DESIGN AND DEVELOPMENT OF CHESTNUT
HARVESTING TECHNIQUES
By
Mark E. De Kleine

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE
Biosystems Engineering
2011

ABSTRACT
DESIGN AND DEVELOPMENT OF CHESTNUT
HARVESTING TECHNIQUES
By
Mark E. De Kleine

A harvesting technique applicable for small chestnut orchards was designed and tested
for chestnuts. The originality of the technique was a single-stage vacuum and separation
process. A harvesting chamber was designed with inline obstructions which separate chestnuts
from debris. The chestnuts were sorted into a deposit chamber and debris and waste was
discharged to the environment.
Tests were made to determine chestnut harvesting performance as affected by the
proportion of chestnuts to debris, and the feed rate into the system. The validation of the
harvesting process was characterized by chestnut loss and separation efficiency. The quantity of
chestnuts and the feed rate interaction significantly affected the harvesting performance
parameters. Chestnut losses were as low as 1.3% while the separation efficiency was as high as
81%.

Copyright by
MARK E. DE KLEINE
2011

ACKNOWLEDGMENTS
Agriculture has been a part of my life, all of my life. My first exposure to the media was
sitting on a giant pumpkin in Grand Haven, MI, when I was 3 or 4 years old. My experience
with orchards and machinery has led me through many exciting adventures in Agricultural
Engineering, including this. I received my B.S. in Biosystems Engineering from Michigan State
University in the Spring of 2005. Six years later, I have had the fortunate, and unique,
opportunity to re-walk those steps with the help of some amazing people.
I would like to express my appreciation and thanks to the flowing people who contributed
to this project:
Dr. D. Guyer, who as my major professor provided insight, supervision, and assistance
throughout this entire project,
Dr. A. Srivastava, Dr. C. Radcliffe, and Dr. D. Fulbright, for providing guidance and
encouragement as a committee in both graduate studies and personal growth,
Dr. G. Van Ee and Dr. F. Bakker-Arkema, for their personal mentorship and advice,
De Kleine Orchards LLC., Chestnut Growers Inc., MSU‟s Biosystems and Agricultural
Engineering Services staff, Mr. Jaime Burns, Mr. Irwin Donis-Gonzales, Mr. Mario Mujadanio
and Ms. Lindsey S. New, for her support and exceptional knowledge of linguistics.
A small percentage of the world‟s population is involved in agriculture. Some sources
say less than two percent of the population is responsible for feeding the rest. My far sighted
goal for this project was to ultimately lay a foundation for engineering education. Those who
understand engineering, at its‟ core, understand how engineering an educational system can
benefit people everywhere. It is my dream that one day an engineered educational agriculture
program will thrive in Michigan.
iv

The picture of Sparty above my desk, signed by Sparty himself, reads:
dream BIG
-Sparty #1

Thank you for taking an interest in me, my thesis, and agricultural engineering!
Mark De Kleine
dekleine@msu.edu

v

Table of Contents
List of Tables ............................................................................................................................... viii
List of Figures ................................................................................................................................ ix
1

Introduction ............................................................................................................................. 1

2

Literature Review .................................................................................................................... 2

3

Objective .................................................................................................................................. 4

4

Justification .............................................................................................................................. 5

5

Design Process ......................................................................................................................... 9
5.1

Separation Concepts and Strategies ................................................................................. 9

5.1.1

Fluidized Beds ........................................................................................................ 14

5.1.2

Cyclone Separation Systems ................................................................................... 15

5.1.3

Physical Properties of Chestnuts and Burs ............................................................. 18

5.2

Machine Design.............................................................................................................. 20

5.3

Platform .......................................................................................................................... 21

5.3.1
5.3.2

Tubing ..................................................................................................................... 23

5.3.3
6

Frame and Engine/Fan drive ................................................................................... 22
Materials used for separation chambers .................................................................. 23

Iterative Designs .................................................................................................................... 23
6.1
6.2

Air Flow Separator ......................................................................................................... 25

6.3

Dowel Grid Separator..................................................................................................... 27

6.4

Axial-Dowel Wheel Separator ....................................................................................... 31

6.5

Radial-Dowel Wheel Separator ...................................................................................... 34

6.6

Deflection Separator ....................................................................................................... 37

6.7

Momentum Transfer Separator ...................................................................................... 39

6.8
7

Procedure ........................................................................................................................ 24

Saltation Sieve Separator (SSS) ..................................................................................... 41

Experiment Design ................................................................................................................ 44
7.1.1

Performance Evaluations ........................................................................................ 46
vi

8

Results and Discussions......................................................................................................... 47
8.1

Chestnut Loss ................................................................................................................. 48

8.2

Separation Efficiency ..................................................................................................... 52

9

Conclusions ........................................................................................................................... 57

10

Future Research ................................................................................................................. 57

APPENDICIES ............................................................................................................................. 59
11

Appendix A: Engine hp and Air Flow calculations ........................................................... 60

12

Appendix B: Velocity Profiles ........................................................................................... 63

12.1 B.1: Air Flow Separator ................................................................................................. 63
12.2 B.2: Dowel Grid Separator ............................................................................................. 69
12.3 B.3: Axial-Dowel Wheel Separator ............................................................................... 73
12.4 B.4: Radial-Dowel Wheel Separator .............................................................................. 76
12.5 B.5: Deflection Separator ............................................................................................... 76
12.6 B.6: Saltation Sieve Separator ........................................................................................ 76
REFERENCES ............................................................................................................................. 82

vii

List of Tables
Table 1. Acreage estimated to economically afford a harvesting machine at various yields. For
interpretation of the references to color in this, and all other figures, the reader is referred to the
electronic version of this thesis. ...................................................................................................... 9
Table 2. Reproduced from Marcus et al. 1990.............................................................................. 13
Table 3. Air velocities sufficient for vacuuming chestnuts, chestnuts in burs, and burs.
Reproduced from Guyer and Kang (2009) ................................................................................... 18
Table 4. Physical properties reported by Yildiz et al. (2009) for wild Turkish chestnuts ............ 19
Table 5. Air Flow Separator testing data. ..................................................................................... 27
Table 6. Dowel Grid Separator testing data. ................................................................................. 30
Table 7. Axial-Dowel Wheel Separator testing data. ................................................................... 33
Table 8. Radial-Dowel Wheel Separator testing data. .................................................................. 36
Table 9. Deflection Separator testing data. ................................................................................... 38
Table 10. Momentum Transfer Separator testing data.................................................................. 40
Table 11. Independent Variables and their levels ......................................................................... 45
Table 12. Testing data for chestnut loss and separation efficiency as a percentage ..................... 48
Table 13. Analysis of variance table for chestnut loss data .......................................................... 49
Table 14. Analysis of variance table for separation efficiency data ............................................. 52
Table 15. Linear regression analysis for chestnut loss based on the number of chestnuts and feed
rate................................................................................................................................................. 56
Table B.16. Air Flow Separator air velocities. ............................................................................. 64
Table B.17. Dowel Grid Separator air velocities. ......................................................................... 72
Table B.18. Axial-Dowel Wheel Separator air velocities............................................................. 75
Table B.19. Momentum Transfer Separator air velocity measurements. ..................................... 81

viii

List of Figures
Figure 1. $7,000 machine pay off prediction point for 3000 lbs/acre. ............................................ 7
Figure 2. $40,000 machine pay off prediction point for 3000 lbs/acre........................................... 8
Figure 3. Solids velocity for various air velocities as geometric diameter increases ................... 11
Figure 4. Reproduced from Srivastava et. al. 2006, "A pneumatic conveying state diagram". .... 12
Figure 5. Material flow and characteristics inside of a conveying duct. ...................................... 13
Figure 6. Cyclone separation systems: (a) depositing chamber (b) depositing chamber with cross
current flow ................................................................................................................................... 16
Figure 7. A component view of the harvesting platform showing: (1) trailer, (2) engine, (3)
Fan/Blower, (4) Saltation Sieve Separator, (5) flexible harvest tube and nozzle; (6) fan/blower
discharge; (7) fan inlet. ................................................................................................................. 21
Figure 8. The harvesting platform. ............................................................................................... 22
Figure 9. Air Flow Separator and regions: (a) conveying region, (b) deposit chamber, (c) inlet. 26
Figure 10. Air Flow Separator ...................................................................................................... 26
Figure 11. Dowel Grid Separator and regions: (a) conveying region (CR), (b) deposit chamber
(DC), (c) inlet. ............................................................................................................................... 28
Figure 12. Dowel Grid Separator .................................................................................................. 29
Figure 13. Dowel Grid Separator conveying regions 1, 2, and 3.................................................. 29
Figure 14. Clogging in the conveying region of the Dowel Grid Separator. ................................ 31
Figure 15. Axial-Dowel Wheel Separator: (a) conveying region, (b) deposit chamber, (c) rotating
dowel cylinder, (d) deflection plate. ............................................................................................. 32
Figure 16. Axial-Wheel Separator. ............................................................................................... 32
Figure 17. Conveying region of the Axial-Dowel Wheel Separator, with components: (a) dowel
rod rotating cylinder and (b) deflection plate. .............................................................................. 33
Figure 18. Radial-Dowel Wheel Separator: (a) conveying region, (b) deposit chamber, (c) inlet.
....................................................................................................................................................... 35
Figure 19. Radial-Dowel Wheel Separator conveying region and (a) slotted inlet plane, (b) dowel
rod spindle, (c) deflection plate and (d) self cleaning apparatus. ................................................. 35
Figure 20. Deflection Separator: (a) conveying region, (b) deposit chamber and (c) inlet. ......... 37
ix

Figure 21. Deflection Separator conveying region with (a) dowel rods and (b) deflectors.......... 38
Figure 22. Momentum Transfer Separator: (a) momentum skids, (b) separation grid, (c)
deflection point. ............................................................................................................................ 40
Figure 23. Saltation Sieve Separator conveying region. ............................................................... 42
Figure 24. Saltation Sieve Separator (SSS). ................................................................................. 43
Figure 25. Saltation Sieve Separator components: (a) momentum skids, (b) hinged paddle, (c)
conveying region, and (d) deposit chamber. ................................................................................. 44
Figure 26. Chestnut loss percentage as affected by the number of chestnuts (Ch) and weight of
debris (D). ..................................................................................................................................... 49
Figure 27. Chestnut loss as affected by the number of chestnuts (Ch) and the feed rate (FR). .... 50
Figure 28. Chestnut loss as affected by weight of debris (D) and feed rate (FR). ........................ 51
Figure 29. Response surfaces for separation efficiency as affected by number of chestnuts (Ch)
and weight of debris (D). .............................................................................................................. 53
Figure 30. Separation efficiency as affected by the number of chestnuts (Ch) and feed rate (FR).
....................................................................................................................................................... 54
Figure 31. Separation efficiency as affected by weight of debris (D) and feed rate (FR). ........... 55
Figure B.32. Velocity measurement positions in the Air Flow Separator. ................................... 63
Figure B.33. Air Flow Separator air velocities at position 1. ....................................................... 65
Figure B.34. Air Flow Separator air velocities at position 2. ....................................................... 66
Figure B.35. Air Flow Separator air velocities at position 3. ....................................................... 67
Figure B.36. Air Flow Separator air velocities at position 4. ....................................................... 68
Figure B.37. Dowel Grid Separator air velocity measurement positions. .................................... 69
Figure B.38. Dowel Grid Separator air velocities at position 1. ................................................... 70
Figure B.39. Dowel Grid Separator air velocities at position 2. ................................................... 71
Figure B.40. Air Flow Separator air velocities at position 3. ....................................................... 72
Figure B.41. Axial-Dowel Wheel Separator air velocity measurement positions. ....................... 73
Figure B.42. Axial-Dowel Wheel Separator air velocities at position 1. ..................................... 74

x

Figure B.43. Axial-Dowel Wheel Separator air velocities at position 2. ..................................... 75
Figure B.44. SSS air velocity measurement positions. ................................................................. 76
Figure B.45. SSS velocity profile at position 1. ........................................................................... 77
Figure B.46. SSS velocity profile at position 2. ........................................................................... 78
Figure B.47. SSS velocity profile at position 3. ........................................................................... 79
Figure B.48. SSS velocity profile at position 4. ........................................................................... 80

xi

1

Introduction
In the past five years, chestnut (Castanea sp.) production in the United States has risen to

the highest levels since the early twentieth century. In 2007, the United States Department of
Agriculture (USDA) census lists 1200 chestnut farms totaling 3,335 acres of chestnuts; Michigan
chestnut growers account for twenty four percent, or 813 acres, of the total chestnut acres in the
US (United States Department of Agriculture: Census of Agriculture 2007). Forty percent of
these acres reported are non-bearing age. With the increase in chestnut production, and rising
labor costs associated with harvesting, an economical harvesting machine is desired.
Chestnuts grow in a protective shell, called a bur, and ripen on the tree during a summer
growing season. They are harvested from the orchard floor after they mature and fall to the
ground. Mature chestnuts either leave the bur completely or are partially contained inside the
bur. During harvest, orchard debris is additionally collected and harvested material typically
includes chestnuts, chestnuts partially in burs, chestnuts completely in burs, empty burs, and
foreign materials such as leaves, twigs, dirt/stones, and grass. Decisions on whether a bur
encased chestnut lying on the orchard floor is “good”, is subjective amongst Michigan chestnut
growers. Some growers consider these nuts to be immature and can be discarded. Others prefer
to retain these chestnuts.
Michigan chestnut growers need a reliable and cost effective method to harvest chestnuts
from their orchards. For most Michigan chestnut growers, current machines are unaffordable
based solely upon their chestnut yield. The average size chestnut orchard in Michigan is 5 acres;
the average size US orchard is less than 3 acres. Worldwide, chestnuts are typically harvested by

1

one of the following methods: 1) hand harvesting with hand held tools or gloves, 2) mechanical
sweeping, or 3) mechanical sweeping and vacuum combination.

2

Literature Review
Literature for harvesting chestnuts was not found in any published articles; however,

vacuuming systems are widely used throughout the agricultural industry. Coates and Lorenzen
(1990) successfully designed and built two vacuum harvesting machines for jojoba seeds.
Although the jojoba seed is smaller than the chestnut, the machine functionality can be
considered for chestnuts. Their harvesting system consisted of vacuuming heads traveling above
the ground, beneath the jojoba bushes. Harvested material is moved through a high velocity air
stream which separates material based on density. Seeds and other material fall to the bottom of
the separation chamber and exit through an airlock system. A positive pressure air system
conveys the seeds to a seed hopper for short term storage. Their conclusions are: a vacuum head
system of this type sufficiently collects seed from the ground, harvesting efficiency varied with
field conditions, and an automatic lateral control parameter was needed for the head design.
They replaced fan blades multiple times during harvest due to excessive wear from sand, dirt,
and rocks traveling through the fan. Finally, they recommend improving the air-cleaning system
as they deemed theirs “inadequate”.
Chestnut harvesting machines typically windrow material into a ground collection
system. Ground collection systems have two primary functions: material engagement and
delivery control. A common mechanical ground collection system uses fingers or paddles made
of rubber, to lift material from the ground. The material is then offloaded onto a conveyor,
which is typically slotted, holed, or made of rollers and chains. Cross flow air systems are used
to pneumatically sort lighter debris from the desired product. Vacuum ground collection systems
2

use negative air pressure and aerodynamic drag to lift material from the ground, through a nozzle
and hose. The harvested material is then conveyed through flexible tubing to the main
processing system of the machine. In some cases, these processing systems include a series of
mechanical scuffing devices which loosen chestnuts from debris.
FACMA, an Italian agricultural equipment manufacturer, has developed a machine
currently being used for harvesting chestnuts, hazelnuts, almonds, walnuts, coffee beans,
macadamia nuts, and olives. This machine consists of two rotating paddle disks in front of the
machine which windrow orchard material into a vacuum collection tube. The harvested material
is collected in an airlock sorting chamber. Material entering the airlock chamber encounters a
rigid mounted rubber baffle which drops the material to the air lock system, and prevents
material from entering the fan inlet. The material leaves the chamber, via an air lock paddle,
where it is dropped over an air stream. The material not separated is moved to a dual rotational
drum sieve for separation and shucking. A second air stream separates debris loosened in the
drum sieve and places it on the orchard floor. A positive pressure system is used to transport
chestnuts to a bagging station at the rear of the machine.
Compared to the current mechanical and vacuum harvesting systems, a single-stage
vacuum harvesting system has potential. Negative air pressure can be used to pick-up and
convey debris, and eliminate the need for a mechanical ground collection system. Because the air
is the product carrier, the need for additional components such as belts and chains is minimized.

3

3

Objective
The overall objective of the research was to design a single-stage vacuuming technique to

harvest chestnuts and sort field debris, and analyze the system performance based on chestnut
loss and separation efficiency.
Michigan‟s chestnut growers, and similarly sized producers, currently rely on hand labor
as the most economical form to harvest their chestnuts. These labor wages for harvest and
handling procedures are increasing resulting in higher operational costs. The availability of labor
is also decreasing and often hard to find. In order to help Michigan chestnut growers progress
and remain profitable, farm level harvesting economics must be addressed. The first objective
for this research is:
1. Define an economic affordability range for purchasing a harvesting machine, strictly
related to chestnut yield and orchard size.
A simple economic model can be used to predict a point in which purchasing a machine
can be more economic than paying labor wages for harvesting. Harvesting duration, labor wages
and pick up rates, along with machine costs impact this prediction point.
2. Design a single stage vacuuming system to harvest chestnuts and sort field debris.
Because costs associated with machinery depend largely on complexity, a single-stage
system is highly desirable. A single-stage system decreases the number of components;
potentially making a harvesting machine more economical for Michigan growers.
3. Evaluate the harvesting technique using two performance parameters: chestnut loss
and separation efficiency.

4

A vacuum harvester can be designed to harvest chestnuts using a single-stage separation
technique. The technique will sort field debris from chestnuts while minimizing chestnut loss.
The system needs to be a viable economic option for Michigan, and similar sized, chestnut
orchards.
4

Justification
In order to help developing chestnut growers progress and remain profitable, farm-level

harvesting economics were addressed. The economic model developed for a chestnut harvesting
was used to predict a point where purchasing and operating a harvesting machine is more
economical than hiring manual labor, based on varying parameters. The average sized chestnut
orchard in Michigan is 5 acres and less than 3 acres for the United States.
Harvesting chestnuts is a value-added process for Michigan growers and the economic
benefits are substantial for growers with smaller orchards. For example, purchasing an efficient
$7000 harvesting machine can be more economical than paying labor wages when a grower has
2 1/2 acres producing 2000 lbs/acre (based on economic model below). An in-orchard harvest
and separation process can lower a grower‟s cost associated with less efficient manual harvesting
and sorting. Chestnuts are typically harvested every two days, resulting in a high demand for
labor followed by an idle period. A harvesting machine can potentially reduce the peak labor
demand.
The economics of a vacuum harvester were calculated using procedures given by
Srivastava et al. (2006) and Bainer, Kepner and Barger (1955). For this analysis, the following
assumptions were used, and based on current economic values and information provided from
Michigan chestnut growers. These assumptions are:

5

1) Machine annual fixed costs are 15% of the purchase price.
2) An acre yield is 3000lbs.
3) A person can harvest 38lbs/hr.
4) A person is paid $8/hr.
5) The harvest season is 21 days.
6) A day consists of 8hrs of harvest time.
7) The initial cost of a machine is $7,000.00.
8) Repairs are necessary.
9) Annual usage is 150 hrs/yr.
10) A harvesting machine can harvest 6 acres in one day.
11) Machine function is limited to 12 acres.

The model used to predict an economic point for purchasing a machine is described
below:
Machine

(1)

(2)

(3)

6

Labor

AnnualWages[

$
Yield (lbs / acre)
]
*Wage($ / hr )
acre PickupRate(lbs / hr )

(4)

Figure 1 depicts the intersection point, or acreage needed to economically purchase a
$7,000 machine based on 3000 lb/ac yield and a manual labor pickup rate of 38 lbs/hr. The
acreage needed was approximated at 2 acres. Decreasing the orchard yield will increase the
acreage needed to economically afford the same machine, based on the same assumptions.

Figure 1. $7,000 machine pay off prediction point for 3000 lbs/acre.

When the purchase price of the machine was increased to $40,000 the acreage needed to
purchase a harvesting machine was approximated at 12 acres as shown in Figure 2. This
assumes a yield of 3000 lbs/ac and a manual labor pick up rate of 38 lbs/hr.

7

Figure 2. $40,000 machine pay off prediction point for 3000 lbs/acre.

Due to the size of Michigan chestnut orchards (~5ac) yield is an important factor in
determining economics. Various yields were evaluated using the economic model to account for
typical orchard yields in Michigan. Table 1 shows yield comparison with varying machine
purchase price. A harvesting machine which is economically feasible for a producer of average
2-12 acres is not known to exist. This is the target range for the economic justification.
A chestnut grower producing 1500 lbs per acre, on 4 acres, could purchase a $5,000.00
harvesting machine. That same grower would need approximately 33 acres to offset a
$40,000.00 machine, which are currently commercially available. The productivity of labor
work influences the break-even point and should be considered.

8

Table 1. Acreage estimated to economically afford a harvesting machine at various yields. For
interpretation of the references to color in this, and all other figures, the reader is referred to the
electronic version of this thesis.

Assumed Purchase Price of a Machine ($)
Yield
(lb/acre) $2,000.00 $5,000.00 $7,000.00 $15,000.00
500
none
none
none
none
1000
~ 6.5
~ 10
~ 15
~ 29
1500
~ 1.5
~4
~6
~ 12
2000
~ 1.5
~3
~3
~8
2500
<1
~2
~ 2.5
~6
3000
~ 0.5
~ 1.5
~2
~5

$40,000.00
none
> 50
~ 33
~ 21
~ 16
~ 12

When a grower‟s yield is 500lbs or less, hand labor is the most economic way to harvest,
based on this model. The target acreage for this model was less than 12 acres, which is common
for Michigan orchards. Predictions above this acreage need more emphasis on machine capacity
and if numerous machines would be needed. Grey cells in Table 1 are outside of the range of
this economic model and need reconsideration.

5
5.1

Design Process
Separation Concepts and Strategies
A vacuuming system was initially decided upon based on achieving our criteria for a

single-stage system. The design of this type of system is based on horizontal and vertical
pneumatic conveying of materials. Research has proven useful and these conveying systems are
used in various industries, including agriculture. Some advantages of pneumatic conveying are:
flexibility of routing from horizontal to vertical, multiple distribution locations, generally low
maintenance, and ease of user control. Some disadvantages are: high power consumption,

9

wearing of fan components due to material discharge, limited distance of conveyance, and the
complexity and unpredictable nature of fluid flow.
Vertical conveyance depends upon the terminal velocity of the chestnut. Terminal
velocity is defined as the force required to overcome gravitational and drag forces acting on an
object. It should be noted that the terminal velocity is not adequate for transport; rather it is a
quantitative description useful for initial design and fluidization of a particle. To suspend or
fluidize a chestnut in a vertical conveyance region, the air velocity should never be less than the
terminal velocity. A relationship exists between the air velocity and solids velocity which
contains a geometry variable, shown in equation 5. This relationship was used to design
geometry specific to sorting by density.
Marcus et al. (1990) reported an equation estimating the solids velocity (c):

c
 1  0.68d 0.92p0.5a 0.2 D0.54
v
Where:
c = solids velocity, m/s
v = velocity of air, m/s
ρa = density of air, kg/m

3

ρp = density of particles, kg/m

3

d = particle mean diameter, m
D = diameter of conveying tube, m

10

(5)

MATLAB was used to calculate and plot the velocity for solids as shown in Figure 3.
The diameter of the conveying tube was varied from 1 to 36 in. (0.0254 to 0.9 m). Diameters
over 36 in. (1 m) where considered too large for this scale system.

Figure 3. Solids velocity for various air velocities as geometric diameter increases
A distinguishable relationship exists between the terminal velocity and saltation velocity.
Chestnuts entering an air stream are subjected to aerodynamic drag effectively determining the
pick-up capability. The saltation velocity, or critical air velocity required for horizontal
conveyance, has been published for several agricultural commodities. Figure 4 shows a graph of
superficial air flow versus pressure drop per unit length. The inflection point separates the two
11

flow phases, dense and dilute. This inflection point was used to predict velocities required to
keep chestnuts in the dilute phase.

Figure 4. Reproduced from Srivastava et. al. 2006, "A pneumatic conveying state diagram".

Horizontal conveyance is classified in three phases: dilute phase, dense phase, and fixed
bed. Dilute phase refers to fully suspended particles moving separately through a conveyor pipe;
this occurs when the air velocity is much greater than the minimum saltation velocity for that
particular particle. Fixed bed formation is considered as material not moving, or settling, in the
conveyor pipe; this occurs when the air velocity drops below the minimum saltation velocity
needed. Dense phase conveyance refers to the in-between. In horizontal conveyance, material
generally bounces along the bottom of a duct. The distance between a bounce or wall interaction
is random and hard to predict, especially with non-uniform chestnut shape. Tilting a conveying

12

duct will increase the chestnut interaction with the wall, as depicted in Figure 5. Less air flow
results in dense phase conveyance, leading to bed formation.

Figure 5 Material flow and characteristics inside of a conveying duct.

Marcus et al. (1990) developed a table relating air velocities for horizontal and vertical
transport. They stated that in the case of fine particles, the horizontal conveying velocity ranges
from 3 to 5 times larger than vertical conveying velocities. They also state that the difference in
conveying velocities is much smaller for coarser particles. The velocities for various agricultural
products are shown in Table 2 (Marcus et al. 1990).

13

Table 2. Reproduced from Marcus et al. 1990.

Material
Alum
Calcium carbonate
Coffee beans
Hydrated lime
Malt
Oats
Salt
Starch
Sugar
Wheat

Bulk density
(kg/m3)
800
440
672
480
449
400
1440
640
800
769

Vertical
Conveyance
Velocity (m/s)
19.8
19.8
13.7
12.2
16.8
16.8
25.3
16.8
18.3
16.8

Horizontal
Conveyance
Velocity (m/s)
33.5
33.5
22.9
27.4
30.5
30.5
36.6
27.4
33.5
32

5.1.1 Fluidized Beds
A fluidized chamber design was initially considered as one potential method for
separating chestnuts from debris. Using an upward directed fluid passing through harvested
material was tried by Guyer and Kang (2009), in efforts to fluidized lighter particles. Fluidized
beds can be described as the Winkler process from the 1930‟s, or gas solid fluidization (Douglas
and Walsh, 1966). Coal is typically fluidized to separate undesirable materials or minerals
before entering a combustion process. Fluidized beds have been used to sort agricultural
products spanning back 100 years or more.
The fluidization process allows separation to take place according to varying densities.
Zaltzman et al. (1983) studied and tested fluidized bed mediums as a separation process. An
analytical model was developed for this separation process (Zaltzman 1986) based on
gravitational motion of a sphere in a fluidized bed, which was tested in a laboratory by Mizrach
et al. (1984). Further work was conducted for separating flower bulbs from stones and clods
using the fluidization process (Zaltzman et al. 1985). Zaltzman and Schmilovitch (1986)
14

leveraged this process for sorting potatoes from stones and clods. Research expanded into other
agricultural commodities investigating the separation potential related with density and quality.
The relationship between density and quality has been observed in potatoes, tomatoes, peaches,
peas, pecans, citrus fruits, watermelons, and small seed and grains (Zaltzman et al. 1987). They
concluded that quality changes due to maturity or quality can generally be associated with a
change in density.
From the previous research studies, sorting chestnuts from debris based on density
characteristics seems feasible. Although density associated with chestnut fruit quality has not
been published, Michigan growers routinely dip harvested chestnuts in a water bath and scoop
away the “floaters”. This processing practice leans toward Zaltzman‟s observations and general
predictions associated with other agricultural commodities.
5.1.2 Cyclone Separation Systems
Agriculture has employed the use of cyclone separation systems along with material
processing industries, product handling industries and food industries. Cyclone separation
systems are widely studied and are extremely efficient. These systems separate material from an
air stream by exerting forces on the material. Typically the air stream enters a cylinder at a
tangential point in the sidewall. Gravitational forces, inertia forces, and friction, contribute to the
effectiveness of a cyclone separation system.
Cyclone separation chambers used for sorting include two basic types: depositing
chamber, and depositing chamber with cross-current flow. Inlets vary from tangential, spiral,
and axial entrances. Each system is designed according to a specific need. The depositing
chambers separate materials based on their densities and differing velocities. For a standard
depositing chamber, the air velocity drops due to a diameter increase and gravitational forces
15

acting on the material become influential. The heavier material tends to drop from the air stream
first. Figure 6 represents a depositing chamber and a depositing chamber with cross current
flow.

Air Discharge
Air Discharge

Material In

Material In

Solids Out

(a)

(b)

Figure 6. Cyclone separation systems: (a) depositing chamber (b) depositing chamber with cross
current flow

A depositing chamber with cross current flow closely resembles the initial design
consideration for this research. Air flow enters the depositing chamber from the bottom and
exits through the top, carrying lighter material and debris. A single-stage vacuuming system
allows for this type of design but mass flow must be taken into consideration. Drawing air from
two places increases the pressure drop across each path, requiring additional horsepower to
maintain sufficient air velocity at the suction nozzle.

16

In equilibrium state, the forces on a particle are the aerodynamic or drag force and the
weight of the particle minus the air lifting force (Marcus et al. 1990). This assumption holds true
for a spherical shape. Theoretically, a chestnut was assumed to be spherical for calculations.
Marcus et al. (1990) provide the following force balance representation:
FD  FG  Flift

(6)


FD  CD A w f 2
2

(7)

Where:

FD = drag force
FG = gravitational force
Flift = lifting force
CD = drag coefficient
A = area of sphere
ρ = density of air
wf = settling velocity

For a cross current separation chamber the separation efficiency depends upon the
settling velocity. Marcus et al. (1990) calculated the theoretical efficiency of three different
sized separator diameters: 3m, 6m, 25m. The efficiency of separation decreased as diameter
increased due to geometry and the forces acting upon a particle inside of a depositing chamber.
This research builds upon these principles and applies them specifically to harvesting chestnuts
in Michigan.
17

5.1.3 Physical Properties of Chestnuts and Burs
Guyer and Kang (2009) reported the following air velocities related to chestnuts,
chestnuts in burs, and burs, shown in Table 3. The air velocity required to pick up chestnuts was
reported as 19.3 m/s. From Figure 3, the 19.3 m/s air velocity plot yields a solids velocity of
approximately 15 m/s, in a 8 inch tube. This velocity agrees with the terminal velocity of
Turkish chestnuts, reported in Table 4.. The difference can be attributed to the particle or
geometric mean diameter between the chestnut species being tested.
Table 3. Air velocities sufficient for vacuuming chestnuts, chestnuts in burs, and burs.
Reproduced from Guyer and Kang (2009)
Weight
(g)
6.06
8.86
Empty
21.16
burs
46.24
49.49
13.3
Burs with
25.23
Nuts
39.46
7.24
Nuts
8.49
13.54
Material

Length
(mm)
78.56
77.91
73.68
82.26
91.17
64.37
66.34
65.54
33.83
33.95
35.2

Width
(mm)
48.76
79.19
71.86
79.76
84.42
63.73
59.67
64.21
32.39
30.59
33.47

Suction
Air
Thickness M.C. Pressure Velocity
(mm)
(%,w.b.) (mm,wg)
(m/s)
34.1
10.73
82
12.2
54.64
10.61
82
12.2
33.31
63.56
104
13.1
59.96
73.68
104
13.1
74.84
68.11
104
13.1
51.53
32.11
104
13.1
47.7
66.71
104
13.1
59.8
57.2
104
13.1
18.93
15.88
129
19.3
17.07
42.29
129
19.3
23.68
46.6
129
19.3

Yildiz et al. (2009) reported physical properties of wild chestnuts grown in Turkey.
Included in their analysis is terminal velocity for chestnuts, bulk densities, and geometric
averages. Table 4 is reproduced from a collection of tables by Yildiz.

18

Table 4. Physical properties reported by Yildiz et al. (2009) for wild Turkish chestnuts
Turkish Chestnut Properties at 51.32% m.c.d.b.
Length (mm)
Weight (mm)
Thickness (mm)
Geometric mean diameter (mm)
Sphericity

21.79
23.94
14.55
19.62
0.889

Bulk Density (kg/m3)
Terminal Velocity (m/s)

585
14.51

Two varieties of chestnuts were used in this study: Colossal and Chinese. Both varieties
differ from the wild Turkish chestnuts in all categories listed in Table 4. Colossal chestnuts are
generally larger with more mass and Chinese chestnuts are smaller than both the wild Turkish
chestnuts and Colossal. Michigan chestnut growers have several varieties currently planted;
although the majority of chestnuts produced are either Colossal or Chinese.
Chestnuts were assumed to be round for computational purposes but typically chestnuts
are flat on one side; the actual projected area of a chestnut in the air stream can be rectangular
and significantly smaller than a chestnut‟s spherical projected surface area. A chestnut‟s obscure
shape has a different saltation velocity than a round chestnut and will vary, depending on the size
and shape of the chestnut. As the size of chestnuts increase the difference between the terminal
velocity and saltation velocity decreases. Smaller chestnuts will have a larger saltation velocity
than larger chestnuts.

19

5.2

Machine Design
Saltation velocity for wheat was determined through experiments conducted by Shen,

Haque, Posner (1994). They used stepwise regression to formulate the best multi variable
2

regression equation. The equation provided a R value of 0.974.
V  22.70  2.01* m.c.  3.61* Q  1.19*103 * (d p * m.c.)
8.5*104 ( Ap * T )  0.28*( Ap * m.c.)

(8)

Where:
V = Saltation velocity (m/s)
m.c. = Moisture content (%, w.b.)
Q = Feed rate (kg/s)
dp = Particle geometric mean diameter (m)
T = Conveying air temperature (K)
2
Ap = Specific surface area of the grain (cm /g)

Using numerical values reported by Yildiz et al. (2009) for the moisture content, particle
geometric mean diameter, and specific surface area for wild Turkish chestnuts, combined with
standard air temperature and an assumed feed rate of 0.15 kg/s, the saltation velocity required to
keep chestnuts in the dilute phase was approximated at 4660 ft/min (23.7 m/s). Adequate
conveyance velocity should not be less than the saltation velocity or drop below 4660 ft/min
unless separation or bed forming is desired.

20

5.3

Platform
A component view of the harvesting prototype platform is shown in Figure 7. A gasoline

engine (2) was used to transmit power to a fan/blower (3). The fan/blower inlet (7) was used to
create the vacuum needed to pick up chestnuts. Chestnuts and debris were vacuumed from the
ground and conveyed through flexible tubing (5) and through the separation device (4). Debris
was expelled to the orchard floor through the blower discharge (6).

(3)

(7)
(2)

(1)
(4)

(5)

(6)

Figure 7. A component view of the harvesting platform showing: (1) trailer, (2) engine, (3)
Fan/Blower, (4) Saltation Sieve Separator, (5) flexible harvest tube and nozzle; (6) fan/blower
discharge; (7) fan inlet.

21

Figure 8 The harvesting platform.

5.3.1 Frame and Engine/Fan drive
A Honda GX670cc, 24.0 hp V-twin engine (http://engines.honda.com/) was used to drive
a HP-8D18 high pressure radial blower, from Cincinnati Fan (http://www.cincinnatifan.com).
These two components were significantly larger than necessary but were available from the
department. The blower has an 18 in. wheel, 8 in. inlet and outlet flanges, and a capacity of 7000
CFM. The blower wheel was mounted to a fan shaft which was supported by two pillow block
bearings. The blower base and engine were mounted to a three-wheel pull type trailer using
various brackets made from angle iron. A 7 in. dual-belt drive pulley was mounted to the engine
drive shaft. Twin V-belts transmit engine power to the blower drive shaft, which had a 4 in. dual

22

drive pulley. The transmitting speed ratio is 1.75. This ratio was designed so both the engine
and blower are operating in their most efficient operating range.
5.3.2 Tubing
A 22 ft., 4 in. diameter, flexible tube was used for the suction hose. A nozzle was made
of a 4 in. piece of 4 in diameter, 1/4 in. steel tube, and functioned as the product engagement
device. A handle, carrying strap, and suction hose support structure were fabricated to ease
operation and testing comfort. The suction hose was connected to the inlet port using a 4 in.
length of 4 in. diameter schedule 40 PVC pipe and a standard 4-6 in. diameter hose clamp. The
sorting chamber discharge port consists of (1) 6 in. diameter schedule 40 PVC molded flange, (1)
90deg elbow, and (4ft.) 6 in. schedule 40 PVC pipe. A port adapter was fabricated to bolt on to
the blower inlet, allowing the 6 in. PVC discharge pipe to fit securely inside.
5.3.3 Materials used for separation chambers
Each testing chamber was constructed from 5/8 in. particle board, having dimensions of
30 in. x 36 in. x 6 in. A 30 in. x 36 in. sheet of 3/8 in. plexiglass was used for one outer face.
Standard 1 1/4 in. deck screws were used to mount the internal components and the plexiglass to
the chamber base. A 4 in. PVC plumbing flange was used for the inlet of the chamber. A 6 in.
schedule 60 PVC molded flange was used as the discharge port. Foam, 1/2 in., weatherproofing
seal was used between the mounting faces of the SSS and the inlet, outlet, and plexiglass face.

6

Iterative Designs
Each chamber was designed to a specific hypothesis and approach for chestnut harvesting

and separation, and after testing a chamber, the hypothesis was evaluated and either accepted or
rejected. A rejected hypothesis was reformulated based on the knowledge and observations from
23

the testing and applied to the design concepts for the next chamber. Listed below are the
chamber designs in order of progression throughout this research project. A total of six
chambers were designed and tested.
Two specific characteristics used to completely reject a hypothesis included:
1) Clogging or jamming which requires manual intervention.
2) High chestnut loss during a test.
If either of the above occurred, the chamber design was consider inadequate and the
hypothesis was rejected.
6.1

Procedure
The following describes the testing procedure:
1) Set the engine rpm.
2) Record engine drive shaft and fan motor shaft rpm using a digital tachometer.
3) Lower the velometer into the chamber through drilled holes at specific, geometric,
points of interest and record velocities.
4) Prepare chestnuts and materials for harvesting and record weights and number of
chestnuts.
5) Begin harvesting.
6) Record weights of separation chamber material and blower discharge collection lug,
and count the number of chestnuts collected.
7) Return to step 1 after adjusting the rpm (if necessary).

24

To measure the air stream velocity and examine the velocity profile, the velometer was
lowered into the harvesting chamber. Specific geometric points of interest include: areas behind
obstruction devices, at bends in the air stream, the top and bottom of the conveying region, and
various locations in the sorting chamber. Measurements were taken at the closest inch intervals.
Each hole was plugged after taking the velocity measurement. Air flow data and air speed
corresponding to engine RPM are found in Appendix B: Velocity Profiles.
6.2

Air Flow Separator
The air flow separator design, shown in Figure 9 and 10, resembles a cyclone separation

system. A direct manipulation of the conveying air stream utilizes momentum as a separation
technique. Material is moved from left to right: the suction port is horizontal and the discharge
port is vertical.
The hypothesis for this design is based on air velocity and physical geometric
relationships. As an air stream passes through changing geometry, or orifices, the velocity of the
air changes. Increasing a diameter of a hose, or geometry inside the sorting chamber, lowers the
velocity of the air below the saltation velocity of chestnuts. If the velocity of the air can be
lowered below the saltation velocity of the chestnuts, but not below that of debris, separation of
chestnuts from debris can occur.
In order to shorten the amount of distance needed for adequate air stream separation, a
bend, or redirection, was used for the incoming material. This redirects chestnuts towards an
area of lesser air velocity. These chestnuts should not continue in the air stream however, lighter
material should be removed.

25

Material In
Material Out

(a)

(c)

(b)

Figure 9. Air Flow Separator and regions: (a) conveying region, (b) deposit chamber, (c) inlet.

Figure 10. Air Flow Separator

26

Table 5. Air Flow Separator testing data.
Engine
RPM
Air Flow
Separator

Material input
Total
Marked
Chestnut
Weight Chestnuts
Weight

Collected
Marked Discharge
Weight Chestnuts
Weight

1580

7lbs 10oz

50

1lb 14oz

7lb 10oz

50

0

2580

7lbs 10oz

50

1lb 14oz

7lb 10oz

50

0

3010

7lbs 10oz

50

1lb 14oz

7lb 10oz

50

0

Discharge weight was negligible in all three tests as shown in Table 5. Separation of
material from the air stream was one hundred percent. Air flow problems are complex and air
layer shrinkage was not initially considered. An observation made during the velocity profile
testing shows a shrunken air layer profile specifically at the first bend inside of the chamber.
The velocity of the air, at the bend, is greater in magnitude but smaller in depth. The shrunken
air layer was not adequate to support debris, or chestnuts.
Engine RPM‟s were varied to examine the effects on the air velocity profile and
performance parameters. For the Air Flow Separator, the variation in engine RPM did not have
any effect on the separation performance.
Should a chestnut grower prefer no sorting, the Air Flow Separator is effective. Due to
the lack of sorting, this chamber did not meet the initial objectives and the hypothesis was
reformulated.
6.3

Dowel Grid Separator
The Dowel Grid Separator design consists of wooden dowel rods placed horizontally in

the conveying region, as shown in Figure 11. This design exploits momentum change of the
27

conveying material and uses a sizing grid to separate chestnuts into the deposit chamber. Figure
13 depicts the dowel rods and their respective region; dowels are spaced 1 inch apart in region 1,
1.5 inches in region 2, and 2 inches in region 3. In the y-plane of region 3, no dowel rod was
within 3 inches of another. This design attempted to force chestnut burs to deflect around the
dowel rod. The upward incline of dowel rods in the middle portion of the conveying region was
designed to influence the flow of burs and material upward. Chestnuts should drop out of the air
stream while debris will only be deflected, but still carried out of the separation chamber.

Material Flow

(a)
(c)
(b)

Figure 11. Dowel Grid Separator and regions: (a) conveying region (CR), (b) deposit chamber
(DC), (c) inlet.

28

Figure 12. Dowel Grid Separator
Region 1

Region 2

Region 3

Figure 13. Dowel Grid Separator conveying regions 1, 2, and 3.

29

Table 6. Dowel Grid Separator testing data.

Material In
Engine
RPM
Dowel Grid
Separator

Collected

Total
Marked Chestnut
Weight Chestnuts Weight

Weight

Marked
Chestnuts

Discharge
Weight

2075

7lb 8oz

50

1lb 13oz

1lb 1oz (DC)
4lb 3oz (CR)

23 (DC)
27 (CR)

2lb 4oz

2526

7lb 6oz

50

1lb 13oz

3lb 13oz (DC)
2lb 2oz (CR)

26 (DC)
24 (CR)

1lb 7oz

3030

7lb 8oz

50

1lb 13oz

2lb 6oz (DC)
3lb 1oz (CR)

19 (DC)
31 (CR)

2lb 1oz

Chestnut burs frequently became lodged between the dowel rods, initiating a clog as
shown in Figure 14. Chestnut burs became “Velcro” like, because of their spines, in a clogging
situation. The separation based on momentum change was adequate when clogging was not
present. No marked chestnuts were discharged through the blower. Due to the clogging, this
chamber did not meet the initial objectives and the hypothesis was reformulated.

30

Figure 14. Clogging in the conveying region of the Dowel Grid Separator.
Three RPM‟s were used to evaluate the harvesting performance for the Dowel Grid
Separator. Based on the results, increasing the engine speed beyond 2075 RPM had no impact
on separation. The air velocity was 3800 ft/min measured at the chamber inlet.
6.4

Axial-Dowel Wheel Separator
The Axial-Dowel Wheel Separator utilizes the horizontal dowel rod design from the

Dowel Grid Separator but incorporates it into a rotating cylinder, as shown in Figure 15. The
rotation of the dowel rod cylinder is done by hand when a clog starts to form. The cylinder is
mounted 2.5 inches from the top of the conveying region to produce a shucking effect. A small
deflection plate is mounted downstream of the cylinder to ensure no material can pass unaffected
through the conveying region. An incline plane funnels chestnuts passing through the
obstruction wheel back to the sorting chamber.

31

Material Flow

(d)
(a)
(c)
(b)

Figure 15. Axial-Dowel Wheel Separator: (a) conveying region, (b) deposit chamber, (c) rotating
dowel cylinder, (d) deflection plate.

Figure 16. Axial-Wheel Separator.

32

(b)

(a)

Figure 17. Conveying region of the Axial-Dowel Wheel Separator, with components: (a) dowel
rod rotating cylinder and (b) deflection plate.

Table 7. Axial-Dowel Wheel Separator testing data.
Material In
Engine
RPM
Axial-Dowel
Wheel
Separator

Collected

Total
Marked Chestnut
Weight Chestnuts Weight

Marked Discharge
Weight Chestnuts Weight

1976

8lb 10oz

50

1lb 14oz

2lb 1oz

50

6lb 8oz

2467

8lb 10oz

50

1lb 13oz

2lb 1oz

50

6lb 2oz

3030

8lb 7oz

50

1lb 14oz

2lb

48

6lb

Small chunks of chestnut bur became lodged between the dowel rods of the rotating
cylinder. Although chestnut separation occurred, the buildup of debris over a harvesting period
could produce undesirable sorting effects. A self-cleaning procedure for this type of separation
system should be considered to ensure proper system functionality. Dowel rods inside of the
rotating cylinder could also rotate to help facilitate self cleaning. This would require a gearing
mechanism similar to a standard planetary and sun gear set-up. While this design may be
suitable for harvesting and sorting chestnuts, the addition of moving components and clogging
did not meet our initial requirements. The hypothesis was reformulated to address clogging.

33

Engine RPM‟s were varied to examine the effects on the air velocity profile and
performance parameters. For the Axial-Dowel Wheel Separator, the variation in engine RPM
did not have any effect on the separation performance.
6.5

Radial-Dowel Wheel Separator
The Radial-Dowel Wheel Separator design was to prevent clogging by using the self

cleaning apparatus shown in Figure 18. Dowel rods are placed vertical in the conveying air
stream and are mounted on a rotating spindle. Mounting the dowel rods vertically utilizes the
momentum principles, described above, for sorting. The spacing for the spindle wheel dowel
rods is one inch, which was arbitrarily selected. A second set of dowel rods is used to clean, or
dislodge, any material becoming stuck between the spindle wheel dowels, as shown in Figure 19.
Downward sloping planes in the conveying region funnel separated material to the sorting
chamber. Slots are cut in the inlet planes to allow rolling chestnuts to pass into the deposit
chamber (a), as shown in Figure 19.

34

Material Flow

(c)
(a)
(b)

Figure 18. Radial-Dowel Wheel Separator: (a) conveying region, (b) deposit chamber, (c) inlet.

(c)
(d)

(a)
(b)

Figure 19. Radial-Dowel Wheel Separator conveying region and (a) slotted inlet plane, (b) dowel
rod spindle, (c) deflection plate and (d) self cleaning apparatus.

35

Table 8. Radial-Dowel Wheel Separator testing data.

Material In
Engine
RPM
Radial-Dowel
Wheel
Separator

Collected

Total Marked Chestnut
Weight Chestnuts Weight

3223

7lb 6oz

100

3lb 4oz

3223

9lb 2oz

94

3lb

3223

-

-

-

Weight
3lb 2oz (CR)
3lb 2oz (DC)
3lb 12oz (CR)
2lb 9oz (DC)
-

Marked Discharge
Chestnuts Weight
94

7oz

85

10oz

-

-

Clogging was apparent in both the conveying region (CR) and deposit chamber (DC)
despite the efforts of a self cleaning system. Frequently, chestnuts became lodged between the
dowel rods on the spindle, causing a physical jam. The torque required to dislodge a stuck
chestnut is undesirable when considering harvesting quality. Chestnuts also became stuck in the
self cleaning apparatus itself. Because chestnuts differ in size and shape, spacing of fixed
components becomes difficult. Slots designed for large chestnuts increase the opportunity for
debris to fall through. Conversely, slots designed to limit the debris based on a smaller size,
promote clogging. No suitable slot size or dowel rod spacing was determined to prevent
clogging. Due to the addition of moving components and the clogging, the hypothesis was
reformulated. Only two test runs were needed to reject this hypothesis.
Tests were run using the same RPM, compared to varying RPM‟s in the previous three
chambers. The previous test results do not vary significantly for differing RPM‟s. The inlet air
velocity was 7500 ft/min. A large inlet velocity was used to keep burs from becoming stuck in
the conveying region. Although 7500 ft/min is larger than what was predicted, 4660ft/min, this
design was deemed unsuccessful based on the high chestnut loss.
36

6.6

Deflection Separator
The design of the Deflection Separator, shown in Figure 20, consisted of 3/8 inch dowel

rods mounted in the center of the conveying region, at an angle of 67.5 degrees with respect to
the top of the chamber. Mounted on the sides of the conveying region are six deflectors. The
deflectors have holes to allow air to pass. The first dowel is located 4 inches into the conveying
region. Spacing for mounting the deflectors and dowel rods was selected to be 4 inches. This
spacing was based on a full bur, partially opened, being less than four inches.
Chestnuts and debris entering the conveying region are subjected to momentum changes
from impacting the obstruction devices. Chestnuts deflected from the air stream wind up in the
deposit chamber. The bottom planes of the conveying region are angled to allow chestnuts to
funnel back to the deposit chamber.

Material Flow

(a)
(c)
(b)

Figure 20. Deflection Separator: (a) conveying region, (b) deposit chamber and (c) inlet.

37

(a)
(b)

Figure 21. Deflection Separator conveying region with (a) dowel rods and (b) deflectors.

Table 9. Deflection Separator testing data.
Engine
RPM
Deflection
Separator

Material In
Total
Marked Chestnut
Weight Chestnuts Weight

3003

7lb 9oz

100

3lb 14oz

3003

8lb 8oz

94

3lb

3003

7lb 12oz

94

3lb

Weight
3lb 7oz
30% clogging
4lb 10oz
50% clogging
3lb 8oz

Collected
Marked
Chestnuts

Discharge
Weight

94

3lb 8oz

94

3lb 5oz

86

3lb 14oz

Tests were run using the same RPM, compared to varying RPM‟s as with previous
designs. The previous test results do not vary significantly for differing RPM‟s. A large inlet
velocity was used to keep burs from becoming stuck in the conveying region. Frequently, debris
became stuck in the conveying region. Spacing was allowed for individual burs however, during
harvesting conditions burs are typically stuck together. These clusters caused clogging and the
hypothesis was rejected.

38

6.7

Momentum Transfer Separator
The design of the Momentum Transfer Separator utilizes a sieve-type separation

technique, as shown in Figure 22. This design considers the effects of conveying material at an
angle. While dilute phase transport is desired, material separation and bed formation is still
prevalent. Chestnuts and heavier materials were typically conveyed in the bottom half of a tube.
Momentum transfer skids, mounted in the bottom of the conveying region, allow heavier
material to drop from the air stream. Changing the momentum occurs when the chestnuts, or
debris, strike the momentum skids. The skids are angled at 45 degree in reference to the direct
line between the inlet and outlet of the chamber. A deflection point is added to the top surface,
ensuring all material is obstructed. A separation grid, made of dowel rods, was mounted at the
same 45 degree angle to separate chestnuts in the air stream beyond the deflection point. A
declined plane is used to funnel material passing through the separation grid back to the deposit
chamber.
Two sizes of burs were used during the testing of Momentum Transfer Separator: large
debris (LD) and small debris (SD). Large debris is considered to be burs 4 inches or larger;
small debris is anything less. The LD burs were collected from a chestnut orchard in early
March 2011. Many of these burs fell post harvest and are not typical of harvesting season burs.
The angled separation barriers and sorting grid did allow most burs to pass through the
conveying region.
The chestnut recovery for the Momentum Transfer Separator was exceptional. In one
test, one hundred percent of the chestnuts were captured in the sorting chamber: 97/97. Another
test resulted in 96 out of 97 chestnut captured, as shown in Table 10. The chestnut loss in the
Momentum Transfer Separator was significantly less than other designs, excluding the Air Flow
39

Separator, and was selected as the best design based on the performance criteria. The Air Flow
Separator had no chestnut loss and 100% separation of material from the air stream.

(a)

(c)

(b)

Material Flow

Figure 22. Momentum Transfer Separator: (a) momentum skids, (b) separation grid, (c)
deflection point.

Table 10. Momentum Transfer Separator testing data.

Engine
RPM

Collected
Weight
Marked
(lbs)
Chestnuts

3051

8lb 10oz (LD)

94

3lb 1oz

3lb 8oz

92

4lb 9oz

3051

6lb 7oz (SD)

97

3lb 4oz

3lb 10oz

97

2lb 10oz

3051

Momentum
Transfer
Separator

Material In
Marked Chestnut
Total Weight Chestnuts Weight

8lb 11oz (SD)

97

3lb 4oz

3lbs 11oz

96

4lb 6oz

Discharge
Weight

Based on the results of the testing, the Momentum Transfer Separator met the initial
criteria of minimizing chestnut loss with no clogging and was selected for further testing. The
Saltation Sieve Separator (SSS) is the second phase of the Momentum Transfer Separator, which

40

is different by the following: 1) a hinged paddle replaced the static deflection point and 2) the
separation grid was replaced with a momentum transfer skid.
6.8

Saltation Sieve Separator (SSS)
The SSS is designed to remove chestnuts from an air stream by manipulating the

momentum and the air velocity, and capitalizing on the effects. The SSS system is also designed
to manipulate airflow, specifically forcing material interaction. Manipulating the air stream can
produce a desired effect on what material can be conveyed; the saltation velocity differs for
chestnuts and debris. The velocity of the air stream can be changed by installing orifices in the
conveyance ductwork, either increasing or decreasing the air velocity. Likewise, obstructions or
geometry changes can influence the flow behavior index, and also shift the maximum velocity
point. In the case of the SSS system, shifting the maximum velocity point can be very
advantageous when used in conjunction with conveyance.
Because chestnut separation is the overall goal, the SSS was designed to separate using
obstruction devices to compliment dense phase bed formation, common in horizontal
conveyance. Momentum transfer skids, shown in Figure 23, slope upwards and interact with
chestnuts in the bed formation. The chestnuts in the bed are subjected to a physical interaction
which decreases their momentum. Slots allow these chestnuts to fall from the air stream. While
the bed is effectively being removed from the conveying tube, the geometry internal to the SSS
changes slightly (diameter is increased), decreasing the air velocity. This is done to promote a
fixed bed phase stage in the SSS. The SSS was designed to be mounted at an angle, relative to
the ground, to influence and encourage bed formation in the conveying region.

41

Hinged Paddle
Momentum Skids

Figure 23. Saltation Sieve Separator conveying region.

Material in the top half of the conveying tube will experience particle to particle
interaction and particle to wall interaction. Removing the bed gives the chestnuts in the upper air
stream the chance to interact with the obstruction devices. A hinged sheet-metal paddle, shown
in Figure 23, is mounted to the top of the conveying duct. This paddle ensures no material
passes through the SSS without some forced interaction.
Conveyed material entering the SSS system encounters a change in boundary geometry
which influences the air velocity, subsequently altering the aerodynamic relationships between
the two. For sustained horizontal conveyance, the chestnuts must overcome the effects of the
interactions with debris and the walls of the system. Two chestnuts colliding will experience a
change in momentum, effectively slowing their conveyed velocity at that point in time.

42

Figure 24 Saltation Sieve Separator (SSS).

Figure 25 shows the momentum-transfer skids (a) and hinged paddle (b) location internal
to the SSS system. The paddles was mounted in the conveying region, 6 in. upstream from the
discharged port. Fourteen 1/2 in. holes were drilled into the obstruction paddle to allow airflow
through the face. The paddle is mounted to the chamber with a standard 1 in. cabinet hinge and
hardware. Weight was added to the paddle using a 3/8 in. bolt and four nuts to ensure the paddle
disrupts the material and velocity profile during harvesting air flow conditions. Without the
weight the paddle “opens”, or lays parallel to the air flow. A sliding door on the bottom of the
chamber was used to extract sorted material from the harvested material, manually.

43

Material
Flow

(b)
(a)
(c)
(d)

Figure 25. Saltation Sieve Separator components: (a) momentum skids, (b) hinged paddle, (c)
conveying region, and (d) deposit chamber.

The momentum transfer skids are mounted internally to the sorting chamber, and
constructed from 5/8” particle board. The dimensions are constant in two directions: 6 in. length
and 5/8 in. thick. Widths range from 2.5 in. to 4 in. Because the momentum changes when
material impacts the portion of the skid in the air stream, widths were chosen arbitrarily.
The Saltation Sieve Separator initially met our criteria for success. This design was chosen
for further testing and evaluations.

7

Experiment Design
The harvesting performance of the final design (SSS) was evaluated based on the affects

of three primary inputs: 1) the number of chestnuts entering the separation chamber, 2) the

44

weight of debris and 3) the feed rate of material into the nozzle. The validation of the process
was characterized by chestnut loss and separation efficiency.
The objective of the experiment was to determine the harvesting performance of the SSS
as affected by the proportion of chestnuts and debris, and feed rate. Controlled quantities of
chestnuts, marked with white paint, were added to each debris mixture. These marked chestnuts
were of the variety Colossal and harvested in Michigan during the fall 2010 harvest. They were
stored in cold refrigeration until testing. The debris was collected from a Michigan chestnut
orchard in the spring of 2011. This debris was from the fall 2010 crop and had wintered on the
orchard floor. The debris weight was measured before each test and compared to the weight of
debris collected from the blower discharge. Any material blown from the collection bin was not
considered a part of the total discharge weight. The nozzle feed rates were slow, medium, and
fast. These depend upon the operator but a material density at the nozzle of low, medium, and
high, were thought of as corresponding feed rate descriptions.
Table 11. Independent Variables and their levels
Independent Variables

Coded

Number of Chestnuts

x1

Symbol
Uncoded
Quantity

-1
0
1

Levels
Uncoded
20
60
100

Amount of Debris

x2

Weight

-1
0
1

1 lb.
4 lbs.
7 lbs.

Feed Rate

x3

Speed

-1
0
1

slow
medium
fast

45

Coded

A Box-Behnkin response surface design was used to set up and evaluate the experiment
(Myers et. al. 2009). Three independent variables; number of chestnuts, weight of debris, and
feed rate, were defined. The number of chestnuts was divided into three amounts: 20, 60, and
100. Chestnuts were not included or excluded, based on size or shape. Three levels of debris
weight were used: 1 lb., 4 lbs., and 7 lbs. Nozzle feed rates were slow, medium, and fast, which
are subjective depending upon the operator. Based on the principles of a Box-Behnken design
for three independent variables, fifteen tests were suitable for analyzing the experiment. A run
consisted of 15 tests and testing order was generated using R statistical analysis software. A
total of 3 runs (45 tests) were made and each run order was sequenced randomly to further
increase the accuracy of the experiment. The independent variables and their coded levels are
shown in Table 11.
7.1.1 Performance Evaluations
The performance of the chestnut harvester was characterized by the following
parameters.
7.1.1.1 Separation Efficiency
Separation efficiency was defined as the ratio of the weight of debris removed to the
weight of debris which entered the SSS system, expressed as a percentage. Separation efficiency
was computed as follows:

(9)

46

Where:
SepPer = Separation efficiency (%)
DWin = Weight of debris input (lbs.)
DWdis = Weight of debris discharged (lbs.)

7.1.1.2 Chestnut Loss
Chestnut loss was defined as the ratio of the number of chestnuts lost through the blower
discharge to the number of chestnuts which entered the SSS system, expressed as a percentage.
The following equation was used to calculate the chestnut loss:

( 10 )

Where:
C.L. = Chestnut loss (%)
Ch#in = Number of chestnuts input (qty.)
Ch#S.C. = Number of chestnuts collected in the separation chamber (qty.)

8

Results and Discussions
A second order polynomial was used for fitting the response surface to the experimental

data. Contour plots of response surfaces for chestnut loss and separation percentage are shown
in Figure 26 through Figure 31. The effects of the independent variables and their interactions

47

on the performance criteria are shown in the analysis of variance tables; Table 13 and Table 14
are variance tables for chestnut loss and separation percentage, respectively.
Table 12. Testing data for chestnut loss and separation efficiency as a percentage
Variables
Number of Debris
Chestnuts Weight (lb)
20
1
20
4
20
7
20
4
60
1
60
7
60
1
60
7
60
4
60
4
60
4
100
1
100
4
100
7
100
4

8.1

Feed
Rate
Medium
Fast
Medium
Slow
Slow
Slow
Fast
Fast
Medium
Medium
Medium
Medium
Slow
Medium
Fast

Separation Chamber Discharge
Total
Weight Number of Debris Chestnuts Separation
(lb)
Chestnuts Weight (lb) Loss (%) Efficiency (%)
0.50
18.67
0.60
6.67
60.42
0.65
17.00
2.63
15.00
65.63
1.40
17.67
4.31
11.67
61.61
0.83
18.33
2.00
8.33
50.00
2.19
58.67
0.56
2.22
56.25
2.40
55.33
4.67
7.78
66.67
2.15
57.00
0.75
5.00
75.00
2.77
58.33
4.27
2.78
61.01
2.15
56.33
2.46
6.11
61.46
2.10
58.33
2.81
2.78
70.31
2.17
54.00
3.08
10.00
77.08
3.40
94.33
0.65
5.67
64.58
3.90
96.00
3.25
4.00
81.25
3.71
94.00
4.73
6.00
67.56
4.02
98.67
2.56
1.33
64.06

Chestnut Loss
Machine chestnut loss data are presented in Table 12. The analysis of the regression is

given in Table 13. According to the regression analysis, the estimated coefficients for the feed
rate and the weight of debris have low confidence levels. The only significant variable affecting
chestnut loss was the number of chestnuts, with a 97 percentage confidence for the estimated
coefficient. The feed rate and the number of chestnuts interaction coefficient was estimated at an
80% confidence level.

48

Table 13. Analysis of variance table for chestnut loss data

Coefficients
(Intercept)
x1: Number of Chestnuts
x2: Debris Weight
x3: Feed Rate
x1 : x2
x1 : x3
x2 : x3
x1^2
x2^2
x3^2
Multiple R-squared: 0.7596

Estimate Std. Error t value
Pr(>|t|)
6.3000 1.7819 3.5350 0.0166
-3.0813 1.0912 -2.8240 0.0369
1.0837 1.0912 -1.0912 0.3662
0.2200 1.0912 0.2020 0.8482
-1.1675 1.5432 -0.7570 0.4834
-2.3300 1.5432 -1.5100 0.1915
-1.9450 1.5432 -1.2600 0.2632
1.9588 1.6062 1.2190 0.2770
-0.7662 1.6062 -0.4770 0.6534
-1.0888 1.6062 -0.6780 0.5280
Adjusted R-squared: 0.3268

Figure 26. Chestnut loss percentage as affected by the number of chestnuts (Ch) and weight of
debris (D).

49

Figure 26 is a contour plot for the response surface for chestnut loss based on the
interaction between the number of chestnuts (Ch) and the weight of debris (D). It is sliced or
viewed from the feed rate (FR) set equal to medium speed. Between 80 and 100 chestnuts, the
influence of the weight of the debris has little effect on the chestnut loss percentage. However,
when the weight of the debris increased the chestnut loss increased from 8% to 13%, when 20
chestnuts were input. The maximum chestnut loss occurred at low chestnut numbers. The
chestnut loss for the SSS system can be minimized by increasing the density of the chestnuts
going through the system.

Figure 27. Chestnut loss as affected by the number of chestnuts (Ch) and the feed rate (FR).

50

Figure 27 is a contour plot for the response surface for chestnut loss based on the
interaction between the number of chestnuts (Ch) and the feed rate (FR). It is sliced or viewed
from the weight of debris (D) set equal to 4 lbs. The maximum chestnut loss occurs at low
chestnut numbers with a fast feed rate. Chestnut loss is minimized when the number of chestnuts
is above 60 which ranged from 2-6%. This suggests that when a larger number of chestnuts
enter the SSS system, regardless of feed rate, the chestnut losses will be minimized.

Figure 28. Chestnut loss as affected by weight of debris (D) and feed rate (FR).

51

Figure 28 is a contour plot for the response surface for chestnut loss based on the
interaction between the weight of debris (D) and the feed rate (FR). It is sliced or viewed from
the number of chestnuts (Ch) set equal to 60. The chestnut loss varies from 2.5% at a low debris
weight and a low feed rate, to 8% at a high debris weight and low feed rate.
8.2

Separation Efficiency
The analysis of the regression is given in Table 14. According to the regression analysis,

the estimated coefficients for the number of chestnuts and the weight of debris have low
confidence levels. The most significant variable affecting the separation efficiency was for the
feed rate, with a 98% confident estimated coefficient. The number of chestnuts and the feed rate
interaction coefficient was estimated with a 96% confidence level. Contour plots of the response
surfaces for separation efficiency are shown in Figure 29-15. The separation efficiency for the
SSS system ranged from 50% to 81%.
Table 14. Analysis of variance table for separation efficiency data
Coefficients
(Intercept)
x1 : Number of Chestnuts
x2 : Debris Weight
x3: Feed Rate
x1 : x2
x1 : x3
x2 : x3
x1^2
x2^2
x3^2
Multiple R-squared: 0.778

Estimate
50.873843
0.307292
2.662037
22.208333
0.002083
-0.20625
-2.083333
-0.001589
-0.337963
-1.541667

Std. Error
10.792602
0.275318
3.422283
6.756855
0.026462
0.079386
1.058475
0.002066
0.367232
3.305089

52

t value
Pr(>|t|)
4.714
0.00527
1.116
0.31511
0.778
0.47184
3.287
0.02179
0.079
0.9403
-2.598
0.04836
-1.968
0.10616
-0.769
0.47662
-0.92
0.39964
-0.466
0.66051
Adjusted R-sqaured: 0.3779

Figure 29. Response surfaces for separation efficiency as affected by number of chestnuts (Ch)
and weight of debris (D).

Figure 29 is a contour plot for the response surface for separation efficiency based on the
interaction between the number of chestnuts (Ch) and the weight of debris (D). It is sliced or
viewed from the feed rate (FR) set equal to medium speed. The maximum separation efficiency
occurs at 100 chestnuts and 4 lbs. of debris, and appears as a mound or peak in Figure 29. The
separation efficiency decreases as the number of chestnuts decreases because of the characteristic
of dense phase conveyance, specifically where debris is conveyed. During dense phase
conveyance, the heavier material is conveyed along the bottom portion of the tube, essentially
forcing the lighter material higher in the air stream, or top of the tube. When the number of
53

chestnuts is low, the debris has more opportunity to travel throughout the entire diameter of the
tube, and will interact with the momentum skids. More debris separation occurs when the
number of chestnuts is low.

Figure 30. Separation efficiency as affected by the number of chestnuts (Ch) and feed rate (FR).

Figure 30 is a contour plot for the response surface for separation efficiency based on the
interaction between the number of chestnuts (Ch) and the feed rate (FR). It is sliced or viewed
from the weight of debris (D) set equal to 4 lbs. The maximum separation efficiency occurs at a
100 chestnuts and low feed rate; maximum separation efficiency over 75% was achieved. When

54

20 chestnuts were fed into the SSS slowly, the separation efficiency decreased. The separation
efficiency generally decreased as the number of chestnuts decreased.

Figure 31. Separation efficiency as affected by weight of debris (D) and feed rate (FR).

Figure 31 is a contour plot for the response surface for separation efficiency based on the
interaction between the weight of debris (D) and the feed rate (FR). It is sliced or viewed from
the number of chestnuts (Ch) set equal to 60. A ridge of maximum separation efficiency occurs
across the contour plot. When the weight of debris is 1 lb., maximum separation efficiency is at
a large feed rate. When the weight of debris is 7 lbs., the maximum separation efficiency is at a

55

low debris weight. The maximum SSS separation efficiency ridge includes a medium feed rate
and 4 lbs. of debris, or middle range for each variable.
In addition to the multiple variable regression analysis, regression analysis was used to
investigate the chestnut loss solely based on the interaction between the number of chestnuts and
the feed rate. Table 15 shows the linear regression analysis for the chestnut loss based on the
number of chestnuts and the feed rate. The model coefficient describing the interaction between
the number of chestnuts and the feed rate was estimated at a 98% confidence level. The model
coefficient for the number of chestnuts entering the SSS system was estimated at a 96%
confidence level.
Table 15. Linear regression analysis for chestnut loss based on the number of chestnuts and feed
rate.
Coefficients
Estimate Std. Error t value
Pr(>|t|)
(Intercept)
65.5260 1.6020 40.9000 2.2700E-13 ***
x1: Number of Chestnuts
4.9740 2.1940 2.2670
0.0445
*
x3: Feed Rate
1.4410 2.1940 0.6570
0.5247
x1 : x3
-8.2050 3.1020 -2.6450
0.0228
*
Multiple R-squared: 0.5332
Adjusted R-squared: 0.4059
Signif. codes: 0 „***‟ 0.001 „**‟ 0.01 „*‟ 0.05 „.‟ 0.1 „ ‟ 1

Ninety five percent of the time, the chestnut loss model can be used to describe the
chestnut loss in the SSS system. The equation for chestnut loss is:

( 11 )

56

9

Conclusions
The SSS technique used for harvesting chestnuts can be successfully employed as a

single-stage chestnut harvesting system. Chestnut losses average 4-5% when a large number of
chestnuts were entering the SSS system. The separation efficiency was the highest when a large
number of chestnuts were entering the SSS system at a slow feed rate. This characteristic is
advantageous for growers, or processors, to maximize separation efficiency while minimizing
chestnut losses.
Because the average size orchard in Michigan is 5 acres, a grower producing 1500
lbs/acre could economically purchase a machine which costs $6,000.00. The SSS system and its
single-stage design presents a potentially economically feasible harvesting system for Michigan
chestnut growers.

10 Future Research
Future research for the SSS system is needed for optimization. Realizing that this system
is a foundation to be built upon, the author suggests some further ideas for research. A
harvesting challenge still facing chestnut growers is what should be done with chestnuts that are
contained inside burs. The growers and industry have not determined or set a standard to the
quality of these chestnuts. The SSS system does not remove chestnuts inside of burs
specifically. A pre-treatment process to shuck the chestnut from the burs may be a solution,
assuming growers and the industry want to harvest these chestnuts.
1. Testing the performance parameters when the system is up-scaled to larger capacities:
10, 15, and 20 acres.
2. Adding dual hose suction ports to increase harvesting capacity.
57

3. Testing the performance parameters when the system is installed in conjunction with
a windrowing or ground collection system versus no pre-processing based harvesting.
4. Evaluation of chestnut quality for chestnuts harvested using the SSS technique
compared to hand harvesting and mechanical harvesting.
5. Testing the SSS system for other products or processes where a single-stage system
can improve performance or efficiency.

58

APPENDICIES

59

11 Appendix A: Engine hp and Air Flow calculations
%% Pressure Drop Calculations for Chestnut Harvester

%

%-------------------------------------------------------%
% This program calculated the total pressure drop for
%
% a defined geometric systems, based on the aerodynamic %
% properties of chestnuts and standard air. Horsepower %
% requirement is the desired output.
%
%
%
% Mark De Kleine
%
% January 2011
%
%
%
%-------------------------------------------------------%
close all; clear all;
%% Defining Variables
% Design Requirements
Vc = 23.8; %
Vol = 2.45; %

- Conveying Velocity (m/s)
- Air volume (m/s)

% Standard Air Properties and Gravity
p_air = 1.2; %
- Air Density (kg/m^3)
g = 9.81; %
- Gravity (m/s^2)
u_air = 10^-5; %
- Air Viscosity (kg m/s)
% Properties of Chestnuts
pc = 531; %
d = 0.035; %
Pp = 1135.68; %

- Bulk Density (kg/m^3)
- Chestnut Diameter (m)
- Chestnut Density (kg/m^3)

% Chamber Design Configuration
D = 0.1016; %
delta_z = 1.5; %
k = 0.9; %
numb_bends = 3; %
L = 7; %
R_D = 1.05; %
n_b = 0.6; %

-

Tube diameter (m)
Vertical height lift (m)
Fitting Loss Coeff. Srivastava et al.
Number of Bends in system
Length of Conveying duct (m)
Radius to Diameter Ratio
Blower efficiency

%% Capacity
% Maximum Capacity from Economic Analysis
acres = 9.3; %
- Number of Acres (ac)
lb = 3000; %
- Pounds of Chestnuts per Acre (lbs)
days = 2; %
- Days between harvest (days)
wk_hrs = 8; %
- Hours in a standard work day (hrs)
kg_conv = 0.454; %
- conversion to kilogram
hr_conv = 3600; %
- conversion to seconds
% Bell Curve Percent
perc = .2; %
days

- Bell Curve Percentage fallen between

60

Cap= (((((acres * lb)/ days) * perc) / wk_hrs) * kg_conv) / hr_conv; %kg/s

%% Mass Flow Rate
Q = (pi/4)* D^2 * Vc; %
CFM = Q * 35.3 * 60; %

- Volumetric Flow Rate (m^3/s)
- 35.3 ft^3/m^3 and 60s/1min (ft^3/s)

m_dot = p_air * Q; %

- Mass Flow Rate (kg/s)

theta_m = Cap / m_dot; %

- Mass Flow Ratio

%% Reynolds Number Calculation
N_rc = (p_air * Vc * D) / u_air;
%% Line Pressure Loss
lamda = 4 *(0.0014 + 0.125*(N_rc^-0.32)); %

- Line Friction Factor

P_lin = lamda * (p_air / 2) * (Vc ^2) * (L / D);%

- Pressure Loss (Pa)

%% Acceleration Pressure Loss
% To get Velocity of solids:
Vs = Vc *(1 - 0.68*(d^0.92)*(Pp^0.5)*... %
(p_air^-0.2)*(D^0.54));
P_acc = theta_m * Vc * p_air * Vs; %

- Solids Velocity (m/s)

- Pressure Loss (Pa)

%% Lift Height Pressure Loss
p_star = theta_m * Vc * p_air / Vs;
P_vrt = p_star * g * delta_z; %

- Pressure Loss (Pa)

%% Pressure Drop due to Solids and Particle Interaction
lamda_s= (0.0285 * sqrt(g*D)) / Vc;
P_spi = theta_m*lamda_s*(p_air/2)*(Vc^2)*(L/D);%

- Pressure Loss (Pa)

%% Pressure Loss due to Bends
% Equivalent Length
Leq = (k * D) / lamda;%
bends
Leq_b = numb_bends * Leq;

- Equivalent length for

P_bnd = (P_lin * Leq_b) / L;%

- Pressure Loss (Pa)

%% Pressure Loss due to Solids
P_sol = p_air * (Vc^2) *(0.245 *... %
(m_dot / (p_air * Vc * D^2)) * (R_D^-.26));

- Pressure Loss (Pa)

%% Total Pressure Loss
P_tot = P_lin + P_acc + P_sol + P_bnd + P_spi + P_vrt;% - Pressure Loss (Pa)

61

%% Power Requirements
Pwr_hp = ((P_tot * Q) / n_b) *(1.34/1000); %
(hp)

62

- Blower power needed

12 Appendix B: Velocity Profiles
An Alnor Velometer, series 6000P, was used for measuring air velocity at several
locations throughout the harvesting system. Measurements were taken at varying depths to
quantify the velocity profile and examine the air stream size. The velocity profiles were plotted
for each test. These figures are outlined below.
A Velleman DM6234 digital tachometer is used to measure the engine drive shaft rpm
and blower drive shaft rpm. A piece of reflective tape is added to each shaft at the rpm
measurements point.
12.1 B.1: Air Flow Separator
1

2

3

4

Figure B.32. Velocity measurement positions in the Air Flow Separator.

63

Table B.16. Air Flow Separator air velocities.
Engine shaft RPM
Fan Shaft RPM

1580
2580
3010
928
1500
1777
Velocity Measurements (ft/min)
5000
6500
7400

Inlet tube
Position 1 (in)
1
3
4
5

3000
4000
1800
0

1800
4800
5000
2600

3800
4800
7000
3200

1
3
5
7
9

2200
1100
0
0
0

5200
2600
0
0
0

5400
5000
2000
0
0

1
2
3
5
7

3000
1600
800
0
0

5000
2400
1500
1000
1000

5800
6000
3000
1800
1800

1
2
3
5

1100
1300
800
0

1200
1400
1200
300

1800
2200
2000
1000

1

3000
3000
2100

4100
4800
3700

4600
5000
3200

Position 2 (in)

Position 3(in)

Position 4 (in)

Outlet (in)
2 at Focus
2 at Wall

The positions at which the air velocity was taken are shown in Figure B.32. The y-value
of depth is measured from the top of the air flow chamber. The x-values were located at key
points of interest and can be found in Error! Reference source not found.. The air velocity

64

measurements were taken in the middle of the air stream, along the third dimension or z-axis.
The following graphs represent the air layer geometry in the cross section of the chamber.

1

Depth in chamber (in)

1.5
2
2.5
1580 RPM

3

2580 RPM
3.5

3010 RPM

4
4.5
5
0

2000

4000

6000

8000

Air velocity (ft/min)
Figure B.33. Air Flow Separator air velocities at position 1.

At position 1, the maximum velocity point shifts downward, or further down from the top
of the chamber, as the RPM and air velocity are increased. The maximum velocity point shifted
downward 1 inch when the engine RPM doubled. Negative pressure air flow becomes more
streamlined as the velocity increases; that is, non plug-like flow. As the air particles move faster
away from each other, the air layer shrinks and the maximum velocity point becomes more
distinguishable, as in the 3010 RPM velocity profile.

65

1
Depth in chamber (in)

1.5
2
2.5
1580 RPM

3

2580 RPM

3.5

3010 RPM

4
4.5
5
0

1000

2000
3000
4000
Air velocity (ft/min)

5000

6000

Figure B.34. Air Flow Separator air velocities at position 2.

At position 2, the maximum velocity point is at the top of the air chamber. This is
unintuitive based on the air profile described at position 1. The complexity of an air flow
problem is apparent here. As air is transitioning from position 1 to position 3, it passes through
position 2 where the air stream is bending and curving through the chamber. The maximum
velocity point can be shifted using geometry.

66

1

Depth in chamber (in)

1.5
2
2.5
1580 RPM

3

2580 RPM

3.5

3010 RPM

4
4.5
5
0

2000

4000

6000

8000

Air velocity (ft/min)
Figure B.35. Air Flow Separator air velocities at position 3.

The air velocity profiles for the Air Flow Separator at position 3 are depicted in Figure
B.35. The top side of the conveying region has the largest velocity for all three RPM settings.
The increase in air velocity is related to the geometry of the chamber and airflow characteristics
around bends. Measurements taken at position 3 show the air layer shrinkage at the top of the air
chamber.

67

1

Depth in chamber (in)

1.5
2
2.5
1580 RPM

3

2580 RPM
3.5

3010 RPM

4
4.5
5
0

500

1000

1500

2000

2500

Air velocity (ft/min)

Figure B.36. Air Flow Separator air velocities at position 4.

At position 4, the velocity profiles resemble a standard pipe flow profile and are similar
throughout. The highest velocities were more towards the top of the chamber.

68

12.2

B.2: Dowel Grid Separator
The velocity profiles for the Dowel Grid Separator at varying positions are depicted in

the following figures. Figure B.37 show the positions at which the air velocities were measured.

1

2

3

Figure B.37. Dowel Grid Separator air velocity measurement positions.

69

Depth in chamber (in)

0
1
2

2075 RPM
2526 RPM

3

3030 RPM
4
5
6
0

2000

4000

6000

Air velocity (ft/min)
Figure B.38. Dowel Grid Separator air velocities at position 1.

The air velocity profile for the Dowel Grid Separator at Position1 remains relatively
constant across the spectrum of tests. The lowest engine RPM had the highest velocity 3 inches
below the top of the chamber. There is no explanation for this; human error while taking data
was possible.

70

0
Depth in chamber (in)

1
2
3

2075 RPM

4

2526 RPM

5

3030 RPM

6
7

8
0

500

1000
1500
2000
Air velocity (ft/min)

2500

Figure B.39. Dowel Grid Separator air velocities at position 2.

The air layer profile for the Dowel Grid Separator at position 2 is shown in Figure B.39.
As the engine RPM increased each air layer profile increased accordingly. There was no
significant shift in the depth of the maximum velocity point between any air profiles at position
2.

71

Depth in chamber (in)

0
1

2075 RPM

2

2526 RPM
3

3030 RPM

4
5

6
0

500

1000
Air velocity (ft/min)

Figure B.40. Air Flow Separator air velocities at position 3.
Table B.17. Dowel Grid Separator air velocities.
Engine shaft RPM
Fan Shaft RPM
Focus Inlet tube
Position 1 (in)

2075
2526
3030
1230
1479
1828
Velocity Measurements (ft/min)
3800
4600
6500

1
2
3
5

3200
3200
3600
1000

3800
3000
2100
1600

5600
4100
2000
900

1
3
5
7

1500
1500
0
0

2200
1800
1000
0

1700
2200
1600
1000

1
2
3
5

1500
1000
800
1500

1100
1400
1200
1000

1200
1400
1400
1300

1
Focus 2
Wall 2

2600
3000
3000

3000
3300
2700

2800
3300
2700

Position 2 (in)

Position 3(in)

Outlet (in)

72

1500

12.3 B.3: Axial-Dowel Wheel Separator
Air velocity profiles for the Axial-Dowel Wheel Separator are described below.
Measurements were taken at the positions shown in Figure B.41.

1

2

Figure B.41. Axial-Dowel Wheel Separator air velocity measurement positions.

73

Depth in chamber (in)

0
1
1976 RPM
2

2467 RPM

3

3030 RPM

4
5

6
0

2000

4000
Air velocity (ft/min)

6000

Figure B.42. Axial-Dowel Wheel Separator air velocities at position 1.

The air profile layers in the Axial-Dowel Wheel Separator at position 1 are shown in
Figure B.42. The profile for each air layer is similar in the bottom 3 inches of the chamber.

74

0
Depth in chamber (in)

1
2

1976 RPM

3

2467 RPM

4

3030 RPM

5
6
7

8
0

1000

2000
Air velocity (ft/min)

3000

Figure B.43. Axial-Dowel Wheel Separator air velocities at position 2.
Table B.18. Axial-Dowel Wheel Separator air velocities.
Engine shaft RPM
Fan Shaft RPM
Focus Inlet tube
Position 1 (in)
1
2
3
5
Position 2 (in)
1
3
5
7
Outlet (in)
Focus 1
Focus 2
Wall 2

1976
2467
3030
1222
1471
1816
Velocity Measurements (ft/min)
3200
4300
5600
900
3100
2300
1300

3800
3600
3200
1300

4800
4200
3000
1800

0
1200
2000
1000

0
1500
2100
1200

0
2000
2400
1200

2500
3300
2100

3000
3700
3500

3700
4600
4200

75

12.4 B.4: Radial-Dowel Wheel Separator
Air velocity profiles for the Radial-Dowel Wheel Separator were assumed to be similar to
the profiles gathered in the Axial-Dowel Wheel Separator because the boundary planes did not
change geometry in the conveying region between the two designs.
12.5 B.5: Deflection Separator
The air velocity profiles for the Deflection Separator were not taken in this experiment.
The deflectors described in the design of the Deflection Separator were mounted in the middle of
the air flow chamber, preventing consistent measurement positions between the previous
designs.
12.6 B.6: Saltation Sieve Separator

Air Flow

1

2

3

Figure B.44. SSS air velocity measurement positions.

76

4

Air velocity profiles were collected within the SSS system at the positions shown in
Figure B.44. One of the benefits of the hinged paddle is: it influences the air layer to remain
towards the bottom of the conveying region, yet will open when a large amount of debris, or plug
flow, is present. This design ensures that the material will be conveyed along the momentum
skids.

0
0.5
Depth in chamber (in)

1
1.5
2
3070
RPM

2.5
3
3.5
4
4.5
5
0

1000

2000

3000

4000

5000

Air Velocity (ft/min)

Figure B.45. SSS velocity profile at position 1.

The maximum velocity of the air stream at position 1 is shown in Figure B.45, and is
located in the bottom half of the tube. Airflow can pass through the paddle, effectively slowing
the airflow directly downstream from it, or at the top of the conveying region. A higher velocity
airstream, just above the momentum skids, will help move any bed formation that might occur
when material is entering the SSS system.
77

0

Depth in chamber (in)

0.5
1
1.5
2
2.5

307
0
RP
M

3
3.5
4
4.5
5
0

1000

2000

3000

4000

5000

Air Velocity (ft/min)

Figure B.46. SSS velocity profile at position 2.

At position 2, the air profile again shows a maximum velocity in the bottom half of the
conveying region. Position 2 is located above the second momentum skid from the inlet side.

78

0
0.5
Depth in chamber (in)

1
1.5
2
2.5
3

3070
RPM

3.5
4
4.5
5
0

1000

2000

3000

Air Velocity (ft/min)

Figure B.47. SSS velocity profile at position 3.

Position 3 is located 4 in. upstream from the hinged paddle. The maximum velocity has
begun to shift through the artificial orifice created between the paddle and the transfer skids,
shown in Figure B.47. This desired effect produces a smaller air stream directly above the
transfer skids. A faster moving air stream above the transfer skids helps ensure material
movement while sorting from the bottom.

79

0

Depth in chamber (in)

0.5
1
1.5
2
3070
RPM

2.5
3
3.5
4
4.5
5
0

1000

2000

3000

Air Velocity (ft/min)

Figure B.48. SSS velocity profile at position 4.

Interactions were designed to change the momentum of the chestnuts and also to lessen
the impact forces commonly associated with fruit damage; angling the momentum skids while
hinging the paddled minimizes the perpendicular impacts. It is assumed that a perpendicular

80

Table B.19. Momentum Transfer Separator air velocity measurements.
Engine shaft
RPM
Fan Shaft RPM

3070
1822
Velocity Measurements
(ft/min)

Focus Inlet tube

5400

Position 1 (in)
1
2
3
5

0
1200
2900
4200

1
3
5
7

2000
2000
3900
1400

1
2
3
5

1500
1800
2500
2100

1
2
3
5

0
800
2000
2500

1
Focus 2
Wall 2

3600
4500
3700

Position 2 (in)

Position 3(in)

Position 4(in)

Outlet (in)

81

REFERENCES

82

Baker, K. D., R. L. Stroshine, G. H. Foster and K. J. Magee. 1984. Performance of a pressure
pneumatic grain conveying system. ASAE Paper No. 84-3515, ASAE, St. Joseph, MI
49085.
Cernusca, Ina. Fall 2009. U.S. Chestnut Import Trends. The Chestnut Grower, Vol 11. No. 4 (4).
Coates, W., Lorenzen, B. 1990 Equipment for Ground Harvesting Jojoba seed. American
Society of Agricultural Engineers. Vol 6(2):March 1990.
Guyer, D. E., Kang, W., Development of Chestnut Harvesters for Small Farms. Printed article.
2009.
Henderson and Perry., Agricultural Process Engineering. New York, John Wiley and Sons, Inc.
1955.
Langhaar, Henry L., Dimensional analysis and theory of models. New York, John Wiley and
Sons Inc. 1951.
Lapple, C. E. and C. B. Shepherd (1940). Calculation of particle trajectories. Industrial and
Engineering Chemistry. Vol 32:605-617. 1940.
Kepner, R. A., Bainer, R., Barger, E. L., (1972). Principles of Farm Machinery. Westport, CT.
The AVI Publishing company inc.
Klenin, N. I., Popov, I. F., Sarun, V. A., (1985). Agricultural Machines. New Delhi, India.
Amerind Publishing Company.
Kraus, M. N., (1968). Pneumatic Conveying of Bulk Materials. New York, NY. The Ronald
Press Company.
Marcus, R. D., Leung, L. S., Klinzing, G. E., Rizk, F., (1990). Pneumatic conveying of Solids: A
theoretical and Practical Approach. New York, NY. Chapman and Hall.
Mizrach, A., A. Zaltzman, G. Manor and Z. Nir. 1984. Gravitational motion of spheres in a
fluidized bed. TRANSACTIONS of the ASAE 27(6):1674-1678.
Mohsenin, Nuri, N. 1970. Physical Properties of Plant and Animal Materials. New York, N.Y.:
Gordon and Breach, Science Publishers, Inc.
Myers, Raymond H., Montgomery, Douglas C., Anderson-Cook, Christine M. (2009). Response
Surface Methodology: Process and Product Optimization Using Designed Experiments.
3rd ed. New Jersey.
Patzlaff, A. W. 1980. Hydrodynamic blueberry sorting. U.S. Patent No. 4225424.
Quackenbush, Harold E., Pneumatic fruit harvesting and associated fruit characteristics. M.S.
Michigan State University. 1961. Print.

83

Shen, L., Haque, E., Posner s., 1994. Saltation Velocity of Wheat Material in Horzontal Flow.
Transactions of the ASAE: American Society of Agricultural Engineers: VOL. 37(2):577580, print.
Shortley and Williams. Principles of college physics. Prentice-Hall. Englewood cliffs, NJ.
1959.
Srivastava, A. K., Goering, C. E., Rohrbach, R.P., Buckmaster, D. R., Engineering Principles of
Agricultural Machines. 2nd Edition Chapter 1, pp. 1-14. American Society of
Agricultural and Biological Engineers technical library.
Srivastava, A., VanEe, G., Ledebuhr, R., Welch, D., Wang, L., Design and Development of an
Onion-Peeling Machine. Applied Engineering in Agriculture, American Society of
Agricultural Engineers (1997): VOL. 13(2):167-173
United States Department of Agriculture. 2007. Census of Agriculture: Summary and State
Data. Volume 1: Geographic Area Series, Part 51. Printed Article.
Van Ee, G. R., Haffar, I. "A Model to Describe Terminal Velocities of Graded Cucumbers”.
Journal of American Society of Agricultural Engineers (1990):Vol 6(2) Print.
Yildiz, M. U., Ozcan M. M., Calisir S., Demir F., Er F., 2009. Physico-Chemica Properties of
Wild Chestnut Fruit Grown in Turkey. World Applied Sciences Journal 6(3): 365372,2009.
Zaltzman, A., Schmilovitch, Z. 1986. Evolution of a Potato, Fluidized Bed Medium Separator.
ASABE technical libray.
Zaltzman, A., Z. Schmilovitch and A. Mizrach. 1985. Separating flower bulbs from clods and
stones in a fluidized bed. Canadian Agricultural Engineering 27(2): 63-67.
Zaltzman, A., A. Mizrach and Z. Schmilovitch, 1986. Analytical model of a gravitational
separation process in fluidized bed medium. Journal of Agricultural Engineering
Research 34(4):257-273.
Zaltzman, A., R. Feller, A. Mizrach and Z. Schmilovitch. 1983. Separating potatoes from clods
and stones in a fluidized bed medium. TRANSACTIONS of the ASAE 26(4): 987-990,
995.
Zaltzman, A., A. Mizrach and Z. Schmilovitch. 1982. Fluidized bed separator. United States
Patent No. 4,322,287.

84