Analysis and Design of Hemp Fibre Decorticators

by

Jinke Xu

A Thesis submitted to the Faculty of Graduate Studies of

The University of Manitoba

in partial fulfilment of the requirements of the degree of

MASTER OF SCIENCE

Department of Biosystems Engineering

University of Manitoba

Winnipeg

Copyright © 2010 by Jinke Xu

 ACKNOWLEDGEMENTS

I would like to thank Dr. Ying Chen from the bottom of my heart. She is my

advisor, but most of the time, more like a close friend. She is the person who changed

my life. Her wisdom and attitude in scientific research was guiding me during my

master’s study, and will be my lifelong treasure. Her trust and encouragement gave

my great confidence to complete my research.

I would like to sincerely thank my co-advisor Dr. Qingjin Peng, for his

inspirational advice and guidance, and consistent support of my study. I would also

like to thank Dr. Claude Laguëand Dr. Wen Zhong for their advice and for serving as

my thesis committee members.

I would like to thank my colleagues in my research group. Special thanks are

given to Meghan Baker for her great help in experimental data. I also want to

appreciate the support from my friends, and the great time we had together during the

unforgettable two years.

I would like to dedicate my achievement to my parents: my father Jingde Xu and

my mother Yunqiu Wang. They are always the source of my courage and faith to keep

going forward.

i
 GENERAL ABSTRACT

Hemp (Cannabis sativa L.) fibre is a natural renewable material which has been

used in important areas closely related to people’s daily life. The increasing need of

hemp fibre of high quality requires better fibre processing methods, more advanced

facilities with higher machine performance and efficiency. Decortication is the key

procedure to extract hemp fibre, and it significantly affects the output fibre quality

and purity. The machine used for this process, known as a decorticator, needs to be

well designed. The energy requirement is worth being evaluated for decorticators.

This study consisted of two parts. In the first part, the specific energy of using a

hammer mill for decorticating hemp was examined. The experimental data (three

hammer mill screen scenarios and three feeding masses) were used to fit modified

size-reduction theories (Kick’s, Rittinger’s and Bond’s laws). The experimental data

were also used to develop a linear regression model to predict the specific energy

from the ratio of initial and final fibre lengths. Results showed that all modified laws

and the linear model performed equally well for specific energy prediction, and they

had better prediction accuracy at a higher feed rate.

In the second part of the study, integrated with virtual reality (VR) technology,

TRIZ (“Theory of Inventive Problem Solving” in Russian) method was used for

designing and evaluating a new hemp scutcher prototype in virtual environments. An

evaluation system was developed for making comparison of the new design and the

traditional scutchers. The new design is expected to have a better performance in

ii
terms of scale, product quality and energy efficiency. The TRIZ-VR integrated design

has great potential to be a fast, reliable and low-cost design trend.

iii
 TABLE OF CONTENTS

1. General Introduction ..................................................................1

1.1 INTRODUCTION ................................................................................................ 1
1.2 GENERAL OBJECTIVES ................................................................................... 3
1.3 THESIS STRUCTURE ........................................................................................ 3

2. Literature Review .......................................................................4

2.1 HEMP PLANT ..................................................................................................... 4
2.2 HEMP FIBRE PROCESSING ............................................................................. 6
2.2.1 Harvesting...................................................................................................... 6
2.2.2 Retting ........................................................................................................... 7
2.2.3 Baling ............................................................................................................ 8
2.2.4 Pre-cutting ..................................................................................................... 8
2.2.5 Decorticating ................................................................................................. 9
2.2.6 Cleaning ....................................................................................................... 10
2.3 DECORTICATION MACHINERY .................................................................. 11
2.3.1 Roll crusher ................................................................................................. 11
2.3.2 Hammer mill ................................................................................................ 14
2.3.3 Other machines used for decortication ........................................................ 17
2.4 FIBRE PROCESSING LINES ........................................................................... 17

3. Analysis of Energy Requirement for Hemp Fibre Decortication

Using a Hammer Mill ...................................................................20

3.1 ABSTRACT ....................................................................................................... 20
3.2 INTRODUCTION .............................................................................................. 21
3.3 MATERIALS AND METHODS ....................................................................... 26
3.3.1 Experiments ................................................................................................. 26
3.3.1.1 Hemp samples, equipment and experimental design ........................... 26
3.3.1.2 Specific energy measurement .............................................................. 30
iv
 3.3.1.3 Fibre length analysis ............................................................................ 33
3.3.2 Modified size reduction laws ....................................................................... 33
3.3.3 Development of a size reduction model ...................................................... 37
3.4 RESULTS AND DISCUSSION ........................................................................ 40
3.4.1 Fibre length analysis .................................................................................... 40
3.4.2 Specific energy requirement ........................................................................ 43
3.4.3 Fitting modified Kick’s law ......................................................................... 43
3.4.4 Fitting modified Rittinger’s law .................................................................. 48
3.4.5 Fitting modified Bond’s law ........................................................................ 52
3.4.6 Fitting a linear model................................................................................... 55
3.5 CONCLUSIONS ................................................................................................ 58
3.6 REFERENCES ................................................................................................... 59

4. Integration of TRIZ and Virtual Environments for Hemp Scutcher

Design ......................................................................................... 61
4.1 ABSTRACT ....................................................................................................... 61
4.2 INTRODUCTION .............................................................................................. 62
4.2.1 The TRIZ Theory......................................................................................... 64
4.2.2 TRIZ Applications ....................................................................................... 68
4.3 METHODOLOGY ............................................................................................. 70
4.3.1 TRIZ Design integrated with VE ................................................................ 70
4.3.2 Problem analysis and design criteria for hemp scutchers ............................ 72
4.4 PRELIMINARY DESIGN AND EVALUATION IN VE ................................. 77
4.5 VR MODELING AND PROTOTYPE EVALUATION ................................... 84
4.6 CONCLUSIONS ................................................................................................ 94
4.7 ACKNOWLEDGEMENTS ............................................................................... 94
4.8 REFERENCES ................................................................................................... 95

5. Summary ................................................................................. 97

Appendix ..................................................................................... 99

v
 LIST OF TABLES

Table 3-1. Nomenclature for deducing modified laws. ................................................ 35

Table 3-2. Summary of C, X, and b expressions in three modified laws. .................... 39
Table 3-3. Specific energy of decortication. ................................................................ 45
Table 3-4. Summary of CK and bK values. ................................................................... 47
Table 3-5. Summary of CR and bR values. ................................................................... 51
Table 3-6. Summary of CB and bB values. ................................................................... 54
Table 3-7. Summary of CL and bL values..................................................................... 57
Table 4-1. Selected TRIZ contradictions and solutions from TRIZ contradiction
matrix .......................................................................................................... 76

vi
 LIST OF FIGURES

Figure 2-1. Simplified scheme of a hemp stem cross section (not to scale) .................. 5
Figure 2-2. Working elements of the roller machine used for decortication
(Hobson et al. 2001) (A) pair of smooth-faced rollers; (B) three pairs
of fluted rollers follow; (C) shell feed; (D) feed pinned rotor; (E)
main pinned rotor. ..................................................................................... 13
Figure 2-3. Structure of the hammer mill used for decortication (Munder and
Fürll 2004). ............................................................................................... 16
Figure 2-4. Process flow sheet for processing hemp (Munder and Hempel 2004)...... 19
Figure 3-1. Retted hemp sample precut into segments. ............................................... 27
Figure 3-2. Decortication equipment (Baker 2010) (a) hammer mill; (b) swing
hammers. ................................................................................................... 28
Figure 3-3. Staggered perforated screen of the hammer mill (Baker 2010) (a)
opening size 19.28 mm; (b) opening size 25.74 mm. ............................... 29
Figure 3-4. Typical power-time curve; data from the treatment: 25.74 mm screen
with 200 g feeding mass. .......................................................................... 32
Figure 3-5. Output fibres in different length categories (a) 0-20 mm; (b) 20-40
mm; (c) 40-60 mm; (d) 60-80 mm;(e) 80-100 mm; (f) > 100 mm. .......... 41
Figure 3-6. Length distribution of different feeding masses (a) 200 g; (b) 125 g;
(c) 75 g. ..................................................................................................... 42
Figure 3-7. Linear regression for fitting modified Kick’s law (a) feeding mass 200
g; (b) feeding mass 125 g; (c) feeding mass 75 g; (d) combined all
feeding rates. ............................................................................................. 46
Figure 3-8. Linear regression for fitting modified Rittinger’s law (a) feeding mass
200 g; (b) feeding mass 125 g; (c) feeding mass 75 g; (d) combined
all feeding rates. ........................................................................................ 50
Figure 3-9. Linear regression for fitting modified Bond’s law (a) feeding mass
200 g; (b) feeding mass 125 g; (c) feeding mass 75 g; (d) combined
all feeding rates. ........................................................................................ 53
Figure 3-10. Regression for fitting linear model of specific energy and the ratio of
initial and average output fibre length (a) feeding mass 200 g; (b)
feeding mass 125 g; (c) feeding mass 75 g; (d) combined all feeding
rates. .......................................................................................................... 56
Figure 4-1. Flowchart of TRIZ in problem solving. .................................................... 67
Figure 4-2. Flowchart of TRIZ integrated with VE. .................................................... 71

vii
Figure 4-3. Basic principle of the scutching procedure (Hobson et al. 2001;
Patterson 1949). ........................................................................................ 73
Figure 4-4. Alternative schemes of roller layout (a) “S” shape; (b) “U” shape; (c)
“C” shape; (d) improved “C” shape.......................................................... 79
Figure 4-5. Housing design evolution. ......................................................................... 82
Figure 4-6. 2D CAD Model of the new design. ........................................................... 85
Figure 4-7. The framework of the VE. ......................................................................... 86
Figure 4-8. Main menu of the system. ......................................................................... 87
Figure 4-9. Route simulation examples inVE. ............................................................. 90
Figure 4-10. The models of the traditional scutchers................................................... 91
Figure 4-11. The model of the new design in VE (a) front view of the model with
a transparent cast; (b) side view of the model with a transparent cast;
(c) inner structure without cast. ................................................................ 92
Figure 4-12. KeyFrame Node to simulate material route. ........................................... 93

viii
 1. General Introduction

1.1 INTRODUCTION

Hemp plants have been cultivated since ancient time, benefiting human beings in

different aspects. The use of hemp as medicine can be traced back to around 2300 B.C.

in China. Hemp cloth is known as the oldest sample of cloth. Hemp cores (also named

as hurds) are able to be converted to paper products. Hemp seeds can be used for oil

extraction. With the improvement of processing techniques, hemp oil has become an

important original material for biodiesel production in recent years (Wikipedia Online

2009).

Modern industries require sustainable and ecological development in order to

face the challenge of environmental issues. “Green” production and natural materials

have gained more and more attention from people all over the world. Hemp fibre, as

an important natural material, has been studied and used in different areas. Hemp fibre

is stronger than flax and cotton fibre, and it is believed that hemp fibre may substitute

carbon fibre and glass fibre in composite manufacturing (Munder and Fürll 2004;

Mwaikambo 2006). Good thermal insulation property makes hemp fibre be widely

used as thermal insulating materials (Munder and Fürll 2004). Hemp fibre also plays

an important role in textile industry which is closely related with our daily life.

Fibre processing equipments and methods are worth being studied to elevate the

quality of fibres and meet various needs. Decortication which detaches fibres from

cores is an important procedure affecting fibre quality. The machine used to achieve

1
decortication is known as a decorticator. Two types of most commonly used

decorticators are hammer mill and roll crusher. Hammer mill has the advantage in

decorticating short fibres and its capacity is relatively high, while roll crusher is

capable of extracting long fibres (Munder and Fürll 2004; Hobson et al. 2000). In

some of the main producing areas, hemp fibre extraction is still a labor-intensive work.

Current technology of hemp fibre processing is not very efficient. For instance,

hammer mill processing has high energy requirement (Baker 2010), so it may

decrease the economical value of the fibre product. In order to select the suitable

decorticators according to their demand, users need to know the analytical data of

energy requirement to operate the decorticators in an efficient way.

Developing new high-standard decorticators is one of the main goals in

improving hemp fibre processing. Incorporating advanced designing methods and

tools, such as TRIZ (“Theory of Inventive Problem Solving” in Russian) method, may

bring novel ideas and prompt the development of fibre processing facilities. TRIZ is a

logic methodology which can be applied in different areas including machine design.

By analyzing the problems of the current decorticators, TRIZ can provide general

solutions to improve the function of the machine. The new designed prototype can be

evaluated in virtual environment instead of building a real model. This trend is able to

achieve the preliminary decorticator design and evaluation with low cost.

2
1.2 GENERAL OBJECTIVES

The general objective was to provide fundamental knowledge and tools for

developing hemp decortication machinery in the future. The specific objectives of this

study were: 1) to investigate the relationship between the specific energy requirement

of fibre decortication by hammer mill and the fibre length change; 2) to fit the

experimental data with the modified size reduction theories (Kick’s, Rittinger’s and

Bond’s laws); and 3) to develop a fast conceptual design trend incorporating TRIZ

design method with virtual reality (VR) technology in hemp decorticator (scutcher)

designing.

The results of specific energy requirement of hammer mill will be helpful to

estimate the energy consumption in using or designing hemp fibre decorticators. The

TRIZ/VR method has the potential be used as a tool for modifying and improving the

hemp processing line in the future.

1.3 THESIS STRUCTURE

This thesis has been structured around two stand-alone engineering papers.

Chapter 1 includes the general introduction and objectives of the study. Chapter 2

consists of general literature review on hemp processing and machinery. Chapter 3

and Chapter 4 focus on energy requirement of hammer mill decortication and TRIZ

design of hemp scutcher in virtual environment respectively. Chapter 5 summarizes

the results and conclusions. References (for Chapter 1, Chapter 2 and Chapter 5) are

presented in appendix.

3
 2. Literature Review

2.1 HEMP PLANT

Hemp (Cannabis sativa L.) is the common name for plants of the entire genus

Cannabis. Cannabis sativa L. is often referred to as true hemp to distinguish it from

other fibre crops. Hemp is frequently confused with marijuana, because of their

similar leaf shapes. THC (delta-9-tetrahydrocannabinol) is the active ingredient in

marijuana as well as a material for making drugs. However, hemp contains virtually

no THC (less than 0.3%), while the content of THC in marijuana can reach from 6 or

7% up to 20%. Therefore hemp cannot be used as a drug (Wikipedia Online 2009).

The innermost layer of hemp plant is the pith, surrounded by woody material

known as core. The separated core is also called as “hurd” in some papers (Declerck

et al. 2008). Outside of this layer is the growing tissue which develops into core on

the inside and into the bast fibres on the outside (Beckermann 2007). Figure 2-1

shows the non-scaled simplified scheme of a hemp stem cross section.

Hemp can be grown on a range of soils, but tends to grow best on land that

produces high yields of corn. The soil must be well drained, rich in nitrogen, and

non-acidic. Hemp grows fast and it requires limited pesticides and attracts few pests.

In northern latitudes, hemp is usually planted between early March and late May.

Hemp averages between 2-4 m in height in about four months of growth. The stem

diameter can range from 6-16 mm in the field according to a research conducted by

Chen et al. (2004).

4
 Epidermis

Bast fibre

Hurd

Hollow pith

Fig. 2-1. Simplified scheme of a hemp stem cross section (not to scale).

5
 Hemp is sensitive to day length. The plant matures (sets seed) as days get shorter

in the fall. The seeds of hemp are smooth. The seeds usually contain from 29-34% oil.

The composition of the oil mainly consists of three fatty acids: linoleic (54-60%),

linolenic (15-20%), and oleic (11-13%) (Wikipedia Online 2009). There is a wide

range of applications on both the fibre and seed.

2.2 HEMP FIBRE PROCESSING

Basically, processing hemp for the purpose of fibre needs to follow the procedures

of harvesting, retting, baling, pre-cutting, decorticating and cleaning. Retting in the

field helps later decorticating. But recent research has proved that “green

decortication” without retting is also acceptable for some industrial applications

(Munder and Fürll 2004; Hobson et al. 2001; Riddlestone et al. 2006). However,

decortication of green hemp required more power and energy than that of retted hemp,

and decortication of green hemp was less effective when a hammer mill was used,

when compared to decortication of retted hemp (Baker 2010).

2.2.1 Harvesting

Hemp crops are harvested at different times for different hemp products. When

grown for textile fibre, the crop should be harvested when the fibre is at its highest

quality. The best time to harvest stalks for high quality primary fibre is when crop is

in flower. Harvesting for seed production and stalks occurs 4-6 weeks after flowering,

when male plants begin to shed pollen. Hemp harvest is usually performed with the

use of general-purpose agricultural harvesting machines such as haybines and

6
combines. Because of the large biomass, the whole plant cannot be fed into the

harvester such as a conventional combine machine (Chen and Liu 2003). For the

hemp cultivated for single-purpose (fibre) only, haybines can be used for swathing.

For the hemp cultivated for dual-purpose (seed and fibre), a windrower has been

developed by Chen and Liu (2003) for swathing seed and fibre separately.

2.2.2 Retting

For high-quality applications, the bast fibres should be separated from the retted

stalks. Munder and Fürll (2004) reported that the tensile strength and modulus of

elasticity of hemp fibres from retted plants were higher compared with the ones from

unretted hemp. Retting can also greatly improve the fineness and the stability of the

hemp fibres (Munder and Hempel 2004).

Traditionally, the hemp stalks are laid in the field to dry for a couple of days after

swathing. This kind of retting is called “dew retting”. The combined action of bacteria,

sun, air, and dew produces fermentation, dissolving much of the stem material

surrounding the fibre bundles. This type of retting is highly depended on the climate

of the location and the weather situation. Over retting may destroy the structure of the

stems and decrease the mechanical properties of the fibres (Akin et al. 2000). Though

varying weather conditions affect the quality of fibres, retting process in the field is

still commonly used especially in European countries because it is inexpensive and

environment friendly.

Instead of leaving the stalks in the field, another retting method is to use water.

Water retting provides more uniform and high-quality fibres. Currently, most hemp

7
fibre used in textiles in China is water retted. Stems can be immersed in rivers or

tanks. Hot water is sometimes used for special fibre quality requirements.

In recent years, chemical retting and enzymatic retting methods have been

developed by some researchers. Akin et al. (2005) compared the effect of a pilot plant

for processing both dew-retted and enzyme-retted flax. The enzyme retting followed

the previous method reported by Akin et al. (2000). The results showed that

dew-retted flax fibres had higher strength than enzyme-retted ones. However, enzyme

retting provided better fineness and higher fine fibre yield.

2.2.3 Baling

Baling is the process which gathers the harvested individual hemp plants into

bales. It can be done by most types of balers. The straws of flax and linseed are

shorter and are usually formed into rectangular bales. Different from that, the hemp

straws are better to be collected into round bales. Round bales have larger surface area

than square bales with the same volume, and they are easier to release moisture and

are able to keep the bales dry during storage. Bales are then sent to the central unit for

further processing.

2.2.4 Pre-cutting

Hemp stalk length can be up to 3 m. Such long straw may cause wrapping of

machine rotating parts in the following decorticating and cleaning procedures. Thus,

pre-cutting may be necessary before processing for fibre. Hemp was cut into a length

of about 300-400 mm previously by Munder and Hempel (2004). A commercial bast

fibre processing line designed by Van Dommele Engineering (VDE) can accept the

8
maximum straw length of 600 mm (Declerck et al. 2008). In order to protect the

machine, the optimum fibre length for decortication was selected from 80-200 mm by

Munder et al. (2003). Pre-cutting length of hemp straws also affects the average

product length. The results have shown that longer cutting length can produce longer

fibre at the end of the process. Baker et al. (2009) processed the feed hemp stock with

a pre-cutting length of 40 mm in a planetary mono mill because this length was

considered as the most suitable length for the production of biocomposties.

For different agricultural applications, some machines have been used to cut

cereal straws and hays, such as linear knife grid, bale slicer, tub grinder and shredder

(Igathinathane et al. 2006). Many of them are either length-uncontrolled or energy

consuming, and they are not recommended to be used for industrial hemp pre-cutting.

Rotary cutter is more energy saving than linear cutter (Persson 1987). Munder et al.

(2003) used a driven knife disk co-operated with a bale holding and turning device to

pre-cut the straws into 400 mm. A commercial hemp fibre processing line called

Temafa GmbH Lin Line has a bale opener which is able to treat round as well as

rectangular bales (Temafa GmbH 2004).

2.2.5 Decorticating

Decorticating is the most important procedure of hemp fibre processing. An

effective decortication can make the following cleaning process simpler, and can

elevate the fibre quality significantly. Generally, the machine used for decortication is

called “decorticator” by researchers. One important principle on using the decorticator

is that the process should not adversely influence the mechanical properties of the

9
fibre, so that the design of the machines and the choice of operational parameters

should be considered before the process. Decortication machinery is discussed in

details in the following sections.

With the increasing need for long fibres and the development of decorticators, the

cutting procedure has been eliminated in some processing methods. Most of these

decorticators are roll crushers, which have a horizontal path for straws passing

through (Hobson et al. 2001; Booth et al. 2004; Akin et al. 2005). Such methods treat

the whole plants in the format of bundles without cutting, so that longer fibres can be

separated from the straws. Long fibres are useful for textile purpose and mat

manufacturing (Munder and Hempel 2004).

2.2.6 Cleaning

The decortication procedure can separate most of the fibres from the cores. But

there are still free cores and cores attached to the fibres which need further cleaning.

Multiple ultra cleaner was applied in a laboratory line developed by Munder and

Hempel (2004). Free cores may be sieved out using vibratory screens or rotary

screens. Cores which are attached to the fibres are more difficult to clean. Hobson et

al. (2000) used high speed pinned rotor to purify the fibre ejected from the rollers.

Akin et al. (2005) reported the use of scutching wheel to stoke flax to shorten and

clean fibres. The scutching wheel had a diameter of 1168 mm and a rotational speed

of 263 rpm. Lin-Comb Shakers was used for separating loose shives in a commercial

line (Temafa GmbH 2004).

10
2.3 DECORTICATION MACHINERY

2.3.1 Roll crusher

One commonly used decorticator is roll crusher. A simple roll crusher consists of

a pair of rolls rotating in opposite directions. Material is fed in between the rolls and

crushed by forces of the rolls. If the rolls rotate at the same speed, compression is the

primary force. Shearing and compression are the primary forces on the materials

when the rolls rotate at different speeds. If the rolls are grooved, a tearing or grinding

action is introduced. Roll crushers are mainly used for ore and rock comminution in

the industrial sectors such as mineral processing, chemistry, cement, and building

material. Currently, the commercial machines have different scales to meet the

different demands of customers.

Roll crushers have the potential to be used as the decorticators to separate fibres

and cores. An example of a roll crusher model is shown in Fig. 2-2. Cores of the bast

fibre plants are brittle and can be easily broken by pressure. The roll crusher normally

has no effect on the length of fibres. Industrial flax manufacturing applies fluted rolls

to crush and break the woody core into short pieces and then separates them from the

remained fibres. This procedure is generally called as “scutching” in flax industry. In

fact, the designing concept is similar to the roll crushers. Some roll crushers are also

called as scutchers.

When designing the parameters of a roll crusher, the ratio of length to diameter of

the rollers can reach to 4:1 (Feedmachinery 2009). Many lab-scaled machines were

designed based on the structure of roll crushers to separate fibre from core of hemp

11
and flax. Hobson et al. (2001) developed a modified lab-scale flax decorticator to

separate fibres from unretted hemp straws. The decorticator had four pairs of rollers

(Fig. 2-2). Akin et al. (2005) used a pilot plant called Flax-PP to process flax. The line

consisted of scutching roller combines. The sequence of the elements in the line was:

9-roller calender, top shaker, scotching wheel, top shaker, 5-roller calender and top

shaker. The component called scutching wheel removed the shives greatly for unretted

Neche and dew-retted Natasja, but was less effective on enzyme-retted Jordan. A lab

decorticator “Bahmer-Flaksy” was used by Amaducci et al. (2008) to scutch the hemp

stalks, in order to investigate the decorticability of hemp stems obtained at different

plant densities and harvesting times. The hemp bundles were repeatedly put through

the decorticator six times for each test. The results showed that scutching was more

efficient in the lower part of the stems and long stems. Low plant population was

suitable to reduce the loss of smaller and thinner plants during processing. Kovur et al.

(2008) also used lab-scale Bahmer-Flaksy decorticator to extract hemp fibres. The

pressure of the breaking rolls was 10 N, with the linear feeding speed of 13 m/min.

12
Fig. 2-2. Working elements of the roller machine used for decortication
(Hobson et al. 2001).
(A) pair of smooth-faced rollers; (B) three pairs of fluted rollers follow;
(C) shell feed; (D) feed pinned rotor; (E) main pinned rotor.

13
2.3.2 Hammer mill

Another type of fibre decorticator is hammer mill. Hammer mills are widely used

in size reduction processes such as flour grinding and biomass industry. Hammer

mills have large capacity and high energy efficiency compared with other size

reduction equipments (Miu et al. 2006). The basic structure of a hammer mill is that a

steel drum with a vertical or horizontal cross-shaped rotor in it (Fig. 2-3). The purpose

of hammer mills is to shred material into small particles. Pivoting hammers are

mounted on the rotor. The bottom mounted screen controls the product size. Hammers

used inside the hammer mill can have a huge range of configurations, shapes, facings,

and materials. Sometimes a hammer will be designed to have two holes. When one

end is worn, the hammer can be used if sides are switched. The screen design should

allow the greatest amount of open area. Holes will be aligned in a 60-degree staggered

pattern.

When the hammers rotate with the rotor, the ends of them impact on the material

and shred them into smaller particles until the particle size is smaller than the

screen-hole-size. Not like roll crushers, hammer mills can grind fibrous materials. In

such case, a screen with cutting edges will be employed. The design of the hammers

attached to the rotor should meet the requirement of maximum contact with the feed

ingredient. Hammers should be balanced and not trail each other. The common tip

speeds for hammer mills are 5000-7000 m/min, and usually 8500-10000 rpm.

Munder et al. (2003) developed a pilot plant for hemp, flax, and oil seed linen

with the co-operation with the Kranemann Gartenbaumaschinen GmbH. The pilot

14
plant can achieve a capacity of 3 t/h of straw. The decorticator used in this plant was a

modified swing-hammer mill. Figure 2-3 shows the structure of the hammer mill

(Munder and Fürll 2004). The rebound stress of the hammers on the straw stems

separated the fibre from the brittle core. During the decortication process, about 45%

of the shives arose at the decorticator. The final fibre purity of the retted hemp could

reach to over 99%.

15
 Fig. 2-3. Structure of the hammer mill used for decortication
(Munder and Fürll 2004).

⃰ This figure was originally published in ASAE Paper Number 046091. Reprinted with

permission of the American Society of Agricultural and Biological Engineers, St.

Joseph Michigan.

16
2.3.3 Other machines used for decortication

Besides roll crushers and hammer mills, researchers also reported fibre

decortication using other machines. Gratton and Chen (2004) applied modified

cutterhead and scutching bars to decorticate hemp feedstock. The maximum fibre

yield and purity were 61% and 52% respectively. Anthony (2002) used different types

of cotton gin machines to separate fibre from seed flax straw. The results indicated

that the range of the content of pure fibre was from 7.1% to 12.8%. Also, the author

concluded that the raw material had to be retted before decortication. Baker et al.

(2009) conducted a series of hemp decortication experiments using a lab-scale

planetary mono ball mill. The results showed that higher grinding speeds and longer

grinding time could separate more fibres from the cores.

2.4 FIBRE PROCESSING LINES

Some fibre processing lines have been constructed for laboratory tests or industry

manufacturing. A flax fibre pilot plant was developed by USDA and Czech Flax

Machinery (Akin et al. 2005). This line had good product quality and great

commercial potential. A commercial “Lin Line” developed by Temafa GmbH had a

capacity of 2 t/h, and was able to process both hemp and flax straws (Temafa GmbH

2004). A successful commercial line built by VDE had the capability to handle 6.6 t/h

straw bales and technically get 1.92 t/h maximum output fibres (Declerck et al. 2008).

In summary, the hemp fibre decortication consists of field handling and central

unit processing. The field handling includes harvesting, retting, and baling.

17
Mechanisms of pre-cutting, decortication, post-decortication are the main components

in central unit line. Other components such as metal detector, conveyor, and output

fibre baler may be varied in different lines. Munder and Hempel (2004) reported the

use of a pilot-scale fibre processing line which consisted of all necessary functions

from bale opening to product packaging (Fig. 2-4).

18
Fig. 2-4. Process flow sheet for processing hemp
(Munder and Hempel 2004).

19
 3. Analysis of Energy Requirement for Hemp
Fibre Decortication Using a Hammer Mill

3.1 ABSTRACT

Hemp fibre decortication is an important procedure in hemp fibre processing. The

cost of energy during decortication affects the commercial value of hemp fibres for

industrial production. This study investigated specific energy requirement for hemp

decortication using a hammer mill, and the length distribution of output fibre based on

the tests with three screen opening sizes (19.28 mm, 25.74 mm, and no-screen

scenario) and three feeding masses (200 g, 125 g, and 75 g). Results showed that 200

g feeding mass and small screen-hole size required the highest specific energy (73.84

J/g). Screen size affected the length distribution, and for all three feeding mass

scenarios, more short fibres were produced when small size screen was used.

Modified Kick’s, Rittinger’s and Bond’s laws and a linear model were developed to fit

the experimental data in regard to relating specific energy with initial and final fibre

lengths. Results showed that all modified laws and the linear model performed

equally well for specific energy prediction, and they had the best prediction accuracy

at the highest feeding rate. Keywords: hemp, decortication, fibre, energy, hammer

mill.

20
3.2 INTRODUCTION

Hemp (Cannabis sativa L.) fibre is a renewable and environmentally friendly

natural material. The material has been used for different applications since early ages

because of its outstanding mechanical and physical properties, such as high tensile

strength and good thermal insulation character (Kozowski 2000). Because of the

advantages, it becomes an important industrial starting material applied in textile,

pulp and paper manufacturing, biocomposites, automobile accessories, and thermal

insulation materials (Fortenbery and Bennett 2004; Van Wyk 2007). Hemp fibre has

the potential to substitute carbon fibre and glass fibre (Munder and Fürll 2004).

A hemp stem basically consists of the outer layer of fibre and inner layer of

hollow cores. To extract the fibres from the fibre-core attachment, decortication is

required. Decortication is the process that mechanically separates fibre and core. The

equipment used for decortication is called a decorticator. Various types of equipment

have been used to decorticate hemp. Hobson et al. (2000) modified a roll crusher

(scutcher) which was originally for flax and retted hemp decortication to extract long

fibre bundles from unretted hemp. The roll crusher had three pairs of fluted rollers

running at different speeds, followed by a high speed pinned rotor to remove the

remaining cores attached to the fibres. The fibre yield reached up to 30% by mass.

Gratton and Chen (2004) developed a field-going unit with a cylindrical cutterhead to

provide fibres for non-textile applications. The cutterhead from a forage harvester was

modified and installed right after the pick-up entrance, and it shortened the cores by

breaking and scutching of the hemp stalks. Longer fibres and shorter cores were then

21
delivered to a straw walker for cleaning. A processing line which featured a modified

hammer mill as the decorticator was developed by Munder and Fürll (2004) for fibre

production. They obtained a good decorticating effect from the impact stress of the

hammers on the hemp straws and gained the fibre yield of 22-24%. This mechanism

also worked well with the input materials of flax and linseed. Baker (2009)

investigated the effect of low moisture (9.7%) hemp decortication using a lab-scale

planetary mono ball mill. The results showed that grind speed and duration affected

the properties and purity of the outputs. The combination of 200 rpm with 4-min

duration gave the greatest long fibre yield of 51%.

Among the above mentioned equipments, hammer mill and roll crusher are most

commonly used for commercial fibre decortication. For decortication purpose,

hammer mills are more suitable for short fibre production which could be used in

industrial composites or pulp manufacturing, while roll crushers have the potential to

provide long fibres for textile products. A hammer mill requires relatively low initial

and maintenance cost, and can work with large capacity of various materials.

Hammer mills are generally serving as size reduction machine in a wide range of

applications such as biomass handling and flour grinding (Mani et al. 2003; Fang et al.

1998). Using hammer mill as a decorticator could achieve the separation of fibres and

cores, but at the same time, both fibres and cores are shortened. Hammer mill is an

energy consuming device with lower energy efficiency (Lobo 2002; Baker 2010). For

this reason, energy requirement of hammer mill has been studied by many researchers

in different application areas. The major energy input of hammer milling is consumed

22
on size reduction of the feed. The factors affecting the energy requirement include

both material properties of the feed and machine geometrical parameters.

Normally, specific energy is used to assess the energy consumption of hammer

mill. Specific energy is defined as the energy consumed per unit feeding mass.

Vigneault et al. (1992) compared the specific energy, grinding rate, particle size, and

hammer lifetime using a hammer mill with two hammer thickness scenarios: 3.2 mm

and 6.4 mm. The average specific energy for thin hammer tests was 10.2 kW h t-1,

which was 13.6% less than that of thick hammer (11.8 kW h t-1). The grinding rate

was higher for thin hammer scenario. A relationship of wheat moisture, debranning

rate, screen size and the energy requirement of grinding was investigated by Dziki and

Laskowski (2005) based on the experiments using a laboratory hammer mill. The

hammer mill was equipped with a current recorder to calculate the electrical energy

during the grinding process. A multiple linear regression was built to relate specific

energy with the variables of grain moisture, mass of grain, and diameter of screen

openings. They found that higher debranning degree caused a decrease of specific

energy, and more specific energy was required for higher moisture grain. Smaller

screen size led to higher specific energy requirement. Similar results in terms of

screen size were also reported by Arthur et al. (1982) and Mani et al. (2004). Mani et

al. (2004) compared the specific energy of grinding wheat and barley straws, corn

stover and switchgrass by a hammer mill with 22 swinging hammers. A negative

correlation between the specific energy consumption and the hammer mill screen size

was found for each material with the moisture content of 8% (w.b.) or lower. Moisture

23
content affected the specific energy at the same time, that the higher the moisture

content, the higher was the specific energy. Ghorbani et al. (2010) tested the specific

energy of reducing the size of alfalfa chops on a hammer mill. The average specific

energy ranged from 5.65 kJ kg-1 to 30.51 kJ kg-1 over different hammer mill screen

sizes and initial material sizes. Holtzapple et al. (1989) reported that a two-stage

hammer mill required 0.18 MJ kg-1 electrical energy for reducing initial wood chip

length of 25.4 mm into final length of 2 mm.

Researchers have presented different models to simulate the energy requirement

of hammer mill comminution (Austin 2004; Shi et al. 2003). Theories were developed

relating the specific energy of size reduction to the initial and final sizes of the

material. Among these theories, Kick’s, Rittinger’s and Bond’s size reduction laws are

most famous and widely accepted (Lowrison 1974; Henderson and Perry 1976; Pfost

and Swinehart 1970). Their theories are presented as the following equations:
L1
EK = CK × ln 3-1
L2
1 1
ER = CR × ln - 3-2
L2 L1
10 10
EB = CB × ln - 3-3
L2 L1

where EK, ER and EB stand for the specific energy of each law. L1 and L2 are the initial

size and the final size of the material respectively. CK, CR and CB are constant factors.

If the constant factors are known, specific energy requirement can be estimated

based on the initial and final material sizes. For example, Ghorbani et al. (2010)

calculated the constant numbers in Kick’s, Rittinger’s and Bond’s laws respectively

according to the regression of the experimental data in alfalfa comminution by a

24
hammer mill. Rittinger’s law fitted the data best.

In summary, when hammer mill is used as a decorticator, energy requirement is

an important factor to be considered. Little research has been done in the aspect of

specific energy of decortication. No literature was found in relating specific energy to

hemp fibre lengths before and after decortication. The objectives of this study were: 1)

to theoretically modify Kick’s, Rittinger’s and Bond’s laws and relate the specific

energy with the initial and final fibre lengths; 2) to fit the modified laws with the

experimental data to determine the constant factors; and 3) to develop a regression

model of specific energy and the ratio of initial and final fibre lengths based on the

experimental data.

25
3.3 MATERIALS AND METHODS

3.3.1 Experiments

3.3.1.1 Hemp samples, equipment and experimental design

The experiment was carried out as a part of another study by Baker (2010) on

hemp fibre decortication using a hammer mill. Details on hemp samples, equipment

(hammer mill) and experimental design have been described in Baker (2010). Here

these are briefly summarized. The hemp samples (variety: USO 31) were from a farm

in western Manitoba (Canada). The hemp stalks were retted in field for six weeks and

then baled and stored outside. The moisture content of the samples was 9.7%. Before

being loaded into the hammer mill, hemp samples were cut into segments using a

heavy-duty paper cut in order to prevent wrapping in hammer mill (Fig. 3-1).

A swing hammer mill (Model E221.5 – TFA, Bliss Eliminator, Oklahoma, U.S.A)

was used as the decorticator to decorticate the hemp samples (Fig. 3-2). The hammer

mill had 24 swinging hammers symmetrically installed on the rotor. The scale of each

hammer was 175 × 50 × 7 mm. Hammer mill was installed with staggered perforated

plate screens with round openings. The hammer rotor rotating at a speed of 3600 rpm

was driven by a 22 kW motor.

The treatments of the decortication experiments were: three screen scenarios

19.28 mm (Fig. 3-3a), 25.74 mm (Fig. 3-3b), and no screen; three feeding masses 75 g,

125 g, and 200 g. Each treatment was conducted with four replications, which gave a

total of 36 tests (3 screen scenarios × 3 feeding masses × 4 replications).

26
Fig. 3-1. Retted hemp sample precut into segments.

27
 (a) (b)

Fig. 3-2. Decortication equipment (Baker 2010).

(a) hammer mill; (b) swing hammers.

28
 (a) (b)

Fig. 3-3. Staggered perforated screen of the hammer mill (Baker 2010).
(a) opening size 19.28 mm; (b) opening size 25.74 mm.

29
3.3.1.2 Specific energy measurement

A sensor kit was linked to the hammer mill, transmitting data including the motor

torque, speed, power, current, and voltage to a laptop computer every 0.1 second.

9000XDrive Software was applied to record data. The power data recorded from the

sensors were translated and plotted into power-time curves (Fig. 3-4). The ascent of

the bell shape curve was because of the increasing power during decortication. The

power value decreased as the decortication completed gradually. The decortication

process ended within approximately 5 seconds. For each power-time curve, 5-seconds

width of the curve containing the peak was selected to calculate the energy. The

following equation was used to evaluate total decortication energy input during the 5

seconds:
y2 + y1 y + y2 y + yn-1
Etotal = ×∆t + 3 ×∆t + ⋯ + n ×∆t (3-4)
2 2 2
where:

Etotal= total energy (J);

y1= the power value of the first data point (W);

y2= the power value of the second data point (W);

yn= the power value of the nth data point (W); and

Δt= time interval (s).

The horizontal line before the bell shape curve shown in Fig. 3-4 represented for

the no-load power, which was 594 W. This no-load power was translated to a no-load

energy of 2970 J (for 5 seconds). To determine the net energy, the no-load energy was

deducted from the total energy obtained from the above equation. The specific energy

30
was calculated as dividing net energy by the feeding mass (Eq. 3-5).
Enet
ES = (3-5)
M
where:

ES= specific energy (J/g);

Enet= net energy (J); and

M= feeding mass (g).

31
 12600

10600
Power (W)

8600

6600

4600

2600
46 47 48 49 50 51
Time (s)

Fig. 3-4. Typical power-time curve; data from the treatment: 25.74 mm screen
with 200 g feeding mass.

32
3.3.1.3 Fibre length analysis

To determine the initial fibre length, ten fibres from the initial sample without

decortication were randomly chosen and measured. As the hemp fibres were always

tangled, the length distribution of the output hemp fibres was done manually. After

each decortication test, the fibres from the output (mixture of fibres and cores) were

collected, and the final length was characterized with six length categories, and they

were used: 0-20 mm, 20-40 mm, 40-60 mm, 60-80 mm, 80-100 mm, and longer than

100 mm. Each fraction was weighed by Mars digital scale (MS200, North York,

Canada) and the proportion value of each fraction was recorded for fitting energy

models as described below.

3.3.2 Modified size reduction laws

Kick’s, Rittinger’s and Bond’s size reduction laws can achieve good simulation

when the material consists of single component. However, the outputs from the

hammer mill consisted of two components: fibres and cores, and the properties of

these two components were quite different. The three existing laws could not be used

directly. A theoretical modification was made to simply relate the specific energy with

the initial and final lengths of fibres without considering the sizes of the cores. This

was because most cores were rejected (commonly by air flow) to a waste stream

during decortication process and their length distribution cannot be quantified.

To describe the procedure of the modification, a list of nomenclature parameters

is shown in Table 3-1. The initial lengths of fibre and core were assumed to be the

same controlled by the pre-cutting procedure. The ratio of final core and fibre lengths

33
was considered as constant due to the assumption that the ratio was one inherent

character of the hemp samples. The proportions of core and fibre contents were

assumed to be uniform of all samples. These assumptions are:

1) the initial length of the core L1c equals to the initial length of the fibre L1f ;

2) the relation between final length of the core and fibre is L2c = r × L2f , where r

is a constant;

3) the proportion of core μc and the proportion of fibre μf are constants.

The total specific energy could be simplified in two parts: specific energy of fibre

and specific energy of core. Thus:

EK = μc Ec + μf Ef (3-6)

34
 Table 3-1. Nomenclature for deducing modified laws.

Nomenclature
L1c initial core length (mm) E total specific energy (J/g)
L1f initial fibre length (mm) Ec specific energy of core (J/g)
L2c final core length (mm) Ef specific energy of fibre (J/g)
L2f final core length (mm) Cc constant factor for core
μc mass proportion of core fraction Cf constant factor for fibre
μf mass proportion of fibre fraction r, C, b constants

⃰ Subscripts K, R, B and L stand for Kick, Rittinger, Bond, and the linear regression
model developed respectively.

35
 Taking the fibre and core fractions separately, the original Kick’s law (Eq. 3-1)

would be expressed as:

L1c
Ec = CKc × ln (3-7)
L2c
L1f
Ef = CKf × ln (3-8)
L2f
Substituting L1c and L2c by the terms of L1f and L2f, the relationship could be

presented as:
L1f L1f
EK = μc ∙ CKc ∙ ln + μf ∙ CKf ∙ ln
r ∙L2f L2f
L1f 1
= μc ∙ CKc + μf ∙ CKf ln + μc ∙ CKc ∙ ln (3-9)
L2f r
The final fibre length might vary in a large range. For instance, in this study, the

final fibre length varied from shorter than 20 mm to longer than 100 mm. To consider

the length distribution of the final fibre, EK could be calculated as:

n
L1f 1
EK = μc ∙ CKc + μf ∙ CKf × ln δi + μc ∙ CKc ∙ ln (3-10)
L2fi r
i=1

where δi is the mass proportion of the ith length category, n is the number of length

categories, and L2fi (mm) is average output fibre length of the ith length category. The

Eq (3-10) is rewritten in the following general form:

EK = CK ∙ XK + bK (3-11)

where CK is a constant factor to be determined by the experimental data. bK (J/g) is a

L1f
i=1 δi L
n
constant term according to the assumptions. XK represents ln , and it was
2fi

the independent variable in the modified Kick’s law. Equation 3-11 was the modified

Kick’s law for assuming hemp decortication specific energy. This modified law

related the total specific energy with the initial and final length of the fibre fraction

only.
36
 Similarly, the modified Rittinger’s and Bond’s laws are as follows:
n
μc ∙CRc 1 1 μc ∙CRc μc ∙CRc
ER = + μf ∙CRf ∙ δi - + - (3-12)
r L2fi L1f L1f ∙r L1f
i=1

ER = CR ∙ XR + bR (3-13)
n
μc ∙CBc 10 10 10∙μc ∙CBc 10∙μc ∙CBc
EB = + μf ∙CBf ∙ δi – + – (3-14)
r i=1
L2fi L1f L1f ∙ r L1f

EB = CB ∙ XB + bB (3-15)

Variables bR and bB (J/g) were coefficient factors related to the initial fibre length.

In this study, the initial fibre length was considered as unchanged. The expressions of

C, X, and b for the three modified laws are summarized in Table 3-2. To determine the

unknown factors in the modified models, linear regression was conducted based on

the experimental data of specific energy and length distribution.

3.3.3 Development of a size reduction model

A linear model was developed relating the specific energy of decortication and

the ratio of initial and final fibre length. For simplicity, the average final fibre

length L2f (mm) was used and it was calculated by the following equation:
n

L2f = δi ∙ L2fi (3-16)

i=1

The slope of the regression equations CL (J/g) represents the constant factor in the

relationship (equivalent to CK, CR and CB). A constant term bL was included. The

relationship could be expressed as:

L1f
EL = CL ∙ + bL
L2f
= CL ∙ XL + bL (3-17)

37
 L1f
where XL was the independent variable in the model, representing for .
L2f

38
 Table 3-2. Summary of C, X, and b expressions in three modified laws.

Modified law C(K, R or B) X(K, R or B) b(K, R or B)

L1f 1
Kick’s μc ∙ CKc + μf ∙ CKf ln i=1 δi L
n
μc ∙ CKc ∙ ln
2fi r

μc ∙CRc 1 1 μc ∙CRc μc ∙CRc

i=1 δi
n
Rittinger’s + μf ∙CRf L2fi
- L1f L1f ∙r
- L1f
r
μc ∙CBc 10 10 10∙μc ∙CBc 10∙μc ∙CBc
Bond’s + μf ∙CBf i=1 δi
n
- L1f ∙ r
- L1f
r L2fi L1f

⃰ The units of C, X, and b are determined by the parameter units listed in Table 3-1.

39
3.4 RESULTS AND DISCUSSION

3.4.1 Fibre length analysis

The average initial fibre length was 111 mm. The manually separated output fibre

samples in different length categories are shown in Fig. 3-5. Average fractions for

each screen scenario under different feeding masses are presented in Fig. 3-6. Results

indicated that for screen opening size of 19.28 mm and 25.74 mm, majority fraction

of the output was short fibres in the range of 0-20 mm. With the screen size of 19.28

mm, all fibres were shorter than 80 mm, while product from the 25.74 mm screen

scenario contained only small proportion (less than 3%) of fibres longer than 80 mm.

The no-screen scenario produced fibres spreading in all length categories, and

approximately half of the fibres were within 40-80 mm. With screens (19.28 and

25.74 mm), the decortication process produced more short fibres comparing with the

no screen scenario. This result agreed with conclusions from other studies that smaller

screen opening size provided finer products (Ghorbani et al. 2010; Mani et al. 2003).

Feeding mass in the range from 75 g to 200 g did not show obvious effect on the final

fibre length.

40
(a) (b) (c)

(d) (e) (f)

Fig. 3-5. Output fibres in different length categories.

(a) 0-20 mm; (b) 20-40 mm; (c) 40-60 mm;
(d) 60-80 mm;(e) 80-100 mm; (f) > 100 mm.

41
 90
19.28mm
80

Mass proportion (%)

25.74mm
70
no screen
60
50
40
30
20
10
0
0 20 40 60 80 100 120
Fibre length (mm)
(a)
90
19.28mm
80
Mass proportion (%)

25.74mm
70
no screen
60
50
40
30
20
10
0
0 20 40 60 80 100 120
Fibre length (mm)
(b)
90
80 19.28mm
Mass proportion (%)

70 25.74mm
60 no screen
50
40
30
20
10
0
0 20 40 60 80 100 120
Fibre length (mm)
(c)

Fig. 3-6. Length distribution of different feeding masses.

(a) 200 g; (b) 125 g; (c) 75 g.

42
3.4.2 Specific energy requirement

The highest net energy and the highest specific energy was 14.77 kJ and 73.84 J/g

respectively when the small opening size (19.28 mm) screen was used to decorticate

200 g feeding hemp sample (Table 3-3). The no screen scenario with 75 g feeding

mass required both lowest average net energy (0.70 kJ) and lowest average specific

energy (9.31 J/g). The trends of specific energy were that higher feeding mass resulted

in higher specific energy and smaller screen opening resulted in higher specific

energy to complete decortication. Similar trends were observed for net energy. This

result agrees with the conclusions of some studies using hammer mill in size

reduction of crops, grains, and residues (Mani et al. 2004; Ghorbani et al. 2010;

Arthur et al. 1982; Dziki and Laskowski 2005).

The standard errors of average specific energy for 75 g feeding mass cases were

relatively high. It might be because that machine energy output was not stable when

processing smaller amount of feeding mass, and it would cause fluctuant of motor

torque and data transmitted to the recording facility. With higher feeding mass, the

recorded energy data approached to a relatively stable status for each replication.

3.4.3 Fitting modified Kick’s law

In the modified Kick’s law shown in Eq. 3-11, the values of specific energy EK

were measured, and XK were derived from the data of fibre length. Thus, values of CK

and bK were obtained through linear regression of EK versus XK. Feeding mass had

little contribution on the final fibre length distribution as discussed above, meaning
L1f
i=1 δi L
n
that the feeding mass had little effect on the value of XK (i.e. ln ) for a
2fi

43
given screen scenario. However, feeding mass affected the specific energy, and it in

turn affected the regression coefficients: CK and bK. For this reason, linear regressions

for three feeding masses were conducted separately to exhibit the effect of screen

opening size (Figs. 3-7 (a), (b) and (c)). A combined data regression including data

points over all feeding masses is also shown in Fig. 3-7 (d). Values of CK were taken

as the slopes of the regression equations and values of bK were taken as the

intersection. They are summarized in Table 3-4.

44
 Table 3-3. Specific energy of decortication.

Screen Feeding Average net Average specific Standard

scenario mass energy energy ES errors
(mm) (g) (kJ) (J/g) (J/g)
19.28 200 14.77 73.84 ±5.80
125 7.13 57.02 ±10.29
75 3.97 52.90 ±15.52
25.74 200 11.56 57.78 ±3.31
125 5.62 44.95 ±9.08
75 3.01 40.06 ±10.73
no screen 200 3.03 15.17 ±1.63
125 2.38 19.00 ±3.51
75 0.70 9.31 ±4.19

45
 90 90
80 80
EK = 52.63XK - 48.26 EK = 29.67XK - 13.90
70 R²= 0.96 70 R²= 0.73
60 60
EK (J/g)

EK (J/g)
50 50
40 40
30 30
20 20
10 10
0 0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
XK XK

(a) (b)
90 90
80 80
EK = 36.99XK - 33.81 EK = 39.19XK - 30.86
70 R²= 0.70 70 R²= 0.72
60 60
EK (J/g)

EK (J/g)

50 50
40 40
30 30
20 20
10 10
0 0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
XK XK

(c) (d)

Fig. 3-7. Linear regression for fitting modified Kick’s law.

(a) feeding mass 200 g; (b) feeding mass 125 g;
(c) feeding mass 75 g; (d) combined all feeding rates.

46
 Table 3-4. Summary of CK and bK values.

Feeding mass CK bK
R2 value Domain of XK
(g) (J/g) (J/g)
200 52.63 -48.26 0.96 1.15≤ XK ≤2.27
125 29.67 -13.90 0.73 0.99≤ XK ≤2.25
75 36.99 -33.81 0.70 1.00≤ XK ≤2.25
Combined data 39.19 -30.86 0.72 0.99≤ XK ≤2.27

47
 The value of coefficient of determination (R2) varied from 0.72 to 0.96 with the

feeding mass of 200 g giving the highest R2 value. Ghorbani et al. (2010) reduced the

size of alfalfa using a hammer mill and fitted the experimental data with Kick’s law.

Their regression gave a 0.87 R2 value.

The negative values of bK indicated that the r value is greater than 1, according to

Table 3-2. If the extreme condition is considered where there is no size reduction, no

specific energy would be cost, and the value of bK would be zero. This is consistent

with the theory. Under such condition, the ratio of L2c and L2f is 1 (i.e. r=1), so that

the term bK is zero. The modified Kick’s law (Eq. 3-11) fit by the hemp decortication

data can be rewritten, for example, for the combined feeding rates, as the following

format:
n
L1f
EK = 39.19 ×ln δi - 30.86 (3-18)
L2fi
i=1
where L1f and L2fi are in mm, and EK is in J/g.
3.4.4 Fitting modified Rittinger’s law

As shown in Eq. 3-12, the bR in the modified Rittinger’s law was related to the

initial fibre length L1f. As the L1f was constant (the average L1f was measured as 111

mm), the value of bR should be a constant. Similar to the fitting exercise of modified

Kick’s law, linear regression was performed between ER and XR (i.e.

1 1
i=1 δi
n
- ) to determine the CR and bR variables under each feeding mass
L2fi L1f

case and the combined data. Table 3-5 summarized the values of CR and bR, and R2

values. Similarly, data from high feeding mass (200 g) scenarios had better fitness

with the regression, giving a R2 value of 0.97.

48
 Theoretically, if extreme condition is considered, bR should be zero. This can be

interpreted as that when there is no size reduction (i.e. L2fi = L1f), the r (the ratio of

final core length to fibre length) will be 1, so the bR in Eq. 3-13 should be zero,

meaning that there is no energy consumed. Having determined the unknown

coefficient factors in Eq. 3-13, the modified Rittinger’s law can be expressed by the

following equation for the combined data:

n
1 1
ER = 793.04 × δi - - 1.47 (3-19)
L2fi L1f
i=1
where L1f and L2fi are in mm, and ER is in J/g.

49
 90 90
80 80
ER = 1029.89XR - 6.53 ER = 614.15XR + 7.34
70 70
R²= 0.97 R²= 0.75
60 60
ER (J/g)

ER (J/g)
50 50
40 40
30 30
20 20
10 10
0 0
0 0.02 0.04 0.06 0.08 0.1 0 0.02 0.04 0.06 0.08 0.1
XR (mm-1) XR (mm-1)

(a) (b)
90 90
80 80
70 ER = 750.28XR - 6.06 70 ER = 793.04XR - 1.47
R²= 0.72 R²= 0.75
60 60
ER (J/g)

ER (J/g)

50 50
40 40
30 30
20 20
10 10
0 0
0 0.02 0.04 0.06 0.08 0.1 0 0.02 0.04 0.06 0.08 0.1
XR (mm-1) XR (mm-1)

(c) (d)

Fig. 3-8. Linear regression for fitting modified Rittinger’s law.

(a) feeding mass 200 g; (b) feeding mass 125 g;
(c) feeding mass 75 g; (d) combined all feeding rates.

50
 Table 3-5. Summary of CR and bR values.

Feeding mass CR bR
R2 value Domain of XR (mm-1)
(g) (J·mm·g-1) (J/g)
200 1029.89 -6.53 0.97 0.019≤ XR ≤0.078
125 614.15 7.34 0.75 0.015≤ XR ≤0.076
75 750.28 -6.06 0.72 0.016≤ XR ≤0.077
Combined data 793.04 -1.47 0.75 0.015≤ XR ≤0.078

51
3.4.5 Fitting modified Bond’s law

The linear regression results of fitting Bond’s law are shown in Fig. 3-9, in

respect of 200 g, 125 g, 75 g feeding masses, and the combined data. Similar to the

10 10
modification procedures of Kick’s and Rittinger’s laws, the term i=1 δi
n
-
L2fi L1f

as a whole was calculated as dependent variable, XB. The regression results showed

that high feeding mass again led to a more accurate regression with R2 of 0.97. The

modified Bond’s law for the combined data is presented in Eq. 3-20:
n
10 10
EB = 34.11 × δi - - 6.44 (3-20)
i=1
L2fi L1f
where L1f and L2fi are in mm, and EB is in J/g.

52
 90 90
80 80
EB = 44.43XB - 13.17 EB = 26.18XB + 3.86
70 R²= 0.97 70 R²= 0.75
60 60
EB (J/g)

EB (J/g)
50 50
40 40
30 30
20 20
10 10
0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
XB (mm-1/2) XB (mm-1/2)

(a) (b)
90 90
80 80
EB = 32.54XB - 11.18 EB = 34.11XB - 6.44
70 R²= 0.72 70 R²= 0.74
60 60
EB (J/g)

EB (J/g)

50 50
40 40
30 30
20 20
10 10
0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
XB (mm-1/2) XB (mm-1/2)

(c) (d)

Fig. 3-9. Linear regression for fitting modified Bond’s law.

(a) feeding mass 200 g; (b) feeding mass 125 g;
(c) feeding mass 75 g; (d) combined all feeding rates.

53
 Table 3-6. Summary of CB and bB values.

Feeding mass CB bB
R2 value Domain of XB (mm-1/2)
(g) (J·mm1/2/g) (J/g)
200 44.43 -13.17 0.97 0.60≤ XB ≤1.95
125 26.18 3.86 0.75 0.47≤ XB ≤1.90
75 32.54 -11.18 0.72 0.51≤ XB ≤1.92
Combined data 34.11 -6.44 0.74 0.47≤ XB ≤1.95

54
3.4.6 Fitting a linear model

Results of regression are shown in Fig. 3-10. The values of CL are summarized in

Table 3-7. According to the results of the combined data, the relationship between

decortication specific energy EL (J/g) and the ratio of initial and average final fibre

length for the combined data was expressed by the equation below:
L1f
EL = 9.11 × - 2.43 (3-21)
L2f
where L1f and L2f are in mm, and EL is in J/g.
The R2 values of the linear regression model were similar to the three modified

laws. The advantage of the linear model is that the model uses the average final fibre

length and no fibre length distribution analysis is required.

55
 90 90
80 EL = 11.13XL - 4.39 80
EL = 7.04XL + 6.21
70 R²= 0.96 70 R²= 0.74
60 60
EL (J/g)

EL (J/g)
50 50
40 40
30 30
20 20
10 10
0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
XL XL

(a) (b)
90 90
80 80
EL = 9.02XL - 8.29 EL = 9.11XL - 2.43
70 R²= 0.76 70 R²= 0.76
60 60
EL (J/g)

EL (J/g)

50 50
40 40
30 30
20 20
10 10
0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
XL XL

(c) (d)

Fig. 3-10. Regression for fitting linear model of specific energy

and the ratio of initial and average output fibre length.
(a) feeding mass 200 g; (b) feeding mass 125 g;
(c) feeding mass 75 g; (d) combined all feeding rates.

56
 Table 3-7. Summary of CL and bL values.

Feeding mass CL bL
R2 value Domain of XL
(g) (J/g) (J/g)
200 11.13 -4.39 0.96 1.87≤ XL ≤7.54
125 7.04 6.21 0.74 1.64≤ XL ≤7.25
75 9.02 -8.29 0.76 1.81≤ XL ≤7.12
Combined data 9.11 -2.43 0.76 1.64≤ XL ≤7.54

57
3.5 CONCLUSIONS

Length distribution of the output fibres from decortication using a hammer mill

with screen scenarios of 19.28 mm and 25.74 mm opening indicated that majority

fibres were shorter than 20 mm. Small screen opening size produced more fine fibres

than larger sized screen and no-screen cases. Feeding mass had little effect on length

distribution. The highest specific energy (73.84 J/g) was required when using small

opening size screen (19.28 mm) to process 200 g retted hemp. The modified Kick’s,

Rittinger’s and Bond’s laws could determine specific energy requirement of

decorticated hemp material which consists of both fibre and core. Those modified

laws also could address different length fractions of hemp fibre. The three modified

laws and the linear model developed described the hemp decortication data using a

hammer mill with the R2 value as high as 0.97 in the case of 200 g feeding mass.

Lower R2 values (0.70-0.76) were observed for the two lower feeding masses. The

linear model was simpler and had comparable accuracy as the other three modified

laws.

The modified laws and the linear model were developed based on certain

assumptions and data from a given hammer mill and a given hemp feedstock. These

may affect the accuracy of the models in energy prediction. Caution should be taken

when using the results.

58
3.6 REFERENCES

Arthur, J.F., R.A. Kepner, J.B. Dobie, G.E. Miller and P.S. Parsons. 1982. Tub grinder
performance with crop and forest residues. Transactions of the ASAE 1488-1494.

Austin, L.G. 2004. A preliminary simulation model for fine grinding in high speed
hammer mills. Powder Technology 143-144: 240-252.

Baker, M.L., Y. Chen, C. Lagüe, H. Landry, Q. Peng and W. Zhong. 2009. Fibre
characteristics of hemp decortication using a planetary mono mill.
CSBE/SCGAB Meeting Paper No. CSBE09-303. Winnipeg, MB.

Baker, M. 2010. Evaluation of a hammer mill and planetary ball mill for hemp fibre
decortication. Unpublished M.Sc. thesis. Winnipeg MB: Department of
Biosystems Engineering, University of Manitoba.

Dziki, D. and J. Laskowski. 2005. Influence of selected factors on wheat grinding

energy requirements. TEKA Commission of Motorization and Power Industry in
Agriculture 5: 56-64.

Fang, Q., E. Haque, C.K. Spillman, P.V. Reddy and J.L. Steele. 1998. Energy
requirements for size reduction of wheat using a roller mill. Transactions of the
ASAE 41(6): 1713-1720.

Fortenbery, T.R. and M. Bennett. 2004. Opportunities for commercial hemp

production. Review of Agricultural Economics 26: 97-117.

Ghorbani, Z., A.A. Masoumi and A. Hemmat. 2010. Specific energy consumption for
reducing the size of alfalfa chops using a hammer mill. Biosystems Engineering
34-40.

Gratton, J.L. and Y. Chen. 2004. Development of a field-going unit to separate fibre
from hemp. Applied Engineering in Agriculture 20(2): 139-145.

Henderson, S.M. and R.L. Perry. 1976. Agricultural Process Engineering, 3rd edition.
Westport, Connecticut. The AVI Publishing Company INC.

Holtzapple, M.T., A.E. Humphrey and J.D. Taylor. 1988. Energy requirements for the
size reduction of poplar and aspen wood. Biotechnology and Bioengineering 33:
207-210.

Lobo, P. 2002. The right grinding solution for you: roll, horizontal or vertical. Feed
Management 53(3): 23-26.

59
Lowrison, G.C. 1974. Crushing and Grinding: The Size Reduction of Solid Materials.
London. Butterworths.

Mani, S., L.G. Tabil, S. Sokhansanj and M. Roberge. 2003. Mechanical properties of
corn stover grind. ASAE Meeting Paper No. 036090. St. Joseph, Mich.: ASAE.

Mani, S., L.G. Tabil and S. Sokhansanj. 2004. Grinding performance and physical
properties of wheat and barley straws, corn stover and switchgrass. Biomass and
Bioenergy 27: 339-352.

Munder, F. and C. Fürll. 2004. Effective processing of bast fiber plants and
mechanical properties of the fibers. ASAE/CSAE Meeting Paper No. 046091. St.
Joseph, Mich.: ASAE.

Pfost, H.B. and C.E. Swinehart. 1970. Feed Manufacturing Technology. Chicago,
Illinois. American Feed Manufacturers Association INC.

Riddlestone, S., E. Stott, K. Blackburn and J. Brighton. 2006. A technical and

economic feasibility study of green decortication of hemp fibre for textile uses.
Journal of Industrial Hemp 11(2): 25-55.

Shi, F., T. Kojovic, J.S. Esterle and D. David. 2003. An energy-based model for swing
hammer mills. International Journal of Mineral Processing 71: 147-166.

Van Wyk, L. 2007. The application of natural fibre composites in construction: a case
study. Six International Conference on Composite Science and Technology.

Vigneault, C., T.M. Rothwell and G.B. Bourgeois. 1992. Hammermill grinding rate
and energy requirements for thin and conventional hammers. Canadian
Agricultural Engineering 34(2): 203-207.

60
4. Integration of TRIZ and Virtual Environments
for Hemp Scutcher Design

4.1 ABSTRACT

TRIZ (“Theory of Inventive Problem Solving” in Russian) method provides useful

theory for users to analyze and solve problems using a logic manner. It is widely used in

different fields and has gained great success in machine design. Virtual reality (VR) is a

computer generated environment with features of visualization and users’ interaction,

which provides a cost-effective tool for product evaluation and design improvement. In

this research, TRIZ is applied in the design of a new hemp scutcher prototype. With the

assistance of VR technology, the design visualization and concept evolution are

achieved in a virtual environment. A VR-based system is developed to evaluate the

mechanism of the conceptual design, and to compare differences with traditional

machines. The new design is supposed to have a better performance in terms of scale,

product quality and energy efficiency. Keywords: hemp scutcher, design, TRIZ, virtual

reality.

61
4.2 INTRODUCTION

Hemp (Cannabis sativa) fibre is a renewable and environmentally acceptable

material. It has been used in different areas including textile industry, civil

engineering industry and automobile industry. Hemp fibre can substitute glass and

carbon fibres in a wide range of industrial products. With the increasing need of high

quality hemp fibres, the processing methods of hemp fibres have been developed

during the past decades. Because the properties of hemp are similar to other bast fibre

plants such as flax and linen, some of the existing processing machines for hemp were

modified based on those used for flax or cotton processing (Hobson et al. 2000; Akin

et al. 2005).

The key procedure in hemp fibre processing is the decortication, which can

detach the fibre (outer layer of hemp plant) from the core (inner layer). Scutching

machines, or scutchers, are used by some industries to achieve this purpose. However,

the effect of current scutching machines on hemp fibre processing is not satisfying.

Problems such as large machine scale, waste of products and energy affect the

economy of the production. An efficient machine should be designed to improve the

processing quality.

Methods used by engineers to produce design ideas include brainstorming,

axiomatic design, TRIZ (Teoriya Resheniya Izobretatelskikh Zadatch), etc

(Shirwaiker and Okudan 2006; Pham et al. 2009). When engineers encounter a

problem or an improvement issue, the most direct way to find a solution may be the

trial-and-error method: try one thing, if it does not work, try something else (Mathis

62
2009). Mostly, people prefer to use brainstorming for quick and direct problem

solving because it is a natural style of human mind. However, this kind of unstable

and unpredictable way may require a significant time before a right solution is

attained. TRIZ is the acronym for the “Theory of inventive problem solving” in

Russian. It was originally developed by G. S. Altshuller and his colleagues between

the 1940s and the 1980s (Barry 2009). Having reviewed more than 40,000 successful

patents, they summarized the common features for different creative work and

patterns to predict solutions of problems. The theory helps users to analyze problems

and contradictions inside the problems, and provide reasonable directions for

solutions with the assistance of tools. During the past decades, research in different

fields involving TRIZ has improved the method by complementing the theory

continuously. Nowadays, the theory is close to maturity and has been successfully

applied in problem solving in a large range of areas.

The objective of this research was to design and evaluate a new hemp scutcher

based on the TRIZ principle in virtual environments (VEs). Virtual Reality (VR)

technology is capable of constructing a user-centered, three-dimensional (3D)

environment in which abstract and complex information is visualized in an intuitive

and realistic manner (Cobb et al. 1995; Shukla et al. 1996; Cecil and Kanchanapiboon

2007). It is an important tool for industry design to save the cost of physical

prototypes. By using VR technologies, designers can build and test virtual prototypes

to forecast usability problems in the design process.

63
 In this research, the problems and weakness of current scutching machines were

analyzed based on the investigation of several hemp and flax fibre scutcher patents.

TRIZ specially concentrates on the use of the experience and knowledge to improve

problems because it is a method developed based on a large number of patents (Pham

et al. 2009). Therefore, TRIZ method was used to find solutions to overcome the

bottlenecks of the design. The conceptual designs of a new hemp fibre scutcher were

developed according to suggested principles by TRIZ. The prototype evaluation and

design improvement are achieved using the VR technology. A VR tool, EON Studio

6.1, was applied to build the virtual environment to provide functions required for the

design evaluation and improvement. The advantage of designing in the VE is that it is

very easy to modify the conceptual model without spending a lot of money on

constructing physical prototypes.

4.2.1 The TRIZ Theory

TRIZ is a type of logic and knowledge-oriented methodology (Hsieh and Chen

2009). One important principle of TRIZ is the use of evolution patterns. A pattern

represents a direction to idealize the current problem. The first step of the solving

problems using TRIZ is to idealize the problem without considering the weakness,

cost or barricades (Cai et al. 2008). The ideal final result is defined as the point when

users get all benefits they want without any of costs or harms (Mann 2004; Ma et al.

2006). Using patterns will help designers to forecast the future development and give

the direction to follow to generate ideal final results. Direct evolution is one of the

64
ways to get the ideal final results before using TRIZ for further development (Barry et

al. 2009; Cai et al. 2008; Ma et al. 2006).

Once ideal final results are settled by designers, TRIZ requires the encountered

technical problems and difficulties be identified, which are in the formats of

contradictions. The basis of the theory is the finding of contradictions in the converted

TRIZ problems. It is believed that the fundamental concept of solving any problems is

to eliminate the contradictions (Barry et al. 2009). Two groups of contradictions are

classified in TRIZ problems: technical contradictions and physical contradictions.

Technical contradiction is the classical engineering “trade-off”. When specific point

of a system needs to be improved, some other parts will be affected and may get

worse simultaneously. Physical contradiction is also called inherent contradiction

which stands for the contradiction of an object with opposite requirements. After

contradictions have been identified, the tools for eliminating contradictions can be

applied. The powerful function of “TRIZ-contradiction matrix” can provide

references as solutions which have been successfully used in patents on solving

problems in specific areas. Using the contradiction matrix, designers can identify up

to 39 engineering parameters summarized in TRIZ, both in improving and worsening

directions. Contradiction matrix then provides the recommended “40 principles” for

specific contradiction combinations. There are also “76 standard solutions” related to

the 40 principles providing more detailed ways of contradiction elimination (Barry et

al. 2009; Hsieh and Chen 2009; Domb 1999). With the development of the theory, the

65
39 improving and worsening features of TRIZ contradiction matrix have been

increased to 48 features recent years (Pham et al. 2009).

Finally, based on the solutions provided by the matrix, TRIZ solutions can be

transferred into a real design for solving actual problems. Figure 4-1 indicates a basic

flowchart of TRIZ in solving problems. However, there are also some weaknesses of

TRIZ method. When improving and worsening features are selected, some of the cells

in the contradiction matrix are empty so that the matrix cannot provide

recommendation for such situations (Pham et al. 2009). In such cases, single

engineering inventive principle can be used (Hsieh and Chen 2009). The principles

appearing most frequently in the contradiction matrix can be applied as the suggested

principles. In addition, because of the complexity of translating the practical

requirements, constraints or criteria of design problems in TRIZ problems, TRIZ

contradiction matrix is known being difficult for applications (Pham et al. 2009).

66
 Problems

Select evolution patterns

Ideal final result

Contradictions

TRIZ Contradiction-solving tools

Contradiction matrix Standard solutions

Single Engineering Inventive Principle …

If no
Preliminary solutions

If yes

Final solutions

Fig. 4-1. Flowchart of TRIZ in problem solving.

67
4.2.2 TRIZ Applications

TRIZ-based studies on machine design have been conducted by many researchers.

Design improvements are also achieved using TRIZ. Cai et al. (2008) applied the

TRIZ method in analyzing the modular fixture design details and forecasting the

development potential. The intelligent modular fixture is difficult to be produced

because of the complexity of machining components. The controllability of

automation degree and feedback has been improved using TRIZ. A conceptual design

of the packaging machine for Chinese traditional medicine pill dropping was

developed by Ma et al. (2006). The authors first converted the real problem into

substantiate function diagram of the process using a substance-field analysis. Based

on suggested solution principles given by the conflict matrix, a design of tumbling

cylinder with filling mechanical structure was invented to meet the package

requirement. A study on the effect of TRIZ to support designers using the descriptive

design framework was conducted by Pham et al. (2009). The descriptive design

framework was developed with TRIZ methodology as the core, which could help a

designer to track his ideas throughout the design process. Shirwaiker et al. (2006)

conducted a review on both TRIZ and axiomatic design and found that the

compatibility of the two methods could be applied synergistically to achieve better

solutions. TRIZ is considered to be capable to generate innovative solutions, while

axiomatic design has the function to analyze the effectiveness of solutions.

A manufacturing process can be improved using TRIZ if the problems in terms of

technical contradictions can be eliminated according to the recommended principles

68
by contradiction matrix. Hsieh and Chen (2009) conducted a research on the design of

friction stir welding process using TRIZ as a problem analyzing and solving tool.

According to the contradiction matrix, “Phase transitions” was selected as the solution

orientation for improving the welding process. The rotating welding tool was

designed as the heat generator by means of the friction between the head and

materials. Additionally, the study conducted other process problems, and used TRIZ

suggestions in the welding process such as a friction stir welding roller press to

maintain a constant welding depth. Zhang et al. (2006) studied the development of the

cutting technology and generated the evolutional technology by means of TRIZ.

Fresner et al. (2009) used TRIZ principles showing similarities and correspondence to

the strategies of Cleaner Production so that the authors realized that the concept of the

ideal final result and Laws of Evolution could be a framework for assisting Cleaner

Production improvement. The effectiveness of TRIZ-based Cleaner Production

concept on improving chemical process was proved by a practical application.

TRIZ has the potential in improving design, and managing the design process.

Mathis (2009) reported an example using the TRIZ method to improve the design of

candy pouch. The research used TRIZ as well as experiments to prove the effect of

each design stage. In a non-technological area, Su et al. (2008) attempted using the

TRIZ methodology to develop a systematic framework for service industry. The

approach successfully supported resolving problems of service quality in the

e-commerce sector.

69
4.3 METHODOLOGY

4.3.1 TRIZ Design integrated with VE

As shown in Fig. 4-1, TRIZ method provides designers the tools to gain

reasonable solutions for design problems. However, not all the recommended

solutions can fit the need for real problems. Before transferring TRIZ solutions to a

real design, there should be a pre-testing procedure in order to evaluate the solutions.

VR, as a visualization tool, can assist to get feedback from the evaluation of a design

built in VEs according to TRIZ solutions. Comparing with common evaluation

methods, VR has the advantage to check TRIZ solutions by users’ interaction with the

model in VEs. The users can evaluate a design by simulating its working processing

for the feasibility. If the feedback proves that the model is working properly, then the

design can go for manufacturing. If not, users can modify the design in VEs. Users

can also go back to the TRIZ procedure to select other solutions for the problem based

on the evaluation in VEs. The whole process is shown in Fig. 4-2. It includes TRIZ

procedures shown in Fig. 4-1. This procedure is proposed for the methodology

applied in this research.

70
 Problems

TRIZ

Final solutions

3D-Models (3)
(2)

Testing (1)

If no
If (1) is no
Feedback If (2) is no

If yes
VEs

Final design
prototypes

Fig. 4-2. Flowchart of TRIZ integrated with VE.

71
4.3.2 Problem analysis and design criteria for hemp scutchers

The principle of hemp fibre scutching is to bend the stalks sharply in one

direction and then re-bend them in an opposite direction to break the brittle core, and

simultaneously to comb the material by bars or brushes rapidly in order to remove the

broken parts from the fibre (Johansen 1946). Many traditional scutchers used this

principle for decortication machine designing and fibre processing (John 1941; Waldo

1956; Patterson 1949; Cary et al. 1956; Horine 1906). Some laboratory-scaled

scutchers have been tested and their effects on the fibre decortication have been

evaluated (Akin et al. 2005; Hobson et al. 2001). Mostly, gears and rollers with

grooves are used to achieve this goal. The machine generally includes a horizontal

path and a couple of roller pairs. When the stalks are introduced to pass between the

roller pairs, the sharp teeth of the rollers bend the stems so that the brittle cores will be

broken into smaller pieces as shown in Fig. 4-3. Some of them may fall down through

the gaps between roller pairs, while others still attach to the fibre and should be

removed by following procedures.

72
 Scutching roller

Hemp stalk Cores and shives

Fig. 4-3. Basic principle of the scutching procedure

(Hobson et al. 2001; Patterson 1949).

73
 By analyzing the existing methods, some problems of the scutcher design have

been found. The main problem was that in order to gain clean fibre (fibre free of

cores) after scutching, large numbers of roller pairs were required. For the horizontal

arranged structure as shown in Fig. 4-3, this would increase the length of the

horizontal path and at the same time increase the machine scale. Simultaneously, with

the increasing roller pair number, the power and energy requirement for driving the

rollers would be increased, making the procedure costly and less efficiently. In

addition, for a single roller, the functioning part was only the touching point

in-between the roller pair which is getting in touch with the stalk, while other teeth

around this roller were running ineffectively. Productivity was also a factor need to be

considered. Processing speed might be increased if the rotational speed of the rollers

is increased. However, more energy was required to run the machine, and there might

be more loss of the fibre due to the higher speed. More broken fibres could fall down

with the shives and be collected as waste.

According to the above analysis, the new design should compromise the following

criteria:

• Fewer roller pairs;

• Relatively smaller in machine scale and machine part dimensions;

• Less fibre waste along the processing line;

• Simple structure to achieve better maintenance ability;

• Better controllability for adjusting machine performance on different hemp

varieties;

74
• Less chance of wrapping or entanglement.

Two evolution patterns were chosen in this study, including:

• Transition toward micro-level and increasing use of fields.

• Increasing dynamism and controllability.

Four pairs of contradictions were abstracted from the real problems, as

summarized in Table 4-1.

75
Table 4-1. Selected TRIZ contradictions and solutions from TRIZ contradiction
matrix (TRIZ40 2010).

Improving feature Worsening feature Solution

#39 Productivity #6 Area of stationary object #17 Moving to a new dimension
#29 Manufacturing #22 Loss of energy #2 Extraction
precision
#9 Speed #22 Loss of energy #35 Parameter change
#14 Spheroidality – Curvature
#20 Continuity of useful action
#9 Speed #29 Manufacturing precision #10 Prior action

76
4.4 PRELIMINARY DESIGN AND EVALUATION IN VE

Before the design prototype was evaluated in the VE, 3D CAD models of the

possible design solutions were built using Auto CAD 2009. EON Studio 6.1, a VE

modeling tool, was used to construct virtual environments for the concept testing and

design evaluation. CAD models were imported into EON Studio in the format of .3ds

for the VE modeling.

According to the TRIZ contradiction matrix, six solutions were selected to form

the design concept as listed in Table 4-1. However, these solutions could only

generate the design directions without exact ideas. Based on these proposed concepts,

possible models were built in the VE, and then a preliminary evaluation was taken to

assess the proposed conceptual model. At the design stage in the VE, no physical

activities could be executed to test and evaluate the effect of the prototype. It was also

difficult to simulate the hemp material reaction in virtual models due to the

complexity of the bio-fibre material properties. The model evaluation was only based

on the geometrical feasibility of machine operations. The machine structure was

improved in the VE based on the simulation feedback. This procedure is illustrated as

the route (1) in Fig. 4-2.

Taking the solutions “#17 Moving to a new dimension” and “#14 Spheroidality –

Curvature” for instance, four alternative schemes of the roller layout were generated

in VE as shown in Fig. 4-4. All of them were curve paths designed to achieve both

space-saving and energy-efficient purposes. Comparing with traditional linear

processing line, hemp cores should be removed more easily with fewer roller pairs

77
because the bending may effectively help to detach the cores from the fibres.

Meanwhile, developed in a new dimension, the machine scale was expected to be

limited to a reasonable size.

78
 (a) (b)

(c) (d)

Fig. 4-4. Alternative schemes of roller layout.

(a) “S” shape; (b) “U” shape; (c) “C” shape; (d) improved “C” shape.

79
 Figure 4-4(a) has an “S” shape curve with 4 pairs of rollers. Rollers with smaller

radius were selected to decrease loading energy of the machine. The total horizontal

length was decreased due to the curved line. The structure might achieve good

bending and scutching force to break the cores, but the sharp turns might damage the

fibre at the same time. It might cause wrapping and entanglement if the turning angle

is not well designed. Figure 4-4(b) has a “U” shape curve, but it contained many pairs

of rollers. It could save the horizontal length of the whole line, and generate

horizontal bending by the side housing. However, the energy consumption might be

high because many roller pairs were applied. Material might choke at the turn. Figure

4-4(c) applies 4 roller pairs to form a “C” shape curve. It was a relatively smoother

curve compared with the previous two alternatives. The arrangement was in a vertical

plane so that horizontally it would not take too much space. However, the large gaps

between roller pairs might cause shorter fibres falling down as a waste. Figure 4-4(d)

was evolved from the C shape scheme. This improved C shape consisted of one big

roller at the centre, and in reaction with four assisting rollers around it. This design

also followed the TRIZ solutions “#2 Extraction” and “#20 Continuity of useful

action”. The central roller could be efficiently used without wasting much energy. So

the improved C shape was selected as the basic scheme of rollers.

The improved C shape had the character of a “planetary layout”. Having selected

the “planetary layout” as the proposed structure, housing was then designed step by

step. Firstly, a simple sector housing was created (Type A in Fig. 4-5). It was noticed

that the material path through the roller gaps could not be regulated, which might be a

80
reason for wrapping. Then the idea of adding a curved housing structure as shown in

Type B came out from the TRIZ solution “#10 Prior action”. Such a structure could

form a relatively stable working path for material. Additionally, high speed air flows

were considered to help in leading material mass stream and remove the shives, so

that three air flow inlets were created as shown in the circles in Type C in Fig. 4-5.

81
Type A

Type B

Type C Air
flow
inlets

Fig. 4-5. Housing design evolution.

82
 Using the similar procedure as mentioned, new design ideas could be evaluated,

compared, and improved in the VE. Based on the feedback from the models according

to the criteria, necessary modifications could be conducted easily. The detailed

designs according to the TRIZ solutions in the VE are summarized below:

• Moving to a new dimension: Instead of traditional horizontal path, a curve path

was designed to save the space taken by the machine.

• Extraction: One single functional roller was “taken-out”, with four smaller

assisting rollers around it. This roller was placed at the centre, so that nearly half

of its perimeter could function the scutching.

• Parameter change: Smaller rollers were chosen to achieve a faster speed and better

controllability.

• Spheroidality: Rotational working mode was selected instead of linear one.

• Continuity of useful action: Around the central roller, four smaller rollers reacted

with it together, making half of the central roller under working condition all the

time. (The traditional designs just used one single working point between a roller

pair.)

• Prior action: The housing of the machine contained curve structure around the

four assisting rollers. The gap between two rollers was dominated by the housing,

so that the path of the passing material was controlled, preventing wrapping and

miss-introducing. Three high speed air flows were applied to lead the material,

and they could help remove unwanted shives from the fibre to purify the product.

83
 In addition, removable teeth on the rollers were designed to make the maintenance

easier. Also, the changeable teeth could fit different plant varieties. For a variety of

tougher stem property, original teeth could be substituted by the sharper teeth to

satisfy the decortication effect expected. Figure 4-6 shows a 2D CAD model of the

final proposed machine. A VR model of the complete machine was then developed to

test and evaluate these concepts in the VE, which will be discussed in following

sections.

4.5 VR MODELING AND PROTOTYPE EVALUATION

The user interface was designed with a menu, including four typical traditional

scutchers and the proposed design. Users could select one of the models to view

details of the structure. It made the comparison of the new design and the traditional

machines in scale, structure, components and working principle. Figure 4-7 is the VE

framework built. Figure 4-8 shows the user interface of the system.

84
 Feeding rollers

Material inlet
Assisting roller

Central roller

Material outlet
Air flow inlet

Changeable teeth

Leading roller

Fig. 4-6. 2D CAD Model of the new design.

85
 Menu

Select
model

Traditional Traditional Traditional Traditional New design

Type A Type B Type C Type D

Housing Run Route Remove Change teeth

transparent housing

Navigate the model in Right click the

360° model

Fig. 4-7. The framework of the VE.

86
Type A, B, C, and D

New design

Fig. 4-8. Main menu of the system.

87
 The model operation in VE was achieved using the prototypes of

“ObjectNavManager” and “ObjectNav” provided in EON Studio. The trigger action

was controlled by the “ClickSensor” connected to the prototypes in the route

simulation shown in Fig. 4-9. When selecting one of the models by left clicking it, the

view would focus on this single model, so that users could view it in detail as shown

in Figs. 4-10 and 4-11. Also, dragging an object by clicking the right mouse button

could provide 360°view of the model from any angle.

There were two ways of showing details of the inner machine structure: making

the housing transparent or removing the outer case. Both functions were programmed

to be a textbox-button-controlled method. For the transparency function, a prototype

called “SmoothOperator” was selected and connected to a textbox-button “Housing

Transparent”. The texture property of the housing material was set to be 50%

transparent once the “SmoothOperator” sending out an order signal. To remove the

housing, “Script” node was edited to control the appearance of the objects connected

to it as shown in Fig. 4-9. The commands were shown as follows:

function On_show()
{
visible.value = false
}
function On_disappear()
{
visible.value = true
}
Using dynamic models was the main character of the system. The running

machine could provide a clear view of the working status of the models compared

with static models, including the movement of the main working components and the

88
materials. Therefore it was easier for designers to evaluate and modify the preliminary

design based on the visualization feedback. “Rotate” and “Place” nodes were used to

dynamically display the models in VE. The main function parts in these scutching

machines were the rotating rollers and gears. Textbox button “Run” was the central

control of this function. “Change teeth” button was connected to a “Place” node.

When it was clicked, the teeth around an assisting roller would be moved out of the

machine, representing the changeable teeth structure as shown in Fig. 4-11 (c).

Moreover, a time-step-motion was added to each model to present the material path

through the scutcher. A red arrow was created as an object, and a “KeyFrame” node

was taken to simulate the material path of each model as shown in Fig. 4-12. The

position of each time step was set in the property of the “KeyFrame” node. The

motion of the red arrow stood for the hemp stalk passing through the gap between

rollers. Three green arrows were created to demonstrate the air flow direction of the

new design.

Comparing with the traditional machines, the new design could compromise the

designing criteria in terms of reduced scale, less energy consumption, and higher

product quality, etc.

In summary, users could easily compare the differences of the new design and the

traditional ones in VEs. With the help of the virtual environment, users could get a

better understanding on the mechanism and working principle of the machine

designed. The design feasibility could be easily analyzed due to the flexibility of

modifying prototypes in VEs.

89
 View control

Housing
appearance
control

Fig. 4-9. Route simulation examples inVE.

90
Type A Type B

Type C Type D

Fig. 4-10. The models of the traditional scutchers.

91
 (a)
Green arrows
showing air flows

(b)
Red arrow showing material
passing route

(c)

Fig. 4-11. The model of the new design in VE.

(a) front view of the model with a transparent cast;
(b) side view of the model with a transparent cast;
(c) inner structure without cast.

92
Fig. 4-12. KeyFrame Node to simulate material route.

93
4.6 CONCLUSIONS

The TRIZ method can be successfully used in designing conceptual prototypes of

a hemp scutcher. The new design of the scutcher consisted of a “planetary layout” and

air flow assistance. It will overcome some bottleneck problems of the existing

machines in large scale, energy and productivity inefficiency. Design evaluation was

conducted in a virtual environment without building physical models, and a dynamic

model navigation which provided a straightforward view to the working status of the

machine was developed. The use of TRIZ methods with assistance of virtual reality

technology in the machine design is a new trend. It has a great potential for a fast,

reliable and low-cost design evaluation.

4.7 ACKNOWLEDGEMENTS

This research is supported by Natural Sciences and Engineering Research

Council of Canada (NSERC).

94
4.8 REFERENCES

Akin, D.E., R.B. Dodd and J.A. Foulk. 2005. Pilot plant for processing flax fiber.
Industrial Crops and Products 21: 369-378.

Barry, K., E. Domb and M.S. Slocum. 2009. TRIZ-What is TRIZ. The TRIZ-Journal,
http://www.triz-journal.com/archives/what_is_triz/ (2009/11/05).

Cai, J., H. Liu, G. Duan, T. Yao and X. Chen. 2008. TRIZ-based evolution study for
modular fixture. Global Design to Gain a Competitive Edge, Chapter 6: 763-772.

Cary, J.E., R.E. Shafer and V. Cary. 1956. Fibre decorticating machine. US Patent No.
2745142.

Cecil, J., and A. Kanchanapiboon. 2007. Virtual engineering approaches in product

and process design. International Journal of Advanced Manufacturing
Technology 31: 846-856.

Cobb, S.V., M.D. Cruz and J.R. Wilson. 1995. Integrated manufacture: A role for
virtual reality. International Journal of Industrial Ergonomics 16: 411-425.

Domb, E., J. Terninko, J. Miller and E. MacGran. 1999. The seventy-six standard
solutions: how they relate to the 40 principles of inventive problem solving. The
TRIZ-Journal, http://www.triz-journal.com/archives/1999/05/e/index.htm
(2009/11/05).

Fresner, J., J. Jantschgi, S. Birkel, J. Brnthaler and C. Krenn. 2009. The theory of
inventive problem solving (TRIZ) as option generation tool within cleaner
production projects. Journal of Cleaner Production, DOI: 10.1016/
j.jclepro.2009.08.012.

Hobson, R.N., D.G. Hepworth and D.M. Bruce. 2001. Quality of Fibre Separated
from Unretted Hemp Stems by Decortication. Journal of Agricultural
Engineering Research 78(2): 153-158.

Horine, E.E. 1906. Fibre scutcher. US Patent No. 810280.

Hsieh, H.J. and J.L. Chen. 2009. Using TRIZ methods in friction stir welding design.
International Journal of Advanced Manufacturing Technology, DOI 10.1007/
s00170-009-2172-y.

Johansen, J.V. 1946. Method of decorticating fibrous materials. US Patent No.

2391977.

95
John, S.L. 1941. Decorticating method and apparatus. US Patent No. 2244971.

Ma, L., R. Tan, H. Zhang and X. Zhang. 2006. TRIZ application in conceptual design
of packaging machine for dropping pill of Chinese traditional medicine. IEEE
International Conference on management of innovation and technology: 600-603.

Mann, D. 2004. TRIZ: An introduction. Report for Systematic Innovation Ltd, UK.

Mathis, F. 2009. TRIZ used to improve the Mars bite-size candy pouch. The
TRIZ-Journal, http://www.triz-journal.com/archives/2009/01/03/ (2009/11/05).

Patterson, D.E. 1949. Machine for decorticating fiber-bearing stalks and leaves. US
Patent No. 2480602.

Pham, T.D., W.K. Ng and C.M. Ang. 2009. Applying TRIZ to support designers in a
descriptive design framework. Proceedings of the ASME 2009 International
Design Engineering Technical Conferences & Computers and Information in
Engineering Conference, DETC2009-87455.

Shirwaiker, R.A. and G.E. Okudan. 2006. TRIZ and axiomatic design: a review of
manufacturing case-studies & their compatibility. PICMET Proceedings:
2510-2520.

Shukla, C., M. Vazquez and F.F. Chen. 1996. Virtual manufacturing: An overview.
19th International Conference on Computers and Industrial Engineering 31(1/2):
79-82.

Su, C., C. Lin and T. Chiang. 2008. Systematic improvement in service quality
through TRIZ methodology: an exploratory study. Total Quality Management &
Business Excellence 19(3): 223-243.

TRIZ40. 2010. TRIZ contradiction matrix. http://www.triz40.com/aff_Matrix.htm

(2010/11/1).

Waldo, W.G. 1956. Decorticating apparatus. US Patent No.2753600.

Zhang, F., Q. He and Y. Xu. 2006. The innovative study of cutting technology based
on TRIZ evolution theory. International Technology and Innovation Conference,
Section I: Advanced Manufacturing Technology: 155-159.

96
 5. Summary

Hemp fibre can be widely used for various industrial applications. Hemp

decorticator is an important component in a hemp processing line. Hammer mill and

roll crusher (scutcher) are two types of decorticator, and their performance in fibre

decortication is worth being studied. Energy requirement affects the cost of the

process, which at the same time determines the availability of the processing method

and equipment.

The results of this study provided analytical resource in estimating specific

energy of hemp decortication using a hammer mill. One of the contributions of this

study is that modified Kick’s, Rittinger’s and Bond’s laws were developed. The

second contribution is the development of a linear model. The constant factors in the

modified laws and the linear model were determined based on the experimental data.

All modified laws and the linear model fit the data equally well. They can be used to

estimate specific energy of hammer mill decortication in the future if the output fibre

length is known.

Another contribution of this study is the application of TRIZ-VR integrated

method in designing a conceptual hemp scutcher. The new design has high potential

to be used as long fibre processor in the future. This tool provides a useful trend of a

fast, reliable and low-cost design evaluation.

For the future work, a physical prototype based on the conceptual design can be

built and tested to evaluate the performance of the design. TRIZ-VR integrated

97
method has the potential to be used for designing other components of a hemp fibre

processing line such as the cleaning component. Virtual environment as a platform

can be integrated with data analysis functions. For instance, with the assistance of

external numerical simulation software, energy or force data can be introduced to the

prototype in the VEs, and the evaluation procedure can be supported by the data

analysis results. Furthermore, the evaluation in VE can contain assembly-disassembly

analysis, manufacturing efficiency assessment, and production cost calculation.

98
 Appendix
(References for Chapter 1, Chapter 2 and Chapter 5)
Akin, D.E., R.B. Dodd and J.A. Foulk. 2005. Pilot plant for processing flax fibre.
Industrial Crops and Products 21: 369-378.

Akin, D.E., R.B. Dodd, W. Perkins, G. Henriksson and K.E. Eriksson. 2000. Spray
enzymatic retting: a new method for processing flax fibres. Textile Research
Journal 70: 486-494.

Amaducci, S., J. Mussig, A. Zatta and G. Venturi. 2008. An innovative production

system for hemp fibre for textile destinations: from laboratory results to industrial
validation. International Conference on Flax and Other Bast Plants 104-117.

Anthony, W.S. 2002. Separation of fiber from seed flax straw. Applied Engineering in
Agriculture 18(2): 227-233.

Baker, M.L., Y. Chen, C. Lagüe, H. Landry, Q. Peng and W. Zhong. 2009.

CSBE/SCGAB Paper No. CSBE09-303.

Beckermann, G. 2007. Performance of hemp-fibre reinforced polypropylene

composite materials. Unpublished Ph.D. thesis. Department of Materials and
Process Engineering, University of Waikato.

Booth, I., R.J. Harwood, J.L. Wyatt and S. Grishanov. 2004. A comparative study of
the characteristics of fibre-flax (Linum usitatissimum). Industrial Crops and
Products 20: 89-95.

Chen, Y. and J. Liu. 2003. Development of a windrower for dual-purpose hemp

(Cannabis sativa). Canadian Biosystems Engineering 45: 2.1-2.7.

Chen, Y., J.L. Gratton and J. Liu. 2004. Power requirements of hemp cutting and
conditioning. Biosystems Engineering 87(4): 417-424.

Declerck, S., S. Amelynck and H. Koether. 2008. Bast fibre processing by VDE. 2008
International conference on flax and other bast plants. ISBN#978-0-9809664-0-4.

Feedmachinery. 2009. Roll Crusher. http:// www.feedmachinery.com (2009/10/28).

Gratton, J.L. and Y. Chen. 2004. Development of a field-going unit to separate fibre
from hemp (Cannabis sativa) stalk. Applied Engineering in Agriculture 20(2):
139-145.

99
Hobson, R.N., D.G. Hepworth and D.M. Bruce. 2000. Quality of fibre separated from
unretted hemp stems by decortication. Journal of Agricultural Engineering and
Research 78(2): 153-158.

Igathinathane, C., A.R. Womac, P.I. Miu, M. Yu, S. Sokhansanj and S. Narayan. 2006.
Linear knife grid application for biomass size reduction. ASABE Paper No.
066170. St. Joseph,Mich.: ASABE.

Kovur, S.K., K.C. Schenzel, E. Grimm and W. Diepenbrock. 2008. Refined hemp
fiber characterization. BioResources 3(4): 1081-1091.

Miu, P.I., A.R. Womac, I. Cannayen and S. Sokhansanj. 2006. Analysis of biomass
comminution and separation processes in rotary equipment – A review. ASABE
Paper No. 066169. St. Joseph, Mich.: ASABE.

Munder, F. and C. Fürll. 2004. Effective processing of bast fiber plants and
mechanical properties of the fibers. ASAE/CSAE Meeting Paper No. 046091. St.
Joseph, Mich.: ASAE.

Munder, F., and H. Hempel. 2004. Results of an advanced technology for

decortication of hemp, flex and linseed. Molecular Crystals and Liquid Crystals
418(1): 165-179.

Munder, F., C. Fürll and H. Hempel. 2003. Results of an advanced technology for
decortication of hemp, flax and oil seed linen. ASAE Publication Number
701P1103e, ed. Graeme Quick.

Mwaikambo, L.Y. 2006. Review of the history, properties and application of plant
fibres. African Journal of Science and Technology Science and Engineering
Series 7(2): 120-133.

Persson, S. 1987. Mechanics of Cutting Plant Material. St Joseph, MI, USA: ASAE.

Riddlestone, S., E. Stott, K. Blackburn and J. Brighton. 2006. A technical and

economic feasibility study of green decortication of hemp fibre for textile uses.
Journal of Industrial Hemp 11(2): 25-55.

Temafa GmbH. 2004. Temafa GmbH lin line. Journal of Natural Fibers 1(1):
119-121.

Wikipedia Online. 2009. Hemp. http://en.wikipedia.org/wiki/Hemp (2009/10/28).

100