-
THE DESIGN, PROTOTYPE DEVELOPMENT
AND CONCEPT VALIDATION OF A
CHILE SORTING MACHINE
BY
RYAN HERBON, B.S.
A thesis submitted to the Graduate School
in partial fulfillment of the requirements
for the degree
Master of Science Industrial Engineering
New Mexico State University
Las Cruces, NM
May 2003
2003 by Ryan Herbon
-
ii
The Design, Prototype Development and Concept Validation of a
Chile
Sorting Machine, a thesis prepared by Ryan Herbon in partial
fulfillment of
the requirements for the degree, Master of Science, has been
approved and
accepted by the following:
Dr. Linda Lacey Dean of the Graduate School Dr. Edward Pines
Chair of the Examining Committee Date Committee in charge:
Dr. Edward Pines, Chair Mr. Anthony Hyde, M.S. Dr. Jim Libbin
Dr. Linda Riley
-
iii
VITA
June 17, 1979 Born at Long Beach, California
1997 Graduated from Aliso Niguel High School, Aliso Viejo,
California
2000-2001 Student Shop Assistant, Manufacturing Technology and
Engineering Center (M-TEC), Student Project Center, New Mexico
State University
2001 Received Bachelor of Science in Engineering Technology From
New Mexico State University Las Cruces, New Mexico
2001-2003 Project and Design Engineer, Manufacturing Technology
and Engineering Center (M-TEC), New Mexico State University
-
iv
ABSTRACT
THE DESIGN, PROTOTYPE DEVELOPMENT AND CONCEPT
VALIDATION OF A CHILE SORTING MACHINE
BY
RYAN HERBON, B.S.
Master of Science Industrial Engineering
New Mexico State University
Las Cruces, New Mexico, 2003
Dr. Edward Pines, Chair
The Design, Prototype Development and Concept Validation of
a
Chile Sorting Machine details the process that went into the
design,
development and fabrication of a prototype machine capable of
sorting sticks
and other foreign material from mechanically harvested red chile
peppers.
The New Mexico Chile Pepper Task Force funded this project in an
effort to
assist the processors and producers of red chile in the State of
New Mexico.
Chile is grown on 20,000 acres in New Mexico and contributes
$418 million to
the States economy.
This project went through four distinct stages consisting of
research,
design, prototype development and testing to arrive at an
effective design.
-
v
After research completion, the design began using the Funnel
Approach to
Design methodology along with extensive solid modeling and
computer
simulation. Prototype development was centered on two major
features, the
gap-belt and the color sorter. Upon completion of the prototype,
testing,
concept validation and result tabulation where completed in
December 2002.
It was found that through the implementation of the
chile-sorting machine,
significant reductions could be made in the amount of sticks
present in
mechanically harvested chile.
-
vi
Table of Contents
LIST OF FIGURES
.........................................................................................
ix
LIST OF TABLES
.........................................................................................
xiv
INTRODUCTION.............................................................................................1
Problem Statement
......................................................................................1
Background..................................................................................................3
Research......................................................................................................7
Cleaning Methods from other
Industries...................................................7
Mechanical
Harvesters...........................................................................10
Chile Trash
.............................................................................................16
Current Cleaning
Methods......................................................................18
Rienk
Table.........................................................................................19
Modified Rienk Table
..........................................................................20
Air Blower
...........................................................................................22
Finger Rake
........................................................................................24
Rock
Tank...........................................................................................25
Counter-Rotating Rollers
....................................................................25
Centrifugal Blower
..............................................................................26
Patented cleaning methods
....................................................................27
Research
Summary................................................................................32
METHOD.......................................................................................................34
Initial Design
Considerations......................................................................34
-
vii
Funnel Approach to
Design........................................................................37
Chile Sorting Design Using
FAD................................................................39
Step 1: Design Requirement
..................................................................39
Step 2: Existing Elements from Research
..............................................39
Step 3: New Elements from
Brainstorming.............................................40
Step 4: Product
Constraints....................................................................45
Step 5: Combined or Mixed Ideas
..........................................................48
Step 6: Project
Constraints.....................................................................49
Non-Tangible FAD Results
.................................................................51
Step 7: Prototype
Development..............................................................52
Gap-Belt Design
.................................................................................52
Color
Sorting.......................................................................................58
Color Sensing
Concepts..................................................................58
Sensors
.......................................................................................58
Machine Vision
............................................................................60
Sensing Methodology
..................................................................61
Color Sorter Removal
Concepts......................................................63
Existing Color
Sorters......................................................................66
Coffee Bean Color
Sorters...........................................................67
WECO Color Sorter
.....................................................................69
Operation
.................................................................................69
Variables
..................................................................................71
Integration of the Design
Elements.....................................................74
-
viii
Step 8: Prototype
Construction...............................................................80
Design Process Summary using
FAD........................................................88
TESTING.......................................................................................................90
Test Procedure
..........................................................................................90
Data
Collection...........................................................................................93
Chile Sorting Machine Results
.................................................................103
Gap Belt Results and
Observations..................................................103
Color Sorter
Results..........................................................................105
CONCLUSIONS
..........................................................................................107
Technical..................................................................................................107
Economic Impact
.....................................................................................110
RECOMMENDATIONS FOR FUTURE WORK
...........................................117
APPENDIX A PROJECT
SCHEDULE......................................................124
APPENDIX B CAD DRAWINGS OF THE CHILE SORTING STATION ...129
APPENDIX C THE CENTRIFUGAL
BLOWER.........................................136
APPENDIX D THE ADJUSTMENT OF PROTOTYPE VARIABLES.........156
REFERENCES............................................................................................159
-
ix
LIST OF FIGURES
Figure 1: Mechanically harvested red chile with high trash
content.................3
Figure 2: Relatively clean mechanically harvested red chile
..........................4
Figure 3: Hand laborers removing sticks from mechanically
harvested red
chile...........................................................................................5
Figure 4: The USDA tumbler
cleaner...............................................................6
Figure 5: A color sorter designed for use on tomatoes
..................................10
Figure 6: The Boese harvesting machine
......................................................11
Figure 7: The open double helix picking head of a Boese
harvesting
machine.........................................................................................12
Figure 8: The McClendon chile
harvester......................................................13
Figure 9: The open-helix picking head on the McClendon
harvester.............14
Figure 10: The Pik Rite pepper
harvester......................................................15
Figure 11: The Pik Rite harvesting head
.......................................................15
Figure 12: Sonora chile
plant.........................................................................17
Figure 13: The popular Rienk table or leaf rail cleaning system
with .625 in. spacing
.....................................................................................20
Figure 14: The modified Rienk table with a spacing of 1.625
inches to allow pods to fall through and walk sticks off the
end....................21
Figure 15: A bottom view of the Boese machine's modified Rienk
table .......22
Figure 16: The air blower on the Boese
harvester.........................................23
Figure 17: The finger rake cleaning system on the Boese
harvester.............24
Figure 18: Counter-rotating rollers cleaning method (Wolf &
Alper, 1984) ....25
Figure 19: The centrifugal blower concept design
.........................................26
-
x
Figure 20: The fabricated centrifugal blower bench model
............................27
Figure 21: The helix picking head covered under 1971 Patent #
3,568,419
......................................................................................28
Figure 22: A top view of the Rienk table patented in 1985 by
Patent # 4,507,911
......................................................................................29
Figure 23: The star-shape Rienk table disc (Patent #
4,507,911)..................29
Figure 24: The cleaning apparatus patented by Rutt and Zook in
1995 ........30
Figure 25: The gapped rod, cleaning device described by patent #
5,287,687
......................................................................................31
Figure 26: A depiction of the blower cleaning system patented by
Oren and Daryl
Urich..............................................................................32
Figure 27: An example of the stick trash in mechanically
harvested red chile
...............................................................................................35
Figure 28: The Funnel Approach to
Design...................................................38
Figure 29: Chile diverting concept
.................................................................41
Figure 30: Air separation gap-belt concept
....................................................42
Figure 31: The concept of winnowing the chile material as is
done in
wheat.............................................................................................42
Figure 32: Vacuum plate concept
..................................................................43
Figure 33: The vacuum plate concept simulated
...........................................44
Figure 34: The gap-belt
concept....................................................................45
Figure 35: A force diagram illustrating the forces acting on
material located on the incline belt
..............................................................54
Figure 36: Gap belt adjustment mechanism
detail.........................................55
Figure 37: The incline belt, variable speed V-belt aligning
system ................57
Figure 38: The R55 color sensor from Banner Engineering
..........................59
Figure 39: Lego machine vision concept
model.............................................60
-
xi
Figure 40: The vacuum concept illustrating the action of the
vacuum arm moving down to pick up a pod from the material
belt.....................65
Figure 41: The vacuum concept illustrating the up stroke of the
air cylinder while the four-bar mechanism sweeper knocks the pod
from the vacuum tip onto the pod belt
....................................65
Figure 42: The Sortex Niagara color sorting machine
...................................67
Figure 43: The Xeltron color sorting
machine................................................68
Figure 44: The operating principles of the WECO color sorter
......................70
Figure 45: The sensor array and reject fingers on the WECO color
sorter ....71
Figure 46: The WECO color sorter adjustment panel
....................................72
Figure 47: The box dumper and draper that feed the chile
cleaning
station............................................................................................75
Figure 48: The design of the incline belt feeding the gap-belt
cleaning
station............................................................................................76
Figure 49: The designed gap-belt cleaning station interface
.........................76
Figure 50: The design of the spreader shield used to spread out
the material that has fallen through the gap over the entire 40
inches of the color sorter belt
........................................................77
Figure 51: The design of the spreader bar used to flatten the
flow of material into the color sorter
..........................................................78
Figure 52: The control box of the chile-sorting machine that
housed the speed controllers for all five motors
...............................................79
Figure 53: The fabrication of the trash belt
....................................................80
Figure 54: The fabrication of the spreader
bar...............................................81
Figure 55: Fabrication of the incline belt
........................................................81
Figure 56: A side view of the completed chile-sorting machine
.....................82
Figure 57: The final design of the chile-sorting machine to
allow for a comparison with the constructed prototype
...................................82
-
xii
Figure 58: A front view of the completed chile-sorting
machine.....................83
Figure 59: Sticks removed vs. pods lost through 20 batches of
the gap belt testing
.....................................................................................96
Figure 60: The average results of the gap-belt testing
..................................97
Figure 61: The Percentage of rejected material through 20
batches during the testing of the color sorter.
...........................................101
Figure 62: The average results of the color sorting
testing..........................102
Figure 63: The retail cost of a chile-sorting machine as
affected by level of
production................................................................................113
Figure 64: Proposed chute method for the alignment of material
on the incline belt
...................................................................................119
Figure 65: Proposed placement of Reject Box to eliminate the
stick length constraint of the WECO Color
Sorter................................123
Figure 66: The design of the centrifugal
blower..........................................136
Figure 67: The manufactured centrifugal blower prototype
.........................137
Figure 68: The drive mechanism of the Centrifugal Blower
.........................138
Figure 69: Detail of the friction drive
wheel..................................................139
Figure 70: The helix shelf of the Centrifugal Blower
....................................140
Figure 71: The intended arrangement of material on the helix
shelf of the Centrifugal Blower
.......................................................................140
Figure 72: The air first nozzle used on the Centrifugal Blower
....................141
Figure 73: The second air nozzle used on the Centrifugal
Blower...............142
Figure 74: Percentage of chile pods exiting through the bottom
of the rotating drum during trials of three separate, random
batches of material through the Centrifugal Blower
..................................146
Figure 75: Percentage of sticks exiting through the bottom of
the rotating drum during trials of three separate, random batches
of material through the Centrifugal Blower
......................................147
-
xiii
Figure 76: Percentage of chile pods stuck on the helix of the
rotating drum during trials of three separate, random batches of
material through the Centrifugal Blower
......................................148
Figure 77: Percentage of sticks stuck on the helix of the
rotating drum during trials of three separate, random batches of
material through the Centrifugal Blower
....................................................149
Figure 78: Percentage of chile pods stuck at the bottom of the
rotating drum during trials of three separate, random batches of
material through the Centrifugal Blower
......................................150
Figure 79: Percentage of sticks stuck at the bottom of the
rotating drum during trials of three separate, random batches of
material through the Centrifugal Blower
....................................................151
Figure 80: Material arranged on the helix shelf of the
Centrifugal Blower ..154
-
xiv
LIST OF TABLES
Table 1: A classification of the trash present in mechanically
harvested
red
chile.........................................................................................36
Table 2: The product constraint filter of the FAD
process..............................47
Table 3: The remaining design concepts before project
constraints..............49
Table 4: The results of the project constraint filter
.........................................50
Table 5: The parts and materials ordered from McMaster-Carr for
the prototype construction of a chile-sorting machine
.........................85
Table 6: The material ordered from Pipe and Metal Supply (PMS)
for the manufacture a chile-sorting system prototype
...............................86
Table 7: Parts loaned to the Chile Task Force for the purpose of
testing the chile-sorting prototype
.............................................................86
Table 8: The labor costs that were used in the construction of
the prototype
.......................................................................................86
Table 9: Total Costs incurred in the fabrication of the chile
sorting machine
prototype.........................................................................87
Table 10: Gap-belt results for removal of sticks longer than 8
inches ...........94
Table 11: Gap-belt results of chile pods (small sticks
removed)....................95
Table 12: Test results of color-sorter stick removal
.......................................98
Table 13: Test results of color sorter removal of discolored
pods..................99
Table 14: Test results of color sorter pod
loss.............................................100
Table 15: The variables used in the testing of the gap belt
.........................104
Table 16: Color sorter variables used at testing
..........................................106
Table 17: Estimated labor cost to manufacture a chile cleaning
station similar to the project prototype assuming 80 hours of labor
........111
-
xv
Table 18: An estimated total cost for a manufacturing company to
manufacture a chile cleaning station similar to the project
prototype
.....................................................................................112
Table 19: Variables used in the calculation of yearly machine
cost for a processor or producer
.................................................................114
Table 20: The total yearly cost of a chile-cleaning machine to a
processor or producer
.................................................................114
Table 21: The variables used in the overall economic impact
calculations..115
Table 22: The economic impact of a chile-cleaning machine to an
individual user and the industry as a whole
.................................115
Table 23: Recommended screening process
..............................................117
Table 24: The first batch test results of the centrifugal blower
testing .........144
Table 25: Batch 2 Test Results of the Centrifugal Blower
...........................144
Table 26: Batch 3 Test results of the Centrifugal
Blower.............................144
-
1
INTRODUCTION
Problem Statement
The chile industry annually contributes $418 million in
economic
activity in New Mexico (Hall and Skaggs, b). However, scarcity
of available
labor, increasing labor costs and a 35% loss of chile acreage to
Mexico and
other foreign competitors since 1994 are threatening the
domestic industrys
survival (Hall and Skaggs, b).
In 1998, the New Mexico Chile Task Force (NMCTF) was formed
to
explore ways that new ideas and technologies could help increase
the
industrys profitability. The Task Force has more than 100 active
members
from private, corporate, state and federal organizations. They
represent
growers, processors, researchers, Extension personnel and
members of
agricultural support industries (Diemer, Phillips & Hillon,
2001). New Mexico
State Universitys (NMSU) College of Agriculture and Home
Economics
coordinates NMCTF efforts.
The NMCTF has identified widespread adoption of machine
harvest
technologies as the most important change that the industry must
make.
Machine harvest efforts to date have realized considerable
success, but that
success has been offset somewhat by the amount of trash and
debris
-
2
introduced into the harvested product. This debris must be
removed by hand.
Therefore, NMCTF members have given a high priority to efforts
to
developing effective methods of mechanical trash removal (Salton
and
Wilson, 2003). In October 2001, the NMCTF asked the
Manufacturing
Technology and Engineering Center of NMSUs College of
Engineering to
assist in developing a mechanical cleaner. M-TEC saw the project
as way to
aid its mission of supporting economic development within the
state through
technical assistance. M-TEC agreed to support a graduate student
who
would lead the project and report the results in a masters
degree thesis.
The focus of this project is to design and build a prototype
machine to
effectively remove trash from mechanically harvested red chile
without
harming the product. The purpose of the prototype machine is to
validate the
design concepts to determine which cleaning processes should
be
incorporated into a production-level machine.
This thesis provides documentation for the Chile Task Force.
The
information it contains will be disseminated to New Mexico chile
producers
and processors.
-
3
Background
Several equipment manufacturers design and manufacture
harvesting
equipment for the red chile industry. While the machines
effectively remove
the chile pods from the plants, they also remove an abundance of
sticks,
leaves and other debris with the product (Martin, 2002). Chile
trash becomes
a liability when it goes to a processing plant because it causes
degradation of
chile quality and poses the possibility of damage to expensive
processing
equipment (Dave Layton, personal communication, 10/4/2001).
Depending
on the time of year that harvest occurs, trash can be a major
liability (Figure
1) or a relatively minor problem (Figure 2).
Figure 1: Mechanically harvested red chile with high trash
content
-
4
Figure 2: Relatively clean mechanically harvested red chile
Since 1965, researchers and scientists have experimented
with
mechanical sorting methods, with limited success (Marshall,
1984).
Currently, sticks are hand-removed from chile pods in the
processing plant.
This method poses two significant problems for processors. The
first is that
the cost of employing approximately 15 laborers to remove debris
from chile
pods (Figure 3) as the pods are off-loaded from the truck via a
conveyor belt.
Hand-labor accounts for 40-60% of chile production cost (Martin,
11/12/2001).
In 2001, it was estimated that hand labor costs $65 a day in the
U.S. versus
$6 a day in Mexico (Martin, 11/14/2001). That cost differential
makes
competing with Mexican imports extremely difficult (Diemer,
Phillips & Hillon,
2001). The second problem is the removal of discarded material
from the
processing plant. Debris left in the field may be plowed under
to amend the
-
5
soil for the following years crop. Once it reaches the
processing plant, it
must be treated as industrial waste, incurring significant
disposal costs.
Figure 3: Hand laborers removing sticks from mechanically
harvested red chile
The NMCTF began the chile debris removal initiative in 1998
to
complement the mechanical harvesting process. The first cleaning
system
developed through the task force was a device, called a tumbler
sorter
(Figure 4), developed by the United States Department of
Agriculture,
Agricultural Research Service (USDA, ARS) Southwestern Cotton
Ginning
Research Laboratory. A thorough description of the machine can
be found in
the Research section of this document. The machine currently
does not
work as effectively the chile industry requires.
-
6
Figure 4: The USDA tumbler cleaner
Because of the sorting machines limited success, the Chile
Task
Force was open to new ideas for solving the problem. This led to
the initiation
of the research phase of this project to develop an
understanding of the
problem that would help generate new, innovative solutions.
-
7
Research
To obtain information needed to design an effective machine,
research
was conducted in five areas: (1) cleaning methods from other
industries, (2)
mechanical harvesters, (3) the type of trash found in
mechanically harvested
chile, (4) cleaning equipment currently being used and (5)
cleaning equipment
that had been tried and patented. The following section contains
the results
of the research in these areas.
Cleaning Methods from other Industries
The problem of having trash mixed in with mechanically
harvested
crops is a problem that is not unique to the chile industry. In
fact, the majority
of mechanically harvested crops have had to overcome this same
problem.
Perhaps the most popular example is that of cotton. It took many
years for
the ginning process to be effective enough to use widespread
mechanical
harvesting. According to Ed Hughs, the Director of the USDA,
ARS
Southwestern Cotton Ginning Research Laboratory, the trash
removal
process in the chile industry is about 40 to 50 years behind
that of the cotton
industry (Ford, 1999). Cotton is one of Americas largest market
crops, grown
on more than 16 million acres in the U.S. (Cotton Background,
2002). Chile,
on the other hand, is grown on only 28,000 acres in the United
States, with
20,000 of those acres being in New Mexico (All About Chiles,
2002). Farm
-
8
implement companies saw a valuable market for cotton
harvesting
equipment. The chile industry, however, is not deemed large
enough to
generate a return on R&D dollars spent in the area of chile
machine harvest
(Lenker, 1984).
When cotton is harvested, sticks and seeds are present in the
product.
This foreign matter makes up between 1,000 and 2,100 pounds per
500-
pound bale of cotton fiber (Ed Hughs, personal communication,
3/17/2003).
Different types of cleaners are used to remove different types
of debris (Holt,
Baker & Brashears, 2002). Small debris, such as leaves, is
removed by
conveying the material across an airline after breaking it up
with a series of
spiked cylinders. The burs and sticks then are removed by
stripping them
from the cotton by using a saw cylinder to seize and pull the
cotton across a
series of mechanical scrubbing devices or cleaning bars
(Research on
Machinery Management and Process Control for Cotton Gins,
n.d.).
Unwanted debris accounts for about 5% of the total lot weight
of
mechanically harvested peanuts (Suszkiw, 2001). To sort peanuts
from the
debris, such as plant stems, rocks and dirt, the material is fed
across three
layers of screens, similar to those used to screen gravel
(Blankenship and
Woodall, 1997). Rotating drums made out of mesh material are
used to
remove small debris from the peanuts. The mesh drums filter
debris attached
to the peanut pods.
-
9
The separation of dirt clods from mechanically harvested onions
can
be accomplished in a unique way. The material falls from the end
of a belt
onto a spinning steel roller. When the onions and clods rebound
from the
roller, they follow different trajectories. This allows for
separation because the
onions travel farther than the clods. This process is then
repeated over three
stages. Each stage removes more dirt clods as well as more
onions (Coble,
1984).
There are many other examples of cleaning and sorting of
crops.
Cranberries are sorted by bouncing them along a bouncing board.
The
material that bounces is kept and the material that doesnt
bounce is not kept
(Cranberry Bog Tour, n.d.). Blueberries are sorted using a
combination of
vibration and an air blower (Donahue, Bushway, Moore and
LaGasse, 1999).
The trash problem was taken care of in baby greens by
changing
management practices, such as changing their rotation crops
(Wood, 2002).
Color based sorting has been used in many crops, although it has
not
been used for debris removal. Typically, color is used to sort
different colors
of a single product, such as green from red tomatoes, red from
green
jalapenos or red from white wheat (Pasikatan & Dowell,
2002). A color sorter
designed for use on tomatoes is shown in Figure 5. Color sorting
machines
are used to sort coffee beans, onion flakes, nuts, rice and
different types of
seeds. Color sorters also have been used to sort fungi-infected
peanuts and
defective pistachios (Pearson & Doster, 1998).
-
10
Figure 5: A color sorter designed for use on tomatoes
Mechanical Harvesters
Even though the harvesting process is out of the scope of this
project,
research was conducted into the harvesters as a way to fully
understand the
origin of the trash problem. Currently, there are three major
manufacturers of
chile harvesting equipment: Boese, McClendon and Pik Rite
Harvesters.
There are numerous differences among these three machine types
in how
they harvest and clean the chile.
-
11
Figure 6: The Boese harvesting machine
The Boese Harvesting Company of Saginaw, Michigan, builds a
self-
propelled machine (Figure 6) tailored specifically for the
harvesting of chile
peppers. The Boese machine uses a counter-rotating, double-helix
picking
head (Figure 7). As the helixes rotate in opposite directions
and the machine
moves along the row of chile plants, the plants pass directly
between the set
of helixes. As this occurs, the helixes pull the material
outward, removing the
pods from the plant. The pods are removed by the combination of
the pulling
motion and the vibration that the helixes cause. The pods then
land on a set
of conveyor belts on each side of the double helix where they
are conveyed
into the machines cleaning apparatus. The Boese machine sends
the
harvested material through numerous cleaning processes,
including a
-
12
conveyor-belt cleaning table, to allow hand labor to remove
trash from the
pods.
Many regard the Boese machine as the most effective
harvester
available because of the multiple cleaning methods incorporated.
However,
the cleaning processes also are the source of most complaints
about the
machine because the many hydraulically actuated motors make
troubleshooting problems difficult and time consuming. A Boese
Harvester
costs approximately $360,000 (Greg Boese, personal
communication,
2/5/02).
Figure 7: The open double helix picking head of a Boese
harvesting machine
-
13
Figure 8: The McClendon chile harvester
McClendon harvesters, made in Tulia, Texas, employ a picking
head
similar to the Boese machine. The McClendon machine (Figure 8),
is a John
Deere cotton picker to which is attached a chile harvesting
picking head
manufactured by McClendon. It uses the same counter-rotating,
double helix
concept. It, however, has four helixes per row (Figure 9), as
opposed to the
Boeses two helixes. The helixes also can be exchanged for
fingers to pick
different types of chile. After harvest, the material is
transported from the
helixes across minimal cleaning equipment that cleans out small
debris
before it is loaded into a holding basket on the machine.
McClendon
machines cost about $240,000 (Jim McClendon, personal
communication,
2/5/02).
-
14
Figure 9: The open-helix picking head on the McClendon
harvester
The Pik Rite machine (Figure 10) uses a picking mechanism that
is
significantly different from that used by McClendon and Boese.
It uses two
belts with finger rakes that comb opposite sides of the chile
plant. The finger
rakes (Figure 11) lift the plant material, stripping the pods
from the plant. This
harvest method is considered most thorough. It leaves few pods
in the field
because it can grab everything that has fallen onto the ground
as well as the
material stripped from the plant. It also incurs highest amount
of trash
content because of its harvesting method. The machine
incorporates
-
15
numerous cleaning methods and a hand cleaning station. The Pik
Rite
machine is not self-propelled and requires a tractor to pull it.
The machine
costs $130,000 (Jim McDonnel, personal communication,
2/5/2002).
Figure 10: The Pik Rite pepper harvester
Figure 11: The Pik Rite harvesting head
-
16
Each harvester brand is effective at getting chile pods off of
the plant
and each introduces large amounts of trash into the product.
This problem is
compounded late in the season, after the first frost kills the
plants (Hall &
Skaggs, a). At that point, it is not uncommon for the machines
to remove
entire plants from the ground.
Chile Trash
There are several factors that affect the amount of trash in
mechanically harvested chile. The two major factors, besides the
brand of
machine used to do the harvesting, are the variety type and the
time of
season harvested. B-18 has higher success with machine
harvesting
because its pods are removed easily from the plant. Sonora
plants (Figure
12) are difficult to separate from the plants (Wall, 2002).
Machine harvested
Sonora contains substantially more plant debris (Hall &
Skaggs, a). Plant
varieties with an upright habit and an even fruit distribution
with fruit sets eight
or more inches above the ground harvest most effectively, with
less trash
(Ford, 2000). This is to say that there is a better ratio of
pods to sticks.
Harvesting is also improved by having a reduced number of
lateral shoots
below the main branching (Palevitch & Levy, 1984).
-
17
Figure 12: Sonora chile plant
While chile harvesting typically takes place from September to
the end
of December, the plant characteristics change distinctly after
the first frost.
The first hard frost in southern New Mexico generally occurs in
October
(DeWees, 1998). The plants become very brittle after the frost,
increasing
exponentially the amount of sticks that break off and are pulled
into the
system by the harvester. Chile harvested after the frost is also
more difficult
to clean because the weight, density and shape of the pods match
more
closely that of the sticks. Throughout the season, the moisture
content of the
pods decreases from 7:1 at the beginning of the season to 2:1 at
the end
(Rich Phillips, personal communication, 2/4/2002). According to
chile
processor Lou Biad, many producers park their mechanical
harvesting
-
18
machines after the frost because it costs more to separate the
trash from the
harvested pods than it does to pay a hand harvest crew to
harvest the crop
(personal communication, 9/11/2002).
Currently processing plants in the region do not have the
volume
capability to process the entire New Mexico chile crop between
September
and the first frost. They must lengthen the season into December
and
sometimes January in order to level production. To effectively
do so, better
cleaning equipment needs to be introduced. The season cannot be
extended
earlier into the year because soil temperatures are too low for
the germination
of the seeds before the middle of March (Bob Bevacqua,
personal
communication, 10/9/2002).
Current Cleaning Methods
The predominant cleaning method, used in almost every situation,
is
hand labor. As the material comes into a processing plant, it
moves down
conveyor belts where hand laborers remove the sticks from the
pods. In
addition, the Boese and Pik Rite harvesting machines incorporate
hand-
cleaning stations. Using hand labor for cleaning is not only
expensive, but
also dangerous for the workers who may encounter poisonous
snakes and
other dangerous elements among the product being sorted.
-
19
Rienk Table
Currently, several types of cleaning equipment are used
widely.
Perhaps the most prevalent is the Rienk table (Figure 13). The
Rienk table,
also known as a square tumbler or a leaf rail, is used to sort
small trash from
the harvested chile. It has been used since the 1940s (Richey,
1961) for
cleaning dirt from sugar beets (Danisco, 1997). The machine has
multiple
rotating shafts with plastic squares mounted on them.
Alternatively, spinning
stars have been used for the cleaning of sugar beets (Smith,
1965). The
plastic squares are offset so that the squares from one shaft
fit between the
squares on the next shaft. The desired spacing can then be
attained by
adjusting both the distance between the shafts and the distance
between the
squares on each shaft. Typically, the spacing interval is small
enough so that
the pods are conveyed along the top and the trash that is
smaller than the
pods falls through and is discarded (Abernathy & Hughs,
2001b). This type of
cleaning equipment is used in most processing plants and on each
of the
McClendon, Pik Rite and Boese mechanical harvesters. It has been
proven
to be a very effective method for removing small trash from the
mechanically
harvested chile.
As the squares rotate, the rotating motion acts as a conveyor,
moving
the material from the input to the output side. Using the square
shape, the
gaps between squares and the adjacent shafts increase and
decrease with
the motion of the squares, allowing the trash to fall through.
The cleaning is
-
20
aided by the bouncing effect attained by the material moving up
and down
with the spinning of the squares.
Figure 13: The popular Rienk table or leaf rail cleaning system
with .625 in. spacing
Modified Rienk Table
A variation of the Rienk table has been used to separate out
trash that
is longer than the pods (Figure 14). This method is employed on
Boese and
Pik Rite harvesters and on the research cleaner designed and
built by the
USDA, ARS Southwestern Cotton Ginning Research Laboratory. On
their
prototype machine, USDA researchers used the same squares as
those on
the standard Rienk table, but added more spacing between them.
The
spacing is set at 1.625 inches, compared to 0.625 inches on the
standard
Rienk table (Abernathy & Hughs, 2002b).
The purpose of the modified table is to convey the sticks that
are
longer than the pods off of the table and allow pods and trash
smaller than
-
21
pods to fall through the gaps. The modified Rienk table on the
Boese
harvester basically uses the same method but with
different-shaped spinning
disks and much more spacing between the shafts. Instead of
squares, the
Boese machine uses a finger-shaped spinning media (Figure 15).
The Pik
Rite harvester uses a similar system.
Figure 14: The modified Rienk table with a spacing of 1.625
inches to allow pods to fall through and walk sticks off the
end
The testing of the modified Rienk table has had mixed results.
It does
not always remove all of the trash that is longer than chile
pods. Sticks that
are long and straight often turn vertically and fall through the
spacing, rather
than being carried along to the discharge section as intended.
The table is
however very effective at removing whole plants or sticks with
multiple pods
attached.
-
22
Figure 15: A bottom view of the Boese machine's modified Rienk
table
Air Blower
Boese and Pik Rite Harvesters and several farms
truck-loading
conveyors use an air blower type of cleaning method. The Boese
machines
blower is depicted in Figure 16. It blows a stream of air across
harvested
material as it falls from a conveyor, diverting lighter
material, such as leaves,
into a different container. The air blower method has acceptable
results for
separating out leaf material and pods gutted by insect
infestation, but it has
not been effective for sorting heavier types of trash (Joel
Tellez, personal
communication, June 11, 2002).
-
23
Figure 16: The air blower on the Boese harvester
Researchers at the USDA, ARS Southwestern Cotton Ginning
Research Laboratory conducted experiments to determine if an air
blower
could be used to sort the sticks from pods based on their
relative cross
section to the airflow and relative masses. The results were
inconclusive,
with varying rates of separation effectiveness. (Abernathy and
Hughs, 2001a)
Moisture content was a major factor in the varying results.
Early in the
season, moisture is as high as 7:1, water to chile, making pods
much heavier
than the sticks. Because the plants are more pliable, the
harvesters are
much more effective and the cleaning is not as critical. Toward
the end of the
season, the moisture drops to about a 2:1 ratio (Rich Phillips,
personal
communication, 2/4/02), making the pods approximately the same
weight as
the sticks. Because plants are brittle, more sticks are
harvested and the
weight similarity of sticks and pods makes air separation
difficult.
-
24
Finger Rake
A cleaning method that has been implemented on the Boese
harvesting machines is the finger rake (Figure 17), which strips
pods that are
attached to sticks. The fingers are bolts protruding outward
from three
counter-rotating drums. The bolts are offset so that as the
material is carried
by one drum, the other strips across it, pulling off the pods.
The material is
brought into this system by offsetting the first drum from the
cleated conveyor
belt that is transporting the harvested material into the
machine. This allows it
to grab anything that is over a certain height.
Figure 17: The finger rake cleaning system on the Boese
harvester
-
25
Rock Tank
Chile processing plants generally include a rock tank in their
cleaning
processes. The rock tanks purpose is to float the chile across a
water bath,
washing the product and allowing heavy trash, such as rocks, to
fall to the
bottom. The rock tank also accumulates much of the dirt from the
product
and generally needs to be drained several times daily (Vince
Hernandez,
personal communication, 9/26/2002).
Counter-Rotating Rollers
Another method of removing attached pods from sticks is to
convey the
material through spring-loaded, counter-rotating, rubber-studded
rollers
operating at 100 rpm. This conveys the attached pods and sticks
into two
pairs of picking elements operating at 1000 rpm (Figure 18). The
detached
material, as well as sticks, then falls back onto a conveyor
with the material
that was not drawn into the cleaner (Wolf & Alper,
1984).
Figure 18: Counter-rotating rollers cleaning method (Wolf &
Alper, 1984)
-
26
Centrifugal Blower
Another cleaner design that was proposed by M-TEC engineers
in
Spring 2002 used a combination of centrifugal force and an air
stream to
separate the material within a cylindrical container. The
premise was that the
centrifugal force would cause all of the material to arrange
itself around the
outside of the container where a ledge was welded to the
container. Due to
the different arrangement of the sticks and pods on the ledge,
the sticks were
to be blown out the top while the pods were to move to the
bottom. A bench
model of this concept, named The Centrifugal Blower, was
designed (Figure
19), built (Figure 20), tested and proved not to work. A report
of this project is
included in Appendix C.
Figure 19: The centrifugal blower concept design
-
27
Figure 20: The fabricated centrifugal blower bench model
Patented cleaning methods
An extensive United States patent search was conducted to
identify
any patented cleaning methods not in use currently. The majority
of the
cleaning patents where part of overall harvester patents. Seven
patents were
found that where related to chile harvesting and cleaning of
chile peppers.
Descriptions of each follow.
The concept of using a helix head to harvest chile plants was
originally
patented in 1971 by W.G Creager Chili Harvester (U.S. Patent
#
3,568,419), detailed the use of flexible coils that were of the
same type and
function as current helix heads.
-
28
Figure 21: The helix picking head covered under 1971 Patent #
3,568,419
In 1985, Isaac Wolf and Yekutiel Alper obtained a patent for
Apparatus for Harvesting and Separation of Produce, (U.S. Patent
#
4,507,911). Wolf and Alper summarized the implementation of a
vegetable
harvester that included variations of the open double-helix
picking head and
Rienk table (Figure 22). It also included a description of a
separation method
for detaching attached pods by contacting them with a pair of
rollers,
essentially running the stem through a ringer. The Rienk table
described is
essentially the same one currently used; however, instead of
spinning
squares, it used spinning stars (Figure 23).
-
29
Figure 22: A top view of the Rienk table patented in 1985 by
Patent # 4,507,911
Figure 23: The star-shape Rienk table disc (Patent #
4,507,911)
Both Larry Rutt with Robert Zook and Robert Cosimati
patented
cleaning machines that use cylinders made of rotating rods.
Rutts patent,
-
30
Cleaning Apparatus, (Patent # 5,427,573) describes a cleaning
system
intended to remove sticks and leaves that consists of a rotating
cylinder made
of spaced-out rods (Figure 24). Some of the cylinders rods are
gear-driven
by the cylinder ends so that they rotate in relation to the
rotational speed of
the cylinder. The remaining rods are idlers that are allowed to
spin freely as
the drum rotates. The machine also has a rotating auger brush in
the center
that conveys the peppers from one end of the cylinder to the
other. The
intention is to allow the trash to fall through the gaps while
the pods are
conveyed to the end of the drum. Cosimatis device, Patent #
5,210,999,
lacks the internal brush and puts the cylinder on an incline. It
also uses
compressed air to aid in the sorting.
Figure 24: The cleaning apparatus patented by Rutt and Zook in
1995
Contained within the patent of a harvester (Patent #
5,287,687),
inventors Oren, Gary and Randy Urich described a similar
cleaning method, a
-
31
table of counter rotating rods. The rods are spaced to allow
sticks but not
pods to be pulled through (Figure 25). It is on a table as
opposed to the drum
shape of the previously described patents. A similar system was
included in
Greg Boeses 2002 patent of a harvest machine (Patent #
6,419,093). Within
his patent, the rollers are described as being adjacent in an
attempt to pull
leaves and sticks through, conveying the pods along the top.
Figure 25: The gapped rod, cleaning device described by patent #
5,287,687
In Patent # 5,930,987, Oren and Daryl Urich describe a method
of
trash removal similar to that which is employed on the Boese
harvester. They
describe using an air stream, acting perpendicularly to the
material flow, to
blow off small trash that is lighter than pods (Figure 26).
-
32
Figure 26: A depiction of the blower cleaning system patented by
Oren and Daryl Urich
Research Summary
Although the chile industry has tried many mechanical solutions,
they
have not yet developed a clear-cut method for removing all
unwanted debris.
The debris problem is a moving target and varies dramatically
depending on
when during the season chile is harvested and which machine type
is used.
In addition, there is no current system that accurately defines
the
classification of the trash material to note the effectiveness
of various
machines.
The following information is a summary of key points of the
research:
All currently used mechanical harvesters introduce trash.
Existing equipment only works well on certain aspects of
cleaning and could be incorporated as part of a screening
process.
-
33
Existing or commercial machines fail to remove sticks that are
longer than pods and sticks that are the same size as pods.
Other industries have successfully used modern technology to
overcome similar problems.
Color Sorting Technology may be adaptable to the red chile
sorting.
Most industries use multiple stages to remove trash.
-
34
METHOD
Initial Design Considerations
This project focused only on finding a solution for removing
trash and
debris from machine harvested red chile. Red chile pods were
considered
good product. Any material brought into the system other than
red chile
pods, was defined as trash. This trash fits into five
categories: leaves, dirt,
rocks, sticks and discolored pods. The amount of each depends on
the type
of harvester used, the variety of chile and the time of year
harvested. After
first frost, the increase in the number of sticks harvested
becomes so
problematic that mechanical harvesting is almost infeasible.
Processors often
reject chile harvested after frost due to the high trash
content. For this
reason, the chile-sorting machine must be designed to work
effectively after
the frost.
Currently, effective solutions exist for separating out the
first three
categories: leaves, dirt and rocks. Therefore, these trash
categories are
beyond the scope of this project. Of the last two categories
(sticks and
discolored pods), the sticks, such as those seen in Figure 27,
constitute the
majority of the trash problem. Sticks have the highest potential
to damage
equipment and to degrade the quality of the red chile product.
They also are
the most difficult to remove.
-
35
Figure 27: An example of the stick trash in mechanically
harvested red chile
Sticks that are harvested with pods can be classified as
follows: sticks
that are shorter than pods; sticks that are longer than pods;
and sticks that
are the same length as pods. Within these classifications, there
are multiple
subcategories: straight sticks; branched sticks (sticks with
multiple nodes);
and sticks with pods attached to them. The latter category also
encompasses
whole plants that are sometimes brought into the system by the
harvester.
Table 1 lists trash classifications and methods currently
available for
separating them from pods.
-
36
Table 1: A classification of the trash present in mechanically
harvested red chile
Category Sub-Category Effective Cleaning Method
Leaves - Rienk Table Dirt - Rienk Table and Rock Tank Rocks -
Rock Tank Discolored Pods - No Solution Exists
Sticks Shorter than pods and straight Rienk Table
Sticks Shorter than pods and forked No Solution Exists
Sticks Shorter than pods with attached pods Modified Rienk Table
Removes Some
Sticks Same Size as pods and straight No Solution Exists
Sticks Same Size as pods and forked No Solution Exists
Sticks Same Size as pods with attached pods Modified Rienk
Table
Sticks Longer than pods and straight No Solution Exists
Sticks Longer than pods and forked Modified Rienk Table removes
some
Sticks Longer than pods with attached pods Modified Rienk
Table
-
37
Funnel Approach to Design
The Funnel Approach to Design (FAD) is based on the premise
that
the more ideas available for exploration at the beginning of the
design
process, the more likely it is that a feasible solution will be
found. An analogy
is that of a job search. A job opening is more likely to be
filled with a qualified
candidate if a large pool of applicants is available. The
timeline of the entire
design, development and concept validation was planned out
before FAD
began (Appendix A).
FAD starts with new and existing ideas or inputs based on the
design
requirements given by the customer. New ideas are generated
through
brainstorming while existing ideas come from research. Both the
new and
existing ideas are then filtered through the product
constraints, eliminating
any that are infeasible or dont fall within the constraints. The
remaining ideas
are analyzed further, researched, combined and filtered through
the project
constraints to arrive at two or three concepts that are
developed into
prototypes. Figure 28 shows how a large number of ideas can be
reduced to
a few workable solutions through FAD.
-
38
Figure 28: The Funnel Approach to Design
-
39
Chile Sorting Design Using FAD
Step 1: Design Requirement
The Chile Task Force, considered the customer for this project,
gave
the design requirement to M-TEC. The specific design requirement
was to
design a machine that could sort red chile pods from sticks. The
categories
of sticks that required immediate attention were straight long
sticks and sticks
that were the same size as pods.
Step 2: Existing Elements from Research
The following existing elements came from research and the
interviews
conducted with processors and producers:
o Rienk Table Uses spinning squares to remove small trash
o Modified Rienk Table Uses spinning disks to remove branched
material
o Rotating rods with small gap between them Pulls sticks through
the gap
o Saw cylinder Pulls cotton across a stripper bar
o Color-based defect sorters Remove a defective material based
on color
o Bouncing Board Accepts material that bounces a certain way
o Best-Management practices Changing the way that the crop is
grown to reduce the
number of sticks o Finger rake
Strips pods from attached sticks
-
40
Step 3: New Elements from Brainstorming
At the beginning of the project, a brainstorming session was
held to
produce as many new ideas as possible. As the project matured,
new ideas
were generated constantly.
One original idea was to bounce all of the material across a
screen and
have a scraper under the screen to pull through any stick ends
that stuck
through the screen. The hope was that an end of each stick would
protrude
to the underside while the pods would stay completely on the
top. The
scraper arm would then pull the stick completely through or cut
off a portion of
it. If it cut a portion, then the next portion would need to
stick through to be
cut or pulled. In this manner, even a non-straight stick could
eventually be
removed one inch at a time.
Another idea was to attempt to arrange pods and sticks in a
singular
layer and then devise an electronic method to sense pods and
sticks and
divert them one way or another (Figure 29). A plus side to the
concept is that
de-stemming could be integrated easily. De-stemming is a process
that is out
of the scope of this project but is another that the chile
industry would like to
mechanize.
-
41
Figure 29: Chile diverting concept
Another possible sorting method was to drop the harvested
material off
a belt while quickly pulsing an air stream up and toward another
belt. The
idea behind it was that material like chile, with a larger cross
section would be
blown onto the second belt while other material would fall
through the gap
(Figure 30). A similar concept explored was that of winnowing,
such as wheat
is separated from chaff. This idea was to put all of the
material in a large
container and puff large amounts of air from the bottom to keep
the material
suspended (Figure 31). Again, the idea was that by having a
larger cross-
section, the pods would eventually filter to the top while the
sticks would be
suspended lower in the container.
-
42
Figure 30: Air separation gap-belt concept
Figure 31: The concept of winnowing the chile material as is
done in wheat
-
43
Another idea explored was a vacuum plate (Figure 32).
Numerous
small holes would be drilled in a metal plate and enough vacuum
pressure
applied to the holes to hold a pod to the plate if the pod
covered more than
four or five of the holes (Figure 33). It was thought that the
flat shaped pods
would cover more holes than the narrower sticks. The plate could
then be
moved elsewhere and the air pressure released, dropping off the
pods. An
alternate method for using a vacuum plate would be to have
sensors tell
exactly where pods were and then selectively control which
vacuum holes
were to have vacuum applied via electronic valves.
Figure 32: Vacuum plate concept
-
44
Figure 33: The vacuum plate concept simulated
An alternate brainstormed idea was to run parallel V-belts with
small
spacing between them, while vibrating the whole assembly. The
pods were
to travel along the belts and the sticks were to fall through
the gaps. The
premise was that the vibration would force the sticks into the
gaps, where
they could pass through due to their narrower shape.
A similar idea created during the initial brainstorming was the
use of a
gap belt. This concept included an inclined belt that would move
harvested
material uphill toward a second trash belt. Material would be
oriented
parallel to the direction of travel on the belt. Pods would fall
through a gap
between the inclined belt and the trash belt, while sticks that
were longer than
the gap between the belts would travel onto the trash belt
(Figure 34).
-
45
Figure 34: The gap-belt concept
The following is a summary of the new ideas gathered through
brainstorming.
Pull sticks through a screen Divert pods and sticks based on
sensor input Pulse an air stream toward material falling from a
belt Winnow the material with an air stream within a container
Vacuum plate Drop small trash between V-belts Gap-Belt
Step 4: Product Constraints
Three primary product constraints were involved in the design of
the
machine. These constraints were that the sorter could not damage
the pods
in any way; it must be able to maintain a flow rate of 48,000
lbs hr 1; and that
-
46
it must be capable of removing at least 80% of the sticks from
the harvested
material. These product constraints were used to initially
filter all of the
brainstormed and existing ideas.
It was very important not to break the pods with the cleaning
machine.
The producers get paid based on product weight, and the seeds
make up a
large portion of that weight. If the cleaning machine damaged
the pods, there
would be a high likelihood that seeds would be lost from the
product.
The second imposed product constraint was that the system had to
be
adaptable to work with the current volume of chile being
harvested and
processed. This would prevent the cleaner from becoming a
bottleneck in the
production line. The typical volume that goes through a
processing plant is
approximately 40 boxes per hour. Each box is 4 ft. by 4ft. by 4
ft. and
contains roughly 1,200 pounds of harvested material (Phillips
2/4/2002). This
is a total flow rate of 48,000 lbs hr -1.
Perhaps the most important product constraint is the requirement
that
the machine remove 80% of the sticks. This 80% is a moving
target that
changes drastically depending on many factors. Each individual
customer
has a different interpretation of what percentages of sticks
need to be
removed. Each also has been known to change the required
percentage
depending on the current years production.
The results of passing the new and existing ideas through the
product
constraints are summarized in Table 2.
-
47
Table 2: The product constraint filter of the FAD process
Product Constraints
Concepts
Does not harm the
pods
Can maintain 48,000 lb
h-1flow rate
Can possibly eliminate 80% of
the Sticks that are longer
than pods
Can possibly eliminate 80% of
sticks that are the
same size as pods
Does concept Passes Product constraint filter
Bounce on screen and pull sticks through No Yes Yes Yes No
Winnow with air Yes Yes No No No Narrow flow and divert sticks
based on sensor input Yes No Yes Yes No Long narrow screens for
sticks to fall through Yes Yes No No No Blow pods across a gapped
belt Yes Yes Yes Yes No Vacuum plate based on shape Yes Yes Yes Yes
No Sensor guided vacuum plate Yes Yes Yes Yes Yes Centrifugal force
separation Yes No No No No Vibrating v-belts with gaps between them
Yes Yes No Yes Yes Gap-Belt Yes Yes Yes No Yes Rienk Table Yes Yes
No Yes Yes Modified Rienk Table Yes Yes Yes No Yes Rotating Rods
with gap between them No Yes Yes Yes No Saw cylinder No Yes No No
No Color based defect sorter Yes Yes Yes Yes Yes Bouncing board Yes
Yes No No No Best management practices Yes N/A Yes Yes No Finger
rake No Yes No No No
-
48
Step 5: Combined or Mixed Ideas
In order to find an effective design solution, FAD uses the
premise that
if several ideas can be combined into one, then the likelihood
of it being
successful increases. For this project, it was possible to
combine the gap-belt
with aspects of the V-Belts with gaps between them.
The gap belt requires all of the material to be oriented in the
same
way. According to Vince Hernandez of Biad Chili, it is possible
to achieve this
orientation by having a conveyor made of multiple V-belts, with
alternating V-
belts traveling at different speeds. This method of alignment
has been shown
to work in bell pepper processing plants because the faster belt
causes one
end of the material to be grabbed and pulled while the other end
is held in
place by the slower belt. This occurs until the material rests
solely on either a
fast or slow belt and is aligned parallel to the direction of
travel. This method
of alignment allows the integration of the V-belts with gaps
between them
cleaning concept with the gap-belt cleaning concept. The only
aspects that
differ from the original V-belt concept are that the V-belts are
no longer going
to move at the same speed and the entire arrangement will be on
an angle
rather than flat.
The concepts that remained after the combining and mixing of
ideas in
the funnel are summarized in Table 3.
-
49
Table 3: The remaining design concepts before project
constraints
Sensor guided vacuum plate New Idea
Gap-Belt combined with V-Belts with gaps Mixed Idea
Rienk Table Existing Idea
Modified Rienk Table Existing Idea
Color based defect sorter Existing Idea
Step 6: Project Constraints
The next filter the concepts went through was the project
constraints
filter. The three project constraints were: 1) the production
model machine
had to cost less than $50,000; 2) the machine must be complete
for the 2002
harvesting season; and 3) the machine must incorporate ideas
never before
tried in the chile industry.
The cost constraint was based on the fact that the machine had
to be
affordable for an end user to purchase. The target was to have a
production
model that could be purchased by a consumer for less than
$50,000. It would
be infeasible for processors and producers to justify the
purchase of a
machine that costs much more than that.
There was also a time constraint involved. The prototype had to
be
manufactured so that the concept could be validated by the 2002
chile
season. Chile Task Force representatives identified this time
frame to
-
50
demonstrate progress to members and to allow for the thesis to
be completed
by April 2003.
The final project constraint filter was that the prototype had
to use
innovative or new ideas. The Chile Task Force did not want the
same ideas
that had been tried for the last 40 years to be tried again.
Their feeling was
that if there was a mechanical way to remove the sticks, it
would have been
discovered in the last 40 years. Task Force members felt very
strongly that
the solution would be something that had not been tried, most
likely because
of its high-tech nature.
Figure 3 summarizes the results of the conceptual ideas
passing
through the project constraints.
Table 4: The results of the project constraint filter
Project Constraints
Concepts
Would cost less
than $50,000
Could be completed
by the 2002
harvesting season
Is an Innovative
(not previously tried) idea
for the chile
industry
Does the concept pass the Project Constraint Filter
Sensor guided vacuum plate No No Yes No Gap-Belt combined with
V-Belts with gaps Yes Yes Yes Yes Rienk Table Yes Yes No No
Modified Rienk Table Yes Yes No No Color based defect sorter Yes
Yes Yes Yes
-
51
Non-Tangible FAD Results
The Funnel Approach to Design yielded several important
non-tangible
results. After arriving at several existing and new ideas based
on the design
requirement and filtering those ideas through the lists of
product and project
constraints, two ideas were decided upon. The two ideas chosen
were a gap
belt and a color sorter.
The first idea chosen was to attempt to sort the sticks that
were longer
than pods by using a gap-belt concept. The reasoning behind the
gap-belt
was that any material that was longer than the gap would travel
to the other
side while material that was the size of a pod or shorter would
fall through the
gap, assuming that the material could be oriented in a direction
parallel to the
direction of travel of the belt. It was also possible to combine
the gap-belt
with the spaced apart V-belts cleaning method because the
V-belts also could
provide a method for aligning the material.
The second concept was electronic differentiation of the pods
from the
sticks, based on color. This was chosen because there was an
obvious
difference between the two at all times during the season. At
the beginning of
the harvest season, the pods are bright red while the sticks are
green.
Toward the end of the season, the sticks turn brown while the
pods change to
a darker red. In addition, so many mechanical methods had been
tried over
the years that it was felt that perhaps a more high-tech
solution could be the
answer to the problem. Color sorting machines have been used by
many
-
52
industries in sorting out product that is not the proper color,
but have never
been used in the sorting of product from vegetation trash.
The funnel approach does not end once the concepts have
passed
through the constraints. The next step is to develop these
concepts into
prototypes. The development of prototypes for the gap-belt and
color sorter
is described in the following section.
Step 7: Prototype Development
After the funnel process, the machine development involved
design of
physical apparatuses that would be used for the gap-belt and
color sorter.
This was accomplished through use of SolidWorks, solid modeling
software.
The concepts were initiated, simulated and optimized using this
software.
The level of detail involved in the design included clearance
fits, bolt sizing
and production / assembly details. The reasoning behind the
extensive
computer modeling of the machine was to minimize the fabrication
and
assembly problems, therefore eliminating unnecessary labor and
material
costs.
Gap-Belt Design
The first concept that was decided upon was a gap-belt concept
to
attempt to remove sticks longer than pods. As stated earlier,
the concept of
-
53
the gap belt is that sticks longer than pods would cross the gap
and travel on
to the trash belt, while the pods would fall through the gap
(Figure 34).
In order for the gap belt to work, it was decided that the
incline belt
feeding the gap should be at the maximum possible angle. This
would allow
the force of gravity to be overcome by the normal force exerted
by the belt.
The sticks and pods would be held on the incline belt longer,
increasing the
possibility of the sticks ending up on the trash belt.
The coefficient of static friction, s, between the belt material
and the chile pods and sticks was calculated based on experimental
data. This value
was then used to determine the theoretical maximum angle for the
incline
belt. This calculated angle was between 28 and 39, depending on
the material. When using something that is as varied as chile pods
and sticks,
the calculation of the coefficient of friction becomes very
difficult. The
coefficient, s, varies depending on the moisture content of the
material and how much of the material is actually contacting the
belt. This is a variant
because of the differences in shape from one pod to the next.
The force
diagram (Figure 35) illustrates the three forces acting on a
chile pod situated
on the incline belt. The same forces would act on a stick.
-
54
Figure 35: A force diagram illustrating the forces acting on
material located on the incline belt
The gap belt was designed to allow for ease of testing and for
ease of
manufacture. It was designed to allow for adjustment in the
horizontal gap
distance from a value of no gap up to a 10-inch gap. This
allowed for all
different sizes of pods to be accommodated. It also allowed the
drop off point
from the incline belt to be five inches above or below the
pickup for the trash
belt. The angle of the incline belt, the belt feeding the gap,
also was designed
to be completely adjustable from 0 to 50. According to George
Abernathy, an engineer at the USDA, ARS Southwestern Cotton Ginning
Research
Laboratory, the maximum angle at which chile pods would stick to
a belt
without sliding downhill was 30. The concept of the gap belt was
dependant
-
55
on achieving the maximum possible angle of the incline belt. The
gap belt
adjustment mechanism is shown in Figure 36.
Figure 36: Gap belt adjustment mechanism detail
The end of the trash belt that was at the gap interface received
special
design considerations. Besides being designed to allow for all
of the
adjustments to be made, it was necessary to ensure that all of
the sticks that
came in contact with the belt would be pulled to the trash side.
This was
accomplished by using the smallest size roller available. Using
a small roller
also increased the number of possible alignments between the
incline belt
and the trash belt. The limiting factor as to how small the
roller could be was
the tightest radius that the food grade conveyor belting and the
clips that held
that belting together could handle. The manufacturer of the
belting
recommended a two-inch diameter minimum roller.
-
56
The gap belt requires all of the material to be oriented in the
same way
in order for the concept to work. This orientation was achieved
by having the
incline belt made up of multiple V-belts (Figure 37), with
alternating V-belts
traveling at different speeds. According to Vince Hernandez,
this method of
alignment has been shown to work in bell pepper processing
plants. In order
to get the alternating belts to travel at different speeds,
while having a uniform
interface at the gap, the shaft at the upper end of the incline
had 27 idler
pulleys on it. The idler pulleys had bearings pressed into them
to allow each
to spin independently. Two shafts at the bottom of the incline
drove the belts.
The shafts were set several inches apart and had pulleys
set-screwed for
every other belt. This caused half of the belts to be slightly
longer in order to
reach the rear drive shaft and the other half to be slightly
shorter in order to
reach the front drive shaft. The drive system of the gap belt
was designed in
accordance with ASAE 211.5 standard for V-belt drives for
agriculture
machines (American Society of Agricultural Engineers, 2001).
A
horsepower, variable speed, 180 VDC motor powered each drive
shaft.
-
57
Figure 37: The incline belt, variable speed V-belt aligning
system
Strips of Ultra High Molecular Weight (UHMW) plastic were used
on
the underside of the belts to maintain tension. These strips
were crowned so
that the belts in the center of the conveyor system were 0.75
inches higher
than the belts at the outside of the system. Grooves for each
belt were milled
in the UHMW plastic to keep the alignment of the V-Belts
consistent over the
entire width.
-
58
Color Sorting
Color sorter operations were significantly more involved than
the gap
belt concept. In order to devise the proper color sorting
technique, more
experimentation and investigation were required. This
investigation went
through three stages of development as listed below:
o Color sensing operations
o Color material removal concepts
o Off the shelf Color Sorters
The WECO Color Sorter
Color Sensing Concepts
Designing a system to use the difference in color between the
pods
and sticks was a difficult problem. The primary dilemma occurred
in
attempting to find a sensing method that could differentiate
between the sticks
and pods. Multiple Internet searches were conducted and several
sensor
manufacturers where contacted. Several companies were found that
made
sensors that could distinguish color. The majority of the
sensors were
photoelectric.
Sensors
One photoelectric color sensor, a R55 from Banner
Engineering,
similar to that shown in Figure 38, was purchased for $290 and
tested. It was
-
59
found that the sensor could distinguish between the pods and
sticks;
however, it could only do so within a very limited range of
about 5/8.
Through further research, it was found that laser color sensors
were available
that had ranges up to 13 inches. One such sensor was the SA1M
from Idec,
at a price of $1350. It is believed that this sensor could
accomplish the job
but that it was too cost prohibitive. All of the concepts for
removing the pods
or sticks once the color was determined relied on having an
array of the color
sensors aimed at the belt and then having one removal device for
each color
sensor. Conceptually, this could be a 36-inch wide belt with a
color sensor
and removal device at every inch. That would come to a price of
$48,600 for
the sensors alone, putting a production machine far out of the
$50,000
acceptable price range.
Figure 38: The R55 color sensor from Banner Engineering
-
60
Machine Vision
At this point, it was found that LEGO Company manufactured and
sold
a simple machine vision system for use on their RCX robots. The
system
could operate a switch when a target color was sensed and the
system could
distinguish between the color of a pod and a stick. The LEGO
system was
very primitive, allowing only three targets and only having the
ability to switch
one output on at a time. It was only used as a demonstration of
the
possibilities and as a building block from which machine vision
was
researched. The model shown in Figure 39 was built to show the
Chile Task
Force the capabilities of the vision system and to demonstrate a
possible
method for pod removal.
Figure 39: Lego machine vision concept model
Camera for Input
Pod Removal Arm
-
61
Machine vision was found to be a very complicated and
involved
method of sensing. The basic premise behind it is that a camera
is attached
to a processor, which analyzes the array of pixels to make some
sort of
decision about what the camera sees. From that information,
logical
decisions can be executed or measurements can be taken.
Through
research, it was found that the majority of machine vision
systems focus on
black and white camera inputs. They then determine shapes and
sizes based
on the light and dark pixels. Very few color-sensing systems
were found. Of
these, the majority could handle fewer than five targets for
each setup. With
such a limitation, at least eight of those systems would be
required to obtain a
minimum resolution of one-inch over the entire width of a
40-inch wide belt.
The only system that seemed to come close to providing what
was
needed was the CV-700 from Keyence. The CV-700 allows for
four
programs, each having eight target windows operating
simultaneously, giving
the possibility of 32 target areas. The CV-700 costs
approximately $7,000 for
a 1-camera setup or $9,000 for a 2-camera setup. A second camera
may be
necessary to get adequate resolution over the whole width of a
belt and would
feed its data into the same processor used by the first
camera.
Sensing Methodology
It was decided that the color sensing method should sense the
red
color of the pods, not the green or brown color of the sticks.
This was
-
62
decided upon because the presumably larger cross section should
provide an
easier target to sense and extract. A setup to remove sticks
would require a
much higher resolution extraction method and also would sense
the stems
that were the same color as the sticks, kicking out the stems
also.
Unfortunately, stems are attached to pods, causing the pods to
get extracted
to the same area as the sticks. Of course, this means that
absolutely no
separation is taking place.
If, in a future setup, it becomes necessary to sense the green
color of
sticks, it would have to become a two-step processing system.
That is to say
that after the processor checked the color of the pixels, it
also would have to
check the shape of that green color to determine if it was a
stick or a stem.
This shape recognition is an algorithm that is common in the
area of digital
machine vision, yet is out of the scope of this project. For
this project, it was
only desired to sense a red color and then extract that red
color because the
red is unquestionably a pod.
-
63
Color Sorter Removal Concepts
Once a system for identifying the color of material on a belt
was
established, a method for removing the material, either sticks
or pods was
required. Once again, the Funnel Approach to Design (FAD) was
used to
brainstorm multiple ideas before deciding on one to pursue.
Initial
brainstormed ideas included an indexing head with controlled
vacuum holes
or a vacuum arm that could move down, pick up material and kick
it onto
another belt.
The principle behind the indexing head pick up was as follows:
A
sensor could identify all of the sticks and pods on a section of
belt. A large
plate, perhaps 36 inches by 36 inches, could descend on the
belt. That plate
would have vacuum holes every inch or half inch, controlled by
individual
electronic valves. The controller could determine which of the
valves to open
and close to pick up only the pods, based on the input from the
color sensing
method. The head could then rotate out to the side of the belt
where the pods
could then be deposited by releasing the vacuum.
Ideally, while that was occurring, another identical plate would
be
removing material from the next section of belt. This concept
was not
pursued because of the high cost of manufacturing such a setup.
The large
number of valves would place the cost higher than the project
constraint of
having a production model under $50,000. High reliability valves
would cost
-
64
approximately $50 each. A one-inch resolution over a 36-inch by
36-inch
vacuum plate would require 1,296 valves. Each plate would cost
$64,800 or
$129,600 for the above-mentioned two-plate setup. The setup also
would not
pass the product constraint of maintaining a 48,000 lbs h-1 flow
rate. It was
anticipated that the belt would need to stop to accomplish the
task, therefore,
making it difficult to maintain the high flow rate.
The other concept explored for removing the material of a
certain color
was an array of arms that would move vertically, down toward the
belt to
vacuum up individual pods. The controller would actuate each arm
when the
specified color was viewed. A four-bar mechanism attached to
each arm
would operate a sweeper as the arm went down to pick up a pod.
The
sweeper would cross the vacuum tip as the arm moved back up,
kicking the
pod onto a belt moving perpendicularly to the main belt.
A very positive aspect of this proposed method was that it would
not
require the material to be in a single layer. The color sensors
would see
whatever was on the top layer and the arm would