Scholars' Mine Scholars' Mine
Masters Theses Student Theses and Dissertations
Fall 1996
Plastic fiber rolling for concrete reinforcement Plastic fiber rolling for concrete reinforcement
Jeffery S. Thomas Missouri University of Science and Technology, [email protected]
Follow this and additional works at: https://scholarsmine.mst.edu/masters_theses
Part of the Mechanical Engineering Commons
Department: Department:
Recommended Citation Recommended Citation Thomas, Jeffery S., "Plastic fiber rolling for concrete reinforcement" (1996). Masters Theses. 6720. https://scholarsmine.mst.edu/masters_theses/6720
This thesis is brought to you by Scholars' Mine, a service of the Missouri S&T Library and Learning Resources. This work is protected by U. S. Copyright Law. Unauthorized use including reproduction for redistribution requires the permission of the copyright holder. For more information, please contact [email protected].
PLASTIC FIBER ROLLING FOR CONCRETE REINFORCEMENT
by
JEFF SCOTT THOMAS, 1971-
A THESIS
Presented to the Faculty of the Graduate School of the
UNIVERSITY OF MISSOURI - ROLLA
In Partial Fulfillment of the Requirements for the Degree
MASTER OF SCIENCE IN MECHANICAL ENGINEERING
o-Advisor
T7187129 pages
1996
Approved by
Copyright © 1996
JEFFERY SCOTT THOMAS
ALL RIGHTS RESERVED
111
ABSTRACT
Fiber reinforced concrete is gaining in acceptance and usage as its strength and toughness benefits
are realized. Polypropylene fibers are currently the most commonly used fibers. However~ the cost of
virgin plastics limits the percent content that can be economically added to concrete. One would first
believe that recycled plastics could offer a viable alternative to virgin-plastic fibers, but at present time
extra selection and cleaning steps required to reprocess the recycled plastic into fibers makes it more
expensive than virgin-plastic fibers.
A fiber rolling process conceived and developed at the University of Missouri-Rolla has the
potential of making recycled plastic fibers an economic alternative to virgin-plastic fibers in concrete
reinforcement. This research effort has focussed on two primary issues. The first issue involved
designing and fabricating a laboratory-scale rolling mill, the heart of this new process, that has confirmed
the feasibility of the fiber-rolling concept. After a review of conventional flat rolling models for steel,
the second issue involved developing a mathematical model of the mechanics involved with fiber rolling.
The model allows predictions of roll torques and separation forces to be made for the process, which
have been benchmarked against those for flat rolling processes. The model has also allowed a parametric
study to be completed as the first step in eventually optimizing the process.
The potential benefits to society resulting from this study involve finding a new use for plastics
that would otherwise be deposited in landfills and decreasing the cost of plastic fibers, which currently
limits the extent of strength~ toughness, and service life improvements that can be gained \vith fiber
reinforced concrete. The prototype rolling mill and the mathematical model developed in this research
effort are now ready to be used to further develop the recycled-plastic fiber-rolling process. with the goal
of eventually making recycled-plastic fibers a commercially available product.
IV
ACKNOWLEDGMENTS
The conceivement of the fiber rolling process investigated in this study originated with my advisor,
Dr. D. R. Carroll. I wish to thank Dr. Carroll for his guidance and support on this project and others
during my studies at the University of Missouri-Rolla. I also wish to express thanks to Dr. L. R. Dharani
and Dr. D. C. Look, Jr. for their words of instruction and encouragement during my undergraduate and
graduate work.
This investigation relies on the previous research of members involved with the Recycling
Laboratory in the Department of Basic Engineering at the University of Missouri-Rolla. In particular,
Claire-Nechol Sevier provided some initial prototype designs, a literature survey on plastic fiber
production, and experimental help.
Assistance was provided in the design and parts machining of the original fiber-rolling prototype
by Eddie Light, Department ofBasic Engineering machine shop supervisor, and Bob Hribar, Department
of Mechanical and Aerospace Engineering and Engineering Mechanics machine shop assistant.
DemMaTec Foundation provided parts machining for the beginning of a second prototype rolling mill.
This project was partially supported by the University of Missouri Research Board and the
Opportunities for Undergraduate Research Experience program.
I \vould like to also express my appreciation to Dr. D. R. Fannin, Tom Bryson, and the entire
faculty and staff of the Department ofBasic Engineering who have provided me with both undergraduate
and graduate teaching experience and an overall sense of belonging at this university. Their friendship
and guidance will be sincerely remembered.
Of course, my accomplishments would have been impossible without the patience and constant
support of my wife, Cherie, and my faith in God.
v
TABLE OF CONTENTS
INTRODUCTION .
A. BACKGROUND................................................................................................... 1
1. Recycling Plastics. ... ...... 1
2. Plastic Fibers in Concrete Reinforcement. 2
3. Fiber Rolling Process 3
B. OBJECTIVES....................................................................................................... 3
C. ORGANIZATION 4
PREVIOUS RESEARCH 5
A. FIBER-REINFORCED CONCRETE 5
B. TRADITIONAL ROLLING MODELS 5
1. Mechanics of Flat Rolling. 7
2. Assumptions. 8
3. Force Balance 8
4. Determination of the Neutral Point. 9
5. Governing Equations Summary 9
6. Roll Force and Torque 10
7. Constraints.. .............................................................................................. 10
RECYCLED PLASTIC-FIBER ROLLING PROCESS 12
A. RECYCLED PLASTIC PROCESSING 12
B. PROTOTYPE ROLLING-MILL DESIGN 14
1. Current Prototype Development. 14
2. Future Prototype Development. 14
C. MATHEM.l~TICAL MODELING 15
1. Current Modeling Efforts. 15
2. Future Modeling Efforts " 15
D. EXPERIMENTAL WORK 16
III.
II.
Page
ABSTRACT iii
ACKNOWLEDGMENTS iv
LIST OF ILLUSTRATIONS viii
LIST OF TABLES , x
SECTION
I.
VI
1. Current Experimental Efforts. 16
2. Future Experimental Efforts '" 16
IV. MECHANICS OF FIBER ROLLING 18
A. GEOMETRY TERMINOLOGY FOR FIBER ROLLING 18
B. ASSUMPTIONS 27
C. AREA CALCULATIONS 28
D. STRESS CALCULATIONS 30
1. Case I 30
2. Case II. 30
E. FORCE BALANCE 32
1. Differential Element 1. 32
2. Differential Element 2 33
3. Differential Element 3. 34
4. Differential Element 4. 35
F. DETERMINATION OF THE NEUTRAL POINT 35
G. GOVERNING DIFFERENTIAL EQUATIONS SUMMARy 36
1. Case I 36
2. Case II 37
H. NUMERICAL SOLUTIONS TO THE GOVERNING EQUATIONS 39
l. ROLL FORCE AND TORQUE CALCULATIONS 40
1. Case I 40
2. Case II. 40
J. CONSTRAINTS 40
K. THE TAGUCHI METHOD................................................................................ 42
V. RESULTS 46
A. MATERIAL PROPERTIES 46
B. MODEL COMPARISON USING ROLL-FORCE PRESSUREDISTRIBUTION DIAGRAMS 47
C. MODEL COMPARISON USING THE TAGUCHI METHOD 48
D. FIBER ROLLING PARAMETRIC STUDy 59
1. Roll-Face Pressure Distribution Diagrams 59
2. System Parameter Effects on Roll Torque 62
VI. CONCLUSIONS 66
VB
APPENDICES
A. PROTOTYPE ROLLING-MILL DETAIL DRAWINGS 68
B. PROTOTYPE ROLLING-MILL ASSEMBLY DRAWINGS 86
C. FLAT ROLLING CODE 103
D. FIBER ROLLING CODE 109
REFERENCES 117
VITA 119
Vl11
LIST OF ILLUSTRATIONS
Figure Page
1. Schematic diagram for flat rolling. .......................... .............................. .... ................................ 7
2. Flat rolling differential elements. . . .... . .. .. .. ....... ...... .. .. .. 8
3. Overall system layout. . ,. ............ ......... ............................................................... 13
4. Thread terminology 18
5. Schematic diagram for fiber rolling. 19
6. Full fiber being processed. 20
7. Sample differential elements in deformation zone. 21
8. Evenly spaced deformation-zone cross sections. .. 22
9. Side-by-side cross-section comparison. . 23
10. Stacked cross-section comparison. . 23
11. Differential element variables before full-thread depth. 24
12. Differential element variables after full-thread depth. 25
13. Area definitions 28
14. Stresses on differential element. 31
15. Roll-face pressure distribution comparison for flat and fiber rolling 48
16. Roll-face pressure distribution for flat rolling with change in Jl. 49
17. Roll-face pressure distribution for fiber rolling with change in Jl. .. 49
18. Effect of system variables on roll torque for flat and fiber rolling. . 55
19. Model comparison using the Taguchi Method. 56
20. Model comparison using the Taguchi Method 57
21. Model comparison using the Taguchi Method. . 58
22. Roll-face pressure distribution for fiber rolling ",'ith change in Jl. .. ... ..... ... .. . 59
23. Roll-face pressure distribution for fiber rolling with change in N 60
24. Roll-face pressure distribution for fiber rolling with change in to' 61
25. Roll-face pressure distribution for fiber rolling with change in t f . .. 61
26. Roll-face pressure distribution for fiber rolling with change in R 62
27. Roll torque for fiber rolling with change in fl 63
28. Roll torque for fiber rolling with change in N 63
29. Roll torque for fiber rolling with change in to' 64
30. Roll torque for fiber rolling with change in tf' 64
IX
Figure Page
31. Roll torque for fiber rolling with change in R. 65
32. Smooth and threaded roll details. 70
33. Long roll insert details. 71
34. Short roll insert details. 72
35. Roll crank insert details 73
36. Manual crank handle details 74
37. Smooth-roll pillow box details 75
38. Threaded-roll pillow box details. . 76
39. Base plate details 77
40. Primary base angle details 78
41. Secondary base angle # 1 details. 79
42. Secondary base angle #2 details. 80
43. Primary frame member details 81
44. End plate details. . 82
45. Side plate details 83
46. Hopper details 84
47. Hopper bending details 85
48. Roll assembly. . 88
49. Roll assembly with handle 89
50. Roll assembly with pillow boxes 90
51. Base plate. .. 91
52. Base plate with angles. 92
53. Base plate with angles and frame 93
54. Base plate with angles and frame (4 views) 94
55. End and side plates 95
56. End and side plates (4 views) 96
57. Rolls and frame 97
58. Rolls and frame (4 views) 98
59. Pillo\v boxes and frame 99
60. Pillow boxes and frame (4 views). . 100
61. Hopper and frame. 101
62. Hopper and frame (4 vieV\Ts). 102
x
LIST OF TABLES
Table Page
I. System parameters. 43
II. L18(21 x 37) array 43
III. Fiber rolling orthogonal array. . 44
IV. Tensile test results. . 47
V. Flat and fiber rolling variable settings. 50
VI. Flat rolling array results. 51
VII. Average roll torque for flat rolling. .. .. .. .. .. 51
VIII. Variable effects on roll torque in flat rolling. .. .. .. .. ... 52
IX. Optimum combination for flat rolling 52
X. Fiber rolling array results. .. 53
XI. Average roll torque for fiber rolling. 54
XII. Variable effects on roll torque in fiber rolling. 54
XIII. Optimum combination for fiber rolling 55
I. INTRODUCTION
A. BACKGROUND
Some major concerns still exist regarding the concept of recycling. These concerns revolve around
the cost, reclaimability, and efficiency of the entire recycling process. Thus, the only items being
commonly recycled at present are those that can be easily identified and separated from the bulk of the
trash content, such as plastic bags, aluminum cans, and paper bags [1]. In many cases (especially with
paper), recycling also degrades the material making it inferior to virgin material. However, as the natural
resources available to man continues to decline, recycling will gain in importance and acceptance.
1. Recycling Plastics. Waste plastic consists primarily of plastic bottles, bags, and sheeting. The
material is thin and can be ground into irregular flakes at a very low cost in order to reduce the storage
volume and shipping cost [2]. The plastics come in many colors, and the principle contaminants are the
paper labels and adhesives.
Only a small portion of the plastic materials being disposed of daily is currently being recycled.
Clear # 1 two-liter bottles and opaque #2 milk containers are commonly recycled, but most other plastic
bottles (including colored # 1 and #2 bottles), bags, and films generally end up in landfills. These
plastics are commonly referred to as the #3-7's [2].
There are several reasons why most plastics end up in a landfill rather than being recycled.
Companies are very good at recycling their trimmings in the plant, because they know the content and
color and can control the contamination. Trimmings are ground and recycled in the plant. However,
post-consumer waste has a more questionable history. The potential for contamination with paint, oil,
cleaners, solvents, etc. is high, which prevents recycled plastics from being used as food packaging. So
the first problem in recycling plastics is that it cannot be used for food packaging, which is a major use
for virgin plastics.
The second problem is color control. If plastics of many colors are mixed together, the resulting
color is unattractive, and the product made from it will be unappealing. Manufacturers can add pigment
and control the color when using clear two-liter bottles or opaque milk containers, but colored bottles
pose major problems. In order to use the recycled plastic, a special washer to remove the paper labels
and adhesives must be used first [2], and the plastic must be washed several times to remove a majority
of the contamination. The next step is to melt the plastic, run it through a fine screen to remove the
remainder of the contamination, extrude it, and finally cut it into pellets. The pellets are then used just
like virgin plastic resin. Even after all this cleaning, there is still enough contamination in the resin that
light colors, like yellow, cannot be controlled. Dark colors such as blue and black are easily controlled,
2
and medium colors such as red can be reasonably well controlled. Color control is very important for
most consumer products. This is the primary reason colored bottles are not recycled.
A third problem arises if the plastics are to be molded. Virgin resins come in batches, and the
rheological (melt) properties will be consistent within each batch [2]. There are variations from batch
to batch, and the equipment must be adjusted to accommodate the change in properties. This problem
is amplified for recycled plastics because the rheological properties will change constantly, due to
plastics from many different batches being mixed together. When doing a molding process using
recycled plastic, the manufacturer must be willing to accept a slightly higher reject rate and a little more
down time.
In traditional recycling efforts, plastic material is first broken down into its original petroleum-like
state where it can be remixed and meshed with other texturizers and reinforcers for reprocessing [1],
which effectively discards the processing that went into making a finished product in the first place. A
more efficient approach would be to take advantage of this previous processing. This research effort has
sought to find a way of taking full advantage of the recycled plastic material in its existing state, with
the goal of better utilizing man-kind's recyclable resources.
2. Plastic Fibers in Concrete Reinforcement. Recycled plastic can be a valuable raw material if
products are developed to take advantage of its properties. Using it as a filler in asphalt or other low
value materials is not the best use of recycled plastic. In those applications the recycled plastic must
compete on a cost basis with tar, sand, and gravel, which is not a realistic possibility.
It has been determined that virgin-plastic fibers can be used to improve the strength and toughness
ofconcrete [2]. Plastic fibers for reinforcing concrete currently retail for $7 per kg, so this would be a
much more valuable use for recycled plastics. It is a better utilization of resources.
The cost ofvirgin plastic limits the percent content that can be added to concrete, thereby limiting
the performance improvements. One might think that recycled plastics could provide a cost effective
substitute to virgin plastics, but current processing techniques require that recycled plastic must be
carefully selected and thoroughly cleaned before it can be used to produce fibers. The cost of these
additional selection and cleaning processes prevents the use of recycled plastics and encourages the use
of virgin plastics in concrete reinforcement at the present time.
One of the current processes for producing virgin-plastic fibers is similar to making fishing line
[1]. A drawing technique is utilized, where the molten material is drawn through a small opening and
cooled to form hair-thin lines. These lines are later cut into fibers lengths. Another technique involves
fibrillated polypropylene, where thin plastic sheet is perforated to form row after ro\v of fibers ready for
fibrillation. The process developed here is much simpler than the processes currently used in making
fibers and \vill produce lower cost fibers.
3
Making fibers for concrete reinforcement is a good use for recycled plastics. Concrete is a dirty
material, so a little contamination in the fibers will not affect the concrete. The color of the fibers is
unimportant. The process developed in this research effort does not involve molding, so variations in
rheology will be less important than for other products. Therefore, the problems generally associated
with recycling plastics can be overcome with this process.
3. Fiber Rolling Process. The process proposed here could overcome the conventional limitations
of recycled-plastic, fiber-reinforced concrete. This process would welcome many types of normally
unacceptable recycled plastics and eliminate much of the need for cleaning before processing. The
overall effect would be to decrease the amount of potentially useful plastic that is conventionally
discarded into landfills each year and at the same time make it economically feasible to use higher
contents of plastic fiber in concrete production.
The first step involves the chopping of unsorted recycled plastics into small flakes - a process that
is relatively cheap and commonly employed in other recycling operations. The chopped plastic would
then be slightly heated while moving through a hopper system before entering a set of rolls that would
simultaneously elongate and cut the plastic into thin fibers. This part of the process involves one smooth
roll and one threaded roll which are in contact with each other. The close spacing provides for the
needed elongation and reduction in cross-section of the flakes, while the threads cut the flakes into
individual fibers.
The developed process takes advantage of the recycled plastic's previous processing. It is not
necessary to sort by color or remove the paper labels or adhesives. Extensive cleaning is not required~
removing the rocks, sand, and larger dirt particles is adequate. The recycled plastics have been stretched
biaxially by the blowing and extrusion operations which were used to make them into bottles, bags, or
sheeting, and thus exhibit a biaxial orientation of the polymer molecules [2]. This prealignment assists
in aligning the molecules with the fibers, \\7hich makes them stronger and stiffer. Preliminary results
from this new process are promising.
B. OBJECTIVES
The primary goal of this research effort has been to develop a mathematical model of the
simultaneous rolling and cutting action of a machine that processes recycled plastic flakes into thin fibers
for use in concrete reinforcement. The intended purpose for the model is to show quantitatively how the
parameters affect the properties of the fibers produced. Predictions from the models would hopefully
aIlo\\' the machine developed in this research effort to be modified or completely redesigned if needed.
Through this effort a fundamental understanding of the process has been developed, and it is now
possible to optimize the different parameters of the fiber production process.
4
Another objective ofthis research effort has been to develop a useful product for plastics that are
othenvise not recycled at present. Clear two-liter bottles and opaque milk containers are commonly
recycled, but most other plastic bottles, bags, and films generally end up in landfills.
The final objective of this research effort has been to develop a low-cost fiber for concrete
reinforcement. Plastic fibers are widely used for reinforcing concrete, and they retail for approximately
$7 per kilogram. This represents a large market potential for a cheaper recycled-plastic fiber. Hopefully
this research effort will contribute to recycled plastic fibers becoming an economic possibility.
C. ORGANIZATION
A systematic approach is taken in this study to describe the present state of development in fiber
reinforced concrete and how recycled plastics may someday playa major role in this field. Since a fiber
rolling operation will be used, standard flat rolling theory will be reviewed. The present state of
development in the proposed recycled-plastic, fiber-rolling process will then be described, followed by
a presentation of the mathematical model developed for this process. The results of these modeling
efforts will be given, and, finally, some conclusions will be drawn regarding accomplishments of this
study.
5
II. PREVIOUS RESEARCH
A. FIBER-REINFORCED CONCRETE
Fiber-reinforced concrete has a randomly oriented distribution of fme fibers added to a traditional
concrete mix. The fiber size varies, with the length typically less than 50 mm and the diameter typically
less than 1.0 mm. There have been- many types of fibers used to reinforce concrete [3-16]. Steel fibers
provide very good, reasonably-priced reinforcement. Unfortunately, the fibers corrode over time, and
after 6 to 8 years they provide very little reinforcement [14]. Stainless steel makes good fibers, but they
are expensive. Glass fibers provide good reinforcement, but the alkali nature of the concrete causes their
strength to degrade with time. Graphite and kevlar provide excellent reinforcement but are very
expensive. Plastic fibers provide good reinforcement at a reasonable cost. The most commonly used
fiber today is made from fibrillated polypropylene.
There has b-een a substantial amount of research, much of it done at the NSF Center for Science
and Technology of Advanced Cement-Based Materials at Northwestern University, which shows that
adding 1% to 3% by volume fibers in the concrete mix improves the strength of the concrete by ul? to
250/0 and substantially increases the toughness by approximately a factor of four [3-16]. Therefore,
fiber-reinforced concrete is less susceptible to cracking than ordinary concrete and will have a longer
service life. It is a superior material, but it is also a more expensive material. It is currently cost
prohibitive to add 1% to 30/0 by volume fibers to the mix for most applications.
Contractors presently use 0.1 % by volume fibers in the mix. This low volume fraction seems to
stabilize the mix and reduce the shrinkage cracking. There is some improvement in the toughness, so
the concrete will be less susceptible to cracking, but there is no improvement in strength. If fibers could
be produced at a low enough cost, then it would be possible to take full advantage of the strengthening
and toughening of 1% to 3% by volume fibers in the mix. The process investigated in this research effort
has the potential ofreducing the cost of the fibers so that a higher volume concentration of fibers in the
concrete mix could become an economic possibility.
B. TRADITIONAL ROLLING MODELS
An important bulk deformation process is rolling of materials; a mechanical process whereby
plastic defonnation of a material is achieved by passing it through a pair of rotating rolls. Rolling is used
extensively in metals processing. Almost 90% of all steel, aluminum, and copper produced annually is
rolled [1 7]. An overview of the most common rolling practices can be found in most manufacturing
texts.
6
Rolling can be used to produce materials with a variety of shapes and mechanical properties in
relatively long lengths and uniform cross sections. Common shapes include blooms, billets, slabs, plates,
bars, structural shapes, rails, angles, pipes, strip, foil, etc. A rolling mill is the general name given to
rolling machin6yr, consisting of the rolls, support structure, and drive system. The geometry of the final
product depends on the shape of the rolls -- either flat or grooved. Most products, especially metal
products, require many passes through rolling mills before the desired geometry is achieved. The
sequence depends on factors such as the type and amount of material to be worked, the amount of
reduction in thickness, the change in shape, the available roll-mill equipment, and the desired mechanical
properties of the finished product.
Flat rolling, the use of smooth cylindrical rolls, is a well developed field of study. Several
modeling methods [18] and assumptions have been used to predict solutions. These modeling methods
include "work of deformation," "slip-line solution," "upper-bound solution," and various "slab
methods," with slab methods being the most commonly used [17- 21]. The slab method, or free-body
equilibrium approach to rolling, was first developed by von Karman around 1925 [19, 22]. Since that
time almost every aspect of flat rolling has been examinee!.
The rolling of shapes other than flat sheets or bars seems to be based more on experimentation
than mathematical models [23,24]. Judging from the extent of published literature, rolling is one of the
most investigated of all the forming processes [18]. However, little can be found on the mathematical
modeling ofshape rolling. Apparently, mill designers, builders, and operators use the flat rolling models
and years of experience as the basis for their understanding of most shape rolling. This would seem to
make sense \vhen one considers the enormous variety of roll geometries used to produce these non-flat
shapes [25, 26].
When considering the complexity of even a simple roll geometry, the governing differential
equations from the models would have been of little use to a mill smith a few years ago without the
numerical solution capability ofmodem computers. For instance, the differential relationships developed
by von Kannan in 1925 for the relatively simple flat rolling model have not been solved analytically as
of yet [19]. The use of some simplifying assumptions or a numerical approximation method must be
used to obtain useful results.
This lack of established mathematical models for shape rolling has led to a new model being
developed for the rolling of plastic fibers. The procedure used in this new model is similar to the slab
method for flat rolling. Therefore, the basic derivation for representing the roll pressure distribution in
flat rolling [17-21] will be given here for comparison purposes bet\veen flat and fiber rolling.
7
1. Mechanics of Flat Rolling. Figure 1 shows some of the tenninology used in flat rolling~ where
<p is the angular position along the arc of contact,
<p0 is the entry angle,
<Pf is the exit angle (<Pf is always zero for this model),
<PN is the neutral point angle where the material is moving at the same speed as the rolls,
~ is the coefficient of friction between the material and the rolls,
ho is the entry thickness,
hf is the exit thickness,
R is the undeformed roll radius,
p is the roll pressure,
and T is the roll torque.
Figure 1. Schematic diagram for flat rolling.
Depth into the page is assumed to be unity. The material slips relative to the rolls as it is being
dravm through the rolls. The friction forces are in opposite directions on either side of the neutral plane,
due to the material moving slower than the rolls on the entry side of the neutral point and faster than the
rolls on the exit side of the neutral point. A net friction force is required toward the exit to ensure the
material is dra\\l1 through the rolls. Therefore, the neutral point is usually closer to the exit than the entry
[20].
8
From Figure 1 the entry angle can be given as follows.
(h - h 1<P = cos -1 1 - 0 f
o 2 R
Figure 2 shows arbitrary differential elements before and after the neutral point, where
det> is the change in et> over the differential element,
h is the thickness of the element at et>,
dh is the differential change in thickness over the element,
and dx is the differential width of the element.
Notice that the only difference between the two elements is the direction of the friction force.
(1)
x~-----
Ox+dox----II~
h dx
(b)
Figure 2. Flat rolling differential elements. (a) Before neutralpoint. (b) After neutral point.
2. Assumptions. Several assumptions are made to make the problem more manageable [17].
These include:
• plane-strain compression,
• homogeneous deformation,
• rigid, perfectly plastic material,
• constant coefficient of friction along the arc of contact,
• rolls of equal diameter,
• and no elastic deformation of the rolls.
Refer to references [17-21] for a thorough explanation of the reasoning behind these assumptions.
3. Force Balance. Performing a force balance in the x-direction on the elements both before and
after the neutral point in Figure 2, gives
(3)
9
where the element before the neutral point contributes the (+) sign, and the element after the neutral pointcontributes the (-) sign. Neglecting second-order terms, this can be reduced to
d(a h)__x_ = 2pR(sinq, =t= Jlcosq,) .
dq,If the angle <P is assumed to be small, p can be considered a principle stress, with Ox as the other
principle stress. From the maximum-shear-stress theory for plane strain [27], the relationship between
these two stresses and the flow stress Yf is given by
2P - ° = -y = y
x I 'f3(4)
where Y is the modified flow stress. With Ox = 0 at the entry and exit zones, boundary conditions
consist of the dimensionless ratio plY =1 at both <P = <Po and <P = <Pf'
The thickness h can be defined as
h = hi + 2R(1 - cos<p) (5)
Inserting (4) and (5) into (3) and dividing through by Y, the following differential relationship can bederived.
+ 2~R cos<f> ( ~) + 2R sin<f>
hf
+2R(1 -cosq,)(6)
where the (-) sign corresponds to the element before the neutral point and the (+) sign corresponds to the
element after the neutral point.
4. Detennination of the Neutral Point. The differential equation in (6) provides the relationship
for two roll pressure equations, and two boundary conditions have been specified to solve these
differential relationships. However, there is an extra boundary condition that has remained unspecified
as of yet. This unused boundary condition states that the two relationships in (6) meet at the neutral
point. Therefore, setting the two relationships equal at <P=<I>N provides a means of solving for the neutral
point location on the arc of contact. Graphically, the two curves will intersect at the neutral point.
5. Governing Equations Summary. In summary, the following equations in matrix form govern
the flat rolling model.
d[Pl(¢)/Y]P4(¢)/Y
d¢
\vith boundary conditions
hf
+ 2R( 1 - cos <p)(7)
10
(8)
where PI corresponds to roll pressure before the neutral point and P4 corresponds to roll pressure after
the neutral point. The subscripts are added for ease of comparison with the fiber rolling model developed
in Section IV.
The set of differential equations in (7) and boundary conditions in (8) define the dimensionless
ratio of roll pressure over modified flow stress before and after the neutral point. Plots from (7) and (8)
will be presented in Section V as a comparison between flat rolling and the newly derived model for
fiber rolling.
6. Roll Force and Torque. The roll force, F, for each roll can be calculated by multiplying the area
under the roll pressure curves given in (7) by the full width, W, of material being rolled, which is
represented by the following expression.
$0 $N
F = W R Jp\(¢) d¢ + JP4(¢) d¢$N <Pf
Likewise, the roll torque for each roll is represented by the following expression.
(9)
<Po <PN
T = Jl W R 2 JPl(¢) d¢ - JP4(¢) d¢ (10)
$N <Pf
Note that the negative sign in (10) accounts for the opposite direction of the friction forces on either side
of the neutral point.
7. Constraints. There are upper and lower limits on material thicknesses that can be processed
under a given set of rolling conditions. The upper limit on thickness insures the material can be drawn
into the rolls without assistance. The draft, L\h, for flat rolling is defined as the difference between the
initial and final thicknesses. Clearly for large enough drafts compared to the roll diameter, the material
being fed to the rolls will at some point become too thick to be pulled into the rolls, instead slipping
against the roll faces and never entering the rolls. This limitation leads to a maximum <Po, called the
angle ofacceptance, being derived. Edward Mielnik [17] defines this value as
~ = tan-I"¥o,max Jl .
Substituting (1) into (11), the follo\ving inequality for maximum draft can be derived.
Ii.hmax = ( h 0 - hr )max :s 2 R [ 1 - cos ( tan - I Jl ) ]
(11 )
(12)
Notice that maximlUll reduction in material thickness depends on the coefficient of friction between the
11
material and the rolls and the radius of the rolls. If a draft size greater than the maximum given in (12)
is attempted, the friction forces between the material and rolls will not be able to pull the material
through the rolls. This limitation can be relaxed with the use of front and back tension on the material.
However, fiber rolling will not be allowed this luxury, so the characteristics of external tensions on flat
rolling will not be presented here.
The lower limit on material thickness involves a minimum draft commonly encountered in flat
rolling practice [17], which is due to roll deflection (or roll flattening). Reducing the gap between the
rolls any further just creates a greater elastic deformation of the rolls and the mill's support structure,
resulting in the material's thickness not being reduced beyond this minimum thickness. However, this
effect is minimized with rolls possessing a much greater modulus of elasticity than the material being
processed, as in the case of rolling plastic with steel rolls.
Modifying the defmition for draft, the percent reduction for flat rolling is defmed as
=[ hOh-ohf)Percent Reduction · 100% . (13)
This will be used as a comparison tool between the flat and fiber rolling models.
12
III. RECYCLED PLASTIC-FIBER ROLLING PROCESS
One of the first questions posed by this research topic involved whether recycled plastics could
be cut fine enough to mimic virgin-polypropylene fibers for concrete reinforcement. An effort was made
to find a currently-employed cutting process that was capable of producing thin fibers from recycled
plastics. The search turned up only a granulator, which cuts the plastic into small flakes with a surface
area between O. 12 and 1.0 square centimeters and a thickness of the original recycled material [1].
Granulators are already commonly used in conventional plastics recycling but with different intentions.
The purpose of the granulators is to cheaply get the plastic into a more manageable state for shipping,
storage, and further processing. Most recycling collection centers employ granulators as the first step
in recycling plastics.
Another question posed at the beginning of this study was whether or not current plastic-recycling
techniques could be used to produce fibers that are economically competitive with virgin-plastic fibers.
The current trend in recycling is to first break the waste plastic down into its basic polymer state and then
reprocess it just as virgin plastic. Of course, the extra processing costs involved in getting the recycled
plastic ready to reprocess make recycled fibers more expensive than virgin-plastic fibers.
The conclusion was that it is nearly impossible at the current state of recycling technology to
produce fibers from recycled plastics in the same manner as virgin-polypropylene fibers at a comparable
price. It would also be nearly impossible to simply cut the recycled material thin enough to make the
hair-thin fibers. Some kind of forming process would need to be employed.
A. RECYCLED PLASTIC PROCESSING
A new technique for processing recycled plastics into plastic fibers has been conceived at the
University of Missouri-Rolla that could·someday make recycled plastic fibers an economic alternative
to virgin plastic fibers in concrete reinforcement. The plastic would be chopped into flakes and rolled
into fibers with a specially designed rolling mill. A simple flowchart diagram of a possible overall
system layout for this new process can be seen in Figure 3. A truck would deliver a load of recycled
plastics. These could be a mixed variety of full-size plastic items collected at a materials recovery
facility, MRF, or pre-sorted plastics that are unacceptable at other processing facilities.
For this new process, the first step is to granulate the plastic into flakes of approximately one
square centimeter and the thickness of the recycled item it came from. Most recycling collection centers
already perform this step to reduce the plastic's storage volU';TIe and shipping cost. Ho\vever, this
operation could be perfonned in-house if the materials are delivered in uncut form. Selecting the proper
13
Receiving
concrete/PlasticMixture
Bagger
storageHopper
storageHopperstorage
Hopper
Granulator
RollingMill Shipping
Figure 3. Overall system layout.
granulator-screen size will assure that the flakes are suitably sized for the later rolling-cutting operation
will pass out of the granulator. This system would eliminate the cleaning operations that are usually
required at this point to remove labels and contaminants. The chopped flakes could then be stored in a
large hopper until ready for the rolling operation.
A hopper directly preceding the rolls would heat the flakes to the desired processing temperature.
Heating would soften the plastic and make it flow through the rolls more easily. If the plastic is too cold,
the fibers will curl up and appear to be embrittled -- having poor strength. If the plastic is heated too
much, then the flakes will stick to each other and to the rolls and the process will not \vork at all. This
process is being developed to use a mix of different types of plastic bottles, bags, and films, so a
temperature range will need to be identified which works well for all the commonly used plastics (HDPE,
LDPE, PETE, PVC, PP, PS).
The flakes \vill then run through the rolling mill. One of the rolls will be threaded and the other
\vill be smooth. The rolling action will separate the plastic into fibers. The thread depth is designed to
14
be smaller than the thickness of the plastic, so the plastic will be elongated in the direction of the fibers
as it passes through the rolls. This flowing of plastic in the direction of the fibers will help align the
polymer molecules making the fibers stronger and stiffer.
Once through the rolls, the newly produced fibers would again be sent to storage to await further
operations. These operations could include bagging or direct addition to concrete mixtures, if the process
were incorporated on-site with a concrete producer.
Currently, bags of virgin-plastic fibers are sold to concrete producers, who place the desired
number of bags into a concrete transport truck as it leaves the plant. The bags are water soluble and
allow the fibers to mix with the concrete as it is transported to the construction site. This new process
can easily accommodate this technique.
The fiber cross-sectional size and shape can be altered by changing the thread pitch and/or the
shape of the threads on the threaded roll. The length of the fibers can be altered by controlling the size
of the granulated flake, which is done by changing the screen in the granulator. A model has been
developed to allow optimization of these geometric aspects and will be presented in Section IV.
B. PROTOTYPE ROLLING-MILL DESIGN
1. Current Prototype Development. A laboratory-scale prototype rolling mill has already been
designed and fabricated. Currently, the prototype's rolls are powered by a simple hand crank. A hopper
has also been fabricated, but the heating process has not been added to the design as of yet. Therefore,
testing small batches at this point involves sprinkling a handful of plastic flakes directly onto the rolls
and turning the rolls by hand to produce the fibers, effectively bypassing the hopper system. The
processed fibers would simply fall onto the base plate for retrieval and examination.
Several drawings were prepared to aid in the construction and assembly of the prototype.
Appendix A contains a set ofmachining diagrams to the prototype rolling mill, and Appendix B contains
a set of assembly drawings to the prototype rolling mill.
The construction of a second prototype has begun but remains unfinished at this time. The overall
design is essentially the same as that given in Appendix A. The major modification to this rolling mill
is a set of more precisely machined rolls. The intention is to reduce tolerances between the rolls and
ensure complete separation of the individual fibers. The construction of these rolls has been completed,
but time constraints have kept the new rolls from being installed and tested in the first prototype's
support structure at present.
2. Future Prototype Development. As overall development of the process progresses, several
modifications w111 need to be performed to the prototype. The hand crank should eventually be replaced
15
by a motor and gear-drive system to operate the prototype at a greater capacity. Once this is realized,
larger test batches will be possible. For medium batches, the hopper could be partially filled, and a small
cardboard box could be placed under the rolls for fiber collection. For full-scale operation, a conveyor
could continually fill the hopper with flakes from the granulator, while another conveyor could remove
the processed fibers from beneath the mill. The hopper directly preceding the rolls will need a
temperature control unit installedto warm the plastic on its way to the rolls.
Development should ultimately lead to an industrial-scale system involving the granulator, heated
hopper, rolling mill, storage hoppers, packaging equipment, and a conveyor system between each
component in the process.
C. MATHEMATICAL MODELING
1. Current Modeling Efforts. Rolling ofplastic is not a common industrial practice, and this study
has turned up sparse mathematical modeling work published on the rolling of plastics and to no extent
the rolling ofplastic fibers. Models have been developed for the flat rolling of metals, therefore a model
similar to those used for modeling the rolling of metals will be employed.
This frrst model for plastic-fiber rolling will be based on the assumption that the plastic behaves
as an elastoplastic material, even though plastics do not typically behave as elastoplastic materials. The
reason this will be tried is because it is the simplest approach, and it has been shown to work well for
metals. This model will serve as a benchmark for future modeling work.
If the material strain hardens, then the flow stress is a function of the strain history of the process,
and it is necessary to express the amount of strain in the material as a function of the location of the
differential element along the arc of contact. Metals typically behave as either elastoplastic or strain
hardening materials, so the model development process given here is adequate for the processing of
metals. Plastics behave as viseoelastic materials, so the modeling assumptions of elastoplastic or strain
hardening will not be completely accurate for plastics.
Once material properties have been incorporated into the models, it will be possible to solve for
the pressures acting on the plastic and the rolls, and from these values it will be possible to predict torque
and power requirements to turn the rolls. Parametric studies will be perfonned to sho\\! how varying the
system parameters affects the size and shape of the fibers produced and the roll torque required to
produce them. Predictions can be made as how to optimize the process.
2. Future Modeling Efforts. The next incremental improvement of the modeling would be to treat
the plastic as a strain hardening material. In this approach the flow stress will depend on the strain
history of the material, but it will increase as the strain along the arc of contact increases. The entire
16
stress-strain curve for the material being processed would be used in these models. This is a fairly
general approach and may be adequate for plastics. Incorporating viseoplastic properties into the model
would most probably be the next step.
Another part of the future modeling efforts could be to study the effect of different thread profiles.
Triangular threads are the most commonly produced thread patterns, but it is possible to cut other shapes
into the roll. If a different shape makes the process better, then it would be worth some extra cost in
producing the rolls. In order to study different thread profiles, it will be necessary to develop new
differential and defonnation elements, but the same basic procedure can be used to develop these models
as will be given here.
In order to fully understand the mechanics of the process, it will someday be necessary to develop
fmite element models of the process. This will require considerably more effort than the model presented
here. The fmite element method is the most realistic and accurate way to study a flow process like this,
and true viseoplastic material properties can be used, rather than the flow stress used here. The plan for
this model is to refine the process as much as possible, and then later research could use the finite
element method to study the most interesting cases.
D. EXPERIMENTAL WORK
1. Current Experimental Efforts. Preliminary tests indicate that the flakes do not spread laterally
as they go through the rolls. This is an easy situation to see in flat rolling steel plates where the material
width is relatively large compared to the thickness. Here the width to thickness ratio is probably less,
but the threads cutting into the flakes prevents spreading in the lateral direction. Virtually all of the
deformation goes into stretching the fiber along its length and reducing the cross-sectional area.
Dog-bone shaped specimens have been cut from recycled plastics, and tensile tests have been
performed to determine reasonable values for the flow stress in the model. The flow stress will be a
function of temperature and processing speed, but these effects have been neglected at present. With the
absence of a working heated hopper on the prototype resulting in room-temperature operation and with
the mill being hand operated, tensile tests for flow stress were run at room temperature.
2. Future Experimental Efforts. There will be some basic material testing necessary to support
the modeling effort. Tensile test specimens need to be cut from the recycled plastics and pulled to
determine the full stress-strain curve, which will be used in modeling the material as an elastoplastic or
strain-hardening material. Accurate viscoelastic properties will be more difficult to obtain, but will also
need to be obtained eventually.
17
Another part of the experimental effort should be to modify the process according to the
predictions of the modeling effort. New rolls can be designed and built to optimize the shape of the fiber.
A heated hopper can be added to the prototype rolling mill to precisely control the temperature of the
process. The screen size in the granulator can be altered to change the size of the plastic flakes. The
primary purpose of the experimental work should be to validate the model and make sure the predictions
made by the model are reasonable.
Additional experimental work will be to test the strength and stiffness of the fibers produced. This
will be a necessary part of the effort because the goal is to optimize the process to produce the best
quality fibers, and strength and stiffness are two of the important properties that can be compared to
virgin-plastic fibers. Length and cross-sectional area, strength, and stiffness of the fibers will need to
be correlated with the processing parameters.
For the different geometric aspects and temperatures studied, it will be necessary to estimate the
quality ofthe fibers produced and the power required to produce the fibers. By carefully measuring the
torque required to tum the rolls, it will be possible to estimate the power required. The torque required
will depend on the geometry and the viseoplastic properties of the plastic, which will be a function of
temperature.
w~
Figure 4. Thread terminology.
18
IV. MECHANICS OF FIBER ROLLING
The procedure used in developing this mathematical model for fiber rolling is similar to the flat
rolling model presented in Section II. In developing the model, the terminology will be defined and the
assumptions will be presented. Differential elements (slabs) will then be defined, and a force balance
performed to derive the governing differential equations. Finally, a numerical method will be used to
obtain solutions to the governing equations making it possible to estimate roll separation forces and roll
torques. The modeling effort will conclude with the ground-work for a comparison between flat and
fiber rolling and a parametric study.
A. GEOMETRY TERMINOLOGY FOR FIBER ROLLING
Figure 4 shows some of the terminology associated with screw threads [28], which also apply to
the threaded roll, where
N is the number of threads per inch,
D is the overall diameter of the screw threads,
w is the thread pitch,
t f is the height of the thread,
A is the lead angle,
*is the helix angle,
and 8 is the thread angle.
From Figure 4 the following relationships can be defined.
1W=-
N(14)
(15)
w = 2 D tan A (16)
Looking at (16), it can be seen that either a small thread pitch or a large roll diameter will cause
the lead angle to be small. This argument will be used to say that A is negligibly small (or *is nearly
90 degrees) in the fiber rolling model, where the roll diameter is large compared to the thread pitch.
Standard thread profiles are determined by specifying D, w, and 8, and for standard threads e is
approximately 60 degrees. However, this model will let ebe determined by specifying D, \V, and t f . This
\vill aIlo\v for convenience in comparing results in Section V between the flat and fiber rolling models.
19
Figure 5 shows some of the terminology used in fiber rolling, where
<P is the angular position along the arc of contact,
<P0 is the entry angle,
<Pm is the angle at full-thread depth,
<PN is the neutral point angle,
<Pf is the exit angle (<Pf is always zero for this model ),
x is the distance from the vertical axis between roll centers and the point on the arc of
contact corresponding to <P,
to is the material entry thickness,
t f is the material exit thickness and the depth of the threads,
R is the undeformed roll radius,
F is the roll separating force,
and T is the roll torque.
Entry
II( • Fx
Figure 5. Schematic diagram for fiber rolling.
Threadedroll
Exit
Centerlineof rolls
Smoothroll
20
At the beginning of the process, the flake has a rectangular cross-sectional area, and the thickness
is equal to the original thickness of the flake. As the flake is drawn through the rolls, the threads begin
to cut into the plastic. The smooth roll is pressing on the bottom of the element, and the threaded roll
presses on the angled portions of the top of the element. The width of each finished fiber is equal to the
pitch of the threaded roll.
There must be a neutral point along the arc of contact, where the plastic is not slipping relative to
the rolls. Prior to that point, the rolls are moving faster than the plastic. After the neutral point, the
plastic is moving faster than the rolls. The friction forces are in opposite directions on either side of the
neutral plane. A net friction force is required toward the exit to ensure the material is drawn through the
rolls. Therefore, the neutral point is normally closer to the exit than the entry [20].
For thicker films where the flake thickness is greater than twice the thread height, as would come
from plastic bottles, this process would produce fibers that have a triangular cross-section that exits from
the rolls, because the material would fully fill the thread cavity. For thinner films where the flake
thickness is less than twice the thread height, such as would come from plastic bags, this process would
produce fibers that have a trapezoidal shaped cross-section as they exit the rolls, because the material
would not be thick enough to fully fill the thread. To study the full extent of this model, it will be
assumed from here on that the material is thick enough to completely fill the threads on the threaded roll.
If this were not the case, little or no alignment of the plastic molecules would be gained, and the process
would be essentially just a cutting operation.
Figure 6 shows the change in geometry of a fiber as it passes through the rolls. The top roll,
/ First contact wIth rolls
/ Full-thread depth
Figure 6. Full fiber being processed.
21
removed to show the plastic more clearly, would be threaded~ it provides for the triangular shape of the
finished fibers. The bottom roll, also removed in the diagram, would be smooth and would perform
exactly as a roll in a standard flat rolling operation. As seen in the figure, the leading edge of plastic has
been fully processed into fibers, while the undeformed back portion has not come into contact with the
rolls yet The middle portion is in contact with the rolls and is said to be in the deformation zone.
Notice that the top of the fiber remains flat until thefull-thread depth is reached. This top portion
does not contact either roll and therefore has no stress associated with it. Eventually the fiber will come
into full contact with the threaded roll, i.e. the plastic has fully filled the roll thread cavity. Due to this
stress-free surface, one can see that two unique differential elements will need to be used before and after
full-thread depth is achieved.
Figure 7 takes the same deformation-zone portion of the element presented in Figure 6 but divides
it into distinct segments to help envision the differential elements that will be considered. Note that with
enough divisions, the curved surfaces from Figure 6 can be approximated as flat surfaces, as in Figure
7. As the material goes from a rectangle before entering the deformation zone to a triangle exiting the
deformation zone, the first three elements, labeled (i), (ii), and (iii) in Figure 7, will be treated as the
same element because they come before full-thread depth and have the same characteristics. The last
(iv)
(i)
/ Full-threo.d depth
Figure 7. Sample differential elements in deformation zone.
22
two elements, labeled (iv) and (v), will be treated as the same element because they come after full-thread
depth and have the same characteristics.
Figure 8 shows evenly-spaced slices of the deformation-zone portion of the element in Figure 6
to better see the changes in geometry. Figure 9 lays these slices side-by-side and Figure 10 stacks these
slices on top of each other to see how the geometry of each slice is changing with respect to the other
slices.
/ Full-threo.oI oIepth
Figure 8. Evenly spaced deformation-zone cross sections.
From this insight into relative geometry change in Figures 9 and 10, Figures 11 and 12 are used
to define some of the geometry characteristics before and after full-thread depth, respectively, where
d<t> is the differential change in Q> over the length of the element,
dx is the differential length of the element along the centerline of the rolls (x-axis),
8 is the thread angle,
t is the thickness of the element at <P,
dt is the differential change in thickness over the length of the element,
h is the depth of the element into the thread at <1>,
dh is the differential change in h over the length of the element,
m is the distance from the centerline of the rolls (y-axis) to contact with the threaded roll
on either side of the element at <P,
Before Full-Threa.d Depth
23
Centerlineof rolls
After Full-Threa.d Depth
Figure 9. Side-by-side cross-section comparison.
Centerlineof rolls
---~=====::s---Centerlineof rolls
Figure 10. Stacked cross-section comparison.
24
dm is the differential change in m over the length of the element,
n is the distance from the centerline of the rolls (y-axis) to contact \vith the smooth roll on
either side of the element at <t>,
dn is the differential change in n over the length of the element,
v is the distance the element has been displaced by the thread on either side of the element
at <1>,
dv is the differential change in v over the length of the element,
u is the length across the top of the element which has not yet been displaced by the threads
at <1>,
du is the differential change in u over the length of the element,
and w is the width of the element and the thread pitch.
The dashed rectangle corresponds to the original shape of the flake on entry, and the dashed triangle
corresponds to the final shape of the fiber on exit.
I I
~------Jx+dx I
x
- Centerlineof rolls
v u vI...1.... ~I.. "I
I
1,1
II ,, 1
I" w
v+dv u+du v+dv~ 104-'''-~''! ~
--I T
i +h+dhI
t+dt I m+dm
I y +I I n+dn
I i
~~-----'~I
I
I
I 1I I
w
(a) (b) (c)
Figure 11. Differential element variables before full-thread depth. (a) Full element.(b) Front face at x. (c) Back face at x+dx.
From Figure 11, the follo,ving relationships for geometry characteristics before full-thread depth
can be derived.
25
v
w
iIhI
t,Z I'.J \ I m+dm, \ I
'. ----~, -ty I
I
I n+dnI
'"'----_... ~~I
l~v
Centerlineof rolls
xi
x+dx(a) (b) (c)
Figure 12. Differential element variables after full-thread depth. (a) Full element.(b) Front face at x. (c) Back face at x+dx.
x = R sin cP (17)
dx = R dcP cos ¢ (18)
th = -!!.. - R ( 1- cos <P ) (19)
2
dh - R d¢ sin ¢ (20)
m = R ( 1 - cos <P ) (21)
dm = R d¢ sin ¢ (22)
n = R ( 1- cos ¢ ) (23)
dn = R d¢ sin ¢ (24)
t = ~ + R ( 1 - cos ¢ ) (25)2
dt = R d¢ sin ¢
t - 2 R + 2 R cos ¢oV = -------
2 tan 8
dv = - R d¢ sin ¢tan 8
26
(26)
(27)
(28)
u = w -to - 2 R + 2 R cos ¢
tan 8(29)
du = 2R d¢ sin ¢tan 8
¢ = cos- 1 ( 1 + tf - ~ )m R 2R
(30)
(31)
(32)
From Figure 12, the following relationships for geometry characteristics after full-thread depth
can be derived.
x = R sin ¢
dx = R d¢ cos ¢
h=W tan82
m = R ( 1 - cos ¢ )
dm = R d¢ sin <p
n = R ( 1- cos ¢ )
(33)
(34)
(35)
(36)
(37)
(38)
dn = R dq, sin q,
27
(39)
v = w
2(40)
Notice that h and v are independent of <I> after full-thread depth has been achieved. These
variables are determined by the shape of the threaded roll and in tum determine the shape of the final
fiber.
B. ASSUMPTIONS
Several assumptions were made to make the problem more manageable [17]. These include:
• the plastic does not spread laterally as it goes through the rolls,
• homogeneous deformation,
• elastoplastic material,
• constant coefficient of friction along the arc of contact
• rolls of equal diameter,
• no elastic deformation of the rolls,
• the lead angle is small on the threaded roll,
• sharp vee threads, i.e. roots and crests not flattened or rounded, on the threaded roll,
• the material does not flow upward into the roll thread cavity before full-thread depth,
i.e. the top face of the fiber remains flat until full-thread depth is achieved,
• and the original thickness is greater than twice the final thickness, to>2tf .
Several of these assumptions are similar to those in the flat rolling model. The first assumption
says that there is no net flow of material in the y-direction across the width of the flake. This is
reasonable for situations where the material width is much greater than the thickness. Homogeneous
deformation means that the material has the same deformation properties throughout its volume.
Elastoplastic materials undergo plastic defonnation once the yield stress has been reached "rith no elastic
recovery. This will not be completely accurate for plastics, but it will serve as a first estimate for this
model. In actuality, the coefficient of friction will not remain constant along the arc of contact, but this
effect will be neglected for simplicity of the model.
Roll deformation should be negligibly small when rolling plastic flakes with relatively large
diameter and short-length steel rolls. For relatively large roll diameters and small thread pitch, the lead
angle is small, so the assumption of a small lead angle should be accurate. It is physically impossible
to produce thread roots and crests without some degree of rounding, but this assumption is made for a
28
first attempt at modeling the process. Future modeling efforts should probably take this effect into
account, because rounding is amplified in the fine thread pitches used in fiber rolling.
If the original thickness were less than twice the final thickness, full-thread depth would never be
achieved due to the assumption that the top face ofthe fiber remains flat before full-thread depth. If this
\vere the case, the differential elements defined as being after full-thread depth would not exist. For a
more thorough derivation that considers differential elements both before and after full-thread depth, the
requirement is made that to>21f, as mentioned at the beginning of this section. If a situation arises where
this requirement proves unuseful, one can simply disregard the governing equations for elements after
full-thread depth and use the governing equations for elements before full-thread depth along the full arc
of contact.
C. AREA CALCULATIONS
Figure 13 shows labels for each of the faces on the differential element before and after full-thread
depth, \vhere
Ax is the area of the front face at location x,
Ax+dx is the area of the back face at location x+dx,
Adx is the difference in area between the front and back faces,
Ab is the area of the bottom face of the element in contact with the smooth roll,
At is the non-contact area of the top face of the element before full-thread depth,
As is the area of the side face on either side of the element,
and Aa is the area of the angled face in contact with the threads on either side of the element.
A-s
(a) (b)
Figure 13. Area definitions. (a) Before full-thread depth.(b) After full-thread depth.
29
From Figure 13 (a) and (17) through (30), the following area fonnulas can be calculated for each
face of the differential element before full-thread depth.
Ax = [ ; + R ( 1 - cos ¢ ) ] - ~ e [ ; - R ( 1 - cos ¢ )]2 (41)
[t - 2R ( 1 - cos cP ) ] d'"A
dx= A +dx - A = R sin cP w + _0________ '+'
x x tanS
[to - 2R ( 1 - cos cP ) ]
At = R cos cP w - dcPtan 8
As = 2 R 2 cos cP ( 1 - cos cP ) dcP
R [t - 2 R ( 1 - cos <P ) ]A a = - o. J cos 2 8 + sin 2 8 cos 2 cP dcP
2 SIn 8
(42)
(43)
(44)
(45)
(46)
(47)
From Figure 13 (b) and (33) through (40), the following areas can be calculated for each face of
the differential element after full-thread depth.
w 2Ax = 2 w R ( 1- cos <P ) + - tan S
4
w2Ax +dx = 2 It) R ( 1 - cos ¢ + d¢ sin ¢) + - tan 8
4
A dx = Ax+dx - Ax = 2 w R d¢ sin ¢
(48)
(49)
(50)
30
As = 2 R 2 cos ¢ ( 1 - cos ¢ ) d¢
D. STRESS CALCULATIONS
(51)
(52)
(53)
With the coefficient of friction between the material and the rolls given by Jl, Figure 14 shows the
stresses on the differential elements before and after full-thread depth, where
Ox is the normal stress on the front face of the element at location x,
0x+dx is the normal stress on the back face of the element at location x+dx,
a b is the normal stress on the bottom face of the element in contact with the smooth roll,
JlOb is the friction stress on the bottom face of the element in contact with the smooth roll,
a S(+Y) is the normal stress on the positive-y side face of the element,
0s(_y) is the normal stress on the negative-y side face of the element,
0a(+y) is the normal stress on the positive-y angled face of the element,
0aC-y) is the normal stress on the negative-y angled face of the element,
Jl a a(+y) is the friction stress on the positive-y angled face of the element,
and Jl a a(-y) is the friction stress on the negative-y angled face of the element.
Notice that there is no stress on the top face of the element in Figure 14 (a) and (b) since it has no contact
with the rolls.
As can be seen from Figure 14, there are four differential elements that need to be analyzed. These
can be divided into two physically logical cases where only three elements need to be analyzed at a time.
1. Case I. The neutral point is before full-thread depth, i.e. <Po 2 <PN 2 <Pm.
• Let Differential Element 1, corresponding to Figure 14 (a), go from <Po to <PN'
• Let Differential Element 2, corresponding to Figure 14 (b), go from <PN to <Pm.
• Let Differential Element 4, corresponding to Figure 14 (d), go from <Pm to <Pr.
2. Case II. The neutral point is after full-thread depth, i.e. <Pm 2 <PN 2 <Pr.
• Let Differential Element 1, corresponding to Figure 14 (a), go from <Po to <Pm.
• Let Differential Element 3, corresponding to Figure 14 (c), go from <Pm to <PN'
• Let Differential Element 4~ corresponding to Figure 14 (d), go from <P~ to <Pr.
31
a a a (+y) ?a (-y)Jl a (+y) /'I:!",-::::::::::"-/ /
(a) (b)
Jl a a (+y) a a (+y) a a (_y)
ax
+dx l!~?S (-y) °X+dx
-~ Oxas (+y) as (+y)
Jla b \
(c) (d)
-Figure 14. Stresses on differential element. (a) Before full-
thread depth and before neutral point. (b) Beforefull-thread depth and after neutral point. (c) Afterfull-thread depth and before neutral point. (d)After full-thread depth and after neutral point.
These differential elements are defined in such a way to insure the same governing equation will
correspond with its numbered differential element no matter whether Case I or II is being studied. For
example~ the governing equation for Differential Element 4 in Case I will be the same as for Differential
Element 4 in Case II. Just the upper and lower bounds on the governing equations will change.
Case I will rarely occur, due to a net friction force being required in the direction of <Pf to pull the
strip through the rolls, thereby causing the neutral point to be closer to <Pf than <k and due to the
difference between <Po and <Pm being small for to»2t[. This reasoning \vill be supported by the results
presented in Section V.
Notice that Differential Elements 1 and 4 are similar to the differential elements in flat rolling
where full-thread depth has no meaning. This is why the subscripts 1 and 4 were added to (7) and (8)
for flat rolling.
32
E. FORCE BALANCE
1. Differential Element 1. Perfonning a three-dimensional force balance on Differential Element
1 gives the following vector equation.
'LF 1 -1 sinq, 0 0x
'LFy =0 A 0 + °x+dxA x+dx0 +obAb 0 +0 A -1 +0 A 1
x x s s s s
'LFz0 0 cosq, 0 0
sinq,
tan8coscP
- coscP(54)
- cos cP
osinq,
Jl aaAa+-------------J4tan 28cos 2q, +( 1 +tan28)sin2 cP
- 2tan 8 cos cP
- sincP
-tan 8 sin cP
Jl a aAa+-------------J4 tan 28 cos 2q, + ( 1 + tan 28 ) sin 2q,
- 2tan 8 cos cP
sincP
-tan 8 sinq,
ooo
The way the problem is formulated, there is no net flow of material in the y-direction, which is
symbolized by the y-components of (54) canceling each other out along with the effect of OS' Taking
the z-components from (54), one can derive the following relationship between 0a and 0b'
a a
Noticing that
cosq,+
cos cP + Jl sin q,
Jl tan 8 sin cP (55)
and
(56)
(57)
and then neglecting second-order terms~ one can derive the following.
(58)
Substituting (44), (55), and (58) into the x-component relationship from (54), the follo\ving relationship
between Ox and 0b can be derived.
where
d( Ox Ax )
dq, a b W R [ sin q, - Jl cos <P + ( cos <p + Jl sin <p ) F1( <p) ]
33
(59)
sinq,v 4tan 28cos 2<p + ( 1 +tan 2B) sin 2et> - 2 Jl tan Bces <PV 1 + tan 28cos 24>
cos <PV 4tan 28cos 24> + ( 1 + tan 28) sin 24> + fl tan 8 sin<PV 1 + tan 28cos 2<p(60)
With the assumption of no net flow in the lateral direction and if the angle <I> is assumed to be
small, 0b and Ox can be considered the two largest principle stresses. From the maximum-shear-stress
theory [27], the relationship between these two stresses and the flow stress Y f is given by
where Y is the modified flow stress.
2a - a = -y = Yb x .f3 f (61)
Inserting (61) into (59) and dividing through by Y, the following differential relationship for the
dimensionless ratio °b/Y can be derived for Differential Element 1.
(62)
where
G1(<p) = w R [ sin <p + fl cos <p + ( cos <p - Jl sin <p ) F} (<p) ] , (63)
F1(<p) is given by (60), Ax is given by (41), and A dx is given by (43). With Ox == 0 at the entry zone, the
boundary condition for (62) is °b,1 IY== 1 at 4> == 4>0'
2. Differential Element 2. Noticing that the only difference between Differential Element 1 and
Differential Element 2 is that the friction stresses are in the opposite direction, the governing differential
equation for Differential Element 2 is given by the following.
(64)
where
Gi¢) '" w R [ sin ¢ - fl cos ¢ + ( cos ¢ + fl sin ¢ ) F2(¢) ] , (65)
sin <PJ4tan 28 cos 2<P + ( 1 +tan 28 )sin 2<p + 2 ~ tan 8 cos <pJ1 + tan 28 cos 2<p
cos <PV 4 tan 28 cos 2<p + ( 1 + tan 28) sin 2q, - ~ tan 8 sin <PV 1 + tan 28 cos 2q,
34
(66)
Ax is given by (41), and~ is given by (43). With q from Differential Element 1 equal to q from
Differential Element 2 at <Pm, the boundary condition for (64) is 0b,1 /Y=Ob,2/Y at <P = <Pm.
3. Differential Element 3. Perfonning a three-dimensional force balance on Differential Element
3 gives the following vector equation.
LFx 1 -1 sinq, - cosq,
LF =oxAx 0 + °x+dxA x+dx 0 +ObAb 0 +flObAb 0Y
LF 0 0 coscP sinq,z
cos 8sinq, - cos q, 0oA
+ a a sin 8 cos q, +~ °aAa 0 +0 A -1Jcos 28 + sin 28 cos 2<p
s s
-cos Bcosq, - sin q, 0
cos 8sinq, - cos q, 0 0
+°aAa - sin 8 cos q, +J..l°aAa 0 +0 A 1 0
Jcos 28 + sin 28 cos 2<ps s
-cos 8cosq, - sinq, 0 0
(67)
The way the problem is formulated, there is no net flow of material in the y-direction, which is
symbolized by the y-components of (67) canceling each other out. Taking the z-components from (67),
one can derive the following relationship between 0a and 0b'
( cos q, + ~ sin q, ) Jcos 28 + sin 28 cos 2q,
cos 8 cos <p + 11 sin <p Jcos 28 + sin 28 cos 2<p(68)
Substituting (51), (58), and (68) into the x-eomponent relationship from (67), the following relationship
between ax and 0b can be derived.
d( Ox Ax )d¢ = ab w R [ sin ¢ - 11 cos ¢ + ( cos ¢ + 11 sin ¢ ) F3( ¢) ] (69)
where
cos 8 sin <p + ~ cos <p Jcos 2 8 + sin 2 8 cos 2 <p
cos 8 cos <P - J.1 sin <p Jcos 2 8 + sin 2 8 cos 2 <p
35
(70)
Inserting (61) into (70) and dividing through by Y, the following differential relationship can be
derived for Differential Element 3.
(71)
where
G3(<p) = w R [ sin <p + ~ cos cP + ( cos <p - fl sin <p ) F3(cP) ] , (72)
F3(<p) is given by (70), Ax is given by (48), and A dx is given by (50). With ax from Differential Element
3 equal to ax from Differential Element 4 at <An, the boundary condition for (71) is C\ 3 /Y=<\ 4/Y at, ,
4. Differential Element 4. Noticing that the only difference between Differential Element 3 and
Differential Element 4 is that the friction stresses are in the opposite direction, the governing differential
equation for Differential Element 4 is given by the following.
(73)
where
G4(<p) = w R [ sin <p - J.1 cos <p + ( cos <p + Jl sin cP ) F4(<p) ] , (74)
cos 8 sin <p - fl cos <p Jcos 2e + sin 2 8 cos 2 c.P
cos 8 cos <p + fl sin c.P Jcos 2 8 + sin 2 8 cos 2 c.P(75)
Ax is given by (48), and~ is given by (50). With Ox =0 at <Il, the boundary condition for (73) is
o b,4/Y == 1 at <P = <Pr·
F. DETERMINATION OF THE NEUTRAL POINT
For both Case I and Case II, three differential equations have been derived, and a boundary
condition for each equation has been specified. This leaves one extra boundary condition in each case
36
that has remained unspecified as of yet. This unused boundary condition states that two of the three
equations in each case will meet at the neutral point with equal roll pressure values. Therefore, setting
the two relationships directly opposite the neutral point equal to each other with ¢>=¢>N provides a means
of solving for the neutral point location on the arc of contact. Graphically, two of the three curves in
either case will intersect at the neutral point, which will be seen in Section V.
G. GOVERNING DIFFERENTIAL EQUATIONS SUMMARY
The set ofdifferential relationships in (62), (64), (71), and (73) defme the dimensionless ratio of
roll pressure over modified flow stress.
1. Case I. The neutral point is before full-thread depth, i.e. 4>0 2 ~ 2 chn. In summary, the
governing differential equations for Case I in matrix form are given by
(Obo(q,)]
d '~ =[ G i(q,) -Adx,i ( ab,i (<P)] + Adx, i] ,
dq, A . Y A 0
X,l X,l
i = 1,2,and4(76)
Ax,l
Ax,2
Ax,4
Adx,l
Adx,2
Adx,4
t [ t ]2-!: +R( 1 - cos cP ) - _1_ -!: - R( 1 - cos q, )2 tanS 2
t [ t ]2~ +R ( 1 - cos cP ) - _1_ ~ - R ( 1 - cos q, )2 tanS 2
w 22wR( l-coscP) +-tanS
4
[
t - 2R( 1 - cos q, ) ]Rsinq, w + 0 dq,
tanS
. ""[ t - 2R( 1 - cos q,) ]RSln,+, W + 0 dq,tanS
2wRsinq,dq,
(77)
(78)
and
w R [ sin <p + Jl cos <p + ( cos <p - Jl sin <p )F} (<p) ]
wR [sin <p - Jl cos <p + ( cos <p + Jl sin <p )F2( <1» ]
wR [sin <p - Jlcos <p + (cos <p + Jl sin <p )F4( <1»]
37
(79)
sin <l>J 4tan 2ecos 2<1> +(l +tan 2e) sin 2<1> - 2 Jl tan ecos <l>J 1 + tan 2ecos 2<1> ]
cos <1>J 4tan 28 cos 2<1> + ( 1 + tan 28) sin 2<p + Jl tan 8 sin <1>J 1 + tan 28cos 2<p
[sin <I>V4tan 28cos 2<1> + ( 1 +tan 28) sin 24> + 2 fJ. tan ecos <l>J 1 + tan 28cos 2<1>] (80)
cos <1>J 4tan 28 cos 2<1> + ( 1 + tan 28) sin 2<p - fJ. tan 8 sin q,V 1 + tan 28 cos 2q,
[cos 8 sin <p - Jl cos <pJcos 28 + sin 2{} cos 2q,
cos 8cos <P + Jlsin<pvcos 28 + sin 2{} cos 2q,
with boundary conditions
a b,1 (<Po) / y
ab} (<PN)/ Y
ab,2(cPm )/ Y
a b,4 (cPf )/ Y
1
ab,2(cPN)/ Y
ab,4 (<pm)/ Y
1
(81 )
2. Case II. The neutral point is after full-thread depth, i.e. <Pm ~ ~ ~ <I? In summary, the
governing differential equations for Case II in matrix form are given by
i =1,3,and4(82)
Ax,l
Ax,3
Ax,4
Adxl
Adx,3
Adx,4
[ ]
2tIt 0
~ +R( 1 - cos 4> ) - - - - R( 1 - cos q, )2 tanS 2
w22wR( 1-cosq,) +-tanB
4
w 22wR( 1-cos4» +-tanS
4
[
t - 2R ( 1 - cos 4> ) ]RsincP w + 0 d<p
tanS
2wRsin4>dq,
2wRsinq,dq,
38
(83)
(84)
and
wR [sin <p + Jl cos <p +(cos 4> - Jl sin 4> )F1( cP) ]
wR [sin <p + Jl cos cP +(cos cP - 1.1 sin cP )F3( <p) ]
wR [sin cP - Jl cos cP + (cos~ +Jl sin cP) F4( <p)]
(85)
sin 4>J4 tan 2Bcos 24> +( 1 +tan 2B)sin24> - 2 Jl tan 8 cos 4>V1 +tan 28 cos 24> ]
cos 4>J 4 tan2Bcos 24> +( 1 +tan28) sin 24> +Jl tan 8 sin 4>v 1 +tan 28 cos 24>
[
COS Bsin 4> + Jl cos 4>Jcos 2 8 + sin 2Bcos 2cP ] (86)cos Bcos cP - 1.1 sin cPJcos 28 + sin2Bcos 24>
cos Bsin <p - 1.1 cos <pJcos 2 8 + sin 2 Bcos 2 <I> ]
cos Bcos <p + Jlsin<l>Vcos 2B + sin 28cos 2<1>
with boundary conditions
a b, 1( q,0) / y
a b, 1( <Pm) / Y
ab,3 ( <PN ) 1Y
ab,4 (<P[)1Y
1
a b,3 (<Pm) 1Y
ab,4 (4)N) 1Y
1
(87)
39
Refer back to (7) and (8) for a quick comparison between the flat and fiber rolling governing
equations. One will notice that the extra geometry on the differential elements caused by the threaded
roll greatly adds to the complexity of the governing equations when compared to the flat rolling model.
With careful examination, the governing equations for fiber rolling can be simplified to agree with
the governing equations for flat rolling ifeis set to zero. However, direct substitution of 8 equal to zero
into (76) and (82) will make one at first believe otherwise. One must recall that t f and h f have different
meanings in the two models and that the relationship
dhdv =--
tan 8 ' (88)
which was used in deriving (76) and (82), is no longer valid if e is zero. After looking at the original
variable defInitions one would see that (76) and (82) are equivalent to the flat rolling case when eequals
zero.
H. NUMERICAL SOLUTIONS TO THE GOVERNING EQUATIONS
The derivation of a closed fonn solution to the frrst-order linear differential equations given in (76)
or (82) proved futile. Therefore, a numerical algorithm was used to obtain results in tenns of roll
pressure over modified flow stress (Ob,iN, i=1,2,3,4). The IMSL routine IVPRKlDIVPRK [29] was
used to solve the system of equations for Case II given in (82)~ recall that Case I will rarely occur. The
same IMSL routine was used to also acquire numerical solutions to the governing equations for flat
rolling given in (7). The calling Fortran programs, along with the necessary subroutines, for flat rolling
and fiber rolling are provided in Appendices C and D, respectively.
The routine IVPRKJDIVPRK makes use of the Runge-Kutta-Verner fifth-order and sixth-order
method with varying step size for error control to approximate the solution to a set of specified first
order differential equations with initial conditions. The routine uses a varying step size to keep the
global error proportional to a user-specified tolerance. Refer to reference [29] for a more thorough
description of the routine's details. Successive runs were made with varying error control bounds, and
comparisons were made between the calculated errors and the resulting data to ensure reasonably
accurate results were obtained.
Recalling that Case II is more likely to occur than Case I, plots for Case II using numerical solution
data from (82) and (87) will be presented in Section V as a comparison between the models for flat
rolling and fiber rolling.
40
I. ROLL FORCE AND TORQUE CALCULATIONS
After the roll pressure as a function of <P is known, it is an easy calculation to obtain both the roll
force and torque. In theory, the roll force, F, and torque, T, for each roll are calculated from the roll
pressure by the following integral relationships.
1. Case I. The neutral point is before full-thread depth, i.e. <Po 2 <PN 2 <Pm.
$0 $N $m
F = W R f 0b,I(q,) dq, + f 0b,2(q,) dq, + f 0b,4(q,) dq,$N $m ~
2. Case II. The neutral point is after full-thread depth, i.e. <Pm 2 <PN 2 <Pf
~o $m $N
F = W R f 0b,I(q,) dq, + f 0b,3(q,) dq, + f 0b,4(q,) dq,$m $N $1
$0 ~m $N
T = 11 W R 2 f 0b,I(q,) dq, + f 0b,3(q,) dq, - f 0b,4(q,) dq,
$m $N $1
(89)
(90)
(91)
(92)
Note that the negative signs in (90) and (92) account for the opposite direction of the friction forces on
either side of the neutral point.
As can be seen, these relationships require a closed-form solution for roll pressure, which was not
obtained, to evaluate the integrals. Therefore, an integral approximation was used instead. A simple
trapezoidal-rule routine estimated the area under the ablY plots using the data points obtained from the
computer programs. These areas were then substituted into (87), (90), (91), and (92) to obtain the roll
force and torque per roll.
J. CONSTRAINTS
As in the flat rolling model, there will be upper and lower limits on material thickness that can be
processes under a given set of rolling conditions. The draft, d t, for fiber rolling is defined similarly to
the draft for flat rolling, i.e.
(93)
but this is a little misleading because fiber rolling involves more than just a reduction in thickness. The
41
final thickness has a different meaning in fiber rolling than it does in flat rolling, so direct comparison
between the models based on similar drafts cannot be made. In fiber rolling the rolls are in contact with
each other, which is impossible in flat rolling if one is actually going to process any material through the
rolls.
A better comparison between the flat and fiber rolling models involves the percent reduction in
material cross-section as it passes through the rolls in each model. The percent reduction for fiber rolling
is defmed as
Percent Reduction · 100% . (94)
Compare this to (13) for flat rolling. Dividing tf by two in (94) takes into account the triangular shape
of the fiber cross-section. Thus, the results from the flat and fiber rolling models can be compared if the
same percent reduction is used in both.
For this reason, the idea to determine the thread profile on the threaded roll by specifying the roll
diameter, thread pitch, and thread height stated in Section IV-A becomes apparent. Specifying t f instead
of 8 makes it much easier to set up situations where the percent reductions are equal for the flat and fiber
rolling models. In physical terms, a custom-made cutting tool would be needed to cut the threads on the
threaded roll if one were trying to experimentally verify the results from this comparison. However, if
a comparison between the models is not desired, one could just as easily specify 8 and then determine
4through (15) (as would be the case with standard-cut threads) to get results from the fiber rolling model
independent of the flat rolling model.
For the upper limit on material thickness in fiber rolling, it ,¥ill be assumed that the angle of
acceptance can be defined as in flat rolling. The assumption contends that the coefficient of friction
between the thread crests and the material being processes is approximately equal to the coefficient of
friction between the smooth roll and the material. If this is the situation, (31) can be substituted into
(11), which reduces to the following inequality.
to ~ 2 R [ 1 - cos ( tan -1 Jl ) ] (95)
Notice that the maximum initial thickness that can be pulled into the rolls is a function of roll
radius and coefficient of friction. Materials thicker than this will slip against the roll faces and not enter
the rolls without assistance. For fiber rolling of many relatively small plastic flakes at once, front and
back tension on the material would be impossible, so this constraint cannot be relaxed as it can in flat
rolling of large sheets. Also notice that (95) does not include the final thickness, unlike (12). This is
due to the geometry difference between flat and fiber rolling. Recall that the rolls are in contact in fiber
42
rolling, whereas there is a roll gap in flat rolling. Therefore, the start point of the arc of contact, or the
maximum start point <t>O,mIDP is not influenced by the final thickness in fiber rolling.
In practice there will be minimum reduction in thickness as in flat rolling. However, with the roll
modulus of elasticity being much greater than that of the flakes in fiber rolling, the full-thread depth
assumption should provide a greater lower limit. Therefore, the lower limit on the initial material
thickness will be determined by. the full-thread depth assumption, which contends that the initial
thickness is greater than twice the final thickness. Therefore, combining this constraint with (95) the
initial thickness is bounded by the following relationship.
2 tf -< to ~ 2 R [ 1 - cos ( tan - 1 1-1 ) ] (96)
Test cases for fiber rolling presented in Section V will be required to satisfy (96). Then using the same
percent reductions between flat and fiber rolling, the test cases for flat rolling will automatically satisfy
(12).
K. THETAGUCHIMETHOD
A subset of the field termed design for experiments is the Taguchi Method™,· a statistically based
method of parametric design for robustness [30]. The Taguchi Method is useful for experimentally
determining the relative effects of system parameters on a measurable result, or criterion function in
optimization tenninology. By making use of orthogonal arrays [30, 31], one can drastically reduce the
number of experiments required to determine which system parameters most significantly affect the
result, without testing all possible combinations. An orthogonal array is a matrix that shows what
experiments need to be run under specified conditions to determine the relative strength ofinfluence that
each of the variables has on the result. For a more thorough explanation of the Taguchi Method in
parameter interaction and process optimization please see references [30-32].
For preliminary study of the fiber-rolling model, it will be useful to determine the relative affect
of system parameters on an important system characteristic, say roll torque or force. Orthogonal arrays
,,,ill be used in this study to see how the system parameters that appear in the governing equations of
(76) and (82) affect roll torque. These system parameters are given in Table 1. The terms control
factors and noise factors will be defined shortly.
The orthogonal array technique that will be used in Section V assumes that the effects of the
variable factors on the roll torque are independent of each other, or at least their interaction is relatively
small (in Taguchi Method terminology) [30]. The parameters presented in Table I are fundamental (or
· Taguchi Method is a trademark of the American Suppliers Institute, Dearborn, Michigan.
43
Table I. System parameters.
Control Factors Noise Factors
Roll radius, R Flow stress, Y
Number of threads per inch, N Coefficient of friction, J.l
Final thickness, tf Initial thickness, to
design) variables in the fiber-rolling process. Interactions are more important in purely experimental
results where processing conditions, material properties, environmental factors, etc. affect the result.
However, this analysis involves only theoretical results from the derived governing equations. Therefore,
this interaction assumption appears to be valid.
Table II is an L18(21 x 37
) orthogonal array table, which will be used in Section V.
Table II. L,~(21 x 37) array f301.
Variable Setting
F.-xn Nn rl r? r1 r4 r~ rll r7 rRI 1 1 1 1 1 1 1 1
2 1 1 2 2 2 2 2 2
3 1 1 3 3 3 3 3 3
4 1 2 I 1 2 2 3 3
5 I 2 2 2 3 3 1 1
6 1 2 3 3 1 1 2 2
7 1 3 1 2 1 3 2 3
8 1 3 2 3 2 1 3 1
9 1 3 3 1 3 2 1 2
10 2 1 1 3 3 2 2 1
11 2 1 2 1 1 3 3 2
12 2 1 3 2 2 1 1 3
13 2 2 1 2 3 1 3 2
14 2 2 2 3 1 2 1 3
15 2 2 3 1 2 3 2 1
16 2 3 1 3 2 3 1 2
17 2 3 2 1 3 1 2 3
lR ? 3 1 ? 1 ? 1 1
44
The table's title tenninology ofL1/21 X 3) corresponds to 18 experiments being required to study
the relative effect of one variable with two settings and seven variables with three settings on a specified
result. Using the assumption of small interaction between variables, the six variables in Table I can be
assigned randomly to any of the eight C-columns in Table II. However, three settings provide better
results than two settings, so column C 1 will be left free. The six variables will then be assigned to
columns C2 through C7, leaving column C8 free. These free columns will not affect the results obtained
from columns C2 through C7.
Assigning variables in Table I to columns C2 through C7, the modified orthogonal array table for
fiber rolling is presented in Table III.
Table III. Fiber rolling orthogonal array.
Variable Setting
Exp. No. U N t, tf R Y
1 1 1 1 1 1 1
2 1 2 2 2 2 2
3 1 3 3 3 3 3
4 2 1 1 2 2 3
5 2 2 2 3 3 1
6 2 3 3 1 1 2
7 3 1 2 1 3 2
8 3 2 3 2 1 3
9 3 3 1 3 2 1
10 1 1 3 3 2 2
11 1 2 1 1 3 3
12 1 3 2 2 1 1
13 2 1 2 3 1 3
14 2 2 3 1 2 1
15 2 3 1 2 3 2
16 3 1 3 2 3 1
17 3 2 1 3 1 2
18 3 3 2 1 2 3
(97)
45
The six variables \viII each have three specified settings. Usually, two of the settings are assigned
near the extremes ofthat variable's feasible range, and the third setting is assigned near the midpoint of
the feasible range. A table similar to Table III will be used in Section V for flat rolling. The only
difference is that the N-variable column will be left free and the to and trcolumns will be used for ho and
h f, respectively.
Numerical solutions for the roll torque will be calculated using the variable settings in each
experimental nm. The relative effect for each variable factor on the roll torque result will be calculated
by
Effect of Factor A = (Average Result for A at Setting n )max
- (Average Result for A at Setting m )min '
where n and m span the range of variable settings and n*m. From this it can be concluded which
variables have the greatest influence on roll torque. An informal graphic analysis of these variable
effects can also help to detennine an approximation of the best result, lowest roll torque in this case, by
choosing the best compromise of variable settings within the range of variable values studied.
Just as a point of interest, the Taguchi Method can also be a powerful tool when it comes to
optimizing a piece ofmanufacturing equipment under a set of uncontrollable conditions [30]. The goal
is to make the process, and thereby the product, as insensitive to variation as possible -- making it
robust. In the case of fiber rolling, the parameters of initial thickness, coefficient of friction, and flow
stress \\'ould be hard to control because of the variety in materials that would be received through
recycling operations, while the parameters of final fiber thickness, roll diameter, and thread pitch (or its
inverse -- number of threads per inch) would be fixed after the rolling mill is built. With an estimate of
the variation in the hard-to-control parameters, called noise factors, these fixed parameters, called
control factors, could be optimized through the Taguchi Methods for a redesigned mill in order to
minimize the effects of the noise factors. The study could also be expanded to experimentally include
more parameters than are accounted for in the governing equations for fiber rolling, such as processing
temperature, roll lubrication, processing speed, etc. However, this effort would not be \\1arranted until
the results presented in Section V are experimentally verified. This is beyond the scope and time-frame
of this research effort; although, it would someday be beneficial to perform this optimization during the
development of a commercially-oriented prototype.
46
v. RESULTS
A process for producing plastic fibers has been conceived and partially developed, which involves
a granulator, storage hoppers, heated hopper, rolling mill, and conveyor system between each component
in the process. This larger automation model could be more fully examined when the process nears
commercial implementation.
A laboratory-scale rolling mill has been built which can process the recycled plastic into fibers.
There are several geometric parameters of the process that can be modified, as well as the processing
speed and temperature. New processing steps may be added or existing steps modified as the process
is further developed.
Currently, work is still being done on the prototype rolling mill. These efforts involve the
construction of more precisely machined rolls to ensure more accurate comparisons between the
theoretical and experimental work to follow. Machining drawings were prepared for the prototype
rolling mill, and additional drawings have been prepared to better explain how the prototype is
assembled and what the final product looks like.
As of present, minimal testing has been perfonned with the prototype. Current testing includes
analyzing how a standard flake is elongated and cut during the rolling process. Comparing the original
and final dimensions will help to confinn or conflict the standard assumption that the specimen's width
remains virtually constant during the operation. Later testing should compare the theoretical and actual
triangular geometries of the finished fibers. It should also deal with measuring the required motor
torque to operate the rotating rolls and comparing this to the predicted values. Results could also involve
the comparison of the expected and actual roll separation forces. Eventually, testing should examine the
mechanical properties of the recycled fibers and compare these to the properties of virgin-plastic fibers.
Researching of various rolling models has been accomplished, which has helped to provided for
a preliminary mathematical model of the fiber rolling process. Modeling the geometry and analyzing the
differential model of a rectangular-cross section plastic flake as it is processed into thinner triangular
fibers has helped to provide a better intuitive feel for the process. It has also provided a better
understanding of what theoretical fiber elements look like and derive an estimate of how much roll force
and torque will be required to produce fibers under certain conditions.
A. MATERIAL PROPERTIES
Some material property values are needed to obtain numerical results for the flat and fiber rolling
models presented in Sections II and IV, respectively. These properties include estimates of the
47
coefficient of friction between the material and the rolls and the flow stress of plastics that will be
commonly processed through the prototype mill. Tensile tests were performed on a fe\\T samples of
HDPE and PET. The ultimate tensile strengths were determined and will be used to estimate the
modified flow stress given in (4) and (61) for the two models. Results from this testing are provided in
Table IV. Looking up values for the coefficient of friction showed that reasonable static friction values
for plastic on steel commonly ran~e from 0.2 to 0.5 [33].
Table IV. Tensile test results.
Ultimate Tensile Strength (psi)
Test HDPE PET
1 2600 8446
2 2454 7682
3 2036 6859
4 2307 5987
5 9529
6 7118
Average UTS 2350 psi ± 300 7600 psi ± 1900
B. MODEL COMPARISON USING ROLL-FACE PRESSURE DISTRIBUTION DIAGRAMS
It was decided that the first test in verifying the validity of the fiber rolling model in Section IV
would be to benchmark it against the flat rolling model in Section II. Figure 15 shows a comparison of
the roll-face pressure distributions for both models for a test case of arbitrarily chosen input parameters
satisfying the comparison requirement of identical percent reductions in material. These curves are
called hills offriction and are caused by the non-homogenous deformation introduced by the presence
of frictional forces between the material and the rolls [21].
The Clln'es sho,"' the same basic shape, while fiber rolling has a higher pressure distribution than
flat rolling. Both models start and end with a 0b/Y ratio of one due to the boundary conditions on the
governing equations at <Po and <Pr. The pressure distribution for fiber rolling closely follo\vs that for flat
rolling near the entry and exit zones before diverging and predicting a pressure at and around the
neutral-point pressure spike greater than that for flat rolling. This is to be expected when one considers
that squeezing the material into rectangular cross-sections would require less effort than triangular
48
,h o = 0.06 inh f = 0.0045 in
I to = 0.06 in
! t f = 0.009 in!-
I~ I R = 1.5 in II ~ = 0.2
I \% Reduction = 92.5% ~
I \ V' Fiber rollingv
/
fA ~~ vV' Flat rolling
//
if ~r\~
"-~~~
200
oo
<Pf
0.025 0.05 0.075 0.1
Angle along Arc of Contact (rad)
0.125 0.15
<Po
Figure 15. Roll-face pressure distribution comparison for flat and fiber rolling.
cross-sections, where the material near the fiber edges is being reduced to essentially zero thickness. The
fiber rolling model also reaches the neutral point slightly sooner than the flat rolling model, possibly due
to the differences in surface area between the material and the rolls in the two models.
Figures 16 and 17 show how the pressure distributions for the flat and fiber rolling models,
respectively, change with variations in the coefficient of friction, starting with the cases ShO\\l1 in Figure
15. Again, both models show the same trends with the fiber rolling model predicting higher pressures
and neutral point locations slightly closer to the entry zone, than the flat rolling model. For these test
cases it can be concluded that the fiber rolling model produces pressure distributions closely resembling
those for flat rolling.
C. MODEL COMPARISON USING THE TAGUCHI METHOD
Another analysis tool was employed to compare the flat and fiber rolling models. The Taguchi
Method comparison presented here is less straight forward than simply looking as pressure curves, but
it is used to provide a different perspective on comparing the models as \\Tell as saye numerical
computation time by keeping the number of test cases to a minimum.
49
0.15
<Po
0.1250.05 0.075 0.1
Angle along Arc of Contact (rad)
0.025
I 1h o ::::: 0.06 inh f ::::: 0.0045 in I--,R == 1.5 in
111\% Reduction == 92.5%
~
f l\ IJ
,~!\\ /0.20)/ 0.21
,~\\ ~/ 0.22
~~\\\/ 0.23
II 0.24
~i\V 0.25
IV~~
""~~~oo
cPr
50
200
250
!!150>-......~:J(/J
~ 100Q:150::
Figure 16. Roll-face pressure distribution for flat rolling with change in Jl.
0.15
<Po
0.1250.05 0.075 0.1
Angle along Arc of Contact (rad)
0.025
oL~~--l~~~~J...L...L__L__l._Jo
CPf
I-l
>- 0.20; 400 -+---+----+----,H-A---J4--+----t----+-I 0.215 )/ 0.22
~ // I 0.23a. 0.2415 / 0.250:: 200 +----t-__+_
N == 52 threads/in+--~~--+--4-Ir----4---+--+----+-----t--1 to = 0.06 in
t f = 0.009 inR == 1.5 in
_600~~~-~~~~~~-~-~~~%R~~tioo=~.5%
!!
Figure 17 Roll-face pressure distribution for fiber rolling with change in Jl.
50
From the start, let one be warned of the limitations of this method. While the method is extremely
useful in optimizing a system for a given set of parameter variations and determining the effect of each
variable on the system, correctly interpreting the results can be tricky if one is unfamiliar with the
method. With this in mind, the goal here was be to simply compare the flat and fiber rolling models for
a limited set of input parameters. After showing that the fiber rolling model behaved similarly to the flat
rolling model for each variable in their respective governing equations, one could be confident in trying
to optimize the fiber rolling model by itself for a wider set of parameter variations.
Table V provides values for three feasible settings of each variable in the flat and fiber rolling
models. For the given range of settings for roll radius, coefficient of friction, and fmal thickness, (96)
requires that the initial thickness for fiber rolling be between 0.028 and 0.0388 inches, which is satisfied
by all three settings of initial thickness in Table V.
Table V. Flat and fiber rolling variable settings.
Variable
Setting 11 N to tf R Y ho hf
(threads/in) (in) (in) (in) (psi) (in) (in)
1 0.2 40 0.03 0.012 1 2350 0.03 0.006
2 0.225 52 0.0325 0.013 1.5 5000 0.0325 0.0065
3 0.25 60 0.035 0.014 2 7600 0.035 0.007
Plugging the settings from Table V into the orthogonal array in Table III and running the 18
experiments on the governing equations for flat rolling provides the results seen in Table VI. It should
be noted that all force and torque calculations given here and to follow are for a strip of material with
a ooit centimeter width, to simulate an ordinary granulated flake, that completely fills the arc of contact.
The average roll torque for each variable setting in Table V is given in Table VII. These values
are obtained by summing the torque result in Table VI for each variable setting each time that variable
setting is used in an experiment and dividing by the number of times that setting was used in the 18
experiments. For example, the average roll torque for R at setting 2 is the sum of torques in experiments
2,4,9,10,14, and 18 divided by six.
The difference between the largest and smallest average value at the three settings for each
variable as defined in (97) is given in Table VIII. Although these values were derived from average roll
torques and have the units of torque, it is only their relative magnitude that is of importance in the
Table VI. Flat rolling array results.
Variable Setting
Exp. 0/0 <Po <PN 0b(<PN)/Y Roll Force Roll TorqueNo. Reduction (rad) (rad) (lb) (ft-Ib)
1 80 0.155 0.040 12.11 479 2.26
2 80 0.132 0.036 21.24 1934 11.25
3 80 0.118 0.033 32.01 4766 32.53
4 78.33 0.125 0.036 31.70 3841 21.26
5 78.46 0.113 0.033 49.65 2003 13.08
6 82.86 0.171 0.042 19.14 1474 7.18
7 81.54 0.115 0.033 137.19 9263 57.07
8 81.43 0.169 0.044 23.67 2658 12.93
9 76.67 0.124 0.037 39.48 1413 7.83
10 80 0.137 0.037 18.57 1822 11.09
11 80 0.110 0.031 44.09 5619 35.02
12 80 0.161 0.041 10.66 456 2.27
13 78.46 0.60 0.043 13.44 1753 8.63
14 82.86 0.139 0.037 43.38 1560 8.88
15 78.33 0.108 0.032 58.24 4632 28.92
16 81.43 0.119 0.034 110.54 3802 24.49
17 76.67 0.152 0.044 18.37 1397 6.54
18 81.54 0.133 0.037 66.58 6864 37.48
Table VII. Average roll torque for flat rolling.
Average Roll Torque for Each Variable Setting (psi)
Setting ~ ho hf R Y
1 18.88 16.97 24.65 6.63 9.80
2 17.59 21.63 16.85 16.30 20.34
3 29.27 16.18 13.28 31.85 24.64
51
52
Taguchi Method. This data indicates that roll radius most significantly influences flat-rolling roll torque,
followed by flow stress, coefficient of friction, and final thickness. Roll torque is affected least
significantly by initial thickness. Shortly, these results will be used to compare the flat and fiber rolling
models.
Table VIII. . Variable effects on roll torque in flat rolling.
Effect of J.l 11.68
Effect ofho 5.44
Effect ofh[ 11.37
Effect ofR 24.37
Effect ofY 14.84
By looking for the lowest values for each variable, the data in Table VII can also be interpreted
to say that the settings given in Table IX will provide the best compromise of variable settings to achieve
the lowest roll torque for the range of settings in Table V. The disclaimer should again be stressed that
this optimum combination for lowest roll torque is only valid within the prescribed range of variable
settings in Table V. The method can only provide as much robustness as can be obtained given the
variable variations that are studied. However, since the goal is simply to compare the models, the limited
range of settings in Table V should not detract from the conclusions that can be drawn from the Taguchi
Method.
Table IX. Optimum combination for flat rolling.
Variable Setting Value
~ 1 0.2
ho 3 0.03 in
hf 3 0.007 in
R 1 1 in
Y 1 2350 psi
53
The same scheme that was just used for flat rolling was used to derive the variable affects on roll
torque in the fiber rolling model. Plugging the settings from Table V into the orthogonal array in Table
III and running the 18 experiments on the governing equations for fiber rolling provided the results in
Table X. The average roll torque for each variable setting in Table V is given in Table XI. The
difference between the largest and smallest average value at the three settings for each variable as defmed
in (97) is given in Table XII.
Table X. Fiber rolling array results.
Variable Setting
Exp. 0/0 e <Po <Pm abe<Pm) <PN abe<PN) Roll RollNo. Reduct (rad) (rad) (rad) IY (rad) IY Force Torque
Ion (lb) (ft-Ib)
1 80 0.765 0.173 0.078 4.74 0.043 24.38 837 3.43
2 80 0.934 0.147 0.066 11.65 0.039 84.08 5865 26.00
3 80 1.034 0.132 0.060 25.49- 0.036 256.68 26402 125.15
4 78.33 0.805 0.142 0.052 31.78 0.039 91.71 9400 44.51
5 78.46 0.969 0.128 0.047 90.80 0.036 337.01 10092 49.91
6 82.86 0.964 0.187 0.105 4.03 0.047 81.87 4644 16.48
7 81.54 0.765 0.128 0.065 16.63 0.035 470.57 27228 146.48
8 81.43 0.934 0.187 0.095 7.00 0.049 96.21 8218 30.47
9 76.67 1.034 0.142 0.037 693.97 0.041 317.73 9725 70.51
10 80 0.842 0.153 0.068 8.11 0.040 50.69 4096 20.46
11 80 0.895 0.123 0.055 20.84 . 0.034 190.86 19169 94.83
12 80 1.001 0.181 0.081 7.18 0.047 43.20 1338 4.70
13 78.46 0.842 0.181 0.067 11.94 0.048 34.00 3689 15.06
14 82.86 0.895 0.153 0.086 5.89 0.039 179.84 5050 22.50
15 78.33 1.001 0.123 0.045 147.89 0.035 520.11 29767 137.19
16 81.43 0.805 0.132 0.067 16.45 0.036 420.26 12081 66.26
17 76.67 0.969 0.173 0.045 117.56 0.049 79.19 4826 19.48
18 81.54 0.964 0.147 0.075 16.42 0.040 464.67 35387 145.87
54
Table XI. Average roll torque for fiber rolling.
Average Roll Torque for Each Variable Setting (psi)
Setting ~ N to t f R Y
1 45.76 49.37 61.66 71.60 14.94 36.22
2 47.61 40.53 64.67 51.52 54.98 61.01
3 79.85 83.32 46.88 50.09 103.30 75.98
Table Xlf Variable effects on roll torque in fiber rolling.
Effect of ~ 34.08
Effect ofN 42.79
Effect of to 17.78
Effect of tf 21.50
Effect ofR 88.37
Effect ofY 39.76
Again these values were derived from average roll torques and have the units of torque (ft-Ib), but
it is only their relative magnitude that is of importance. This data says that roll radius most significantly
influences fiber-rolling roll torque, followed by the number of threads per inch, flow stress, coefficient
of friction, and final thickness. Roll torque is affected least significantly by initial thickness. These
results agree with those seen in flat rolling, with the addition of threads per inch which is not a factor in
flat rolling.
Figure 18 summarizes the results from Tables VIII and XII. As can be seen, the affect of system
parameters on roll torque in fiber rolling is on a larger scale than those for flat rolling, as would be
expected from the difference in higher roll pressures seen in Figures 15, 16, and 17, but follow the same
trends in both models.
Looking for the lowest values for each variable, the values in Table XI can be interpreted to say
that the settings given in Table XIII ",ill provide the best compromise of variable settings to achieve the
lowest roll torque in fiber rolling for the range of settings in Table V. The same disclaimer that was
given for flat rolling applies here too. The method can only provide as much robustness as can be
obtained given the variable variations that are studied, so this optimization conclusion is only valid if
the system is operating \vithin the range of settings in Table V.
55
100
:c~
80(1):::JCT0t-(5 60c::(1)0>mQ) 40~.f:(1)0> 20cco
.J::.0
01..1. ho ' t 0 hf , t f R y N
II Flat Rolling III Fiber Rolling
Figure 18. Effect of system variables on roll torque for flat and fiber rolling.
Table XIII. Optimum combination for fiber rolling.
Variable Setting Value
Jl 1 0.2
N 2 52 threads/in
ho 3 0.03 in
h f 3 0.007 in
R 1 1 in
y 1 2350 psi
To reinforce the conclusion that the two models agree on variable influence, the data from Tables
VII and XI are plotted in Figures 19 through 21 to show the similarities in influence for each individual
variable setting. Ignoring the difference in scale, the models follow similar trends from one variable
setting to the next for each variable.
------+--+---+-.--+---+----+---+---c-~--+__-_l
0.25
0.21
0.22
0.23
0.24
Co
eff
icie
nt
of
Fri
ctio
n
~"",
~~
~~
o 0.2
120
~10
0~ Q
)
80:J 0
-'- 0 t-
600 0:
: c:40
0 "0 Q)
20~ w
0.25
0.21
0.22
0.23
0.24
Co
eff
icie
nto
fF
rict
ion
o+
-1--+--+--+---~--+-__+_--+----+--+-___i
0.2
35-r-
fi'30
t ~25
0- 02
0t- ~
15c: o
10"0 Q
) iTI5
(a)
(b)
o 0.03
0.03
10.
032
0.03
30.
034
0.03
5In
itial
Th
ickn
ess
(in)
o 0.03
0.03
10.
032
0.03
30.
034
0.03
5In
itial
Th
ickn
ess
(in)
35
ii3
0t Q
)25
:J 0- 02
0t- o
150:
: c: o10
"0 Q) iTI
5
.-.....
..----
---.....
..........
.-~
,.......
--....
........
........
.....""""
'"r--
--.
f--
120
~10
0~ Q
)
80:J 0
-'- 0 t-
60(5 0:
: c:40
0 0 Q)
20~ w
--""'1
00...
........
.. ..........
.....~
--..
--
(c)
(d)
Figu
re19
.M
odel
com
pari
son
usin
gth
eT
aguc
hiM
etho
d.(a
)E
ffec
tof
Jlon
flat
rolli
ng.
(b)
Eff
ect
of
J.lon
fibe
rro
lling
.(c
)E
ffec
toft
oon
flat
rolli
ng.
(d)
Eff
ecto
fto
onfi
ber
rolli
ng.
Vl
0\
---+
---+
-_
..---
+--
--+
--l
~_.
-t--------
....-....
..- ........
.......-
--...... ~
--_
..
35,-----,-----··,..-------T---·-~r_~-..__
~30
=1=_
=t=J
=__~=t=t=t=t::J
Q)
25:J 0
- (;2
0+-
----
---+
--+-
----
-'~-
---T
--+
---+
--+
----+
--t-----i
r- ~15
c o10
-t--
----
-t--
+--
-+--
---+
----
-+-I--+---+--4--+--~
"0 Q) iTI
5t------1
f----t--
a.+
---+
--+
-+I
0.00
60
.00
62
0.0
06
40
.00
66
0.00
680.
007
Fin
alT
hick
ness
(in)
120
~10
0~ Q
)
80::
J 0-~ 0 .....
60
'0 0:: c
400 +J 0 ~
20w
o 0.01
20
.01
25
0.0
13
0.01
35F
inal
Thi
ckne
ss(in
)0.
014
(a)
(b)
--------4
--+
-----i-
-I
~
35
r---
-r
21.
41.
61.
8R
oll
Rad
ius
(in)
1.2
~
.JI'V
"/
V./V
./""
/~
./~
V~V
1---
I~
~
o1
120
~10
0~ ~
800
-~ 0 !::
60
'0 0:: c
400 "0 ~
20
w
21.
41.
61.
8R
oll
Rad
ius
(in)
1.2
o1
~30
~----+---+------+---+-~-+-----!
Q)
25+
---+
--+
-
:J 0- C52
0r- ~
15+
----t--
C o10
+----i-
--::
:ll.
..-t-
---
"0 Q) [1
5
(c)
(d)
Fig
ure
20.
Mod
elco
mpa
riso
nus
ing
the
Tag
uchi
Met
hod.
(a)
Eff
ecto
ftfo
nfl
atro
lling
.(b
)E
ffec
to
ftfo
nfi
ber
rolli
ng.
(c)
Eff
ecto
fRon
flat
rolli
ng.
(d)
Eff
ecto
fRon
fibe
rro
lling
.V
l.....
.:J
---+
----
+--
----
-+--
--+
_I
----+
----+
.--+
--t-
--j
35I-
-+--
---+
----
--+
_--
--+
---+
----
-+--
+--
--+
----
+--
-
;Q30
+--
-+--
----
f---
-f--
----
----
+--
+--
+--
+-+
__
t CD25
:J 0" 02
0r- ~
15c: o
10"0 CD in
5
120
~10
0~ ~
800
" ..... 0 r-60
(5 0:: c
400 .....
,0 ~
20w
r--
-
----~.
..........~
.".,
....
..
-~
"",.
-
-"~~
",,--
o+
--+
----
+--
--+
---+
----
--+
---+
-1~--~_+____+-+---__t
2000
3000
4000
5000
6000
7000
8000
Mod
ified
Flo
wS
tres
s(p
si)
(a)
o 2000
3000
4000
5000
6000
7000
8000
Mod
ified
Flo
wS
tres
s(p
si)
(b)
~-
\
~
V/"
/~
/'
f---
'---
120
~10
0~ ~
800
" ..... 0 ~60
<5 0:: c
400 1:)
~20
w
o 4045
5055
Thr
eads
pe
rIn
ch
(e)
60
Figu
re21
.M
odel
com
pari
son
usin
gth
eT
aguc
hiM
etho
d.(a
)E
ffec
tofY
onfl
atro
lling
.(b
)E
ffec
to
fYon
fibe
rro
lling
.(c
)E
ffec
tofN
onfi
ber
rolli
ng.
VI
00
59
D. FIBER ROLLING PARAMETRIC STUDY
Now that the fiber rolling model has been shown to agree with the flat rolling model by two
different methods~ a parametric study can be perfonned on the fiber rolling model with a greater certainty
of accuracy. Pressure distributions will be studied, with a wider range of variations in all of the
parameters used this time. The analysis will conclude with a look at how all of the system parameters
affect roll torque.
1. Roll-Face Pressure Distribution Diagrams. The variables that appear in the governing
equations for fiber rolling are now varied one at a time while holding all other variables constant to
detennine how each variable affects the system. Figure 22 shows the roll-face pressure distribution for
variations in the coefficient of friction with values ranging from 0.1 to 0.275.
t 0 = 0.03 in -t f =0.013 in
~R = 1.5 in >---
N = 52 threads/in
I 0A> Reduction = 78.3°A>I--
j..L
j \ 1/ 0.275/ 0.250
JA \ / 0.225
I~ V~- 0.200
0.150
II \ 'l v 0.100/
IIJr\\1\/1/V/h~\K"
i~~,...~~ I
600
500
~400!>-......!300::Jfh
~a..= 200ocr:
100
0.025 0.05 0.075 0.1
Angle along Arc of Contact (rad)
0.125 0.15
4>0
Figure 22. Roll-face pressure distribution for fiber rolling ",'ith change in fl·
The peak pressure that occurs at the neutral point increases with increasing coefficient of friction,
\vhich is intuitively correct~ conversely, in the limiting case of no friction between the material and the
roll there would also be no roll pressure. The neutral point location also shifts slightly closer to the entry
zone with increasing coefficient of friction.
Figure 23 shows the roll-face pressure distribution for variations in the thread pitch with values
ranging from 30 to 100 threads/in. The pressure curves for variations in the number of threads per inch
60
0.15
<Po
0.1250.05 0.075 0.1
Angle along Arc of Contact (red)
0.025
II I I II I
IIt 0 = 0.03 in
~
I I ,t f = 0.013 inR = 1.5 in I--
I
II IJ = 0.2
\ I I % Reduction = 78.3%I--
! N IIA\ \ I II (threads/in)
I
I I \ /v100 I
! I90 I
~/
I~ \\~~/
80
II1\\ \17060
IIJ A\ \'\50 I
///J~\\\\40
~../~ 30
I~~~v I I~~"'-.~, I
oo
<Pf
100
600
>--..~ 300:J(/)(/)
~
~ 200o
0:'::
'wc..~ 400~
500
Figure 23. Roll-face pressure distribution for fiber rolling with change in N.
behave much like variations in the coefficient of friction. The neutral point pressure increases with a
greater number of threads per inch due to smaller and smaller cross-sections being produced. The neutral
point location also shifts slightly closer to the entry zone with a greater number of threads per inch.
Figure 24 shows the roll-face pressure distribution for variations in the initial flake thickness with
values ranging from 0.03 to 0.1 inches. The neutral point pressure increases with increasing initial
thickness due to a greater percentage of material being reduced to the same final thickness. The shift in
neutral point location towards the entry zone is more dramatic with changes in initial thickness.
Increases in initial thickness seem only to affect the pressure curves before the neutral point, with
pressure curves after the neutral point all following the same path.
Figure 25 sho\vs the roll-face pressure distribution for variations in the final flake thickness with
values ranging from 0.005 to 0.013 inches. The neutral point pressure increases with decreasing final
thickness due to same amount of material being squeezed into smaller and smaller cross-sections. Here
the neutral point location shifts to\vard the exit zone as final thickness is decreased.
Figure 26 shows the roll-face pressure distribution for variations in the roll radius \vith values
ranging from 0.5 to 3 inches. The pressure distributions here resemble those for changes in final
thickness. The neutral point pressure increases \vith increasing roll radius, and the neutral point location
shifts to\vard the exit zone \vith increases in radius.
61
0.15
<Po
0.1250.05 0.075 0.1
Angle along Arc of Contact (rad)
0.025
1 II
I+- R = 1.5 inT
f--
I t f = 0.013 in\ IT I
to (in)- f-N = 52 threads/in -
j \..L = 0.2
IAI v 0.10 -- >--
I /V 0.09 I
I ~\1 ~ 0.08I
I
0.07
II \' I 0.06 I
I \' 0.05 I0.04 I
I , I~ /0.03
I \~~/
1 I
/ ,~~
i II i
I / ~ I
c:::V ~~
oo
<Pf
120
100
.iii~ 80"iii~
>-~ 60:::J(/J(/JQ)
a:0
40a:::
20
Figure 24. Roll-face pressure distribution for fiber rolling with change in to'
t----+---H -+--~---+--__+_-+__~HR = 1.5 int 0 = 0.03 in
500 +----4---+~-+--+----+-----+---I---+-IN= 52 threads/in -\..L = 0.2
t--t----t1~---r-_r_---t-___t_-_t_-__t_i_-..______.._-__.__----1-
0.15
<Po
0.1250.025 0.05 0.075 0.1
Angle along Arc of Contact (rad)
100 I Ji/h~+--~~~~~~+--l!3lIIK:,-+--.........--+_-l--i---+---+---i~-+-----l------1
OJ.-"'-~=--..J..---L-....l..--=~~ ...i-_~--J_--I. --.J
o<Pf
'en%400 +------..:f------Hf--I---+---+--~__+-_+_-_I__-+______+-~-~
S
~ 300 +--1----1---1-\-+---+-----1 t f (in)
~ J \ /v ~:~~~f-II --+---+---+---t-
I _-+,_---1~ 200 I A\~ 0.009
& I I 1\\ 0.011
\ I A~ ~/'--rvO_.0_13......--'
Figure 25. Roll-face pressure distribution for fiber rolling with change in tf.
62
Obviously, the modified flow stress has no affect on the ratio of roll pressure over modified £1o,v
stress, so no graphical data will be given for attempts at varying flow stress. The roll pressure and
torque will vary linearly with the modified flow stress.
2. System Parameter Effects on Roll Torque. Using the data that was obtained for the pressure
diagrams just presented, it was also possible to plot changes in roll torque for variations in the system
parameters for fiber rolling. Recall that (89) through (92) can be used to calculate the roll force and
torque from the area under the pressure curves. The parameters were varied one at a time while holding
all others constant to determine how each variable affects the roll torque.
0.15
Q>o
0.1250.05 0.075 0.1Angle along Arc of Contact (rad)
0.025
t 0 = 0.03 in r-
t [ = 0.013 inr-
N = 52 threads/in~ = 0.2 r-
% Reduction = 78.3 0;/0 r-I
l......f--,-R (in)
"=1- 3.02.5, 2.0
j l 1.5
I \ 1.0
III A[;/ 0.5
/
IJ' / ~~ /IIV~~ i/'/
~V __~~ Ioo
Q>[
100
700
~:J
~ 300~a.
&200
~500
!>- 400
600
Figure 26. Roll-face pressure distribution for fiber rolling with change in R.
Figure 27 shows the roll torque as a function of the coefficient of friction for values ranging from
O. 1 to 0.275. The roll torque increases with increasing coefficient of friction between the material and
the rolls along a concave-up relationship.
Figure 28 shows the roll torque as a function of the number of threads per inch for values ranging
from 30 to 100 threads/in. The roll torque increases with increasing number of threads per inch along
a concave-up relationship.
Figure 29 shows the roll torque as a function of the initial thickness for values ranging from'O.03
to O. 1 inches. The roll torque increases ,vith increasing initial thickness along a concaVe-dO\\l1
relationship.
I I_ to = 0.03 in I
t f = 0.013 in j~
"-R = 1.5 in 7"-N = 52 threads/iny = 5000 psi I
I
,I
'-0/0 Reduction = 78.3% 7 I
......- \
~{ i1 '/
V ,
~/~ i
~~
I ~~ !
.4~--
120
100
~ 80
~Q):J 60~0t-o0:: 40
20
o0.1 0.125 0.15 0.175 0.2 0.225
Coefficient of Friction0.25 0.275
63
100
80
:0l 60CP:Je-ot- 4000::
20
Figure 27. Roll torque for fiber rolling with change in Jl.
to = 0.03 in
t f = 0.013 in /-R = 1.5 in~ = 0.2 /y = 5000 psi
/r% Reduction = 78.30/0"-
/'./
/t.
~~('
~~I:
-- .......--1~\r---~I:
o30 40 50 60 70 80
Number of Threads per Inch90 100
Figure 28. Roll torque for fiber rolling with change in N.
Figure 30 shows the roll torque as a function of the final thickness for values ranging from 0.005
0.013 inches. The roll torque increases with decreasing final thickness along a concave-down
,ationship.
64
50
45
~40
~CD
5-35ot-oa: 30
25
i !~~~
I--- N = 52 threads/in I
t f = 0.013 in1 ~~I---
R = 1.5 in ~I--~ = 0.2 ~
y = 5000 psi V>--
/'.)V
/~V
//
~IC
200.02 0.04 0.06
Initial Thickness (in)0.08 0.1
Figure 29. Roll torque for fiber rolling with change in to'
0.0140.0120.008 0.01Final Thickness (in)
0.006
~~. I IN = 52 threads/in
'\I--
to = 0.03 in
\I--
I R = 1.5 in
,~~ = 0.2 f---
Y = 5000 psi
\f---
\\~,
',-~~
~~---I
..,~
I200.004
30
70
80
Q.)
5-50ot-oa: 40
;eGO
~
Figure 30. Roll torque for fiber rolling with change in t f ·
Figure 31 shows the roll torque as a function of the roll radius for values ranging from 0.5 to 3
inches. The roll torque increases with increasing roll radius along a concave-up relationship.
65
32.51.5 2Roll Radius (in)
I ) ..~to =0.03 int f =O.013in /N = 52 threads/in
W~ = 0.2 IV IY = 5000 PStI
W% Reduction = 78.3% I / III
V I I
I III
I I /T !i
I
I /" ! II
~V I
~V'"
II............... It.
I -:~ I-"'"o
0.5
50
250
200
Figure 31. Roll torque for fiber rolling with change in R.
An informal examination of the ordinate scales in Figures 27 through 31 shows that the results
here seem to agree for the most part with those obtained using the Taguchi Method, even for the wider
range of parameter settings used in Figures 27 and 31 compared to those used in the Taguchi Method.
Changes in roll radius had the most influence on changes in roll torque, followed by the coefficient of
friction, number of threads per inch, and final thickness. Changes in initial thickness resulted in the least
variation in roll torque or roll-face pressure.
66
VI. CONCLUSIONS
A process for producing plastic fibers has been conceived and partially developed, which involves
a granulator, storage hoppers, heated hopper, rolling mill, and conveyor system between each component
in the process. A laboratory-scale rolling mill has been designed and built which can process recycled
plastic into fibers.
Researching various rolling models has been accomplished, helping to provide a preliminary
mathematical model of the fiber rolling process. Modeling the geometry and analyzing the differential
model of a rectangular cross-section plastic flake as it is processed into thinner triangular fibers has
helped to provide a better intuitive feel for the process. It has also provided a better understanding of
what theoretical fiber elements look like and an estimate of how much roll force and torque will be
required to produce fibers under certain conditions.
The new fiber rolling model was compared to the standard flat rolling model by looking at sample
roll-face pressure distributions. The pressure distribution showed the same basic curve profile with fiber
rolling having a higher pressure distribution than flat rolling for a test case of arbitrarily chosen input
parameters satisfying the comparison requirement of identical percent reductions in material. Pressure
distributions for both models start and end with a a b/Y ratio of one satisfying the boundary conditions
on the governing equations at <Po and <Pr. The pressure distribution for fiber rolling closely follows that
for flat rolling near the entry and exit zones before diverging and predicting a pressure at and around
the neutral-point pressure spike greater than that for flat rolling. The fiber rolling model also reaches
the neutral point slightly sooner than the flat rolling model. For the test cases it can be concluded that
the fiber rolling model produces pressure distributions closely resembling those for flat rolling.
The fiber rolling model was also compared to the standard flat rolling model by looking at
parameter influences using the Taguchi Method over a narrow range of variable settings. This approach
\vas a little less straight forward than the pressure distribution approach, but it was useful in providing
a different perspective on comparing the models. The resulting data shows that roll radius most
significantly influences fiber-rolling roll torque, followed by the number of threads per inch, flow stress,
coefficient of friction, and final thickness. Roll torque is affected least significantly by initial thickness.
These results agree with those seen in flat rolling.
After showing that the fiber rolling model agreed with the flat rolling model using the pressure
distribution and Taguchi Method approaches, a parametric study was performed on the fiber rolling
model. The variables that appear in the governing equations for fiber rolling were varied one at a time
\vhile holding all other variables constant to determine ho\\! each variable affects roll-face pressure
distributions and roll torque.
67
For the parametric study on pressure distributions, the peak pressure that occurs at the neutral
point increases and the neutral point location shifts slightly closer to the entry zone with increasing
coefficient of friction, a greater number of threads per inch, and increasing initial thickness'. Increases
in initial thickness seem only to affect the pressure curves before the neutral point, with pressure curves
after the neutral point all follo\ving the same path. The neutral point pressure increases and the neutral
point location shifts toward the exit zone with decreasing [mal thickness and increasing roll radius. The
flow stress has no affect on the ratio of roll pressure over flow stress and would have a linear effect on
roll pressure if it were removed from the roll pressure over flow stress ratio.
From the parametric study on roll torque, the roll torque increased with increasing coefficient of
friction between the material and the rolls along a concave-up relationship. The roll torque also
increased with increasing number of threads per inch along a concave-up relationship. The roll torque
increased with increasing initial thickness along a concave-down relationship. The roll torque increased
with decreasing final thickness along a concave-down relationship. The roll torque increased with
increasing roll radius along a concave-up relationship.
An infonnal examination of the parametric study shows that the results agree for the most part
with those obtained using the Taguchi Method over a wider range of parameter settings. Changes in roll
radius had the most influence on changes in roll torque, followed by the coefficient of friction, number
of threads per inch, and final thickness. Changes in initial thickness result in the least variation in roll
torque.
As a preliminary attempt at building a working prototype and modeling the mechanics of fiber
rolling, this research effort has been successful. However, it is far from complete. Likely, the model will
eventually need to be modified to incorporate elastic recovery of the plastic fibers as \vell as other
considerations. Hopefully this research effort has contributed to recycled plastic fibers becoming an
economic possibility and an effective alternative to the more expensive virgin-polypropylene fibers in
concrete reinforcement.
APPENDIXA.
PROTOTYPE ROLLING-MILL DETAIL DRAWINGS
69
The following diagrams provide the machining directions for production of all of the parts in the
prototype rolling mill. Figure 32 shows both the smooth roll and the threaded roll. Notice that the rolls
are identical except for one having a fine-thread pattern of 52 threads per inch. Figure 33 shows the
longer inserts that are press-fit into one end of each roll. The intermediate-diameter surface comes into
contact with a sealed roller bearing. The 2 inch extension is presently unused but was designed to
eventually be used by a set of gears between the rolls and another gear on one of the shafts connected
to a drive motor. The fIrst set of added gears would guarantee the rolls turning at the same speed without
any slipping during the rolling operation. The added drive gear and motor would enable a larger number
of fibers being produced at a time ~hen future testing requires it. Figure 34 shows the shorter insert that
is press-fit into the remaining end of the threaded roll. It serves only to provide a bearing support.
Figure 35 shows a long insert that is press-fit into the remaining end of the smooth roll. The hole in the
end opposite the bearing support is to provided for hand crank operation. Figure 36 shows the handle
that is currently used to hand crank the prototype. Figure 37 shows the pillow boxes that hold the
bearings for the smooth roll. Figure 38 shows the adjustable pillow boxes that hold the bearings for the
threaded roll. The slots provide a means to adjust the clearance and pre-load force between the rolls.
Figure 39 shows the base plate for the prototype. Figures 40, 41, and 42 show the six angle
brackets that connect the base plate to the upright structural supports. Figure 43 shows the square tubing
that serves as the main structural members of the prototype. Figure 44 shows the end plates that connect
the pillow boxes to the upright frame members. Figure 45 shows the side plates that provide rigidity to
the frame. The end and sides plates also serve as a protective shield around the rolls to keep fingers from
coming into contact with the rolls and prevent fibers from being discharged towards an operator.
Plexiglass shields could eventually be installed above the end and side plates to further safeguard the
well being of an operator while maintaining a clear line of sight to the rolling operation. Figure 46 shows
the layout for the heated hopper in its current state of development, which does not incorporate any pre
heating operations. The large size of the hopper should ease the incorporation of these additional
mechanisms at a later date. Figure 47 shows how the sheet metal is bent to form the hopper. A narrow
slit is left open at the bottom of the hopper for flakes to fall into contact with the rolls.
I12
.0I
~I
~..J
SMOO
THRO
LLER r-
L_
.1.0
00X
1.0
CO
UN
TER
SIN
K
X2
f-O...E
S
QU
AN
TIT
Y'
1
MA
TE
RIA
LI
ST
AIN
LE
SS
ST
EE
L
THRE
AD
EDRO
LLER
52
TH
RE
AD
S/IN
CH
I12
.0-._-~--~-~----
I
QU
AN
TIT
YI
1
MA
TE
RIA
LI
ST
AIN
LE
SS
ST
EE
L
----....
----
---- ----
I.~I.~u
__
_....--
Figu
re32
.S
moo
than
dth
read
edro
llde
tails
.
.1.0
00X
1.0
CO
UN
TER
SIN
K
X2
t-O..E
S
~IVERSITY
[FM
ISS
IlJR
I-
RO
LLA
PRO
CES
SIN
GR
ECYQ
.ED
PLA
STI
CS
INTO
f1B
ER
S
SKID
TH,
THRE
ADED
RtL
LER
DET
AIL
S
JTI
9-1
0-9
5IR
tLLE
R.D
VG
--....) o
QU
AN
TITY
,2
MA
TERI
AL.
STE
EL
1.62
5
¢1.0
02(P
RE
SS
FIT
ON
EIN
EAC
HR
OL
LE
R)
¢O.7
8125
¢O.7
5
1---2
.0I
0.51-
-j
~IVERSITY
IFM
ISSl
lJR
I-
RlI..
LAPR
OCE
SSIN
GRE
CYCL
ED~
.."=PL
AST
ICS
INTO
FIBE
RS
IlEC
JNM
.I~
LIH
iR
lI..L
ERIN
SERT
DE
lAD.
.S.xx
+/-
.IS
+/-"'.-01
----
-.....-
--..
..--
----
...JO
O(
+/-
-JT
9-1O-~
INSE
RT1.D
VG•
flIT
ICM
.[IM
\IIN
G
Figu
re33
.L
ong
roll
inse
rtde
tails
....
...,)
¢1.0
02(P
RE
SS
FIT
ON
EIN
THR
EAD
ERR
OL
LE
R)
¢O.7
8125
QU
AN
TITY
:1
MA
TER
IAL.
STE
EL
1.62
5
0.5~
~IVERSITY
[FM
ISS[
IJR
I-
RIL
LAPR
OCE
SSIN
GRE
CYCL
EDU
N..£
SSo
n£
JM
S[
SPEC
IFl£
Do
PLA
STIC
SIN
TOFI
BER
SD
lMD
CSID
NS
MIL
..IO
tES
IIE
QW
.S~
SK
lRT
RIJ
..LER
INSE
RT
DET
AIL
S::x
:~:::i
+/-
loS
lEW
D1
-----..
..-----..
.-------tI
JIDN
JTau
:~
JT9
-10
-95
INSE
RT2
.DVG
Figu
re34
.S
hort
roll
inse
rtde
tails
.~ N
01.0
02(P
RE
SS
FIT
ON
EIN
SM
OO
THR
OLL
ER
)
XTH
RU
00
.75
00
.78
12
5
~1.6
25~O
'5~2
'OI~~I
QU
AN
TITY
'1
MA
TER
IALI
ST
EE
L
~IVERSITY
IFM
ISSI
IJRI
-R
[LLA
PRO
CESS
ING
RECY
CLED
UN
..£SS
Dn
£JM
S[
F[C
JrD
aPL
AST
ICS
INTO
FIBE
RS
DK
NSI
DN
S11
I/£..
INCH
ES
IIECJ
MM
.S~
RtL
LER
CRAN
I<IN
SER
TD
ETA
ILS
JOe
+I-
.IS
+/-
..,~
L----.
....---
.....---
----
4,X
)O(
+1
--
JT9-
10-9
SJN
SER
T3.D
VGII
IN
JTIC
M.E
IM\I
DIG
.Fig
ure
35.
Rol
lcr
ank
inse
rtde
tails
......
..:l w
lllO
.625
\1113/8~
rlll
O.1
25
xT
HR
U,
rO.1
875
-<f)
l-.!
=rt
~~O
.687
5'-
1.5
-
24.0
....IV
ERSI
TYIF
MIS
SIIJ
RI
-RO
LLA
PRO
CESS
ING
RECY
CLED
lIL
£S
SIm
DV
JS
[P
£CJF
'EJt
PLA
STIC
SIN
TDfl
iER
SQ
UA
NTI
TY,
1IJ
MD
ISD
INS
ME.
..JN
CH
[S
MA
TER
IALI
STE
EL
TtL£
RM
C£S
MAN
UAL
CRAN
KHA
NDLE
DE
lAIL
SJE
aW
.SN
Q.U
R.xx
+/-
.I:S
+1
-'"
JIE
GR
I[[S
JOO
(+
/-.a
tJT
I9-1O-~
IHA
NIlL
DV
GII
IfI
JTIC
M..E
MM
IDIi
Figu
re36
.M
anua
lcra
nkha
ndle
deta
ils.
'-.J~
+1I
0.62
50.5
-jr
5.5
2.7
5
----I4II~
-----
{l)0
.43
75
XT
HR
UX
4H
OLE
S
{l)1
.65
62
5X
TH
RU
0.2
51
0-3
2U
NF
X0
.62
5LN
G
2.7
5
0.62
5
0.5
QU
AN
TITY
'2
MA
TER
IAL'
ALU
MIN
UM
1.3
75
I~
2.7
5.,
I
tN..
£SS
Dn
DV
JS
[S
PE
CIF
1[JI
IKN
SID
NS
/JIIE
....
N:H
ES
11l.
£RM
CE
SII£
CJM
M.S
NG
.UR
.xx+
/-.C
I+
/-U.-n
.xxx
+/-
...
..
fGT
ICM
.EIM
VII
IG
~IVERSITY
[FM
ISS
[JJR
I-
RO
LLA
PRIlC
ESSI
NG
REC
YCLE
DPL
AST
ICS
INTO
FlIE
RS
SM
[I]T
H-R
I1.L
ER
PR
.LD
WBO
XD
ElA
ILS
JTI
9-12
-95
IS
-PIL
L-R
.DV
G
Figu
re37
.Sm
ooth
-rol
lpill
owbo
xde
tails
.-...
.JfJ
l
5.5 0.3
75
0.87
5
cztO
.437
5X
THR
UX
4S
LOT
Scz
t1.6
5625
XTH
RU
10
-32
UN
FX
1LN
G
0.62
5
0.5-1
r-
0.25
2.7
5 tQ
UA
NTI
TY;
2
MA
TER
IAL.
ALU
MIN
UM
I•
3.12
5•
I
UN
..£S
SIm
£R
V1
S[
P[C
Jf"
'£D
I
DK
NS
JDN
SN
I£...
INC
t£S
TIL£
RM
CES
~M
GL
M.xx
+/-
...
+/-
UII
E8
ED
JOCX
+/-
•
..
IGT
ICA
L£IM
\III
IG
~IVERSITY
[f"'
MIS
SllJ
RI
-R
I1.L
APR
OCE
SSIN
GRE
CYCL
EDPL
AST
ICS
INTO
FIB
ERS
Ti'R
EAIE
D-R
DLL
ERPI
LLD
VBO
XD
ElA
D.S
JTI
9-1
2-9
SI
T-P
D.L
-R.J
Mi
Figu
re38
.T
hrea
ded-
roll
pillo
wbo
xde
tails
.'-
l0
\
~l
I~
~3'0-
2.6
87
5
2.3
75
1
.5-
0.75
l-
1-3
.0
-
J--2
.68
75
'-
2.3
75
~1,5-
--I
'-0
.75
,a,
s
tiltt
tt-H
0.62
5J
1.+
5 2.1
5r
lllLT
IJIl
lJl2
'r5
11.0
IA
~¢5/16-18
UN
CX
THR
UX
18H
OLE
S
~
~
~
TT
5.5
I[0.6
25
I2.
252.
875
l.y5
II
-II
13.2
5I
QU
AN
TITY
I1
MA
TER
IALI
ALU
MIN
UM
~IVERSITY
IFM
ISSI
lJR
I-
ROLL
APR
DCES
SING
RECY
CLED
lIU
SS
Dn
£R
VJ
S[
FE
CJJ
1DPL
AST
ICS
INTD
fliE
RS
IIM
DCSI
DNS
11I£
ININ
CHES
IUJM
M.S~
BASE
PLA
TED
ETA
ILS
~:~:.
+/-
UIIE
&IU
S1
----
--.-
---.
....
,,--
----
--1
10
NIT
ICM
.£~VDG
JT9
-10
-"BA
SE.D
VG
Fig
ure
39.
Bas
epl
ate
deta
ils.
-.J
-.J
¢0
.37
5X
THR
UX
5H
OLE
S
r0
.37
51.
5
r--
L0
.68
75
I5.
5•
I
0.6
25
1.37
5
II
11.0
•I
~0,3125
XTH
RU
X4
HO
LES
0.125~r-
1-O
.l25J
' I
1.3
75
--1
0.6
25
,I
I,'
II
i
QU
AN
TITY
I2
MA
TER
IALI
ALU
MIN
UM
~IVERSITY
[F'
MIS
SOO
RI-
RCl.L
APR
OC
ESSI
NG
RE
CV
Q.E
DlI
US
SD
11
£JM
S[
SP£C
JJ1E
JtP
LAS
TIC
SIN
TDF1
BER
SIK
NS
IIIG
I#IE
...
JNCH
ES
IEC
NL
S~
PRIM
ARY
BASE
AtG
.ED
ETA
D.S
JOe
"1-'
"+
/-..
,~
....
.---
....
.----
...-
----
--1
.xxx
+1
-a
tJT
9-1
0-"
NG
..E
1-R
JMi
..
ICIT
ICN
-E~
Figu
re40
.Pr
imar
yba
sean
gle
deta
ils.
.-.....J
00
0.3
75
..
-.e0 .5
,I
!I!
II!
L.-
-¢
0.3
75
XTH
RU
X2
HO
LES
1,5 L
t0
.37
5-1
I \-0
.75
¢O
,31
25
.-r_
......
..L-.---
}.10.
75
I 1.5
0.87
5-o
J6:6::=d:cd~
~1.8
715-
1
XTH
RUX
2H
OLE
S
0.1
25
-]I-
~O
.l25I
' !I
!
QU
AN
TITY
I2
MA
TER
IAL.
ALU
MIN
UM
\K.£
SS
D1
1£
RV
1S
[S
P[C
JF
IJI
IHN
SD
ICS
fill
£...
H:H
ES
TIL
ER
IlIU
SII
£C»W
.SM
Gl.
AR
.xx+
1-.
1:1
+/-
UIlE
IRE
ES
.X)O
(+
/-•
..
MJT
ICM
..t
MM
IDIG
Figu
re41
.Se
cond
ary
base
angl
e#
1de
tails
.
~IVERSITY
IFM
ISSl
lJR
I-
RllL
LAPR
OCE
SSIN
GR
EC
ya.E
DPL
AST
ICS
INTD
F1BE
RS
SECO
NDAR
YBA
SEA
tG..E
11D
ElA
B..S
JTI
9-1
0-9
5I
AtG
.E2-
R.D
VG
'''J
\0
0.5
1.5 J_ 0.7
5-1
¢0
.37
5X
TH
RU
X2
HO
LES
0.3
75
0.1
25
-11
-
¢O.3
125
XT
HR
UX
2H
OLE
S
0.8
75
l-0.
125T
' II
i
QU
AN
TITY
.2
MA
TER
IAL.
ALU
MIN
UM
lH..£
SS
aTJ£
RV
JS£~
JKN
SJD
NS
lIfE
...
JNQ
£S
1lI
.£R
MC
(SIE
CJM
M.S
NG
I.M
.xx+
/-.1
5+
/-U
.-u
JOe)
(+
/-•
III
NJ1"
ICA
L£ar
ww
DIi
Figu
re42
.S
econ
dary
base
angl
e#2
deta
ils.
~IVERSITY
IFM
ISSW
RI
-RD
LLA
PRO
CES
SIN
GR
ECVQ
.ED
PLA
STIC
SIN
TDfl
iER
S
SECO
NDAR
YBA
SEAN
GLE
.2lE
TA
D.S
JTI
9-1
0-"
IN
«iL
E3-
R.1
Mi
00 o
0.6
25
-1t
rO.3
75 .
II.2.
012.
012.
012.
0
rO.7
5
NO
TEI
MO
UN
TIN
GH
OLE
SFO
RTH
EH
OPP
ERA
REN
OT
INC
LUD
ED.
THEI
RLO
CA
TIO
NS
\JIL
LV
AR
YD
EPEN
DIN
GON
THE
FIN
AL
DIM
ENSI
ON
SO
FTH
ER
OLL
ERS.
:at'*
f'0.
375T
1-1
3.5
~0.37S
XTH
RU-
TYP ----
----
--::
:tff
,3f1L..-~.-
0.37
512
.5
1.1
25
-1t
rO.37~
I.I
QU
AN
TITY
;4
MA
TER
IALI
ALU
MIN
UM
2.01
2.01
2.01
2.0
rO.7
S0.
125
TYPb--
-E1.5
1.5~~
~IVERSITY
IFM
ISSl
lJR
I-
ROLL
A_
PRO
CES
SIN
GR
ECVQ
.ED
lH.£
SS
Dn
£JM
S[
SP
£CIF
"DI
PLA
STIC
SIN
TDFl
IER
SJK
NS
JDN
S/!I
f£.
INH
:HE
S
IIE
QW
.S~
PRIM
ARY
FRAM
EM
EMBE
RD
ETA
ILS
::x:~::.
+/-
UID
US
....
.---
....
..--
--..
.---
----
-11m
NJT
ICM
.£JM
WII
GJT
9-1O-~
FRAM
E.DW
G
Figu
re43
.P
rim
ary
fram
em
embe
rde
tails
.0
0
I_
II
II
II
II(
III
I~Jr--
111-
-fI
rc:
Ip=---J
II
II
33
0.3
75
-1t-
~3.125-
f-1
.75
-f-
1.7
5--3.125~
1.0
--
1.0
IT~'
"~'
" ~
f
'TS 2'rS
2.0
~"
"J
'",~
",~
,~
"f
,~
",~
,~
~""'
2.0
"~RO.5
I5
XT
HR
U--
"Vt
10.0
2.0
HO
LES
t'"
""j
~'"
I'I
'
t"
,~
~,
"
Ir
""
"2.
0 IJ75
2'r5~I\.
...
It 1.5 f
-2
,25
-L
-2
.25
-
03
/8-2
4U
NF
XTH
RU
X8
HO
LES
11.0
,
00.3
12X
10
QU
AN
TIT
Yt
2
MA
TER
IALI
ALU
MIN
UM
~IVERSITY
[J"
MIS
S[J
JRI
-R
ILLA
UN
.£S
SD
n£
JM
S[
SP
[CJF
"ED
PRO
CES
SIN
GR
ECYC
LED
DK
NS
DN
NI£
..IN
CHES
PLA
STIC
SIN
TOfI
BE
RS
TlJ
..E
RA
HC
[S
~~.t:I~
EG
lEE
SEN
DP
LAT
EJE
TA
ILS
.xxx
+/-
•
III
taT
DL
£IM
VII
IiJT
9-1
0-9
5E
ND
-RE
V.D
VG
Figu
re44
.E
ndpl
ate
deta
ils.
00
N
II
,II
;;I
i'
II
¢0.3
125
XTH
RU
X10
HO
LES
0.7
5-4
0.75
~I
1---.
C1.
5
-r--
...-+
--rl
----
----
----
---....
........0,
2511
---w
10.0
Iii'
Iii'
-w ~
~
h.
-~
-'Iii
'
~ '-
tt
2.0 I 1 2.0 I 1 2.0 I t 2.0 J
--1
-1
-
--1
-1
-
-I-f-
I.12
.5.1
QU
AN
TITY
;2
MA
TER
IALI
ALU
MIN
UM
~IVERSITY
IFM
ISS
IJJR
I-
RO
LLA
PRO
CESS
ING
RECY
D..E
DlI
US
SD
n£
RV
JS£
SP
£CIF
1DPL
AST
ICS
INTD
F1BE
RSIK
NS
IDN
S/Il
l£...
JIO
£S
II£
QW
.S~
SID
EP
LATE
DE
TAD
.SJO
e+
/-.1
:1+
/-~
KK
UI-
----
..,.
.---
---r
----
----
t.xx
x+
/-..
.JT
9-1
0-"
SIIE
.DW
GII
IN
nIC
M.E
-.w
ING
Figu
re45
.Si
depl
ate
deta
ils.
00
tj-J
~O.l875
XTH
RU
X3
6H
OLE
S
----------1IIo4l"I-------~
y1
-30
'TY
P
~0.09375
XTH
RU
X3
2H
OLE
S
•I'
-----
•I.
-----
•I'
-----
------
0.7
5T
YP
0.3
75
TY
P
5.0
TY
P
---.#
-1-.-
----------f-~
r-----
--·I•
-----
•I
•
! ~---
----
•I
•-----
•I
•
! ~---
----
•I.
-----
•I.
! L--
----
-.I.
-----
•I
•
0.2
5T
YPr10
.25j-9
.5
1.0
I0.
75TY
P
6.0 6.0
20.0
30.0 QU
AN
TIT
YI
1
MA
TER
IALI
0.0
40
·A
LUM
INU
MS
HE
ET
~IVERSITY
I:FM
ISS
IIJR
I-
RD
LLA
PRO
CES
SIN
GR
ECVQ
..ED
lIU
SS
D1
l£JM
S[
SP
Earo
.PL
AST
ICS
INTD
FlIE
RS
IJM
EN
SII
IGII
l£..
INC
HES
JIE
QW
.S~
.a:»
PER
DET
AIL
SJD
(+
/-.1
:1+
1-
loS
IEII
lIE
D1-
----
.....---
-.,.
.---
----
1JO
O(
+/-
...
JT9-1e-~
tlFP
ER
.DV
GII
IN
JTIC
M.£
JMV
DG
Figu
re46
.H
oppe
rde
tails
.0
0~
• • •
•I
•
•I
•
•I
•
•I
•
•I
•
•I
•
•I
•
•I
•
NO
TE.
BEN
DA
LON
GD
AS
HE
DL
INE
S
•I
•
•I
•
•I
•
•I
•
• • • •
QU
AN
TITY
:1
MA
TER
IAL.
0.04
01
ALU
MIN
UM
SH
EE
T
~IVERSITY
[FM
ISS
OJR
I-
RD
LLA
UlU
SS
Dn
£JM
S[
SP
[CJF
lDPR
OC
ESSI
NG
REC
YD.E
DJJ
MD
CS
IIIG
NI£~
JN
QD
PLA
STIC
SIN
TOnB
ER
Sl1
l..£R
AN
CE
S
~~~..,~
DU
Nn
i'IlP
PER
BEN
DIN
GIE
TA
ILS
.xxx
+/-
.lSI
lIDN
ITIC
M.£
-.v
DIi
JT9
-12
-95
~PER-B.DVG
Figu
re47
.H
oppe
rbe
ndin
gde
tails
.oc V
I
APPENDIX B.
PROTOTYPE ROLLING-MILL ASSEMBLY DRAWINGS
87
The following diagrams were prepared to provide an example of how the prototype rolling mill
is assembled and what the fInished product looks like. Figure 48 shows how the four inserts are press-fit
into the smooth and threaded rolls. Due to the complexity and the required processing times of the CAD
model, the threads have been omitted from these solid-model assembly drawings. The lines seen on the
rolls are CAD tessellation line and should not be confused with thread markings. The shortest insert is
used in the threaded roll; this will help to identify the threaded roll in the remaining discussion. Figure
49 shows how the handle in used in conjunction with the smooth roll's front insert. Figure 50 shows how
the pillo\v boxes are attached to the roll inserts. Sealed bearings are fit into the pillow boxes, and then
the assembly is slid onto the roll inserts to support the rolls. Figure 51 shows the base plate, and Figure
52 shows how the angle brackets are attached to the base plate. Figures 53 and 54 show how the square
tubing frame is attached to the angle brackets. Figures 55 and 56 show how the end and side plates are
attached to the frame members. Figures 57 and 58 show how the rolls are placed through one of the end
plates and await the installation of the pillow boxes, which is shown in Figures 59 and 60. Figures 61
and 62 show how the heated hopper is placed between the frame members and orientated so that flakes
fall directly onto the rolls below.
~~
~IVERSITY
[FM
ISS
WR
I-
RD
LLA
PRO
CES
SIN
GR
ECYQ
.ED
lJU
SS
Dn
DV
JS
[~
PlA
STIC
SIN
TDF1
BER
SIJ
MEN
SJIIt
SN
t£..
INCH
ES
DE
DW
.S~
R[L
LE
RA
SSEM
BLY
.xx+
/-..
,+
/-1
oS
.-nL
----
.....--
-.....
.,......-
----.
...JD
O(
+/-
at
JT1
1-2
8-"
Rl:L
AS
BLY
.DV
GDO
NIT
ICII
L£
..,.
..,.
Fig
ure
48.
Rol
las
sem
bly.
00
00
----~.---------~.----_._.-
------_
._----------
-------~--~----_.-
------~._-----------
Rll.
.LER
ASSE
MJL
YW
ITH
MA
HLE
JT1
11
-28
-93IR
lLA
SLH
.DV
G
~IVERSITY
IFM
ISSW
RI
-Rl
J..LA
PRO
CESS
ING
REC
VQ..E
DPL
AST
ICS
INTO
FlB
ERS
UtL
£SS
D1
l£R
VJS
[S
P£C
IF"D
IKN
SID
NS
11I£~
JIIC
HES
llL£
RA
NC
[SII
£C
IW..
IM
GU
R
::c:~::.
+/-
uID
US.1
-----
10
.aT
ICIlL
£JM
VII
Ii
Figu
re49
.R
oll
asse
mbl
yw
ithha
ndle
.0
0\D
o0
10
0
@I@
Io
01
00
lH-£
SS
D1
lDV
JSt
SP
EC
JF'1
D
IJM
EN
SIII
GII
I£...
KH
ES
TII
.ER
NID
IUI'
W.S
NG
I..A
A.x
x+
/-..
,+
/-&.
:I~
JOO
(+
/-..
.
JIDfC
ITIC
M.£
IMV
IMG
~IVERSITY
IFM
ISSl
lJR
I-
ROLL
APR
OCE
SSIN
GRE
CVD
.ED
PlA
STIC
SIN
TDF1
BER
S
RlL
LE
RA
SSEM
BLV
\lIT
HPI
LLO
WBO
XES
JTI.
11-2
8-95
1PI
LASB
LV.D
WG
Figu
re50
.R
oll
asse
mbl
yw
ithpi
llow
boxe
s.'-..
D o
00
00
00
00
00
00
00
00
00
II
[-----
I
l»4I
VER
SITY
IFM
ISSO
ORI
-RI
1.LA
UtU
SS
on
oV
lSt
SPE
C1f
DI
PRO
CESS
ING
RECY
CLED
DM
NS
IIIG
ME
...
JND
£SPL
AST
ICS
INTO
FIBE
RS
lll.£
RM
I:£
S~
..,~
JIE
8ED
BASE
PLA
TE.xx
x+
/-•
..
NIT
ICIL
EIM
\nN
GJT
U-2
8-9
SB
ASE
-3D
.DW
G
Figu
re5
1.B
ase
plat
e.\0
00
00
00
00
00
00
00
00
00
~~
~Id
BASE
PLA
TEW
ITH
AN
GLE
S
JT-,
u-ee
-95I
BA
SEN
G..D
WG
tMlV
ERSI
TY[F
'M
ISSI
1JR
I-
R[J
..LA
PRO
CES
SIN
GRE
CYCL
EDPL
AST
ICS
INTO
FIB
ERS
lJL
£S
$D
Tt£
RV
1S
[SP
EC
JFD
t
DK
NS
UIG
NIE
...
IOtE
S
TII.E
RANC
ESIl
£aW
..S
NG
UA
~:~:::.
+/-
loS
IIE
8U
J..I
_
•N
ITIC
Ill£
IItIN
DIG
Figu
re52
.B
ase
plat
ew
ithan
gles
.\C t·..
.,)
UN
..£S
SD
1l£R
VJS
£S
P£C
IflE
Da
DK
NS
IDN
SN
t£..
JID
O
llI..E
RN
«:E
SJE
QW
..S
AIG
JI..M
.xx+
I-.I
:S+
1-
U.-
ES
JDO
(+
1-
_
•ta
TIC
M.£
IIW
tID
IG
~IVERSITY
IFM
ISSI
lJR
I-
RlL
LAPR
DC
ESSI
t«i
RECY
CLED
PLA
STIC
SIN
TOFI
BER
S
BA
SE
PLA
TEW
ITH
ANG
LES
AND
F1
WE
JTI
11
-28
-95I
BA
SA
tER
2.D
VG
Fig
ure
53.
Bas
epl
ate
with
angl
esan
dfr
ame.
\D w
[]
~ ~I~
~
UfU
SS
&m
£JM
S[
RC
JF'1
DI
IJM
DIS
IDN
IN
t£...
ICH
ES
TtU
JtM
C[S
JEaw
..I
MG
JlM
JOe
+/-
..+
/-U.-a
JOOC
+/-
•
•N
ITIC
M.£
IMV
IIG
~IVERSITY
IFM
ISS
IlJR
I-
RO
LLA
PRO
CES
SIN
GR
ECYQ
.ED
PLA
STIC
SIN
TDFl
IER
S
BA
SE
PLA
TE
V1T
HN
G.E
SAN
DFR
AM
E
JTI1
1-2
8-9
5IB
AS
EA
WR
.IN
G
Fig
ure
54.
Bas
epl
ate
wit
han
gles
and
fram
e(4
view
s).
\D .+;;..
~IVERSITY
IFM
ISSt
IJR
I-
RtL
LAU
N.£
SS
DT
t£JM
S[
SP
EC
trE
Da
PRO
CESS
ING
RECY
CLED
DJM
ENSU
JGM
E.
INJN
CtD
PLA
STIC
S!N
rcFI
BER
STt
l.ER
ANC
ES~
'*I
~1IE8EES
END
AND
SID
EP
LATE
SJO
OC+
/-..
.II
IN
JrIC
M.I
:IM
VII
Ci
JT1
2-G
6-9
SE
ND
SID
E2.
DV
G
Fig
ure
55.
End
and
side
plat
es.
\D Vl
[I
Ie·
Ie·
.·
.·
II
~IVERSITY
IFM
ISSI
lJR
I-
Rr:L
LAlI
U$
ID
nOV
JSE~
PRO
CESS
ING
RECY
CLED
DK
MII
IN,.
...
N:H
ES
PLA
STIC
SIN
TOFI
BER
S11
L£J
WI:E
S~..,
~1IEIIeS
END
AND
SID
EPL
ATE
S.xx
x+
1-
•..
tar~
IMW
IiJT
12-06~
END
SII£
.DW
G
Figu
re56
.E
ndan
dsi
depl
ates
(4vi
ews)
.\C
;0
\
OO
VE
RS
ITY
[FM
ISSO
OR
I-
R[l.
.LA
lH.£
SS
Dn
£R
V1
S[
$P
[CJn
mPR
OCE
SSIN
GRE
CYCL
EDD
KN
SU
ItS
ME
..JN
CtO
PLA
STIC
SIN
TOFI
BER
ST1
l.ER
AHa:
S~
...~m.n
RI1
.LE
RS
AND
FRAM
EJO
O(
+1
-_
..
NIT
ICM
.l:II
WJD
Ci
JT12-G6~
RlJ
.FR
AM2.
DW
G
Figu
re57
.R
olls
and
fram
e.\.
0-.
J
. ~
·~
·li
e·
~
·
I
LMV
ERSI
TYIF
MIS
SI1J
RI
-R
ILLA
PRO
CESS
ING
RECY
CLED
tIU
UD
TK
JIV
IS[~
PLA
STIC
SIN
DK
NS
UIt
S1#
1£..
JN
aU
TOFI
BER
S
JE
aW
.I~
RlJ
..LE
RS
AND
FR
AtE
~:~:.
.1-u
.-a
1-
....---
....
.----
---1
..
...,
ICIU
IMW
IiJT
12-0
6-95
RllF
RAM
E.D
WG
Fig
ure
58.
Rol
lsan
dfr
ame
(4vi
ews)
.\.D 0
0
~IVERSITY
[FM
ISSI
lJR
I-
R[l.
LAPR
OCE
SSIN
GRE
CYCL
EDlI
US
SD
n£
JM
S[
SP
[CJr
Dt
PLA
STIC
SIN
TOFI
BER
SD
KN
SID
NS
/11I
£..
.K
HE
S
JUN
LS~
PD..L
DW
BOX
ESAN
DFR
AME
.xx./
-.1
5+
1-
UIE
8EE
SI-
----
.....--
---r
----
----
I.xx
x+
1-
•JT
12-06~
Pn.F
'RAM
2.D
VGlID
NIT
ICM
.£~
Fig
ure
59.
Pillo
wbo
xes
and
fram
e.'-C v
PIU
J]W
BOXE
SAN
DFR
AME
JT1
12-G
6-9S
IP
D.rR
AM
E.J
Mi
lI4IV
ERSI
TY[F
MIS
SI:IJ
RI
-R
Il.LA
PRO
CESS
ING
RECY
CLED
PLA
STIC
SIN
TOFI
BER
SlI
US
SD
TtD
VJI
£~
DK
JIIS
IDN
I,.
..IO
U
TtL£
RMCE
SD
EaW
-SM
GU
R
::x:~
=-+
/-u.....1-
-----
•fG
TIC
M.I:...,.
-II- Fi
gure
60.
Pillo
wbo
xes
and
fram
e(4
view
s).
~ o o
KF
PE
RAN
DFR
AME
JT..
,12-G6~I
KP
fRA
M2.
DW
G
~IVERSITY
[FM
ISS
IlJR
I-
RIJ
..LA
PRO
CESS
ING
RECY
CLED
PLA
STIC
SIN
TOFI
BER
SIN
.£S
SD
n£
JN
IS[
SPEC
IF"E
Dt
DK
NS
:DG
/fIl
E.
INJN
CHES
1'tL
£RAH
C£S
JI£
QW
.SN
GU
R::x
:~::.
./-..,J
dH
S..1
----
-II
IIC
JTIC
M.£
IWM
IIi
Figu
re61
.H
oppe
ran
dfr
ame.
o
IO'P
ER
AND
FRAM
EJT
112
-G6-
95I-I
CFf
--RA
ME
-.D-\l
G-
Lt4I
VER
SITY
[F"
MIS
SOO
RI-
Rl1..
LAPR
OCE
SSIN
GRE
CYCL
EDPL
AST
ICS
INTO
FIBE
RS
LlU
SS
an
ov
JK~
DK
MII
DN
I,.
INN
:HE
S
1'Il.
£JtA
NC
DIE
CJM
M.S
IlII
Q.U
A~
:~::.
+/-
..,
IE8E
D..1
----
--•
ND1'
IC#l
LEIM
VII
IG
iI-
Figu
re62
.H
oppe
ran
dfr
ame
(4vi
ews)
.o N
APPENDIX C.
FLAT ROLLING CODE
104
The follo\ving pages of this appendix contain the source code to the IMSL calling routine and
necessary subroutines "vritten in Fortran referred to in Section IV for the flat rolling model.
* FLAT M2.F
* JEFF THOMAS
* 13 JUNE 1996
***********************************************************************
*******
THIS PROGRAM IS TO SOLVE THREE INITIAL-VALUE PROBLEMS OF THE
ORDINARY DIFFERENTIAL EQUATION FORM
SIG_B_PRIME(PHI)=Fl(PHI)*SIB_B(PHI)+F2(PHI)
WHERE Fl AND F2 ARE FUNCTIONS OF PHI ONLY, USING THE
RUNGE-KUTTA-VERNER FIFTH ORDER AND SIXTH ORDER METHOD.
* IT IS ALSO DESIGNED TO FIND THE APPROXIMATE NEUTRAL POINT FROM
* THE NUMERICAL DATA OBTAINED FOR THE THREE ODE CURVES AND THEN
* REARRANGE THAT DATA INTO ONE CONTINUOUS DATA SET, TRIMMING OFF
* UNNECESSARY DATA BEYOND A USEFUL RANGE.
** FROM THIS CONTINUOUS DATA SET REPRESENTING THE DIMENSIONLESS
* ROLL-STRESS-PER-FLOW-STRESS RATIO PER ROLL, A TRAPEZOIDAL AREA
* CALCULATION IS PERFORMED TO DETERMINE THE TORQUE AND ROLL FORCE
* REQUIRED PER ROLL.
************************************************************************
***********************************************************************
* DECLARATIONS
PROGRAM FLAT M2
INTEGER MXPARM,NEQ,IDO,ISTEP,I,J,K,NP,PLOT,LOOP
PARAMETER (MXPARM=SO,NEQ=l)
DOUBLEPRECISION FCN1,FCN4
DOUBLEPRECISION FLOAT,PARAM(MXPARM) ,PHI,PHIEND,TOL,SIG_B(NEQ)
DOUBLEPRECISION MU,H_O,H_F,R,PHI_O,Y
DOUBLEPRECISION SIG_B_l(lSl) ,SIG_B_4(lSl) ,
& SIG_B_I(lS2) ,PHI_I(lS2)
DOUBLEPRECISION AREA_l,AREA_4, TORQUE, FORCE
INTRINSIC FLOAT,SQRT
EXTERNAL FCN1,FCN4,IVPRK,SSET
COMMON/GLOBALS!MU,H_F,R
* OPEN INPUT, OUTPUT, AND ERROR FILES
LOOP=O
OPEN (7,FILE='flat_m2.err' ,STATUS='unknown')
OPEN (8,FILE='flat.in l ,STATUS='old l)
READ (8,*,END=998,ERR=998) PLOT
OPEN (9,FILE='flat.txt' ,STATUS='unknown')
* READ INPUT PARAMETERS
5 LOOP=LOOP+1
READ (8,*,END=998,ERR=998) MU
IF (MU.EQ.777.777) GOTO 999
IF (LOOP.GE.500) THEN
WRITE (7,' (A40) I) 'Too many loops error!'
GOTO 999
ENDIF
READ (8,*,END=998,ERR=998) H 0
READ (8,*,END=998,ERR=998) H F
READ (8, *, EOO=998 ,.ERR=998) R
READ (8,*,END=998,ERR=998) Y
* FIND BEGINNING OF PHI GRID
PHI_O=ACOS(1.0-(H_O-H_F)/R/2.0)
* INITIALIZE SIG B'S
DO 1=1,152,1
SIG_B_1(I)=1001.0
SIG_B_4(I)=1001.0
ENDDO
* SET TOLERANCE CONTROL
TOL=0.0000000005
* SET PARAM DEfAULT
CALL SSET(MXPARM,O.O,PARAM,l)
* SELECT ABSOLUTE ERROR CONTROL
PARAM(lO) =1.0
IDO=l
* CALL DE SOLVER FOR SECTION ONE
PHI=PHI 0
SIG_B(1)=1.0
SIG_B_l(l)=l.O
DO ISTEP=l,lSO,l
PHIEND=PHI_O-FLOAT(ISTEP)*PHIO/150.0
CALL DIVPRK(IDO,NEQ,FCN1, PHI, PHIEND, TOL, PARAM, SIG_B)
IF (SIG_B(l) .GE.1000.0) THEN
WRITE (9,' (A40) ') 'STIFF DATA CHOPPED 777.777'
GOTO 20
ENDIF
SIG B 1(ISTEP+1)=SIG_B(1)
ENDDO
* FINAL CALL TO RELEASE WORKSPACE
105
106
20 IDO=3
CALL DIVPRK(IDO,NEQ,FCN1,PHI,PHIEND,TOL,PARAM,SIG_B)
IDO=l
* CALL DE SOLVER FOR SECTION FOUR
PHI=O.O
SIG_B(l)=l.O
SIG_B_4(151)=1.0
DO ISTEP=1,150,1
PHIEND=FLOAT(ISTEP)*PHIO/150.0
CALL DIVPRK(IDO,NEQ, FCN4, PHI, PHIEND,TOL, PARAM,SIG_B)
IF (SIG_B(l) .GE.1000.0) THEN
WRITE (9,' (A40) ') 'STIFF DATA CHOPPED 777.777'
GOTO 25
ENDIF
SIG_B_4(151-ISTEP)=SIG_B(1)
ENDDO
* FINAL CALL TO RELEASE WORKSPACE
25 IDO=3
CALL DIVPRK(IDO,NEQ,FCN4,PHI,PHIEND,TOL,PARAM,SIG_B)
* FIND NEUTRAL POINT, ARRANGE DATA, & TRIM UNNECESSARY DATA
DO K=1,152,l
IF (SIG_B_4(K) .LE.SIG_B_1(K)) THEN
SIG_B_I(K)=(SIG_B_1(K-1)*SIG_B_4(K)-SIG_B_4(K-1)
& *SIG_B_1(K))/(SIG_B_4(K)-SIG_B_4(K-1)
& +SIG_B_1(K-1)-SIG_B_1(K))
PHI I(K)=PHI_O/1SO.0*(SIG_B_4(K)-SIG_B_1(K))
& /(SIG_B_4(K)-SIG_B_4(K-1)+SIG_B_1(K-1)
& -SIG_B_1(K) )+PHI_O-PHI_O/150.0*FLOAT(K-1)
NP=K
DO J=K+1,152,1
SIG_B_I (J)=SIG_B_4 (J-1)
PHI I(J)=PHI_O-PHI_O/150.0*FLOAT(J-2)
ENDDO
GOTO 50
ELSE
SIG_B_I(K)=SIG_B_1(K)
PHI I (K)=PHI_O-PHI_O/150.0*FLOAT(K-1)
ENDIF
ENDDO
50 CONTINUE
* CALCULATE AREAS UNDER THE CURVE
AREA 1=0.0
AREA 4=0.0
DO 1=l,NP-l,l
AREA l=AREA 1+0.S*(S1G_B_1(1+1)+SIG_B_I(I))
& *(PHI 1(I)-PHI 1(1+1))
ENDDO
DO I=NP,151,1
AREA 4=AREA 4+0.5* (SIG_B_I (1+1) +SIG_B_I (I) )
& *(PHI I(I)-PHI 1(1+1))
ENDDO
TORQUE=1/2.54*(AREA_I-AREA_4)*Y*MU*R**2/12.0
FORCE=1/2.54*(AREA_l+AREA_4)*Y*R
* OUTPUT RESULTS
WRITE (9/100) MU
100 FORMAT ('Coeff. of Friction ... ',F12.6)
WRITE (9/104) R
104 FORMAT ('Roll Radius '/F12.6/' in')
WRITE (9,106) H 0
106 FORMAT ('Original Thickness ... '/F12.6/' in')
WRITE (9/108) H_F
107
108 FORMAT ('Final Thickness , , F12 . 6 /' in')
WRITE (9/120) H_O-H_F
120 FORMAT ('Draft '/F12.6/' in')
WRITE (9/122) (H_O-H_F) /H_O*100. 0
122 FORMAT ('Percent Reduction .... '/F12.6/' %')
WRITE (9/112) PHI_O
112 FORMAT ('Beginning Phi
WRITE (9,116) PHI I(NP)
. , / F12. 6/' rad' )
116 FORMAT ( 'Neutral- Point Phi ' / F12. 6/' rad' )
WRITE (9/118) SIG B I (NP)
118 FORMAT ( 'Sig_b / Y at NP ...... ' ,F12.6)
WRITE (9,119) Y
119 FORMAT ( 'Flow stress .......... ' / F12. 6/' psi' )
WRITE (9/124) FORCE
124 FORMAT ( 'Roll Force ........... ' / F12. 6 / ' lb' )
WRITE (9/126) TORQUE
126 FORMAT ( 'Roll Torque .......... ' / F12. 6, , ft-Ib')
WRITE ( 9 / ' (/) , )
IF (PLOT.EQ.l) THEN
WRITE (9/128)
128 FORMAT (4X/"ISTEP' ,8X, 'PHI' ,9X, 'SIG_B/Y')
DO K=1/152/1
WRITE (9/130) K/PHI_I(K) ,SIG B I(K)
130 FORMAT (I6/2(3X/F12.6))
ENDDO
WRITE (9, 1 (/) 1 )
ENDIF
GOTO 5
*998 WRITE (7,1 (A40) I) 'Read / Write Error'
999 CLOSE (8)
CLOSE (9)
END
*SUBROUTINE FCN1(NEQ,PHI,SIG_B,SIG_B_PRIME)
INTEGER NEQ
DOUBLEPRECISION PHI, SIG_B (NEQ) ,SIG_B_PRIME(NEQ)
DOUBLEPRECISION MU,H_F,R
COMMON/GLOBALS/MU,H_F,R
* FUNCTIONS FOR SECTION ONE
SIG_B_PRIME(1)=(-2.0*R*MU*COS(PHI)*SIG_B(1)+2.0*R*SIN(PHI))
& /(H_F+2.0*R*(1.O-COS(PHI)))
RETURN
END
*SUBROUTINE FCN4(NEQ,PHI,SIG_B,SIG_B_PRIME)
INTEGER NEQ
DOUBLEPRECISION PHI,SIG_B(NEQ) ,SIG_B_PRIME(NEQ)
DOUBLEPRECISION MU,H_F,R
COMMON/GLOBALS/MU,H_F,R
* FUNCTIONS FOR SECTION FOUR
SIG_B_PRIME(1)=(2.0*R*MU*COS(PHI)*SIG_B(1)+2.0*R*SIN(PHI))
& /(H_F+2.0*R*(1.O-COS(PHI)))
RETURN
END
108
APPENDIX D.
FffiER ROLLING CODE
110
The following pages of this appendix contain the source code to the IMSL calling routine and
necessary subroutines written in Fortran referred to in Section IV for the fiber rolling model.
* SIG B M2.F
* JEFF THOMAS
* 14 JUNE 1996
***********************************************************************
*******
THIS PROGRAM IS TO SOLVE THREE INITIAL-VALUE PROBLEMS OF THE
ORDINARY DIFFERENTIAL EQUATION FORM
SIG_B_PRIME(PHI)=F1(PHI)*SIB_B(PHI)+F2(PHI)
WHERE F1 AND F2 ARE FUNCTIONS OF PHI ONLY, USING THE
RUNGE-KUTTA-VERNER FIFTH ORDER AND SIXTH ORDER METHOD.
* IT IS ALSO DESIGNED TO FIND THE APPROXIMATE NEUTRAL POINT FROM
* THE NUMERICAL DATA OBTAINED FOR THE THREE ODE CURVES AND THEN
* REARRANGE THAT DATA INTO ONE CONTINUOUS DATA SET, TRIMMING OFF
* UNNECESSARY DATA BEYOND A USEFUL RANGE.
** FROM THIS CONTINUOUS DATA SET REPRESENTING THE DIMENSIONLESS
* ROLL-STRESS-PER-FLOW-STRESS RATIO PER ROLL, A TRAPEZOIDAL AREA
* CALCULATION IS PERFORMED TO DETERMINE THE TORQUE AND ROLL FORCE
* REQUIRED PER ROLL.
************************************************************************
***********************************************************************
* VARIABLE DECLARATIONS
PROGRAM SIG B M2
INTEGER MXPARM,NEQ,IDO,ISTEP,I,J,K,NP,PLOT,LOOP
PARAMETER (MXPARM=50,NEQ=l)
DOUBLEPRECISION FCN1,FCN3,FCN4
DOUBLEPRECISION FLOAT,PARAM(MXPARM) ,PHI,PHIEND,TOL,SIG_B(NEQ)
DOUBLEPRECISION MU,N,T_O,T_F,R,THETA,PHI_O,PHI_M,Y
DOUBLEPRECISION SIG_B_l(Sl) ,SIG_B_3(lS1) ,SIG_B_4(lSl),
& SIG_B_I(lS2) ,PHI_I(lS2)
DOUBLEPRECISION AREA_l_3,AREA_4,TORQUE,FORCE
INTRINSIC FLOAT,SQRT
EXTERNAL FCN1,FCN3,FCN4,IVPRK,SSET
COMMON/GLOBALS/MU,N, T_O,T_F,R, THETA
* OPEN INPUT, OUTPUT, AND ERROR FILES
LOOP=O
OPEN (7,FILE='sig_b_m2.err' ,STATUS='unknown ' )
OPEN (8,FILE='thread.in' ,STATUS='old')
READ (8,*,END=998,ERR=998) PLOT
OPEN (9,FILE='thread.txt' ,STATUS='unknown')
* READ INPUT PARAMETERS
5 LOOP=LOOP+1
READ (8,*,END=998,ERR=998) MU
IF (MU.EQ.777.777) GOTO 999
IF (LOOP.GE.500) THEN
WRITE (7,' (A40) ,) 'Too many loops error!'
GOTO 999
ENDIF
READ (8,*,END=998,ERR=998) N
READ (8,*,END=998,ERR=998) T °READ (8,*,END=998,ERR=998) T F
READ (8,*,END=998,ERR=998) R
READ (8,*,END=998,ERR=998) Y
* FIND NEEDED PARAMETERS
THETA=ATAN(2.0*T_F*N)
PHI_O=ACOS(1.0-T_O/2.0/R)
PHI_M=ACOS(1.0+(T_F-T_O/2.0)/R)
* INITIALIZE SIG B'S
DO 1=1,152,1
SIG_B_1(I)=1001.0
SIG_B_3(I)=1001.0
SIG_B_4(I)=1001.0
ENDDO
* SET TOLERANCE CONTROL
TOL=O.000000005
* SET PARAM DEFAULT
CALL SSET(MXPARM,O.O,PARAM,l)
* SELECT ABSOLUTE ERROR CONTROL
PARAM(lO) =1.0
IDO=l
* CALL DE SOLVER FOR SECTION ONE
PHI=PHI 0
SIG_B(l)=l.O
SIG_B_1(1)=1.0
DO ISTEP=l,50,1
PHIEND=PHI_O-FLOAT(ISTEP) * (PHI O-PHI_M)/50.0
CALL DIVPRK(IDO,NEQ,FCN1,PHI,PHIEND,TOL,PARAM,SIG_B)
IF (SIG_B(l) .GE.IOOO.O) THEN
WRITE (9,' (A40) ') 'STIFF DATA CHOPPED 7777.7777'
GOTO 20
III
ENDIF
SIG B l(ISTEP+l)=SIG_B(l)
ENDDO
* FINAL CALL TO RELEASE WORKSPACE
20 IDO=3
CALL DIVPRK(IDO,NEQ, FCN1,PHI, PHIEND, TOL, PARAM, SIG_B)
IDO=l
* CALL DE SOLVER FOR SECTION THREE
PHI=PHI M
SIG_B_3(51)=SIG_B(1)
DO ISTEP=l,lOO,l
PHIEND=PHI_M-FLOAT(ISTEP)*PHI M/100.0
CALL D1VPRK(IDO,NEQ,FCN3,PHI,PHIEND,TOL,PARAM,SIG_B)
IF (S1G_B(l) .GE.1000.0) THEN
WRITE (9,' (A40) ,) 'STIFF DATA CHOPPED 777.777'
GOTO 25
ENDIF
S1G_B_3(ISTEP+S1)=SIG_B(l)
ENDDO
* FINAL CALL TO RELEASE WORKSPACE
25 IDO=3
CALL D1VPRK(IDO,NEQ,FCN3,PHI,PHIEND,TOL,PARAM,SIG_B)
IDO=l
* CALL DE SOLVER FOR SECTION FOUR
PHI=O.O
SIG_B(l)=1.0
SIG_B_4(151)=1.0
DO ISTEP=1,100,1
PHIEND=FLOAT(ISTEP)*PHI M/100.0
CALL DIVPRK(IDO,NEQ,FCN4,PHI,PHIEND,TOL,PARAM,SIG_B)
IF (SIG_B(l) .GE.1000.0) THEN
WRITE (9,' (A40) ') 'STIFF DATA CHOPPED 777.777'
GOTO 30
ENDIF
SIG_B_4(151-ISTEP}=SIG_B(l)
ENDDO
* FINAL CALL TO RELEASE WORKSPACE
30 IDO=3
CALL DIVPRK(IDO,NEQ,FCN4,PHI,PHIEND,TOL,PARAM,SIG_B)
* FIND NEUTRAL POINT, ARRANGE DATA, & TRIM UNNECESSARY DATA
DO I=l,Sl,l
SIG_B_I (I) =SIG_B_1(I)
PHI I(I)=PHI_O-(PHI_O-PH1_M)/SO.O*FLOAT(I-1)
112
ENDDO
DO K=52,152,1
IF (SIG_B_4(K) .LE.SIG_B_3(K)) THEN
SIG_B_I(K)=(SIG_B_3(K-l)*SIG_B_4(K)-SIG_B_4(K-l)
& *SIG_B_3(K))/(SIG_B_4(K)-SIG_B_4(K-l)
& +SIG_B_3(K-l)-SIG_B_3(K))
PHI I(K)=PHI_M/IOO.O*(SIG_B_4(K)-SIG_B_3(K))
& /(SIG_B_4(K)-SIG_B_4(K-l)+SIG_B_3(K-1)
& -SIG_B_3(K))+PHI_M-PHI_M/100.0*FLOAT(K-51)
NP=K
DO J=K+1,152,1
SIG_B_I(J)=SIG_B_4(J-1)
PHI I(J)=PHI_M-PHI_M/100.0*FLOAT(J-52)
ENDDO
GOTO 50
ELSE
SIG B I(K)=SIG_B_3(K)
PHI I(K)=PHI_M-PHI_M/100.0*FLOAT(K-S1)
ENDIF
ENDDO
50 CONTINUE
* CALCULATE AREAS UNDER THE CURVE,
AREA 1 3=0.0
AREA 4=0.0
DO I=1,NP-1,1
AREA 1 3=AREA 1 3+0.5*(SIG_B_I(I+1)+SIG_B_I(I))
& *(PHI I(I)-PHI_I(I+l))
ENDDO
DO I=NP,151,1
AREA 4=AREA 4+0.5*(SIG_B_I(I+1)+SIG_B_I(I))
& *(PHI_I(I)-PHII(I+1))
ENDDO
* CALCULATE ROLL TORQUE AND FORCE
TORQUE=1/2.54*(AREA_l_3-AREA_4)*Y*MU*R**2/12.0
FORCE=1/2.54*(AREA_l_3+AREA_4)*Y*R
* OUTPUT RESULTS
WRITE (9,100) MU
100 FORMAT ('Coeff. of Friction ',F12.6)
WRITE (9,102) N
102 FORMAT ('Threads per Inch ',FI2.6)
WRITE (9,103) 1.0/N
103 FORMAT ('Thread Spacing ',F12.6,' in')
WRITE (9,104) R
113
104 FORMAT ('Roll Radius ',F12.6,' in')
WRITE (9,106) T_O
106 FORMAT ('Original Thickness ... ',F12.6,' in')
WRITE (9,108) T_F
108 FORMAT ('Final Thickness ',F12.6,' in')
WRITE (9,120) T_O-T_F
120 FORMAT (, Draft " F12 .6,' in')
WRITE (9,122) (T_O-0.5*T_F)/T_O*100.0
122 FORMAT ('Percent Reduction .... ',F12.6,' %')
WRITE (9,110) THETA
110 FORMAT ('Theta ',F12.6,' rad')
WRITE (9,112) PHI_O
112 FORMAT ('Beginning Phi ',F12.6,' rad')
WRITE (9,114) PHI_M
114 FORMAT ('Full-Thread Phi ',F12.6,' rad')
WRITE (9,115) SIG_B_I(51)
115 FORMAT ('Sig_b I Y at M ',F12.6)
WRITE (9,116) PHI I(NP)
116 FORMAT ('Neutral-Point Phi ',F12.6,' rad')
WRITE (9,118) SIG_B_I(NP)
118 FORMAT ('Sig_b I Y at NP ',F12.6)
WRITE (9,124) FORCE
124 FORMAT ('Roll Force ',F12.6,' lb')
WRITE (9,126) TORQUE
126 FORMAT ('Roll Torque ',F12.6,' ft-lb')
WRITE' (9, , (I) , )
IF (PLOT.EQ.1) THEN
WRITE (9,128)
128 FORMAT (4X, 'ISTEP' ,8X, 'PHI' ,9X, 'SIG_B/y')
DO K=l,I
WRITE (9,130) K,PHI_I(K) ,SIG B I(K)
130 FORMAT (I6,2(3X,F12.6))
ENDDO
WRITE (9,' (/) ')
ENDIF
GOTO 5
*998 WRITE (7,' (A40) ') 'Read I Write Error'
999 CLOSE (8)
CLOSE (9)
END
*
114
INTEGER NEQ
DOUBLEPRECISION PHI, SIG_B (NEQ) ,SIG_B_PRIME(NEQ)
DOUBLEPRECISION F,AREA,DAREA
DOUBLEPRECISION MU,N,T_O,T_F,R,THETA
COMMON/GLOBALS/MU,N, T_O,T_F,R, THETA
* FUNCTIONS FOR SECTION ONE
F=R/N*(SIN(PHI)-MU*COS(PHI)+(COS(PHI)+MU*SIN(PHI))*
& (COS(THETA)*SIN(PHI)*SQRT(l.O+(SIN(ATAN(O.S*TAN(PHI)/
& TAN (THETA) )) )**2+(SIN(ATAN(O.S*TAN(PHI))) )**2)
& -MU*SQRT((COS(THETA))**2+(SIN(THETA) )**2*(COS(PHI))**2))/
& (COS(THETA)*COS(PHI)*SQRT(l.O+(SIN(ATAN(O.S*TAN(PHI)/
& TAN(THETA))))**2+(SIN(ATAN(O.S*TAN(PHI))))**2)+MU*
& SIN(ATAN(O.S*TAN(PHI)))*SQRT((COS(THETA))**2+
& (SIN (THETA) ) *.*2* (COS (PHI) )**2)))
DAREA=R*SIN(PHI)*(1/N+(T_O-2.0*R*(1.O-COS(PHI)) )
& /(2.0*T_F*N))
AREA=1/N*(T_O/2.0+R*(1.O-COS(PHI)))
& -(T_O/2.0-R*(1.O-COS(PHI)))**2/(2.0*T_F*N)
SIG_B_PRIME(l)=(F-DAREA)/AREA*SIG_B(l)+DAREA/AREA
RETURN
E~
*
SUBROUTINE FCN3(NEQ,PHI,SIG_B,SIG_B_PRIME)
INTEGER NEQ
DOUBLEPRECISION ~HI,SIG_B(NEQ) , SIG_B_PRIME(NEQ)
DOUBLEPRECISION F,AREA,DAREA
DOUBLEPRECISION MU,N, T_O, T_F,R, THETA
COMMON/GLOBALS/MU,N,T_O, T_F,R, THETA
* FUNCTIONS FOR SECTION THREE
F=R/N*(SIN(PHI)-MU*COS(PHI)+(COS(PHI)+MU*SIN(PHI))*
& (COS(THETA)*SIN(PHI)-MU*COS(PHI)*SQRT((COS(THETA) )**2+
& (SIN(THETA))**2*(COS(PHI))**2))/
& (COS(THETA)*COS(PHI)+MU*SIN(PHI)*SQRT((COS(THETA))**2+
& (SIN(THETA))**2*(COS(PHI))**2)))
DAREA=2.0/N*R*SIN(PHI)
AREA=2.0/N*R*(1.O-COS(PHI))+T_F/N/2.0
SIG_B PRIME(l)=(F-DAREA)/AREA*SIG_B(l)+DAREA/AREA
RETURN
E~
*SUBROUTINE FCN4(NEQ,PHI,SIG_B,SIG_B_PRIME)
INTEGER NEQ
DOUBLEPRECISION PHI,SIG_B(NEQ) ,SIG_B_PRIME(NEQ)
115
DOUBLEPRECISION F,AREA,DAREA
DOUBLEPRECISION MU,N,T_O,T_F,R,THETA
COMMON/GLOBALS/MU,N,T_O,T_F,R,THETA
* FUNCTIONS FOR SECTION FOUR
F=R/N*(SIN(PHI)+MU*COS(PHI)+(COS(PHI)-MU*SIN(PHI))*
& (COS (THETA) *SIN(PHI) +MU*COS (PHI) *SQRT( (COS(THETA))**2+
& (SIN(THETA))**2*(COS(PHI))**2))/
& (COS(THETA)*COS(PHI)-MU*SIN(PHI)*SQRT( (COS(THETA))**2+
& (SIN(THETA))**2*(COS(PHI))**2)))
DAREA=2.0/N*R*SIN(PHI)
AREA=2.0/N*R*(1.O-COS(PHI))+T_F/2.0/N
SIG_B PRIME (1) =(F-DAREA)/AREA*SIG_B(l) +DAREA/AREA
RETURN
E~
116
117
REFERENCES
[1] C. Sevier, "Processing Recycled Plastics Into Fibers," Proceedings of the 1994-95 OUREProgram, University of Missouri-Rolla, 1995.
[2] D. R. Carroll, "Processing Recycled Plastic into Fibers for Reinforced Concrete," NSF ENG DMIIManufacturing Processes and Equipment Program proposal, University of Missouri-Rolla, 1996.
[3] S. P. Shah and B. V. Rangan, "Fiber Reinforced Concrete Properties," ACIJ, Vol. 68, No.2, p.126, 1971.
[4] Council on Tall Buildings and Urban Habitat - Committee 2ID, "Cast-in-Place Concrete in TallBuilding Design and Construction," McGraw-Hill, pp. 25-62, 1991.
[5] P. Soroushian and Z. Bayasi, "Fiber Type Effects on the Performance of Steel Fiber ReinforceConcrete," ACIMater. J, Vol. 88, No.2, p. 129, 1991.
[6] S. P. Sh~ et al., "Measurement of Properties of Fiber Reinforced Concrete," ACI Mater. J., Vol.85, No.6, p. 583, 1988.
[7] V. S. Gopalaratnam, et aI., "Fracture Toughness of Fiber Reinforced Concrete," ACIMater. J,Vol. 88, No.4, p. 339, 1991.
[8] A. S. Ezeldin and S. R. Lowe, "Mechanical Properties of Steel Fiber Reinforced Rapid SetMaterials," ACIMater. J, Vol. 88, No.4, p. 384, 1991.
[9] D. A. Fanella and A. E. Naaman, "Stress-Strain Properties of Fiber Reinforced Mortar inCompression," ACI Mater. J, Vol. 82, No.4, p. 475, 1985.
[10] V. S. Gopalaratnam and S. P. Shah, "Properties of Steel Fiber Reinforced Concrete Subjected toImpact Loading," ACIMater. J, Vol. 83, No.1, p. 117, 1986.
[11] R. J. Ward and V. C. Li, "Dependance of Flexural Behavior of Fiber Reinforced Mortar onMaterial Fracture Resistance and Beam Size," ACIMater. J, Vol. 87, No.6, p. 627,1990.
[12] A. Nanni, "Design for Torsion Using Steel Fiber Reinforced Concrete", ACIMater. J., Vol. 87,No.6, p. 556, 1990.
[13] C. D. Johnston and R. W. Zemp, "Flexural Fatigue Performance of Steel Fiber ReinforcedConcrete - Influence of Fiber Content, Aspect Ratio, and Type," ACIMater. J, Vol. 88, No.4,p. 374, 1991.
[14] K. Kosa and A. Naaman, "Corrosion of Steel Fiber Reinforced Concrete", ACIMater..J., Vol. 87,No.1, p. 27,1990.
[15] M. R. Nouri and J. Morshedian, "Tensile and Flexural Behavior of Fibre Reinforced CementitiousComposites,'!' Iranian J ofPolymer Sci., Vol. 4, No.1, p. 56, 1995.
[16] D. J. Hannant, "Fibre Reinforcement in the Cement and Concrete Industry," Mater. Sci. andTech., Vol. 11, No.9, p. 853,1995.
118
[17] E. tvt Mielnik, Metalworking Science and Engineering, McGraw-Hill., Ne\v York., NY, pp. 34377, 1991.
[18] E. G. Thomsen., et aI., Mechanics ofPlastic Deformation in Metal Processing, MacMillian, NewYork, NY, pp. 373-83, 1965.
[19] B. Avitzur, Handbook ofMetal-Forming Processes, John Wiley & Sons, New York, NY, pp.771-811,1983.
[20] S. Kalpakjian, Manufacturing Processes for Engineering Materials, 2nd ed., Addison-Wesley,Reading, MA, pp. 344-62, 1992.
[21] E. C. Larke and M. Cook, The Rolling ofStrip, Sheet, and Plate, Chapman and Hall, Norwich,Great Britain, pp. 186-207,261-94,1963.
[22] Th. von Karman, "Contribution to the Theory of Rolling," Z. Angew. Math. Mech., Vol. 5, No.2, pp. 139-41, 1925.
[23] R. B. Palme and C. W. MacGregor, Rolling ofMetals, ASME Research Committee on PlasticFlow of Metals, New York, NY, 1958.
[24] Z. Wusatowski, Fundamentals ofRolling, Pergamon Press, Oxford, Great Britain, 1969.
[25] J. Puppe and G. Stauber, Walzwerkswesen, Handbuch Des Eisenhuttenwesens, Zweiter Band,Berlin, Germany, 1934.
[26] Committee of Hot Rolled and Cold Finished Bar Producers, Cold Finished Steel Bar Handbook,American Iron and Steel Institute, New York, NY, 1968.
[27] R. C. Hibbeler, Mechanics ofMaterials , Macmillan, New York, NY, pp. 486-91, 1991.
[28] J. E. Shigley and C. R. Mischke, Mechanical Engineering Design, 5th ed., McGraw-Hill, NewYork, NY, pp. 325-33, 1989.
[29] IMSL, Inc., User's Manual, IMSL Math/Library, FORTRAN Subroutines for MathematicalApplications, Vol. 2, Version 1.1, Houston, TX, pp. 633-39, 1989.
[30] J. R. Dixon and C. Poli, Engineering Design and Design for Manufacturing, A StructuredApproach, Field Stone, Conway, MA, pp. 19-1 to 19-29, 1995.
[31] G. Taguchi and Y. Wu, Introduction to Off-Line Quality Control, American Supplier Institute,Center for Taguchi Methods, Dearborn, MI, 1980.
[32] G. Taguchi and S. Konishi, Orthogonal Arrays and Linear Graphs, American Supplier Institute,Dearborn, MI, 1987.
[33] R. C. Weast and M. J. Astle, CRC Handbook ofChemistry and Physics, 60th ed.,CRC Press,Boca Raton, FL, p. F-21, 1980.
119
VITA
Jeffery Scott Thomas was born on November 4, 1971 in Springfield, Missouri, as the first of four
sons to James and Kaye Thomas. A product of the Fair Play, Missouri, public school systems, he
received the George Washington Carver Award and the Presidential Academic Fitness Award and
graduated as salutatorian of his class in 1990.
He then attended Southwest Baptist University in Bolivar, Missouri, where he completed a pre
engineering curriculum. While at SBU, Thomas received the President's Scholarship and Dependent of
Minister Scholarship and was appointed as a Southwest Baptist University Scholar and a Polk County
Community Scholar.
He was married to Cherie Joann Thomas in May 1992, before transferring his studies to the
Department of Mechanical and Aerospace Engineering and Engineering Mechanics at the University of
Missouri-Rolla in Rolla, Missouri. As an undergraduate teaching assistant in the Department of Basic
Engineering, Thomas aided in the instruction of computer-aided drafting and various software packages
to college freshmen. As a member of the Opportunities for Undergraduate Research Experience
Program, he investigated possible uses for recycled paper products, particularly as a lumber substitute.
Thomas also received departmental scholarships funded by the Kaiser Aluminum Company and the
Robert F. Davidson Scholarship Fund and was appointed as a Curator's Scholar. He received a Bachelor
of Science degree in mechanical engineering from UMR in May 1995 and graduated magna cum laude.
He continued his stay at UMR to pursue graduate work in mechanical engineering, where he held
a Graduate Fellowship and a Chancellor's Fellowship. Additionally, as a graduate teaching assistant in
the Department of Basic Engineering, he taught undergraduate courses in engineering design with
computer applications.
Thomas' professional-society affiliations included ASME, SAE, NSPE, Order of the Engineer,
and the honor societies of Phi Kappa Phi and Tau Beta Pi. His pastimes included playing computer
games, painting, wood carving, writing poetry, tinkering with various mechanically-oriented projects,
and enjoying time with family and friends.