, D. (2001) “Mechanical properties of the skin and the flesh of apples,” International Journal of Food Properties, 4 (1), 149-161.
Goud, W. A. (1983) “Tomato production, processing and quality evaluation,” AVI Publications Co Inc. Westport, Connecticut.
209
Gupta, R. K. and Das, S. K. (1998) “Friction coefficient of sunflower seed and kernel on various structural surfaces,” Journal of Agricultural Engineering Research, 71, 175-180.
Harding, G. J. (2000) “melon peeler,” U. S. Patent No. 6,116,155.
Harding, G. J. (2001) “Peeler for fruits and vegetables,” U. S. Patent No. 6324969.
Harker, F. R., Stec, M. G. H., Hallett, I. C., & Bennett, C. L. 1997. Texture of parenchymatous plant tissue: a comparison between tensile and other instrumental and sensory measurements of tissue strength and juiciness. Postharvest Biology and Technology, 11, 63-72.
He, S. L. and Tardif, P. (2000) “vegetable peeling apparatus,” U. S. Patent No. D422173.
Helmy, M. A. (1995) “Determination of static friction coefficient of some Egyptian agricultural products on various surfaces” Misr J. Agr. Eng., 12 (1), 267-282.
Holt, C. B. (1970) “The measurement of tomato firmness with a Universal Testing Machine,” Journal of Texture Studies, 1, 491-501.
Hunter cleaner production program: success stories [Homepage of Department of State and Regional Development, NSW, Australia] [Online]. 01 Aug, 2005 – last update. Available:http://www.smallbiz.nsw.gov.au/smallbusiness/Resources/Success+Stories/Cle aner+Production+Success+Stories/.
Jackman, R. L. and Stanley, D. W. (1994) “Influence of the skin on puncture properties of chilled and nonchilled tomato fruit,” Journal of Texture Studies, 25, 221-230.
Jackman, R. L. and Stanley, D.W. (1992) “Area- and perimeter-dependent properties and failure of mature-green and red-ripe tomato pericarp tissue,” Journal of Texture Studies, 23, 461-474.
Janser, E. (1996) “Enzymatic peeling of fruit,” Food Processing, 3, 1-4.
Boyce, J., San Jose, and Calif (1961) “Apparatus for processing fruit,” US patent: 3013595.
Lankler, J. and Morgan, O. (1944) “How wetting agent improves the chemical peeling process,” Food Ind., 16 (5), 888.
Luh, B. S. and Woodroof, J. G. (1988) “Commercial vegetable processing,” 2nd ed-n., AVI Book, New York, USA.
Marakogˇlu, T., Arslan, D., Özcan, M. and Hacıseferogˇulları, H. (2005) “Proximate composition and technological properties of fresh blackthorn (Prunus spinosa L. subsp dasyphylla (Schur.)) fruits,” Journal of Food Engineering, V.68, Issue 2, 137-142.
Martin, R. (2000) “Fruit peeler,” U. S. Patent No. 6125744.
210 Mohsenin, N. N. and Gohlich, H. (1961) “Techniques for determination of mechanical properties of fruits and vegetables as related to design and development of harvesting and processing machinery,” Journal of Pennsylvania Agricultural Experiment Station, 2687, 300-315.
Mohsenin, N. N. (1970) “Physical properties of plant and animal materials,” Second revised and updated edition, Gordon and Breach Science Publishers, NY.
Odigboh, E. U. (1976) “A cassava peeling machine: development, design and construction,” Journal of Agricultural Engineering Research, 21, 361-369.
Ohwovoriole, E. N., Oboli, S., and Mgbeke, A. C. C. (1988) “Studies and preliminary design for a cassava tuber peeling machine,” Transaction of the ASAE, 380-385.
Oje, K. and Ugbor, E. C. (1991) “Some physical properties of oil bean seed,” Journal of Agricultural Engineering Research, 50, 305–313.
Olsen, I. (1941) “Wetting agents speed chemical peeling: experiments with peaches show value of adding sulphonates to alkaline or acid peeling solutions,” Food Ind., 13(4), 51.
Polk, R., Jr. (1972) “Fruit peeling knife assembly,” US patent: 3680614.
Prakash, S., Singhal, R. S., and Kulkarni, R. R. (2001) “Enzymic peeling of Indian Grapefruit (citrus paradise),” Journal of the science of food and agriculture, 81, 1440- 1442.
Pretel, M. T., Lozano, P., Riquelme, F., and Romojaro, F. (1997) “Pectic enzymes in fresh fruit processing: Optimization of enzymic peeling of oranges,” Process Biotechnology, 32, 43-49.
Protte, C. (1999) “Peeling machine,” U. S. Patent No. 5857404.
Kliamow, K., Genczew, L., and Kafedzwiew. (1977) “Bulgarian vacuum method of peeling” Prazemysl Spozywczy, 31 (3), 82-84.
Kunz, P. (1978) “German Federal Republic Patent Application,” 2, 639117.
Radhakrishnaiah settee, G., Vijayalakshmi, M. R., and Usha Devi, A. (1993) “Methods for peeling fruits and vegetables: A critical evaluation,” Journal of Food Science Technology, 30 (3), 155-162.
Ridler, D. G. (2000) “Fruit and vegetable peeling apparatus,” U. S. Patent No. 6082253.
Rouhana, A. and Mannheim, C. H. (1994) “Optimization of enzymatic peeling of grapefruit,” Lebensmittel-Wissenschaft und Technologie, 27, 103-107.
Roy, R. K. (1990) “A primer on the Taguchi method,” Dearborn, Mich : Society of Manufacturing Engineers.
Rouschning, K. (2001) “Peeler for root vegetables,” U. S. patent No. 6167801B1.
211
Rybczynski, R. and Dobrzanski, B. (1994) “Mechanical resistance of apple in different place of fruit,” International Agrophysics, 9 (3), 455-459.
Saif, A. O. and Bahnasaway, A. H. (2002) “Some physical properties of Yemeni coffee grains,” Egypt Journal of Applied Science 17 (9), 57-69.
Singh, K. K. and Shukla, B.D. (1995) “Abrasive peeling of potatoes,” Journal of Food Engineering, 26, 431-442.
Shen Long, H. and Tardif, P. (1999) “Machine for peeling vegetables and fruits,” U. S. Patent No. 5957045.
Smith, D. A. and Harris, H. (1986) “Thermal blast peeler increases food-processing efficiency,” Highlights of Agricultural Research, 33 (2), 9.
Smith, D. A. (1984) “Peeled yield and quality of delicious apples peeled by superheated steam, saturated steam, and caustic peeling,” Journal of the American Society of Horticultural Science, 109 (3), 364-367.
Soffer, T. and Mannheim, C. H. (1996) “Optimization of enzymatic peeling of oranges and pomelo,” Lebensmittel-Wissenschaft und Technology, 27, 245-248.
Sommer, F. (1997) “Device for peeling elongated vegetables,” U. S. Patent No. 5669293.
Somsen, D., Capelle, A., and Tramper, J. (2004) “Manufacturing of par-fried French-fries, part 2: Modelling yield efficiency of peeling,” Journal of Food Engineering, 61, 199-207.
Srivastava, A., VanEe, G., Ledebuhr, R., Welch, D., and Wang, L. (1996) “Design and development of an onion-peeling machine,” Transaction of ASAE, 13 (2), 167-173.
Su, C. S. and Humphries, E. G. (1972) “Rupture properties of cucumber skin,” Pickle Pak Sci., 2, 1-10.
Suter, M. L. (2002) “Peeling apparatus having feeder control based upon load and associated methods,” U. S. Patent No. 6431061B2.
Thompson, R. L., Fleming, H. P., and Hamann, D. D. (1992) “Delineation of puncture forces for exocarp and mesocarp tissues in cucumber fruit,” Journal of Texture Studies, 23, 169-184.
Thomas, W. M., Stanley, D. W., and Arnott, D. R. (1976) “An evaluation of blanch, lye and freeze-heat methods for tomato peel removal,” Canadian Institute of Food Science and Technology Journal, 9, 118.
Toker, I. and Bayindirli, A. (2003) “Enzymatic peeling of apricots, nectarines, and peaches,” Lebensm.-Wiss. U.-Technol., 36, 215-221.
212 Tsang-Mui-Chung, M.; Verma, L.R. and Wright, M.E. (1984). A device for friction measurement of grains, Transaction of the ASAE, 27, pp. 1938–1941.
Voisey, P. W. and Lyall, L. H. (1965a) “Methods of determining the strength of tomato skin in relation to fruit cracking,” Proc. J. Amer. Soc. Hort. Sci., 86, 597-609.
Voisey, P. W. and Lyall, L.H. (1965b) “Puncture resistance in relation to tomato fruit cracking,” Canadian Journal of Plant Science, 45, 602-603.
Voisey, P. W., Lyall, L. H., and Kloek, M. (1970) “Tomato skin strength-its measurement and relation to cracking,” Journal of American. Society of Horticultural Science, 95, 485-488.
Walter, Jr. W. and Schadel, W. (1982) “Effect of lye peeling conditions on sweet potato tissue,” Journal of Food Science, 47 (3), 813.
Weaver, M., Huxsoll, C., and NG, K. (1980) “Sequential heat-cool peeling of tomatoes,” Food Technology, 34, 40.
Willard, M. J. (1971) “Infra-red peeling,” Food Technology, 25, 813-817.
Zittel, D. R. (1991) “Method for treating a product,” U.S. Patent No. 5989614.
3M Australia [Homepage of 3M Australia] [Online]. 01 Aug, 2005 – last update. Available: http://www.3m.com/intl/au/home/index.jhtml.
213