the 1 international symposium on smart material and mechatronics

the 1 international symposium on smart material and mechatronics
THE 1ST INTERNATIONAL SYMPOSIUM
ON SMART MATERIAL AND MECHATRONICS
Makassar-Gowa, 23-24, September, 2014
Kampus II Fakultas Teknik Universitas Hasanuddin,
Jl. Poros Malino No 72, Gowa, Sulawesi Selatan, Indonesia
Editor :
•
•
•
Rafiuddin Syam, PhD – Hasanuddin University—Japan
Prof. Keigo Watanabe-Okayama University-Japan
Prof. Mitsuhiro Okayasu-Ehime University-Japan
Graduate School of Mechanical Engineering
Faculty of Engineering
University of Hasanuddin
ii
ii
PROCEEDING OF
THE 1ST INTERNATIONAL SYMPOSIUM
ON SMART MATERIAL AND MECHATRONICS
ISBN 978-602-71380-1-8
© 2014 Graduate School of Mechanical Engineering, Faculty of Engineering, University of Hasanuddin
This work is copyright. no part may be reproduced by any process without prior written
permission from the Editors. Requests and inquiries concerning reproduction and rights should be
addressed to Rafiuddin Syam, PhD – Graduate School of Mechanical Engineering, Faculty of
Engineering, University of Hasanuddin –Makassar-Indonesia
email to [email protected]
The intellectual property of each paper included in these proceedings remains vested in the
Authors as listed on the papers.
Published by :
Graduate School Mechanical Engineering Engineering Faculty of Hasanuddin University
Jl. P. Kemerdekaan Km 10 Makassar
Sulawesi Selatan, Indonesia 90221
Telp/Fax : (0411) 586015
Email : [email protected]
Website: pasca.unhas.ac.id
ii
iv
Foreword
First, we would like to thank all participants who are willing to send the results of scientific
research papers and participated in the International Symposium on Smart Material and
Mechatronics 2014 As the first International Symposium conducted by Graduate School of
Mechanical Engineering, Hasanuddin University, and our challenge theme is Smart Material and
System Mechatronics.
The International Symposium on Smart Material and Mechatronics 2014 presented as gifts
birthday of Hasanuddin University to 58 years old. We hope, in this symposium some steps are to
conduct research and publications acceleration in the field techno-science include Smart Materials
and System Mechatroncis. The both field of science that became one of the sections that need to be
encouraged to become an advanced nation of Indonesia in the field of technology. Furthermore, the
results of research are good input for accelerating industry.
In this symposium we invite on the field of research area, but not limited to:
• Metal Material, Smart Material, Concrete Material, Composite Material, Strength and stress
of Material, Structure Analysis, Cad and Cam, Vibration and Acoustic, Transportation
System, Environmental Study, Mining, Chemistry, Naval Architecture, Hydrodynamics,
Machining, Production, Heat and Mass Transfer, Thermodynamics, Fluid Mechanics,
Agriculture Engineering, Education Engineering, conservation energy, new energy and
renewable energy, internal and external combustion engine, Civil Engineering.
• Mechatronics, Mobile Robot, Manipulator Robot, Intelligence Systems, Softcomputing,
Artificial Intelligent System, Simulation System, Modeling Systems, Industrial engineering,
Ergonomics, Physics, Applied Mathematics, Computer Science, Information Science,
Smart System on Building, Mechanical System, Design systems, Control System, Control
Practice, Adaptive Control, Sensor Engineering, Electrical and Electronics Engineering,
Material on Electrical and Electronics, Environmental Engineering, Rescue Systems, Smart
Vehicle, Smart Building, Biological Engineering, medical Engineering, Artificial Systems,
Fuzzy Logic Theory and Application.
Thanks to all of my college and all of students of graduate school of Mechanical Engineering
Hasanuddin University.
Makassar-Gowa, September 23, 2014
Yours
Rafiuddin Syam, PhD
Chairman
v
vi
Contents
ST
PROCEEDING OF THE 1
INTERNATIONAL SYMPOSIUM ON SMART MATERIAL AND
MECHATRONICS
No
Title
Pages
1
Measurement and Control for Unmanned Ground, Aerial and Underwater Vehicles
Keigo Watanabe, Okayama University, Japan
Mechanical Properties of Composite Materials,
Mitsuhiro Okayasu, Ehime University, Japan
The Effect of Tool Dimension, Tool Overhang and Cutting Parameters Towards Tool Vibration and
Surface Roughness on Turning Process Zuingli Santo Bandaso, Johannes Leonard
ATIM, Indonesia, University of Hasanuddin, Indonesia
Effect of Water Volume and Biogas Volumetric Flowrate in Biogas Purification Through Water
Scrubbing Method Hendry Sakke Tira, Yesung Allo Padang, Mirmanto and Rio Cristovan Mantiri
Mechanical engineering, Mataram University, Mataram, Indonesia
Effect of Solution Treatment Process on Hardness of Alumina Reinforced Al-9Zn Composite
Produced by Squeeze Casting
Dwi Rahmalina, Hendri Sukma, I. Gede E.Lesmana, Asrin HalimPancasila University, Jakarta,
Indonesia
Review of Carbon Fiber Reinforced Polymer Reinforced Material in Concrete Structure
Ayuddin Gorontalo State University, Indonesia
Material properties of various light metals produced by heated mold continuous casting
Yuta Miyamoto, Mitsuhiro Okayasu, Japan
Microstructure and Mechanical Properties of Al-10Zn-4.5Mg-xCu Turbine Impeller Produced by
Investment Casting,
Muhammad Syahid, Bondan T. Sofyan1,Insani Mukhlisa University of Indonesia, Indonesia
Study of Performance Improvement of Various Stoves with Waste Biomass Briquettes Fuel
Effendy Arif Sallolo Suluh*, Unhas, *ATI Dewantara Palopo, Indonesia
Position Control of an X4-Flyer Using a Tether
Yusuke Ouchi, Keigo Watanabe, Keisuke Kinoshita, Isaku Nagai, Okayama Univercity, Japan
Development of a Mobile Robot as a Test Bed for Tele-Presentation
Diogenes Armando D. Pascua, Sherwin A. Guirnaldo, Mindanao State University Philippines
Intelligent Machine Vision for Automated Fence Intruder Detection Using Self-organizing Map
Veldin A. Talorete, Jr., Sherwin A. Guirnaldo MSU–Iligan Institute of Technology, Philippines
Simulation and Experimental Works of Quadcopter Model for Simple Maneuver
Rafiuddin Syam and Mustari, Hasanuddin University and Dayanu Iksanuddin university, BaubauIndonesia
Design of Wheeled Mobile Robot with Tri-Star Wheel as Rescue Robot
Rafiuddin Syam, Wahyu H. Piarah and *Paisal, Hasanuddin University, *State Polytechnic of
Ambon, Indonesia
Application of Genetic Algorithm for Determining the Optimum Ship Route
Faisal Mahmuddin, Rahmad Patarru, Rahimuddin Samad, Hasanuddin University, Makassar,
Indonesia
Fatigue Life Prediction In Journal Bearing
Irsyadi Yani Hasan Basri and Hafizd Ibrahim Marsil, Sriwijaya University, Indonesia
K01
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
vii
1-6
7-13
14-18
19-22
23-30
31-35
36-38
39-45
46-49
50-58
59-66
67-71
72-76
77-82
83-86
17
18
19
20
21
22
23
24
Development of 5-DOF Robot Arm Manipulator
Ismail Thamrin, Irsyadi Yani, Sriwijaya University, Palembang, Indonesia
Kinodynamic Motion Planning for an X4-Flyer Using a 2-Dimentional Harmonic Potential Field
Kimiko Motonaka, Keigo Watanabe, and Shoichi Maeyama, Okayama University Okayama, Japan
Underactuated Control for a Blimp with Four-Propellers by a Logical Switching Method,
Yoshikazu Nakamura, Keigo Watanabe, Isaku Nagai, Okayama University Okayama, Japan
New Waste Beverage Cans Identification Method,
Firmansyah Burlian, Yulia Resti, Ihsan Budiman, Sriwijaya University,Indonesia
Experimental Test of the Thermoelectric Performance on the Dispenser Cooler
Zuryati Djafar, Amrullah, Wahyu H. Piarah, Syukri Himran, Hasanuddin University, Indonesia
Potential Coir Fibre Composite for Small Wind Turbine Blade Application
Bakri, S.Chandrabakty, R. Alfriansyah, A. Dahyar, Tadulako University, Palu, Indonesia
A new development of thermosiphon solar hot water with paralel-serpentine tube configuration,
Mustofa, Yuli Asmi Rahman, Basri, University of Tadulako, Indonesia
Optimal Design of V-shaped Absorber Plate to the Performance of Solar Water Heater
Jalaluddin, Effendi Arief, Rustan Tarakka, Hairul Arsyad, Andi Mangkau, Labusab, University of
Hasanuddin, Indonesia
vii
87-89
90-93
94-97
98-101
102-106
107-109
110-113
114-119
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN: 978-602-71380-1-8
Measurement and Control for Unmanned Ground,
Aerial and Underwater Vehicles
Keigo Watanabe
Department of Intelligent Mechanical Systems
Graduate School of Natural Science and Technology, Okayama University
Okayama, Japan
[email protected]
Abstract—In this keynote, we overview latest representative research results in our
laboratory on several Unmanned Vehicles (UVs): i.e., Unmanned Ground Vehicles
(UGVs) such as a mobile robot with two-wheeled independent driving mechanism, a carlike four-wheeled mobile robot on the ground, etc.; Unmanned Aerial Vehicles (UAVs)
such as a VTOL aerial robot with four rotors in the air etc.; and Unmanned Underwater
Vehicles (UUVs) such as a robotic manta as one kind of Autonomous Underwater
Vehicles (AUVs) in the underwater etc.
First, we introduce an obstacle avoidance problem for a nonholonomic four-wheeled
mobile robot as UGVs using an image-based control approach, where a fuzzy controller
is designed for controlling a target line extracted from the camera image, together with
the information on the potential field of the environment. In addition, we develop a
stabilizing controller for such a mobile robot to realize an automatic parking system, by
applying an invariant manifold method.
Secondly, we show a measurement system in 3D space for UAVs, where an indoor X4
Flyer, which is a VTOL type aerial robot with four rotors, is considered. In this research,
a position measurement system in an indoor 3D space is developed by using a stereo
camera. In particular, to enable measurement in a dark place, the position measurement
system is built by attaching an infrared LED marker to an object and using two cameras
equipped with an infrared transmitting filter.
Thirdly, among UUVs, we develop a robotic manta as a kind of fish robot with pectoral
fins. Such a biomimetic thruster is expected to provide noiseless propulsion, and to be
more maneuverable in complex near-shore environments and highly efficient in energy
consumption, compared to the conventional AUVs with a propeller-based thruster.
Index Terms—Unmanned vehicles, nonholonomic wheeled mobile robots, VTOL aerial
robots with four rotors, fish robots with pectoral fins.
K-01
.
ISBN: 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
Mechanical Properties of Composite Materials
Mitsuhiro Okayasu
Dept. Materials Science and Engineering
Ehime University
Matsuyama, Ehime, Japan
[email protected]
Abstract— An examination has been made of the mechanical
and failure properties of several composite materials, such as a
short and a long carbon fiber reinforced plastic (short- and
long-CFRP) and metal based composite material. The short
CFRP materials were used for a recycled CFRP which
fabricated by the following process: the CFRP, consisting of
epoxy resin with carbon fiber, is injected to a rectangular plate
cavity after mixing with acrylonitrile butadiene styrene resin
with different weight fractions of CFRP. The fatigue and
ultimate tensile strength (UTS) increased with increasing
CFRP content. These correlations, however, break down,
especially for tensile strength, as the CFPR content becomes
more than 70%. Influence of sample temperature on the
bending strength of the long-CFRP was investigated, and it
appears that the strength slightly degreases with increasing the
temperature, due to the weakness in the matrix. Broken fiber
and pull-out or debonding between the fiber and matrix were
related to the main failure of the short- and long-CFRP
samples. Mechanical properties of metal based composite
materials have been also investigated, where fiber-like high
hardness CuAl2 structure is formed in aluminum matrix.
Excellent mechanical properties were obtained in this alloy,
e.g., the higher strength and the higher ductility, compared to
the same alloy without the fiber-like structure. There are
strong anisotropic effects on the mechanical properties due to
the fiber-like metal composite in a soft Al based matrix.
use CFRP seems to be thrown away into landfill without any
consideration of environmental problems [3]. This occurrence
will be a problem in the future, because the amount of waste
CFRP will increase [4]. Up to date, several researchers have
investigated the mechanical properties of CFRP including
recycled CFRP, the information available appears to be
insufficient.
On the other hand, metal matrix composite material
(FRM) is also important material as engineering material. This
is because of their outstanding mechanical properties. Metal
matrix composite with silicon carbide particle (SiC) are one of
the widely known composites, which have high strength, high
hardness, high wear resistance and high corrosion resistance
[5]. Effect of clustering on mechanical properties of aluminum
alloy 2024-SiC metal matrix composite has been investigated.
Fracture toughness and tensile tests were carried out, and their
mechanical strengths were estimated well by a model [6].
Although CFRP and FRM are excellent materials to use
in various engineering application, there would have still
technical issue for recycling technique and lack of information
regarding their mechanical properties. In this study, our
experimental results obtained previously for the material
properties of long-CFRP, short-CFRP [7] and FRM [8]-[9]
were summarized to consider the mechanical properties of the
composite materials.
Index Terms— CFRP; Carbon fiber; Mechanical property;
Crack growth; Failure mechanism
II. EXPERIMETNAL PROCEDURE
II-1. Long-CFRP and short-CFRP
The long-CFRP, consisting of epoxy resin (thermosetting
high polymer) with a volume fraction of 60% carbon fiber,
was used. Fig. 1 shows the photograph of the long-CFRP
samples showing the carbon fibers and matrix. The shortCFRP samples were made by the following process. The longCFRP was first crushed using a rotating blade to make small
fragments for which the average length by width is 3.4 mm 
0.4 mm. The crashed long-CFRP pieces were then separated
individually into fiber and epoxy resin after the ball milling
process. Most part of the surface of separated carbon fibers is
not already coated by epoxy resin, while some fiber bundles
were present that contained epoxy resin [3]. After the grinding
process, it was found that the mean length of the carbon fibers
is about 200 m. The short-CFRP samples, consisting of
acrylonitrile butadiene styrene resin and CFRP pieces, were
fabricated using standard mixing, grinding and injection
molding procedures. In this case, the CFRP pieces were added
to the ABS resin before the injection process with five
I. INTRODUCTION
In recent years, composite materials have received
special attention because of the excellent mechanical
properties. In particular, production amount of carbon fiber
reinforced plastics (CFRP) have been increased due to their
high strength and low specific weight [1]. CFRP material has
come into practical use for the aerospace and automotive
industries, because of their contribution to higher fuel
efficiency. In fact, the demand for CFRPs has dramatically
increased in recent years [2]. As aerospace and automotive
parts are sometimes employed in atmosphere with high
temperature, examination of mechanical properties of CFRP at
high temperature would be required. In addition, development
of the recycling technology for CFRP has been significantly
important due to their high production amount. Indeed, post-
1
International Symposium on Smart Material and Mechatronics
ISBN: 978-602-71380-1-8
different weight fractions of 0 (i.e., pure ABS), 10, 30, 50 and
70 wt.%. The injection molding process was carried out to
make the short-CFRP with simple rectangular plates 150 mm
 150 mm  3 mm.
Dumbbell-shaped specimen and compact tension (CT)
specimen were used in this test, which obtained from the
center area of the rectangular plate as shown in Fig. 2. In this
case, the rectangular plate cut in two different directions from
the mid-section, either with the loading direction (longitudinal
axis of the specimen) in the direction perpendicular (Type T)
or parallel (Type L) to the flow (or carbon fiber) direction.
The dimension of the parallel area in the dumbbell-shaped
specimen is 7 mm (l) × 3 mm (w) × 1 mm (t), and that of CT
specimen is W = 24.5 mm and B = 3 mm. The CT specimen
was designed based upon the ASTM standard E399 [10]. In
the mid-section of the CT specimens, a through-slit (15 mm in
length with a V-notch root angle of 45 degrees) was machined.
gives a schematic diagram of the heated mould continuous
casting apparatus, consisting of a graphite crucible with runner,
a graphite mould, a cooling device and pinch rolls for
withdrawal of the cast metal. The cast samples in the shape of
a long round bar (4 mm  1 m) was made. The casting
pressure was controlled by the level of molten metal in the
crucible, controlled by furnace displacer block. The
temperature of the molten metal was maintained at about
843K, which is 20K above the melting point of its Al alloy.
The molten metal was cast through the runner and graphite
mould before the cooling process. The graphite mould was
heated to approximately 853K, which is just above the
liquidus of the Al-Cu alloy. For the solidification process, the
aluminum alloy was cooled directly by water flowing to the
exit just out of the graphite mould (see Fig. 3). Interestingly,
with this casting process, a unidirectional growth
microstructure was created, which could be associated with
metal composite material. Fig. 4 depicts microstructure of Al33%Cu sample with the axial and transverse directions. The
primary -Al phase is visible as a dark region. A fine fiberlike eutectic structure of CuAl2 phases with unidirectional
growth along its axial direction can be observed.
Fig. 5 displays the test specimens formed with a
rectangular shape. Note, in this case, tiny special specimen
was designed to examine the metal composite effect on the
mechanical properties, namely anisotropic microstructual
effects. The specimens are denoted as (i) axial direction (OL)
and (ii) transverse direction (OT), as indicated in Fig. 5.
Fig. 1 Photograph of long-CFRP plate.
(a) Type T
Flow
direction
(b) Type L
Fig. 2 Schematic diagrams showing the test specimens in
the short-CFRP samples [7].
Fig. 3 Schematic diagrams of the heated mould continuous
casting system and cooling system [8].
II-2. Metal composite aluminum alloy
In the present study, an attempt was made to create FRM
materials via our heated mould continuous casting technology
(HMC) with a eutectic aluminum alloy. Concept of this
technology is as follows: unidirectional microstructure with
thin fiber-like phases was created by the unidirectional rapid
solidification process. In this approach, An Al-33%Cu eutectic
alloy was selected to make metal composite Al alloy. Fig. 3
Because of the tiny specimen, finite element analysis was
conducted to verify the stress-strain distribution before the
testing. Fig. 5 also indicates the FEA stress distribution on the
loading direction (x-axis). From this result, it is clear that the
high stress level is uniformly distributed in the sample of
parallel area. Thus, the material properties can be estimated to
understand their material characteristics
2
ISBN: 978-602-71380-1-8
International Symposium on Smart Material and Mechatronics
samples would be fractured by the crack growth between
the fibers, namely delamination between the fiber and
matrix. On the other hand, fibers are completely fractured
for 0-CFRP sample, which makes high bending strength.
Fig. 8 presents bending stress - strain curves for the
CFRP tested at different sample temperatures, e.g., 20C,
50C and 100C. There is clear temperature effect on the
bending properties, where the higher the mechanical
properties are obtained for the specimen tested at the lower
temperature. Similar trends were also seen in their fatigue
properties. Fig. 9 indicates the S-N curves for their CFRP
samples. It is obvious that high fatigue strength is detected
for the CFRP at low temperature. Their fracture
characteristics ware further investigated. Fig. 10 shows the
fracture surfaces of the CFRPs after the bending test at
different temperatures. It is interest to mention that there are
different dense of the epoxy. It is seen that low density of
epoxy is obvious for the samples at the higher testing
temperatures. This result infers that the epoxy may have
been melted during the heating process.
Fig. 4 SEM images of the HMC Al-33%Cu alloys, showing
microstructure [8].
(a) Axial direction (OL)
4 mm
(b) Transverse direction (OT)
0
45
4 mm
Fig. 5 Schematic illustration of the specimens and their
position; and FEA model to determine the stress distribution in
the specimen and stress distribution to loading direction [8].
90
III. RESULTS
III-1. Long-CFRP materials
Fig. 6 shows representative bending stress - strain
curves for the long-CFRP with different fiber direction. As
seen, different tensile properties are obtained depending on
the fiber direction. It is clear that low bending properties are
obtained as the CFRP with fiber direction of more than 45
against the loading direction, while high mechanical
properties are detected for the CFRP with 0 fiber direction.
Fig. 7 depicts the fracture surfaces of their specimens after
the bending tests. As seen, fiber surfaces are observed for
the specimens with fiber direction of 45 and 90. Those
Fig. 7 SEM images showing the fracture surface of the specimen.
Fig. 8 Stress-strain curves for the long-CFRP with
different sample temperature.
Fig. 6 Stress-strain curves for the long-CFRP with
different fiber direction.
3
International Symposium on Smart Material and Mechatronics
ISBN: 978-602-71380-1-8
mean UTS value for CFRP 50%-Type L is more than 1.6
times higher than the CFRP 50%-Type T one. This
corresponds to the anisotropic effect in the sample, where
the fiber direction prevails for the strength although the
fiber length is as short as about 200 m as mentioned above.
20C
50C
100C
Fig. 9 S-N curves for the long-CFRP tested at different temperature.
Fiber direction (L) = 0°
20C
Fiber
50C
Vacancy
Fig. 11 Ultimate tensile strength and fracture strain of the
short-CFRPs: (a) Type T and (b) Type L [7].
Fiber
Fig. 12 shows the relationship between the stress
amplitude and cycle number to failure (S-N curve) of the
short-CFRP. It should be noted first that the arrows in this
figure indicate the specimens which did not fail within 10 7
cycles. From Fig. 12(a), the S-N relationships, including the
endurance limit (en), seem to be similar level for all Type
T samples, while the slope of their S-N relationships is
slightly different depending on the CFRP content. For
example, the higher the CFRP content (e.g., CFRP 70%),
the lower the slope of S-N relations, in which S vs. N for
CFRP 70%-Type T crosses those for the other Type T
samples around 103 ~ 104 cycles as indicated in Fig. 12(a).
For CFRP 70%-Type T, the lowest slope of the S vs. N
curve is obtained for CFRP 70%-Type L (Fig. 12(b)),
which also crosses the other ones at around 103 cycles but
only for 0%- and 10%-Type L. Interestingly, the endurance
limit for both CFRP 70% is the same level of about 15.4
MPa. The S-N curves for CFRP 30%- and 50%-Type L are
located at a higher level compared to the others, even
though the endurance limits for CFRP 30%- and 50%-Type
L are close to that for CFRP 70%-Type L. An important
observation from Fig. 12(a)(b) is that relatively high
endurance limit was obtained for both CFRP 70% in spite
of the low tensile properties (Fig. 11). Such fatigue
properties for CFRP 70% are associated with different
100C
Vacancy
Fiber
Fig. 10 SEM images of long-CFRP showing the fracture
surfaces of the specimen tested at different temperatures.
III-2. Short-CFRP materials
Fig. 11 shows the ultimate tensile strength (UTS) for
all the short-CFRP samples. Different tensile properties are
obtained depending on the CFRP content and type (fiber
direction). There is no clear anisotropic effect on the tensile
properties for the CFRP 0% samples: UTS = 38.8 MPa for
Type T and UTS = 40.2 MPa for Type L. For both samples
Type T and L, the tensile strength increases with increasing
CFRP content, but a considerable drop in the tensile
strength was detected for CFRP 70%. The overall tensile
strength for Type L is higher than that for Type T,
particularly CFRP 30%- and CFRP 50%-Type L, e.g., the
4
ISBN: 978-602-71380-1-8
International Symposium on Smart Material and Mechatronics
crack growth rates. For instance, a rough fracture surface
makes low crack growth rate due to the low crack driving
force arising from severe crack closure [11].
To understand clearly the fatigue behavior of the
short-CFRP samples, the S-N relationships were
quantitatively evaluated by a power law dependence of
cyclic stresses and cycles to failure:
(1)
 a   f Nf-b , MPa
obviously high in the OL sample compared to the other one
due to the fiber-like composite effect. The tensile properties of
both Al-33%Cu alloys for OL and OT are UTS = 489.8 MPa
and f = 4.9% and UTS = 384.1 MPa and f = 4.2%,
respectively. Such different tensile properties (OL vs. OT) are
reflected to the reinforcement by the fibrous structure. The
fine eutectic structure in the longitudinal direction can
enhance the material properties (OL), and that tensile strength
is much higher than that for the conventional cast Al alloys.
On the other hand, the high material ductility obtained for the
OL samples can be explained using the failure mechanism.
where a is the stress amplitude, Nf represents the cycle
number to final fracture, f is the fatigue strength
coefficient and b is the fatigue exponent. Those values (f
and b) were obtained by least square analysis. In this case,
an increased fatigue life is expected for a decreasing fatigue
strength exponent b and increasing fatigue strength
coefficient f. In the present case, the f and b for the CFRP
50%-Type L sample shows high fatigue strength, e.g., f =
51.3 and b = 0.07. On the other hand, different fatigue
properties were obtained for both CFRP 70% samples with
lower b and lower f values (f = 21.3 and b = 0.02).
Applied tensile stress, MPa
600
500
OL sample
OT sample
400
300
200
100
0
104 cycles
0
2
4
6
8
Strain, %
Fig. 13 Stress-strain curves for the Al-33%Cu samples, obtain
from axial (OL) and transverse (OT) directions [8].
Fig. 14 shows SEM images of the microstructure of both
Al-33%Cu alloys. As seen in the OL sample, elongated
microstructural characteristic was obtained near the crack,
which would be caused by the fibrous structure. On the
contrary, the crack, created in-between the fiber and matrix,
can be observed for OT samples.
(a)
OL
(b)
OT
Fig. 14 SEM images of the Al-33% Cu alloys showing the crack
paths in the mid-section of the samples: (a) axial direction
(OL) and (b) transverse direction (OT) [8].
Fig. 12 S-N curve for the short-CFRPs: (a) Type T and (b)
Type L [7].
Fig. 15 shows the relationship between the stress
amplitude and fatigue life for both the OL and OT samples. It
should be noted first that the arrows in the S-N diagrams are
the specimens which did not fracture within 50,000 cycles. It
is clear from the S-N diagrams that the S-N relationship for the
OL is located at higher values compared to the OT sample.
The endurance limit for the OL samples is approximately 400
III-3. Metal composite materials (Al-33%Cu alloy)
Fig. 13 shows the stress-strain curves for the Al-33%Cu
alloy: OL and OT samples. It can be seen that the relatively
linear stress vs. strain relations are obtained for the both
samples. The tensile strength and elongation to failure are
5
ISBN: 978-602-71380-1-8
International Symposium on Smart Material and Mechatronics
MPa, which is about twice as high as that of the OT samples.
It should be pointed out that S-N relationship for the OL is
formed more plateau compared to the other one. This may be
affected by the material brittleness for the OL samples. In fact,
such S-N relations is sometimes seen in relatively brittle
materials, e.g., ceramics [12][13].
REFERENCES
[1] G. Ben, “What we have accomplished in nedo project for
automotive structures,” Proceedings of 16th Int. Conf. Comp.
Mater. (2007–7).
[2] E. Kappel, D. Stefaniak, T. Spröwitz, C. Hühne, “A semianalytical simulation strategy and its application to warpage of
autoclave-processed CFRP parts,” Compo.: Part A, 2013, vol.
42, pp. 1985–1994.
[3] K. Ogi, T. Nishikawa, Y. Okano, I. Taketa, “Mechanical
properties of ABS resin reinforced with recycled CFRP,” Adv.
Composite Mater., 2007, vol. 16, pp. 181–184.
[4] T. Kirihara, T. Kawashima, J. Takahashi, T. Matsuo, K. Uzawa,
“Demand and disposal forecast for carbon fibre by bottom-up
approach,” Proceedings of 18th Int. Conf. Comp. Mater. TH32
(2011-8) 1–4.
[5] T. Ozhen, E. Kilickap, O. Cakir, “Investigation of mechanical
and machinability properties of SiC particle reinforced AlMMC,” J. Mater. Process. Technol., 2008, vol. 198, pp. 220–225.
[6] S-J. Hong, H-M. Kim, D. Huh, C. Suryanarayana, BS. Chun,
“Effect of clustering on the mechanical properties of SiC
particulate-reinforced aluminum alloy 2024 metal matrix
composites,” Mater. Sci. Eng. A, 2003, vol. 347, pp. 198–204.
[7] M. Okayasu, T. Yamazaki, K. Ota, K. Ogi, T. Shiraishi,
“Mechanical properties and failure characteristics of a recycled
CFRP under tensile and cyclic loading,” Int. J. Fatigue, 2013,
vol. 55, pp. 257–267.
[8] M. Okayasu, R. Sato, S. Takasu, “Effects of anisotropic
microstructure of continuous cast Al-Cu eutectic alloys on their
fatigue and tensile properties,” Int. J. Fatigue, 2012, vol. 42, pp.
45–56.
[9] M. Okayasu, S. Takasu, S. Yoshie, “Microstructure and material
properties of an Al-Cu alloy provided by the Ohno continuous
casting technique,” J. Mater. Process. Technol., 2010, vol. 210,
pp. 1529–1535.
[10] Annual Book of ASTM Standards, in: ASTM E399, vol. 03.01,
American Society for Testing and Materials, 2008, pp. 497–529.
[11] M. Okayasu, Z. Wang, “Experimental investigation of the
effects of artificial wedges on fatigue crack growth and crack
closing behavior in annealed SAE1045 steel,” Int. J. Fatigue,
2007, vol. 29, pp. 962–976.
[12] M. Okayasu, S. Aoki, M. Mizuno, “Effects of silver-based metal
electroplate on fatigue properties of PZT ceramics,” Int. J.
Fatigue, 2008, vol. 30, pp. 1115–1124.
[13] M. Okayasu, M. Hitomi, H. Yamazaki, “Mechanical and fatigue
strengths of silicon nitride ceramics in liquid aluminum alloys,”
J. Eur. Ceram. Soc., 2009, vol. 29, pp. 2369–2378.
[14] ES. Puchi-Cabrera, MH. Staia, DT. Quinto, C. VillalobosGutiérrez, E. Ochoa-Pérez, “Fatigue properties of a SAE4340
steel coated with TiCN by PAPVD,” Int. J. Fatigue, 2007, vol.
29, pp. 471–480.
Fig. 15 S-N curve for the Al-33%Cu samples, obtained in
the axial direction (OL) and transverse direction (OT) [8].
IV. CONCLUSIONS
An examination has been made of the mechanical and
failure properties for the composite materials. The results have
yielded the following conclusions.
1. Mechanical properties (tensile strength and fatigue
strength) of the CFRP samples are directly attributed to the
sample temperature and fiber directions. The epoxy seems to
be melted when heated to the higher temperature, leading to
the low mechanical properties.
2. The tensile strength of the short-CFRP is found to depend
not only on the CFRP content, but also on the fiber direction.
The tensile strength increases with increasing CFRP content,
but drops suddenly for short-CFRP with higher fiber content,
i.e., 70%. In addition, under the same CFRP content, the
higher tensile strength is detected as the fiber direction is
parallel to the loading direction.
3. A clear anisotropic microstructure was obtained, namely
a fine lamellar eutectic structure with unidirectional growth
along its axial direction. The eutectic structure was formed by
the primary -Al phase and CuAl2 phase, i.e., fiber-like
reinforcement. The tensile and fatigue properties of the
samples in the longitudinal direction of the loading are more
than 30% higher than those for the cast samples perpendicular
to the casting direction.
ACKNOWLEDGMENTS
The author would like to acknowledge many useful
advices and suggestions of Prof. K. Ogi. The author also
appreciates technical support of Mr. K. Ota, Mr. T. Yamazaki,
Mr. R. Sato and H. Iwai.
6
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
The Effect of Tool Dimension, Tool Overhang and
Cutting Parameters Towards Tool Vibration and
Surface Roughness on Turning Process
Zuingli Santo Bandaso
Mechanical and Industrial Engineering Study Program
Akademi Teknik Industri Makassar
Makassar, Indonesia
[email protected]
Johannes Leonard
Mechanical Engineering, Engineering Faculty
University of Hasanuddin, Makassar
Makassar, Indonesia
[email protected]
Abstract— Turning process is the removal of metal from the
outer diameter of a rotating cylindrical workpiece. Turning is
used to reduce the diameter of the workpiece, usually to a
specified dimension, and to produce a smooth finish on the metal.
This research investigates the effect of feed rate, spindle speed,
tool overhang and tool dimensions toward vibration amplitude
and surface roughness on turning process. This study uses both
statistical and graphical analysis of the data collected. The
experimentation was carried out on conventional lathe machine
with straight turning operation. Material used as workpiece was
St.60 carbon steel which was turned with HSS tool bit with the
dimension of 3/8 Inches and ½ Inches. Cutting parameters varied
by spindle speed, feed rate, and tool overhang, while the depth of
cut is maintained at a depth of 0.5 mm. The vibration data of
cutting tool obtained from a transducer (vibrometer) mounted at
a distance of 10 mm from the tip of the cutting tool during the
cutting process takes place, whereas the surface roughness data
obtained from measurements of surface roughness apparatus
after turning process. The results showed that, The effect of feed
rate, spindle speed, tool overhang, and tool dimension
simultaneously towards vibration amplitude and surface
roughness has a grater effects on the use of 3/8 inches cutting tool
than ½ inches cutting tool. With the use of the same tool
dimensions obtained that, The most influential parameters on the
vibration amplitude is tool overhang while the most influential
parameter on surface roughness value is feed rate.
Nowadays, a standard procedure used to avoid vibration
during turning is planning the selection of cutting speed, feed
rate, and depth of cut carefully. A method applied is usually
based on an operator’s experience and also trial and error
method to gain a proper cutting parameters in the machining
process.
Vibration in machining process occurs throughout the
cutting process takes place which is derived from some
sources, such as frame structure of the machine, cutting tool
type, types of material that are cutting, etc. Vibration on
machining process is very complicated because it involves lots
of variables. Nevertheless, at least two kinds of vibrations
occurred on machining process, this covers both force
vibration and self-excited Vibration. Force vibration is usually
gained from components in the machine itself, for example
because there are damaged gear components, imbalances on
machine components, misalignment of the shaft, the electrical
motor rotation, and etc. Self-excited vibration which is socalled Chatter caused by the interaction between the release of
chips and cutting tool which causes interference with the
cutting area. Chatter or self-excited always affects on the
surface roughness of machining product. Therefore a vibration
which is caused by self excited vibration related to the surface
roughness as the result of machining [2]
Some previous research, both descriptive and experiment
have studied how vibration affects surface roughness towards
surface roughness as a product of machining process
[1][2][3][4][5]. One of those researches is the effect of spindle
speed, feed rate, and depth of cut towards tool vibration
amplitude and surface roughness of the workpiece on the lathe
machine, in which concludes that spindle speed is the most
influential towards vibration and so does the surface roughness,
then followed by the feed rate and the last is the depth of cut.
Based on that research, the writer conducted a further research
towards some variables which has not been experimented
before, it is the effect of tool overhang and cutting tool
dimension in which involving the same variation of cutting
conditions that that have been studied before.
Key words— Turning, vibration, surface roughness, cutting
tool
I. INTRODUCTION
Challenges faced by today's modern machining industry
primarily focused on achieving a high quality product. One of
them are the quality of surface roughness. Surface roughness of
a product of machining process can affect some functions of
these products such as surface friction, heat transfer, spreading
capabilities of lubrication, coating, and others. Thus, in
practical field,the desired of surface roughnes value will be the
reference of cutting parameters selection [1]. Turning is one of
the main machining processes used in the process of cutting a
rotating cylindrical workpiece. Lots of machinery components
made through turning process. Problems are often encountered
in all of the machining process especially on turning, is the
vibration during the material cutting process. This vibration
will affect the quality of the products, one of them is the
surface roughness [1].
II. THEORITICAL BACKGROUND
A. Vibration in metal cutting process
Vibration is a back and forth motion about its fixed
equilibrium position. The equilibrium position means a
7
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
C. General overview of turning process
Turning process is the removal of metal from the outer
diameter of a rotating cylindrical workpiece. Turning is used to
reduce the diameter of the workpiece, usually to a specified
dimension, and to produce a smooth finish on the metal.
Three main parameters in turning operation are cutting
speed, feed rate, and depth of cut. Other factors such as
workpiece material and type of cutting tool actually has a
considerable influence. However, the three parameters
mentioned above are parts that can be set by the operator
directly on the lathe machine.
Cutting speed (also called surface speed or simply speed)
may be defined as the rate (or speed) that the material moves
past the cutting edge of the tool, irrespective of the machining
operation used. The equation of cutting speed can be determine
from the equation below
condition in which an object is on motionless position if there
is no force acting on the object. Vibration has similar amplitude
(distance deviation furthest to the midpoint) [6]. All objects
that have mass and elasticity are able to vibrate.
In the machining process, there are three mechanical
vibrations caused by insufficiency of dynamic stiffness in
machinery equiptment. The vibrations are free vibration, force
vibration, and self-excited vibration. A tool holder, workpiece
and the machine itself are parts of machinery which causes
vibration. The free vibration is usually caused by shocks effect,
such as, the presence of impulse waves that are transferred to
the cutting tool or at the time between the beginning of the
cutting tool with the workpiece. Force vibration is caused by
periodic force which occurred in the system, for example due
to the imbalance of machine components for instance, gears
system, spindle or bearing. Self-excited vibration usually
occurs as a result of dynamic instability that occurs during
metal cutting processes. As it shows, the self-excited vibration
is the most uncontrolled vibration while two other vibrations
can be controlled through arranging cutting parameters on the
machine [7].
(1)
Where Cs = cutting speed; m/minute, D = workpiece
diameter /mm, N= spindle speed, revolution / minute. The
spindle speed (N) in eq.(1) is a measure of the frequency of a
rotation. It annotates the number of turns completed in one
minute around a fixed axis. The preferred speed is determined
by working backward from the desired surface speed (sfm or
m/min) and incorporating the diameter (of workpiece or cutter).
Feed rate, Vf , refers to how fast a lathe-tool should move
through the material being cut. This is calculated using the
Feed per Revolution for the particular material. It is expressed
in units of distance per revolution. Feed rate is determined
based on machine power, material properties of workpiece, tool
material, tool shape, and the most important is the expected
surface roughness.
Depth of Cut, the thickness of the material that is removed
by one pass of the cutting tool over the workpiece. [8]
B. Surface roughness
Surface roughness is a measurable characteristic based on
the roughness deviations as defined in the preceding. Surface
finish is a more subjective term denoting smoothness and
general quality of a surface. In popular usage,surface finish is
often used as a synonym for surface roughness.
The most commonly used measure of surface texture is
surface roughness. Withrespect to Figure 1, surface roughness
can be defined as the average of the vertical deviations from
the nominal surface over a specified surface length. An
arithmetic average (AA) is generally used, based on the
absolute values of the deviations, and this roughness value is
referred to by the name average roughness. In equation form
(1)
D. Regression Analysis
Regression Analysis is applied to study and measure the
scatistical relationship among two or more variables. In simple
regression analysis two variables are analyzed, whereas in the
multiple regression analysis more than two variables are
analyzed. In regression analysis, a regression equation was
about to set and used to describe a pattern or a function of the
relationship between variables. Variables to be called the
dependent variable is usually plotted on y-axis. While the
independent variable is the variable that is assumed to give
effect to the variation in the dependent variable and it is usually
plotted on x-axis.
Multiple linear regression, On this multiple linear
regression, there are several independent variables
(X1,X2,X3...Xn) which are connected to one dependent variable
(Y), those are parts of multivariate analysis to estimate the
regression coefficient to describe the effect of independent
variable towards dependent variable. In a multiple regression
test, all of the predictor variables are included in the regression
calculation simultaneously.
where Ra = arithmetic mean value of roughness, m (in); y = the
vertical deviation from nominal surface (converted to absolute
value), m(in); and Lm = the specified distance over which the
surface deviations are measured.
The AA method is the most widely used averaging method
for surface roughness today. Analternative, sometimes used in
the United States, is the root-mean-square (RMS) average,
which is the square root of the mean of the squared deviations
over the measuring length. RMS surface roughness values will
almost always be greater than the AA values because the larger
deviations will figure more prominently in the calculation of
the RMS value.
Figure 1. Deviations from nominal surface used in the two
definitions of surface roughness.
8
Proceeding of International Symposium on Smart Material and Mechatronics
If there are two independent variables (X1) and (X2) and the
dependent variable (Y), then the coefficients of the multiple
regression equation is determined by the following equation:
Hypothesis test of multiple correlation using F- test with
degrees of freedom (df) consists of:
df1 = df numerator = k (k = total of independent variable)
df2 = df denominator = n – k – 1 (n = numbers of pairs of
data or sample)
Value conversion of correlation coefficient R to Fcount uses
the following equation :
Y = a + b1X1 + b2X2 +... bnXn
(2)
E. Correlation Analysis
Analysis of correlation is an inferential analysis used to
determine the degrees of freedom or strength of the
relationship, shape or causal and correlation among research
variables. Type of statistical hypothesis testing correlations
includes simple correlations (bivariate), multiple correlation
and partial correlation.
Pearson’s Product Moment Correlation, this correlation is
used for interval/ratio data in which must meet the following
requirements:
 The sample is taken randomly
 Each variable of data is normally-distributed
 A linear regression equation
 Equation :
rxy 
n. xy   x y
[n x  ( x) 2 ][n y 2 ( y)2 ]
2
ISBN 978-602-71380-1-8
Fb 
R2 / k
(1  R 2 ) /( n  k  1)
(5)
Hypothesis testing criteria, namely:
Accept H0 if Fcount< Ftable, dan reject H0 if Fcount > Ftable
III. RESEARCH METHODOLOGY
A. Experimental set up
Material used as workpiece was St.60 carbon steel which
was turned with HSS tool bit (The Bohler Super Mo Rapid
Extra 1200 Brand) with the dimension of 3/8 Inches and ½
Inches. The cutting tool angles used were Side Relief = 11o,
Front Relief = 8 o, Side Rake = 12 o, Back Rake = 8 o [10]. The
experimentation was carried out on conventional lathe machine
and the method of cutting is shown in Fig. 1, referring to the
experimental set up that has been conducted by previous
researchers [11].
Figure 2, shows that workpiece which will be turned then
divided into four segments separated by grooves. The purpose
of this segmental separation is to minimize the affects of tool
wear which can effects surface quality towards the
effectiveness of measurement. Thus, in collected data
measurement, tool bit cuts four times before being substituted
by a sharpened cutting tool.
(3)
Coefficient of Determination, The Coefficient of
Determination is denoted as r2. This value states the proportion
of the overall variation in the value of the dependent variable
that can be explained or caused by a linear relationship with the
independent variables, the rest is explained by other variables
(errors or other variables). Coefficient of determination
expressed as the square of the correlation coefficient
r2 x 100% = n% meaning that the value of the dependent
variable can be explained by the independent variables of n%,
while the residual value of (100 - n)% explained by an error
(error) or the influence of other variables.. Meanwhile, for
correlation analysis with more dependent variables, there is a
correlation coefficient which is significantly sensitive with
amount of variables. Usually for multiple correlation analysis,
adjustment coefficients of determination are often used.
Multiple Correlation, multiple correlation is the
correlation between two or more independent variables
together with the dependent variable. Value which shows
directions and strength of the relationship between two or more
independent variables on the dependent variable is called
multiple correlations and denoted as R.
The Equation of multiple correlation of two independent
variable X1 and X2 with one dependent variable (Y) as [9] :
Figure 2. Schematic of experimental set-up for turning
R y.12 
r  r  2ry1 .ry 2 .r12
2
y1
2
y2
1  r122
(4)
B. Method of Collecting data
Measurement of datas is undertaken by a well-trained and
experienced lathe operator in the using of vibration
measurement instrument and surface roughness devices.
Collecting of measurement datas was arranged as factorial
designed so that the interactions between independent variables
can be observed more effectively. The independent variables in
the study are feed rate(Vf), spindle speed, tool overhang, and
the tool dimension. While dependent variables is the result of
Where Ry.12 = coefficient of multiple correlation
among X1, X2 and Y, ry1 = correlation coefficient between X1
and Y, ry2 = correlation coefficient between X2 and Y, ry12 =
correlation coefficient between X1 with X2
The tested hypothesis is two tailed test:
H0 : ρy.12 = 0
H1 : ρy.12 ≠ 0
9
Proceeding of International Symposium on Smart Material and Mechatronics
cutting tool vibration (Vrms) and the surface roughness (Ra) of
workpiece. The details of cutting condition and the groups of
experimental testing are shown in Table 1.
ISBN 978-602-71380-1-8
obtained the highest vibration on cutting tool, that is
Vrms= 0.65 cms/s while the vibration of 3/8 inches cutting tool
is Vrms= 0.78 cms/s at spindle speed 880 rpm, feeding 0.24
mm/rev, and cutting tool overhang 40 mm.
Table 1 Groups of Experimental testing variables
abel 1. Variabel Pengambilan data
Figure 3.
Comparison plot between vibration and feed rate on variation
of spindle speed and tool overhang by using ½ inches cutting
tool dimension.
Figure 4.
Comparison plot between vibration and feed rate on variation
of spindle speed and tool overhang by using 3/8 inches
cutting tool dimension
C. Method of Data Analysis
As gained from cutting process on the lathe, measuring
data for vibration amplitude (Vrms) and surface roughness (Ra)
is calculated graphically and statistically. The data were
attained through being plotted in graph. Further analysis is
undertaken by using both manual calculation [12] and
Statistical Package for Social Sciences (SPSS) by regression
and correlation methods to determine how influential the
relationship among the variables (tool dimension, spindle
speed, feed rate, and tool overhang) to the value of the
vibration amplitude and surface roughness of workpiece[13].
IV. MODEL ANALYSIS AND DISCUSSION
A. Result
The result of cutting tool vibration towards feed rate and
variation of spindle speed and tool overhang can be seen in
Figure 3, where it appears that the use of ½ inches cutting tool
at the same length of tool overhang, addition of feed rate on
the turning process will increase cutting tool vibration, where
the highest vibration value occurs in 0,24 mm/rev feed rate.
Moreover, conditions where feed rate is maintained at a
fixed speed and spindle speed varied at different value, the tool
vibration will be higher if spindle speed is increased. However,
the magnitude of vibration is smaller with a vibration caused
by variations in the parameters of feeding. On the use of 3/8
inches cutting tool as shown in Figure 4 , vibration value is
higher compared to vibration value generated by the use of ½
inches cutting tool. On the use of ½ inches cutting tool
The effects of variation on feed rate towards surface
roughness using ½ inches cutting tool as shown in Figure 5,
indicates that increasing the feed rate will increase surface
roughness on workpiece where the highest value obtained on
feeding 0.24 mm/rev. On the variation of cutting tool overhang,
10
Proceeding of International Symposium on Smart Material and Mechatronics
the longer the tool bit the higher the surface roughness value
where 40 mm tool overhang has the highest roughness value.
By using 3/8 inches tool bit as shown in Figure 6, the
surface roughness value will be greater than the roughness
values obtained in the use of ½ inches in every similar cutting
parameter condition used. On the use of ½ inches tool bit the
highest surface roughness value is Ra= 8.54 μm while on 3/8
inches tool bit reaches Ra= 10.34 μm at a spindle speed of 170
rpm, feeding 0.24 mm/rev, and tool overhang 40 mm.
ISBN 978-602-71380-1-8
regression equation of each variables as shown in Table 2 and
Table 3.
Table 2. The result of correlation statistical analysis and tool vibration
regression dan ½ incheses and 3/8 inches.
Table 3. The result of correlation statistical analysis and turning result of
surface roughness regression by using ½ inches tool and 3/8
inches
B. Discussion
This study shows that there is an effect of feed rate, spindle
speed, tool overhang, and tool dimension toward vibration
amplitude and surface roughness on turning process where the
most influential parameter towards vibration amplitude is tool
overhang, while the most affective parameter towards surface
roughness is feed rate.
The experimental result of the effect of independent
variables ( spindle speed, feed rate, tool overhang) towards
vibration amplitude found that, the addition of feed rate on
turning process will increase the tool vibration because the
more feeding given the faster cutting tool cut of the workpiece,
as the result, the frictional forces that occur will be greater due
to the magnitude of compressive force on the tip of the cutting
tool and workpiece. Moreover, a condition where feeding
keeps being constant and the spindle speed varied on different
values resulting the increasing of tool vibration if spindle speed
is increased. However, vibration obtained will not be as higher
as vibration by feeding parameter variation. This vibration is
caused by radial cutting force that occurs as a result of
interraction between the tip of cutting tool and the rotating
workpiece.
The effect of vibration on the cutting tool overhang
variation shows that, the more the tools overhang from the tool
holder, the greater the vibration results. The length of cutting
tool overhang contributes to the deflection which is caused by
cutting force that lead to vibration on cutting tool. At the same
conditions of cutting parameters, the vibration that occurs in
the use of ½ inches tool bit larger than 3/8 inches. This is
caused by the stiffness of the ½ inch tool bit larger than 3/8
inches. Material stiffness is determined by volume and
elasticity modulus of material, while the magnitude of the
deflection is inversely proportional to the value of the rigidity,
while the vibration tends to be even greater if the value of
deflection increases [14].
Figure 5. Comparison plot between surface roughness and feed rate on
variation of spindle speed and tool overhang by using 1/2
inches cutting tool dimension
Figure 6.
Comparison plot between surface roughness and feed rate on
variation of spindle speed and tool overhang by using 3/8
inches cutting tool dimension
To determine a detailed characteristics effect between
dependent and independent variables, the data processing is
done with statistical methods to determine the correlation and
11
Proceeding of International Symposium on Smart Material and Mechatronics
Statistically, the coefficient of determination (R2y.123 ) for
½ inches cutting tool is 0.9442. This shows that, 94.42% of the
variation of the amplitude of vibration is the result of the
influence of spindle speed, feed rate and tool overhang
simultaneously. While the rest 5,6% is caused by other factors.
Because the R-square value close to 1, then this indicates the
strength of the relationship between the independent and
dependent variables in this case. On the use of 3/8 inches
cutting tool shows that R square is 0,963. This clearly reveals
that the effect of spindle speed, feed rate, and tool overhang
variables all at once toward tool vibration amplitude is greater
than the use of ½ inches tool. On the regression equation of
turning vibration amplitude on ½ inches tool, the Constants of 0.3844 states that, if the independent variables like spindle
speed, feed rate and tool overhang equals to zero, then the
vibration amplitude is -0.3844 μm. Regression coefficient (X1)
of 1.351.10-4 states that, the addition of spindle speed of 1 rpm
will decrease the vibration amplitude of -1.351.10-4 cm/s.
Regression coefficient (X2) of 0.015488 states that, every
1mm addition of tool overhang, will increase vibration
amplitude 0.015488 cm/s. Regression coefficient (X3) of
1.14454 states that, addition of every 1 mm/rev feed rates will
increase vibration amplitude 1.14454 cm/s. From the two
equations of regression above, it can be seen that coefficient of
X1, X2 and X3 on the use of 3/8 inches cutting tool is higher
than independent variables coefficient on ½ inches, on the
other words, response of independent variable towards
vibration amplitude for the use of 3/8 inches tool is greater than
the use of ½ inches tool.
To determine the most influential independent variables on
the magnitude of the vibration amplitude, the correlation
analysis tests is conducted. From the calculation of the
correlation coefficient on ½ inches tool shows that the most
influential factor on lathe vibration amplitude is tool overhang
with correlation value 0.6968 at 99% convidence level, the next
is feed rate 0.6182 and the last is spindle speed 0.2767.
In addition, correlation analysis on the use of 3/8 inches
cutting tool, shows that the most influential factor on the tool
vibration amplitude is tool overhang on the correlation value of
0,718, then the feed rate 0,602 and the last is spindle speed
which reaches 0.293. as all correlation coefficients are positive,
it means that, by increasing the amount of spindle speed, tool
overhang, and feed rate will increase tool vibration amplitude.
Nevertheless, value of correlation coefficient on the use of tool
is 3/8 inches which has a higher correlation value than the use
of ½ inches for its all independent variables.
The test result of independent variable effect (spindle
speed, feed rate, and tool overhang) towards roughness value
found that, addition of feed rate on turning process will
increase the value of surface roughness, this is due to the
construction of the pointed end ( very small nose radius) of the
cutting tool, thus, if feeding is slowed down then the distance
between the grooves cut by a cutting tool tip at one revolution
of workpiece will be even greater. This distance will stimulate
such serration effect which is distantly spaced and if measured
with surface tester, so, the distance among these serrations is a
representation of surface roughness of a turning object. On the
variation of tool overhang and tool dimension is seen that
surface roughness depends on vibration, thus, the more
vibrations the roughness value is greater.
ISBN 978-602-71380-1-8
From the results of statistical data processing, for ½ inches
tool bit, shows that 95.1% of the variation in surface roughness
value is a result of the influence of variable spindle speed, feed
rate and cutting tool overhang simultaneously. While the
remained of 4.9% is the result of other factors. On the use of
3/8 inches tool bit found that the effect is greater than ½ inches
usage ie 95.9%. From the regression equation generated by the
use of both types of cutting tool seen that the coefficient of X 1,
X2 and X3 on the use of 3/8 inches tool bit is greater than the
coefficient of the independent variable on the use of 1/2 inches
tool bit, in other words, the response to the independent
variable of vibration amplitude at the use of 3/8 inch tool bit is
much greater than the use of ½ inches tool bit.
The most influential independent variable towards surface
roughness value based on correlation analysis test among
variables, both for ½ inches and 3/8 inches tools are feed rate ,
then spindle speed, and the last is tool overhang. Minus sign on
the correlation coefficient on the influence of spindle speed
states that, the greater the spindle speed is, the smaller the
value of surface roughness (Ra) to be generated. where the
smaller the value of Ra, the smoother is the surface [15].
C. Conclusion
The effect of feed rate, spindle speed, tool overhang, and
tool dimension simultaneously towards vibration amplitude by
using ½ inches tool bit is 94.4 %, while the use of 3/8 inches
tool bit is 96.3%. The effect of feed rate, spindle speed, tool
overhang and tool dimension all at once towards surface
roughness by using ½ inches tool bit is 95.1%, while the use of
3/8 inches tool bit is 95.9%. Amplitude correlation of tool
vibration by using ½ inches tool towards tool overhang
69.68%, towards feed rate 61.82% and spindle speed 27.67%.
On the use of 3/8 inches tool bit, amplitude correlation of tool
vibration towards tool overhang 71.8% towards feed rate
60.2% , spindle speed 29.3%. Surface roughness correlation by
using 1/2 inches tool bit towards feed rate 73.0% , spindle
speed -55.2% , and tool overhang 33.6%. On the use of 3/8
inches tool bit, correlation of surface roughness towards feed
rate 80.9%, spindle speed -42.0% and tool overhang 38.5%.
The most influential parameters on the vibration amplitude is
tool overhang while the most influential parameter on surface
roughness value is feed rate.
REFERENCES
[1] Lazuardhy Muchammad T, Endi S, Erwin S. 2012, “pengaruh
feed motion kondisi chatter terhadap kekasaran permukaan
benda kerja proses bubut”, Jurusan Teknik Mesin Fakultas
Teknik Universitas Brawijaya.
[2] Huang Luke, “A Multiple Regression Model to PredictInprocess Surface Roughness in Turning Operation Via
Accelerometer” , Journal of Industrial Technology, Vol. 17 No.2
April 2001.
[3] M. Siddhp aco anTić, DuŠan kovačevićura 2011, “Wear level
influence on chip segmentation and vibrations of the cutting tool
Tesis”, RMZ – Materials and Geoenvironment, Vol. 58, No. 1,
pp. 15–28, 2011
[4] Koten Viktus K. (2006). Analisis Pengaruh Kondisi
Pemotongan Pada Mesin Bubut Terhadap Amplitudo Getaran
Pahat Dan Kekasaran Permukaan Benda Kerja, Tesis. Pasca
Sarjana Universitas Hasanuddin Makassar.
12
Proceeding of International Symposium on Smart Material and Mechatronics
[5] Haans Anthonius Ls. (2006). Analisis Korelasi Getaran
Terhadap Kekasaran Permukaan Baja Karbon Pada Mesin
Frais Vertikal Dengan Variasi Sudut Tatal Pahat, Tesis. Pasca
Serjana Universitas Hasanuddin Makassar.
[6] Anonim, “Getaran ” (On-line), diakses pada tanggal 12 oktober
2013, http://id.wikipedia.org/wiki/Getaran
[7] K. Khalili, M. Danesh,2013, “Investigation of overhang effect
on cutting tool vibration for tool condition monitoring”, The
University of Birjand, Iran
[8] B.Sentot Wijanarka. 2010, Teknik Permesinan Dasar, Jurusan
pendidikan T. Mesin, FT-UNY
[9] Santoso Singgih, 2014, SPSS 22 from Essential to Expert Skilss,
PT Elex Media Komputindo, Jakarta.
[10] Kibbe Richard R. (2010). Machine Tool Practices,Pearson
Education.Inc, New Jersey
[11] Kassab Safeen Y., Khoshnaw Younis K. (2007). The Effect of
Cutting Tool Vibration on Surface Roughness of Workpiece in
Dry Turning Operation. Mechanical Eng. Dep.University of
Salahddin. Eng. & Technology, Vol.25, No.7, 2007 .
[12] Supardi U.S. (2011). Aplikasi Statistika Dalam Penelitian,
Smart, Jakarta
[13] Santoso Singgih. (2014). SPSS 22 from Essential to Expert
Skilss. PT Elex Media Komputindo, Jakarta
[14] Martin Wenham. (2005). Stiffness and Flexibility. 200 science
investigations for young students, p. 126, ISBN 978-0-76196349-3
[15] Rusnaldy., Setiawan Joga Dharma., Arivian Anggi.(2011).
Monitoring Kondisi Pahat Dengan Sinyal Getaran Pada Proses
Bubut. Jurnal Teknik Mesin ROTASI – Vol. 13, No. 3, Juli
2011: 1-4
13
ISBN 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Effect of Water Volume and Biogas Volumetric
Flowrate in Biogas Purification Through Water
Scrubbing Method
Hendry Sakke Tira, Yesung Allo Padang, Mirmanto and Rio Cristovan Mantiri
Mechanical engineering, Mataram University
Mataram, Indonesia
[email protected], [email protected], [email protected]
Abstract—Energy supply is a crucial issue in the world in the
last few years. The increase in energy demand caused by
population growth and resource depletion of world oil reserves
provides determination to produce and to use renewable
energies. One of the them is biogas. However, until now the use of
biogas has not yet been maximized because of its poor purity.
According to the above problem, the research has been carried
out using the method of water absorption. Under this method it is
expected that the rural community is able to apply it. Therefore,
their economy and productivity can be increased. This study
includes variations of absorbing water volume (V) and input
biogas volume flow rate (Q). Raw biogas which is flowed into the
absorbent will be analyzed according to the determined
absorbing water volume and input biogas volume rate.
Improvement on biogas composition through the biogas
purification method was obtained. The level of CO2 and H2S was
reduced significantly specifically in the early minutes of
purification process. On the other hand, the level of CH 4 was
increased improving the quality of raw biogas. However, by the
time of biogas purification the composition of purified biogas was
nearly similar to the raw biogas. The main reason for this result
was an increasing in pH of absorbent. It was shown that higher
water volume and slower biogas volume rate obtained better
results in reducing the CO2 and H2S and increasing CH4
compared to those of lower water volume and higher biogas
volume rate respectively. The purification method has a good
promising in improving the quality of raw biogas and has
advantages as it is cheap and easy to be operated.
energy promotion scheme as well as an alternative for
reduction of greenhouse gases emissions.
Biogas, a clean and renewable from of energy can be a
good substitution of conventional sources of energy which are
causing ecological-environmental problems and at the time
depleting at a faster rate [3, 4]. Biogas is the combustible gas
produced through an anaerobic digestion at low-temperature
and without oxygen. Thus it application includes electricity,
heating and cooking. On the other hand, there is lack of good
management of the ever-increasing amounts of manure solid
and liquid waste in many communities. Most of the rural
communities discharged the manure without treatment directly
onto wasteland or into rivers and streams. This behavior leads
to unhygienic environment with attendant bad odors and flies
[5]. With appropriate treatment, the manure can be converted
into biogas. Biogas is defined as mixture of gases, consisting of
methane (CH4), carbon dioxide (CO2), hydrogen sulfide (H2S)
and traces of other gases like nitrogen (N2), oxygen (O2),
hydrogen (H2) and ammonia (NH3). The composition of biogas
depends on the organic material as well as on the conversion
technology used, varying between 50-75% CH4, 25-45% CO2,
and 0-20 000 ppm of H2S [6].
From the constituents of biogas, CH4 and CO2 are the main
compounds in determining the quality of biogas. If the level of
CH4 is high, the biogas will has higher calorific value. On the
other hand, if the level of CO2 is high, the quality of biogas will
be worse, marked by lower calorific value. Therefore, to
improve the calorific value of biogas in order to be used
effectively as fuel, the level of CO2 should be reduced or
eliminated [6]. On the other hand, H2S, a kind of highly toxic
and corrosive gas, inhibits the biogas process directly, as well
as indirectly in the case of higher H2S concentrations in
digester. To avoid the negative effects of H2S, a reduction of
H2S concentration in biogas is required before combustion [6,
7].
Removal of CO2 and H2S from biogas is the main factor
to improve the biogas quality [8]. To pursue this aim, a
purification method is required to treat raw biogas. Some
biogas purification methods have been performed, and water
scrubbing method can be a solution as it is a simple and cheap
Keywords—Biogas, CH4, CO2, H2S, water absorption
I. INTRODUCTION
In the last few years the continuously uninterrupted supply
of energy has been a crucial problem in Indonesia. The
increase of energy demand which is caused by population
growth and acceleration in industries has pressured the
government to explore much new and alternative energy to
maintain the development. One of the alternative energy is
biogas. Biogas is a promising energy among other alternative
fuels as it is renewable with abundant feedstock and can be
produced in rural area with relatively low operational cost [1,
2]. Therefore, biogas can be the solution for this renewable
14
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
method among the methods [9].
Apart from cheap and simple, the water scrubbing method
is easy to use specifically for cattlemen in rural areas. It is
possible as the method use simple technology. Therefore,
applying the biogas purification method is expected triggering
the productivity and economy of rural community.
II. EXPERIMENTAL PROCEDURE
A. Raw biogas
The raw biogas used in the experiment come from
anaerobic process in a digester located in the Renewable and
New Energy Laboratory, engineering faculty Mataram
University. The raw material for the biogas is from cow dung.
The ratio of cow dung and water for biogas production in
digester is 1 : 1. According to the previous experiment, it was
found that this ratio could produce maximum biogas volume in
relatively shorter period of anaerobic process. The produced
biogas is then flowed to a receiving-station before directed to
biogas scrubbing unit. Before colleting the experimental data
of purified biogas, data of raw biogas components such as CO2,
H2S and CH4 was taken. The value for each component was
33.6%, 208.33 ppm and 59.36% respectively for CO2, H2S and
CH4.
Figure 1. Schematic diagram of water scrubbing unit
The scrubbing unit was set up and connected to measurements
apparatus and other components as can be seen in Figure 2
below.
B. Experimental variables and equipments
The experiment was performed by applying some variations
such as biogas volume flow rate and water volume. The biogas
volumetric rates were Q1 = 1 lt/min, Q2 = 2 lt/min, and Q3 = 3
lt/min while the water volume was 10, 15 and 25 liters. The
data were taken continuously for 30 minutes long for each
operating condition. The research also has a purpose to know
the relative humidity of the purified biogas as the contact with
water may rise the relative humidity of biogas.
The component of biogas was measured using a biogas
tester (GEO TECH) which measured compounds such as CH4,
CO2, O2 and H2S with accuracy level of ± 0.5% vol. The
relative humidity of purified biogas was measured using
humidity sensor (SHT 11) which has ability to measure
humidity under temperature range from -40 to 123 oC. The
range of humidity measurement starts from 0 to 100%. To
increase the biogas stream pressure which goes to scrubbing
unit, biogas vacuum pump model BP-01 equipped with
double-stage-pump was used. Biogas volume rate was
measured accurately using biogas dedicated flow meter which
has ability to measure flow rate until 4 m3/hr.
Figure 2. Schematic diagram of experimental setup
III. RESULTS AND DISCUSSION
A. Methane (CH4)
Methane is the main component in biogas and can reach as
much as 55% after anaerobic process in digester [10].
C. Biogas scrubbing unit.
The biogas scrubbing unit used for the research has 250 mm
long, 250 mm wide and 750 mm high. The unit was made
from glass in order to observe visually the flow pattern
governed by biogas in absorbent. The biogas input channel
was located downstream at the top of the unit to allow the
absorbent and biogas has longer contact. The schematic
diagram of the unit is shown in Figure 1 below.
Figure 3. Methane concentration at 15 liter of water volume
15
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
This percentage is considered relatively low for heating
and as a fuel purposes. The experiment results show that the
level of methane before purification is 59.36%. However, after
the purification process, the level of CH4 was increased.
Results show that the concentration of CH4 was improved to
60.4% where reached in the operating condition of 25 liter of
water volume and 1 lt/min for biogas volumetric flow rate
respectively. Additionally, the lowest improvement for CH 4, it
was of 59.36%, was gained in operating condition of 15 liter
of water volume and 3 lt/min for biogas volumetric flow rate
respectively (Figure 3).
purified biogas. However, the dissolved rate of methane is
higher compared to those of carbon dioxide and hydrogen
sulfide. Methane solubility rate in water at 10 oC is 42 mg/liter
of water. The solubility of methane in water decreases along
with the increase of water temperature [11].
The results show that in the early minutes of data
collection, from 10th to 18th minute, the methane concentration
is relatively low. The air concentration in the top of scrubbing
unit is still exist at the beginning of measurement. Therefore,
from early minutes to 10th minute, most of the gas measured
comes from air.
It was shown that the higher water volume, the higher
content of methane in biogas (Figure 4). It is resulted by the
longer contact time occurred between water and biogas
molecule in higher water volume operating condition. The
contact between molecules leads to the absorption of the
impurities substances in biogas into water molecule. Under
higher water volume, the rate of reaching the higher methane
concentration is faster compared to that of lower water volume
(Figure 5).
B. Carbon dioxide
Carbon dioxide is an impurities gas which has relatively
high percentage in biogas. Carbon dioxide is an unburned gas
and therefore resulted in the calorific value of biogas reduce
significantly. The more CO2 in biogas, the less calorific value
of biogas is. In order to improve the biogas calorific value,
reducing the content of CO2 through biogas purification is a
must. Carbon dioxide is a water-soluble gas. At room
temperature, the solubility of carbon dioxide is about 90 cm3 of
CO2 per 100 ml water (Shakashiri, 2014). The application of
water scrubbing method is expected to improve the quality of
biogas by reducing carbon dioxide content.
The research showed that the most effective condition to
reduce CO2 was gained at 25 liter of water volume and 1 lt/min
of biogas volumetric flow rate. While 15 liter of water volume
and 3 lt/min of biogas volumetric flow rate showed the least
effective to improve biogas quality as shown in Figure 6 and
Figure 7.
Figure 4. Methane concentration at 25 liter of water volume
However, the level of methane is maintained at constant
rate even without more improvement. This is resulted by the
maximum portion of acid substances absorbed into biogas
molecule. This is indicated by the increase of acidity level of
absorbent water. The change of water acidity brings the ability
of absorbent water down to minimum. Therefore, it is required
to maintain the absorbent acidity, pH 7 is the best for biogas
scrubbing, in order to keep the absorbent water performance.
Figure 6. Carbon dioxide concentration at 15 liter of water
volume
Slower biogas volumetric flow rate leads carbon dioxide
dissolve in water faster than that of higher biogas volumetric
flow rate. It is shown in Figure 8. This condition provides
longer contact time between water and biogas which results
more carbon dioxide molecule in biogas absorbed into water.
This could be happen as biogas stream is upward to the top
where the biogas outlet inside the scrubbing unit is positioned
at the bottom of the unit and thus develops high mixing rate
between biogas and water.
Figure 5. Methane concentration at different water volume with
varied biogas volumetric flow rate
The increase rate of methane concentration in the
experiment is not too high. This is due to methane is dissolved
into water molecule and hence reducing the methane content in
16
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
biogas purification specifically in the early measuring time.
For longer continuous-measurement, a modification and
improvement are required not only to scrubbing unit but also
to operating conditions to obtain better results.
Figure 7. Carbon dioxide concentration at 25 liter of water
volume
Figure 9. Hydrogen sulfide concentration at 25 liter of water
volume
Figure 8. Carbon dioxide concentration at different water
volume with varied biogas volumetric flow rate
Higher mixing rate occurred mainly due to slower biogas
flow rate triggered faster reaction of molecules bonding among
water, CO2 and H2S molecules. This increased the partial
pressure of the molecules to water molecules creating relatively
strong bond among the molecules. Supported by higher water
volume, the previous acid water created by CO2 and H2S
content in water was minimized. This condition made the
concentration of CH4 in biogas increased.
Figure 10. Hydrogen sulfide concentration at different water
volume with varied biogas volumetric flow rate
Despite the water scrubbing method has an advantage in
removing CO2 and H2S from raw biogas, it has a weakness as
the moisture content in biogas increases (result not shown). A
physical reaction between water and biogas resulted in more
water molecules are attached to the biogas molecules. For this
reason an improved water management technique is desirable
to obtained better biogas quality.
On the other hand, like the moisture content, pH samples
from the purified biogas were increased, ranging from 8 to 6
(result not shown). The most significant pH drops were
noticed at slower biogas volumetric flow rate (1 l/min) and
lower water volume (15 liter). One possible explanation for
this phenomenon is longer contact between biogas and water.
A drop in pH is an indication of an increase the water acidity
due to CO2 and H2S dissolve in absorbent. The largest
decrease of pH were noticed when the CO2 and H2S in
purified biogas were high. The major reason for this is the
absorbent has reached the ultimate point to absorb the
compounds.
C. Hydrogen sulfide (H2S)
Although the concentration of hydrogen sulfide is very
small in biogas but it has adverse effect both to environment
and health [12]. Combustion of H2S leads to sulfur dioxide
emissions which have harmful environmental effects.
Removing H2S as soon as possible is recommended to protect
downstream equipment, increase safety, and enable possible
utilization of more efficient technologies such as combustion
engines and fuel cells [13].
Hydrogen sulfide dissolves in water under 437 ml/100 ml
of water at 0 oC and 186 ml/100 ml of water at 40 oC [14].
Before purification process, the concentration of H2S in
biogas is 208.33 ppm, and reduced to 151 ppm after the
treatment. Similar to carbon dioxide result, the highest
reduction of H2S is gained in slower biogas volumetric flow
rate (1 lt/min) and higher water volume (25 liter) as shown in
Figure 9. Other results at 15 and 20 lt/min are shown in Figure
10. However, after about 20 minutes the productivity of
absorbent decreased as the measured H2S in biogas almost
similar to initial raw biogas. It indicated that the absorbent has
been filled with H2S, CO2 and other impurity gases from
biogas. Therefore, this method is feasible to be applied for
IV. SUMMARY
Biogas purification through water scrubbing method is
promising. This method is feasible as it can improve the
biogas quality by reducing the biogas impurities compounds
such as CO2 and H2S. Moreover, this method has low
17
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
[5] E. D. Aklaku, K. Jones, K. Obiri-Danso, “Integrated biological
treatment and biogas production in a small-scale slaughterhouse
in rural Ghana,” Water Environment Research, vol. 12, pp.
2335, 2006.
[6] H. Naegele, J. Lindner, W. Merkle, A. Lemmer, T. Jungbluth, C.
Bogenrleder, “Effects of temperature, pH and O2 on the removal
of hydrogen sulfide from biogas by external biological
desulfurization in a full scale fixed-bed trickling bloreactor
(FBTB),” Int Agric & Biol Eng, vol. 6, no. 1, pp. 69, 2013.
[7] W. Barhost, L. Gupta, Benefits of digester gas scrubbing at the
Dayton WWTP, Ohio, USA, Water Environment Association,
2011.
[8] D. Shannon, H. Kalipcilar, L Yilmaz, Development of zeolite
filled polycarbonate mixed matrix gas separation membranes,
department of Chemical Engineering, Middle East Technical
University Ankara, Turkey, 2006.
[9] A. Dubey, Water scrubbing for carbon dioxide removal from
biogas, Annual report of central institute of agricultural
engineering, Bhopal, India, 2000.
[10] B. Richards, F. Jewell, W. Cummings, R. White, “In situ
methane enrichment in methanogenic energy crop digesters,”
Biomass and Bioenergy, vol. 6, no. 4, pp. 275–274, 1994.
[11] M. McGowan, Water Processing. Third Edition, Water Quality
Association. Water Technology Volume 32. International
Occupational Safety and Health Centre; University of
Wisconsin, 2009.
[12] R. Robert, P. John, P. Brice, The Properties Of Gases and
Liquids. 4 ed. Boston: McGraw-Hill, 1987.
[13] E. Kovacs, R. Wirth, G. Maroti, Z. Rakhely, K. Kovacs, “Biogas
production from protein-rich biomass: fed-batch anaerobic
fermentation of casein and of pig blood and associated changes
in microbial community composition,” PLoS ONE, vol. 8, no.
10, 2013
[14] Y. Lisafitri, Penggunaan Biotrickling Filter Biotrickling Untuk
Mengatasi
Polutan
H2S.
From
http:
//
www.academia.edu/3881807/
Penggunaan_Biotrickling_Filter_Biotrickling_Untuk_Mengatasi
_Polutan_H2s. Downloaded at 18 August 2014.
operational cost, durable and easy to be operated. This allows
rural communities to apply this method for gaining their own
energy resources. The research has found that the most
effective condition to achieve the best results is running the
experiment at slower biogas volumetric flow rate and higher
water volume. At this operating condition the CO2 can be
improved up to 10.5%, and H2S up to 27.5%. A further
research is required to make this method more promising and
efficient as the reduction of CO2 and H2S is occurred at the
early minutes of measurement. Maintaining the pH of
absorbent and eliminating moisture content of purified biogas
are also a challenge in order to obtain high quality of biogas.
ACKNOWLEDGMENT
The authors would like to thank the Department of National
Education of the Republic of Indonesia for supporting this
project under MP3EI research grant with a contract number:
DIPA-023.04.1.673453, 5 December 2013, DIPA Revision 01
29
April
2014
contract
number:
012/SP2H/PL/DIT.LITABMAS/V/2014, 05 Mei 2014.
REFERENCES
[1] L. Jian, “Socioeconomic Barriers to biogas development in rural
southwest China: an ethnographic case study,” Human
Organization, vol 68, No. 4, 2009
[2] A. Weiss, V. Jerome, D. Burghardt, L. Likke, S. Peiffer, E.
Hofsetter, R. Gabler, R. Freitag, “Investigation of factors
influencing bogas production in a large-scaletThermophilic
municipal Biogas Plant,” Appl Microbiol Bioethanol, vol 84, pp.
987-1001, 2009.
[3] T. Kaosol, N. Sohgrahok, “Enhancement of biogas production
potential for anaerobic co-digestion of wastewater using
decanter cake,” American Journal of Agricultural and Biological
Sciences,” vol. 7, no. 4, pp. 494-502, 2012.
[4] Y. Santosh, T. Sreekrishnan, S. Kohli, V. Rana, “Enhancement
of biogas production from solid substrates using different
techniques-a review,” Bioresources Technol, vol. 95, pp. 1-10.
DOI: 10.1016/j.biortech.2004.02.010.
18
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Effect of Solution Treatment Process on Hardness
of Alumina Reinforced Al-9Zn Composite
Produced by Squeeze Casting
Dwi Rahmalina, Hendri Sukma, I. Gede E.Lesmana, Asrin Halim
Mechanical Engineering Department
Faculty of Engineering, Pancasila University
Jakarta, Indonesia
[email protected]
Abstract—Characteristics of aluminium matrix composites
reinforced by alumina have been developed to improve
mechanical properties. One of the determining factors in the
development of this material is parameter of solution treatment
process. This study discusses the performance of the composite
matrix of Al-9Zn-6Mg-3Si reinforced by alumina powder of 5 %
volume fraction. Composite are manufactured by squeeze
casting process with the pressure of 20 Ton in the metal mould.
To improve mechanical properties, the precipitation hardening
process is conducted through variation of temperature of
solution treatment of 450, 475 and 500 °C and holding time of
solution treatment of 30, 60 and 90 minutes. Materials are
characterized by hardness testing and microstructure
observation. The results showed that the optimum condition of
hardness was produced by solution treatment temperature of
500 °C and 90 minutes holding time of 86 HRB.
The drawback is this material has lower strength than other
commercial material such as cast iron, steel or copper.
However the strength of aluminium can be increased through
alloying, cold working and heat treatment through precipitation
hardening process.
One of the aluminium alloys is Al-Zn, which exhibits the
highest strength and in many cases they have higher strength
than steel. This alloys is the combination of zinc and
magnesium that produce the AA7xxx alloy which can be
heated and formed a very high strength. For example, AA 7075
with composition of: Zn 5.0-6.0%, Mg 2.0-3.0%, Cu 1.0% 2.0% gives specific tensile strength of 580 MPa [3]. Zn
element will increase hardness optimally, after the heat
treatment precipitation hardening process [4], The increase in
Zn element up to 9% can increase hardness particularly after
precipitation hardening, however Zn content in present study
was limited to 9% because the higher the content, the
possibility for hot cracking to occur is increase [5]. Based on
this, in order to increase the hardness, Zn and Mg alloy
elements added. Then heat treatment is conducted by
precipitation hardening process on the composite to optimize
the hardness. This research studies the influence of
precipitation hardening parameter such as solution treatment
temperature and holding time towards hardness of the
aluminium matrix composite.
Keywords—aluminium matrix composite, alumina, squeeze
casting, hardness
I. INTRODUCTION
Various engineering components in manufacturing
technology need of material with good mechanical
characteristics. Many innovations have been made to create a
new kind of lightweight materials with excellent mechanical
properties. Composite materials can be used as one of the
alternatives to answer this challenge. Composite material can
combine the excellent properties of consisting constituents to
produce a new material with better characteristic, which offer
several properties excellence such as high strength and
stiffness, good toughness, good strength in high temperature,
high wear resistance and having high ration strength to weight
[1,2].
II. EXPERIMENTAL METHODS
Material used in this research is Al-3Si ingot added with
alloying element, with 6 wt. % Mg and 9 wt.% Zn. Melting
process conducted in the crucible furnace on the temperature
of 850-870 °C. Powder alumina reinforcement with the size of
0.45 micron was poured with 5 % volume fraction, then stirred
with the velocity of 7500 rpm. Composite produced by
squeeze casting process with the application of pressure of 20
ton in a preheated mould, in order to optimize the
solidification process and minimized defects in the interface.
The composite was then underwent heat treatment by
precipitation hardening process to enhanced its mechanical
properties. Precipitation hardening process on cast composite
In the current development, aluminium matrix composite is
very promising, not only its good mechanical properties but
also relates to its low density. Aluminium alloy is chosen as the
matrix because this metal is light in weight, relatively cheap
and easy to fabricate. Moreover, aluminium is a metal that have
been produced independently in Indonesia, therefore it can be
developed even further as many application for domestic
needs.
19
Proceeding of International Symposium on Smart Material and Mechatronics
plate is started by solution treatment with temperature
variation of 450, 475 and 500 °C and variation of holding time
of 30, 60 and 90 minutes; and followed by quenching in water,
and then aging process is performed in temperature of 200 °C
for 2 hours. Material characterization is conducted with
chemical composition, microstructure observation with optical
and electron microscope and hardness testing.
ISBN 978-602-71380-1-8
dendritic structure. The higher solution treatment temperature
the hardening process started earliest and proceeded fastest.
The figure also shows the squeeze casting technology seen to
be succesfully in preventing the occurance of shrinkage due to
the solidification process.
a
III. RESULTS AND DISCUSSION
Composite characteristic is significantly influenced by the
content of alloying elements. Alloying of the matrix was
conducted to improve its mechanical properties and the quality
of the cast metal. Ingot aluminium with silicone element used
for enhanced castability of the casting material. Adding Mg
content improved wettability on the interface area between the
matrix and alumina reinforcement [6]. It was renowned that
the interface condition is very important in determining the
needed properties of the composite systems because it
functions as the media for transferring the load from reinforcematrix-reinforce [7]. Whereas, the addition of Zn will increase
hardness and strength of aluminium alloy after precipitation
hardening process [8].
b
Table 1. The compositions of the matrix.
Zn
9.16
Mg
6.12
Si
2.90
Content (wt. %)
Fe
Mn
Cu
0.22 0.003 0.003
Ni
0.206
Al
Balance
c
Table 1 shows the chemical compositions of the composite.
As listed in the table, besides the intentional additions, other
elements such as Fe, Mn and Cu are also present. Even small
amount of these impurities causes the formation of a new
phase component [9]. Through casting various intermetallic
phases are formed between aluminium dendrites [9,10]. These
intermetallic phases have different structures, stabilities and
mechanical properties. Based on this reason, the cast
composite require solution treatment to improve mechanical
properties. During this treatment some transformation of
intermetallic phases such as plate-like β-Al5FeSi into more
rounded discrete α-Al12(FeMn)3Si particles; and dissolution of
β-Mg2Si particles [11]. These transformations will give
maximum hardness after aging process [11,12]. The hardening
process is conducted in order to improve hardness and
toughness of a matrix. The heat treatment process will
decrease Mg2Si on grain boundary and increase volume
fraction of -aluminium on matrix, which then will produce
precipitate MgZn2, which settle in the grain, due to aging
process [8].
Figure 1: The microstructure of the composite with
variation of solution treatment temperature:
(a) 450 °C; (b) 475 °C; (c) 500 °C.
To observe the presence of presipitate more clearly
SEM/EDS examination was conducted, as seen in Figure 2.
Micro analysis using the SEM-EDS result on the Table 2
showed that there was MgZn2 presipitations present in the
matrix that inhibit dislocation process resulting in the increase
in mechanical properties of the Al-9Zn-6Mg-3Si composite
[13]. Alumina in the composite showed to be distributed
evenly so that it can be concluded that the stirring process with
7500 rpm seen to be distributing the alumina particles evenly.
The microstructure of the studied alloys is given in Figure
1. The heat treatment process will optimize the Zn function
though precipitation mechanism, this could be seen from other
particle morphology/second phase, distribution and shape. The
figure shows that the matrix have more globular shaped
structure with solution temperature of 500 °C. It indicates that
the higher temperature of solution treatment will dissolve
20
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
1
4
3
4
1
2
1
3
3
1
50μm
Figure 2. SEM examination on the composite with Al-9Zn6Mg-3Si matrix with 5 % alumina reinforcement, dissolution
temperature of 500 °C, holding time duration of 30 minutes,
and aging temperature 200 °C .
Figure 3: Effect of solution treatment temperature to matrix
hardness of Al-9Zn-6Mg-3Si reinforced with 5 % alumina
fraction volume, with various holding time of solution
treatment of 30,60 and 90 minutes.
Table 2. EDX on Al-9Zn-6Mg-3Si composite with alumina
reinforcement, on position according to Figure 3.
No.
Composition (wt. %)
The higher solution treatment temperature maximalized the
dissolved Zn in the matrix, whereas increasing the hardness by
forming MgZn2 phase precipitated in matrix, which inhibits
the dislocation movement. In the present study, the Zn content
was limited to only 9 wt. % Zn because increasing Zn content
will lead to hot cracking.
Phase
1
O
-
Mg
2.49
Al
85.55
Si
-
Zn
11.97
2
11.49
17.37
44.69
20.04
6.41
3
4.45
3.15
77.36
4.16
10.87
4
3.14
4.16
78.07
3.91
10.72
MgZn2, α-al
Matrix,
alumina
Matrix,
alumina
Matrix,
Alumina
IV. CONCLUSIONS
From the testing and analysis on Al-9Zn-6Mg-3Si matrix
composite with 5% volume fraction of alumina reinforcement
it can be concluded that:
Figure 3 shows the effect of solution temperature and
holding time on hardness after aging process of the composite.
Increasing solution treatment temperature of 450 to 475 and
500 °C improves hardness for all holding time of 30, 60 and
90 minutes. It has been discovered that the hardness is well
correlated by the solution heat treatment temperature with
linear characteristic. This condition is owing to fact that the
amount of alloying elements in a supersaturated solution will
form the hardening particle of MgZn2 phase precipitated
during aging process, rises along with increasing of solution
treatment temperature. The figure also illustrates that the
duration of dissolution as much as 60 or 90 minutes did not
exhibiting significant influence. This was possible due to the
Zn has been diffused completely into the Al matrix during the
duration of dissolution as much as 60 minutes, thus with the
reason for energy efficiency the optimum condition design for
the duration of dissolution taken was 60 minutes.
1. Increasing solution treatment temperatur from 450 °C to
475 and 500 °C improved the hardness of the composite
with the highest value was HRB 86.
2. Increasing solution treatment holding time from 30
minutes to 60 minutes improved the hardness of the
composite, while the improvement was slight by rising
holding time from 60 to 90 minutes.
3. The improving hardness after heat treatment process was
caused by the formation of MgZn2 phase precipitation in
composite.
ACKNOWLEDGMENT
This work was funded by DIKTI under MP3EI Research
Grant 2014. The authors are grateful to Ahmad Ashari for die
design and pneumatic machine support for the squeeze casting
21
Proceeding of International Symposium on Smart Material and Mechatronics
process. The authors also thank Ari Noor P, Lutfi Dwi P. and
Fajar H. for assisting the research.
[7]
REFERENCES
[8]
[1] _____, ASM Handbook 21: Composites, ASM International,
The Materials Information Company, 1992.
[2] J.C. William, Progress in Structural Materials for Aerospace
Systems (ed. 51st). Acta Materialia. 2003, pp. 5775–5799.
[3] R. Cobden, Alcan, Banbury TALAT Lecture 1501 Aluminium:
―Physical Properties, Characteristics and Alloys‖, pp. 60, Basic
Level.
[4] B.T. Sofyan, S. Susanti, R. R. Yusfranto, Role of 1 and 9 wt.%
Zn in Precipitastion Hardening Process in Aluminium Alloy
AA319, Makara Teknologi, 12 (1), 2008, pp. 48-54.
[5] B.T. Sofyan, D. Rahmalina, Eddy S. Siradj, dan Hery Mochtadi,
Effect Zn Alloying Element on Ballistic Performance of Al-Zn6Mg Matrix Composite, Proceeding of Insinas Conference,
2012, pp. 141-145.
[6] F.L. Matthews, dan R.D Rawlijns, in: Composite Material:
Engineering & Science. Chapman & Hall, London, 1994.
[9]
[10]
[11]
[12]
[13]
22
ISBN 978-602-71380-1-8
M. Noor, Mazlee, et al.: Microstructural Study of Al-Si-Mg
Alloy Reinforced with Stainless Steel Wires Composite via
Casting Technique, American Journal of Applied Sciences
5(6), 2008, p. 721-725.
D. Rahmalina, B. T. Sofyan, Narana Askaningsih, Sigma
Rizkyardiani., Effect of treatment process on hardness of AL7Si-Mg-Zn Matrix composite reinforced with Silicon Carbide
Particulate , Advanced Materials Research Vol. 875-877
(2014), p. 1511-1515
r u e
ieni s i
d
ro
n
t
137, 2003, 821–824.
N.C.W. Kuijpers, W.H. Kool, P.T.G. Koenis, K.E. Nilsen, I.
Todd, S. van der Zwaag, Mater. Charact. 49, 2003, 409–420.
N.C.W. Kuijpers, W.H. Kool, S. van der Zwaag, Mater. Sci.
Forum 396–402, 2002, 675–680.
.K. Gupta, D.J. Lloyd, S.A. Court, Mater. Sci. Eng. A 301,
2001, 140–146.
S.A. Balogun, et.al, The Effect of Cold Rolling and Heat
Treatment on Al 6063Reinforced with Silicon Carbide
Granules, Journal of Materials, Vol 61 No. 8, 2008, pp. 43-47.
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Review of Carbon Fiber Reinforced Polymer
Reinforced Material in Concrete Structure
Ayuddin
Department of Civil Engineering
Gorontalo State University
Gorontalo, Indonesia
[email protected]
However, the material for structural applications is updated by
an expert building or construction material in the world along
with the emergence of the material weakness. Bamboo
materials such as those used for the construction proved to
have some drawbacks, such as bamboo has a very low
resistance, so the potential for bamboo powder beetles
attacked, so the buildings are made of durable bamboo.
Therefore the order of the building from bamboo that is not
preserved only seen as a temporary building components
simply do not hold more than 5 years.
For wood materials are also widely used for the
construction of houses, offices, and bridges in particular in the
village, but this material was also proved to have weaknesses.
The disadvantage is extremely flammable, can be eaten by
termites, may expand and creep, stretch for apps roof with
wood construction is often limited due to the size of the timber
on the market is only about 4 meters, and other weaknesses in
terms of procurement, then in the long run the price of wood
are increasingly expensive because of the decreasing
availability of natural wood materials. Other materials such as
concrete have been widely used in Indonesia, but it also had
shortcomings in terms of usage for construction. Weaknesses
were found among forms that have made it difficult to change,
the weak against the strong pull, has a heavy weight, great
sound reflections power, and execution of work requiring high
accuracy.
Meanwhile, the material of the steel used for the
construction of buildings is still relatively large and very
dominant utilized in Indonesia to date. This material has many
advantages such as high tensile strength, not eaten by termites,
able to carry a heavy burden, resistant to high temperatures,
low maintenance costs, and easily molded according to the
needs of the construction. However, it turns out, according to
experts looking at the field of construction materials still have
a shortage of them can be rusty, weak against the compressive
force, not as flexible as wood can be cut and shaped a variety
of profiles, not solid, and not resistant to fire, and in the case
of slender structures harmless against buckling. Therefore, the
various shortcomings of the steel material and other materials
such as bamboo, wood, and concrete. So the development of
Abstract— Carbon Fiber Reinforced Polymer (FRP) is a
material that is lightweight, strong, anti-magnetic and
corrosion resistant. This material can be used as an option
to replace the steel material in concrete construction or as
material to improve the strength of existing construction.
CFRP is quite easy to be attached to the concrete structure
and proved economically used as a material for repairing
damaged structures and increase the resilience of
structural beams, columns, bridges and other parts of the
structure against earthquakes. CFRP materials can be
shaped sheet to be attached to the concrete surface.
Another reason is due to the use of CFRP has a higher
ultimate strength and lower weight compared to steel
reinforcement so that the handling is significantly easier.
Through this paper suggests that CFRP materials can be
applied to concrete structures, especially on concrete
columns. Through the results of experiments conducted
proved that the concrete columns externally wrapped with
CFRP materials can increase the strength. This treatment
is obtained after testing experiments on 130 mm diameter
column with a height of 700 mm with concentric loading
method to collapse. The experimental results indicate that
a column is wrapped externally with CFRP materials can
achieve a load capacity of 250 kN compared to the
concrete columns externally without CFRP material which
only reached 150 kN. If the column is given internally
reinforcing steel and given externally CFRP materials can reach
270 kN. It shows that CFRP materials can be used for concrete
structures can even replace reinforcing steel that has been widely
used in building construction in Indonesia.
Key words—CFRP material, concrete structure,
increase strength.
I. INTRODUCTION
Technological developments, especially the rapidly
growing field of materials characterized by the appearance of
materials such as FRP Composite. For building construction
such as bridges or buildings previously used material of
bamboo, wood, and steel that serves as reinforcement.
23
Proceeding of International Symposium on Smart Material and Mechatronics
civil engineering construction material appears more
promising and superior to other materials that CFRP material.
Superiority in terms of stress and strain in the CFRP material
compared to other materials can be seen in Figure 1 below
(Tamon U, 2004)
ISBN 978-602-71380-1-8
structural element to be attached to the CFRP materials are
free from contaminants oxide, oil, grease, and dust.
In this research, testing the strength of the CFRP material
serves as reinforcement in concrete columns. The columns
were tested circular column with a diameter of 130 mm and
length is 700 mm column. The focus of research is directed at
improving the lateral voltage such that it adds strength, slow
collapse process as well as a wide cross-section of the column
is ductile collapse. The results of this study, results in a
significant force in the column. This suggests that the use of
CFRP materials can be used as a function of reinforcement
and become an alternative to steel reinforcement that has been
most widely used in building construction in Indonesia.
II. LITERATURE REVIEW
A. Research results of CFRP materials for Construction
Elnabelsy G and M Saatciouglu conduct research related to
the improvement in the round concrete columns designed with
FRP materials in 2004 in Canada. They perform tests of three
large-scale bridge columns are reported in this paper. All
columns had a 508 mm circular cross-section and were
designed to have predominantly flexural response with a shear
span of 2.0 m Measured to the point of application of load,
Consisting of 1.7 m of concrete column height and 0.3m of top
loading beam height. They were reinforced with 12 - 19.6 mm
diameter, 400 MPa grade deformed reinforcement, Equally
distributed along the section perimeter. Each bar was spliced
near the base with a splice length of 390 mm, corresponding to
20 times the bar diameter. Each column had a diameter of 11.3
mm deformed transverse reinforcement, spaced at 300 mm in
the form of circular hoops with overlapping ends. The first
column tested (BR-C8) was the reference column reflecting
as-built conditions, without any seismic retrofit. The
companion two columns were retrofitted with MBrace CF 130
carbon fiber sheets. Column BR-C8-1 had four plies of CFRP
sheets, whereas Column BR-C8-2 had two plies wrapped
around the column. The surface of columns was first treated
with an epoxy-based primer. The CFRP sheets
were pre-cut to the required length and applied on columns
with epoxy saturant. Coupons were made from CFRP jacket
and tested to establish the actual stress-strain relationship of
composite materials. Accordingly, the jackets had
approximately 60,000 MPa elastic modulus and ultimate
strength of 700 MPa, with linear elastic behavior. From these
tests yield data shown in Figure 2, 3, and 4.
Figure 1. Strength/stiffness and fracturing
strain relationship
For the construction of the developing world of technology
in the field of civil engineering materials and structural
systems running very fast. This is demonstrated by the
increasing proliferation of research and discoveries are
oriented to the use of high performance materials coupled with
the structure of the research system, the better. The
combination of the use of high-performance materials in
structural components is reasonable and in certain
circumstances can not be avoided anymore. These conditions,
among others, due to the demands of mechanical performance,
durability, ease of construction, environmental and economic
aspects. Application of CFRP materials can function as a
repair and strengthening of concrete structures. Retrofitting
with CFRP system function can improve strength and provide
increased flexural capacity, shear, axial, and ductility. CFRP
materials for building construction has many advantages,
including high durability CFRP and more economical use in
corrosive environments than are easily corroded steel material.
The use of CFRP is more popular considering the number
of benefits that can be obtained as the weight of the unit is a
small, easy to apply and are handled, the cost of installation
and low maintenance. Material can provide the most
economical solution in retrofitting problem because it can
dramatically reduce the cost of labor. CFRP can be used to
increase the bending and shear capacity of reinforced concrete
beams, bending plate, push, shear and flexural column. CFRP
in the form of sheet, plate or bar can be mounted on the
surface of the beam or plate having a stretch as a flexure
reinforcement. As the beam shear reinforcement, CFRP sheets
can be glued to the side of the beam. Usage on columns,
CFRP sheets can be placed on the outside of the column to
increase the ductility and strength.
CFRP material that can stick to the structural elements
such as beams, columns, plates, then given the adhesive epoxy
resin has the basic ingredients. This adhesive is made from a
mixture of two components. Its main component is an organic
liquid that is loaded into the epoxy, binding arrangement or
oxygen atoms and two carbon atoms. Was added to the
reaction mixture to obtain the final mix. The surface to be
attached should be prepared to obtain an effective
juxtaposition including cleanup efforts on the surface of the
Figure 2. Reference column (non-retrofitted) BR–C8,
24
Proceeding of International Symposium on Smart Material and Mechatronics
reflecting as built conditions
ISBN 978-602-71380-1-8
amount According to the number of composite layers, the
concrete properties and the cross-section shape. thus the use of
Carbon Fiber Reinforced Polymer is an efficient means of
providing confinement of concrete for strength and ductility
enhancement.
Samdani S and AS Sheikh also conduct research related to
concrete columns given CFRP confinement. Twenty-eight
nearly full-scale concrete columns were tested under
monotonic concentric load at the University of Toronto. The
variables tested in the experimental study included the type of
FRP (glass or carbon), the number of layers of FRP, the
orientation of fibers in the FRP shell and the amount of lateral
reinforcement. All specimens were 356 mm in diameter, 1524
mm high standing. The response of the concrete confined with
FRP Showed two slopes of the ascending branch before the
peak stress. The first slope was approximately equal to that of
unconfined concrete. The second slope, being less steep,
started near the peak stress of the unconfined concrete and
continued until the peak. This was Followed by a significant
post-peak response that continued until the FRP shell was
sufficiently ruptured, resulting in a sudden drop of stress in
concrete. Figure 5 shows the axial stress-axial strain curves for
some of Toronto specimens, confined with 1 and 2 layers of
CFRP and GFRP.
Figure 3. Column BR-C8-1, retrofitted with 4 plies
of CFRP sheets
Figure 5 Experimental stress-strain curves
for Toronto specimens
By looking at figure 2, 3, and 4 can be reported that the
Columns with spliced longitudinal reinforcement in their
potential plastic hinge regions near the base have limited drift
capacity. The circular column tested in this investigation
developed lateral drift ratio of 1% prior to significant decay
strength. The failure resulted from slippage of spliced
reinforcement. Then, the circular columns retrofitted with
CFRP sheets Showed Significantly improved hysteretic
behavior. Hoop tensioned developed in the CFRP jacket
maintained the bond between reinforcement and concrete in
the splice region. The column with four plies of CFRP sheets
(jacket thickness of 3.6 mm) was Able to sustain in excess of
6% lateral drift ratio without significant decay strength. The
companion column with two plies of sheets (jacket thickness
of 1.8 mm) Showed 4% to 5% drift ratio with pinched
hysteresis loops.
Benzaid R and Mesbah AH also conduct research
investigations on round and rectangular concrete columns of
CFRP externally supplied. The experimental program was
carried out shorts column specimens with a square cross
section of 140x140 mm and a height of 280 mm. For all RC
specimens the diameter of the longitudinal and transverse
reinforcing steel bars were respectively 12 mm and 8 mm. The
longitudinal steel ratio was constant for all specimens of
2.25% and equal to .The yield strength of the longitudinal and
transverse reinforcement was 500 MPa and 235 MPa,
respectively.
The results of this investigation reported that in all cases
the presence of external CFRP jackets Increased the
mechanical properties of PC and RC specimens, in different
Figure 4 Column BR-C8-2, retrofitted with 2 plies
of CFRP sheets
From figure 5 we see that the concrete was not given
CFRP and GFRP rebars was experiencing rapid collapse.
However, if given the CFRP and GFRP materials externally,
the stress and strain increased significantly from concrete
without reinforcement of CFRP and GFRP. If given 2 layers
of CFRP and GFRP then continue to experience an increase of
1 layer of CFRP and GFRP rebars. The incidence of
laboratory results indicate that given the additional CFRP
layers, the more stress and strain increased.
Ongpeng CMJ doing research for improvement in the
column using CFRP materials in 2006 in Manila. In his study,
ninety four specimens of sizes 180mm diameter by 500mm
height were fabricated and tested. This means fully wrapped
CFRP specimens were used with the unconfined compressive
strength of 30 MPa, 120 mm spacing for the steel ties, using
two plies of CFRP, and the first specimens out of a total of
three identical specimens. In wrapping CFRP to the concrete
specimens cured for 28 days, the fibers were oriented 90 °
with respect to the longitudinal axis of the concrete column. In
the preparation of the epoxy matrix, the resin and hardener
25
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
ratio is 4: 1 and was hand mixed for at least 5 minutes. The
overlap of CFRP is 35 mm and 70 mm for the one-and twolayer of CFRP respectively.
The results of this study can be seen in Figure 6 the stressstrain diagrams of the three specimens that have no steel ties
and 40-mm spacing of steel ties, respectively, with increasing
amount of CFRP ply used from zero to two plies. It can be
observed that the confinement effect of using CFRP and steel
ties had Increased the compressive strength and the average
longitudinal strain Represents that the ductility of the
specimen.
Figure 9. Hoshikuma 1997 with addition of CFRP sheets
as confinement (D=500mm, d=500mm,
L=1500mm,
=1.01%, fys=295Mpa,
and f’c=28.8 Mpa)
In figure 10. using closer tie spacing, which results to an
increase in ρs, led to a gradual increase in f’cc. On the other
hand, increasing the over-all thickness of the CFRP by varying
the number of CFRP ply also shows significant enhancement
of compressive strength. Except for the RC column having no
steel ties, an addition of 1-ply led to no increase in f’cc.
Figure 7 Stress-Strain diagram of specimens with steel ties
Furthermore, Ongpeng CMJ and Oreta CW also conducted
research on the effect of CFRP on restraints on a column by
using Artificial Neural Networks. In this study, there are three
sets of the data collected from references. It was Categorized
as follows: SC (Steel Confinement) data sets that steel ties
used alone as confining material, CC (CFRP Confinement) set
- that the data used alone as confining CFRP material, and
SCC (Steel and /or CFRP Confinement) data sets that used
both, steel ties and /or CFRP, as confining materials.
From these results, it was found the effect of CFRP on
concrete columns. Effect of CFRP materials can be seen in the
image below. Figures 8 and 9 that there is an abrupt increase is
of at least 65% to 100% in f’cc from zero-ply to one-ply of
CFRP Regardless of the spacing of lateral steel ties. However,
by adding another ply of CFRP to a total of two plies, the
increase of was minimal. One common geometric property
between the column by Mander et al 1988b Hoshikuma et al
1997 and is the outer diameter D. Both columns have are
relatively large outer diameter D = 500mm for both, and the
core diameter, d = 438 mm and d = 500 mm respectively .
Figure 10. Sakai et al 2000 with addition of CFRP sheets
as confinement (D=200mm, d=185mm,
L=600mm,
=1.18%, fys=376Mpa,
and f’c=29.8 Mpa)
In figure 11 can be seen throughh enhancement due to
superposition effect of each material are less than that of the
actual experimental data for 1-ply and 2-plies of CFRP. On the
other hand, the ANN model SCC9-7-1B, which assumed no
superposition of strength enhancement on each confining
materials, but rather the total enhancement due to the
interaction of both.
Figure 8. Mander et al 1988b with addition of CFRP sheets
as confinement (D=500mm, d=438mm,
L=1500mm,
=1.6%, fys=340Mpa, lateral steel
bar diameter=12mm, and f’c=28 Mpa)
26
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Figure 11. Strength Enhancement using Li and Fang 2004
Data (D=300mm, d=250mm, L=600mm,
=0.76%, fyh=274.7Mpa,
t=0.11mm, fyfrp=4120.2Mpa, fc’=17.2Mpa)
Figure 12. Retrofitting of railroad bridges using FRP wraps,
SBVR, Moorefield, WV (July 2010)
B. Application of CFRP in the construction
In research Hota G and Liang R in 2011 relating to the use
of CFRP Civil Infrastructures. This study is an example of
CFRP retrofitting has been Widely used successfully to
Strengthen the structures as an effective disaster prevention
approach or to restore the damaged structures after disasters
such as hurricanes and Earthquakes. In the United States,
many of the existing highway or railroad bridges have either
reached the end of Reviews their service life or require
rehabilitation to continue in service. Due to Decreased funding
levels for new constructions, government agencies are
interested in utilizing FRP wraps to rehabilitate structures at a
fraction of the outright replacement cost and extend the
structural service life for few more decades. The advantages of
CFRP wraps include a minimum of traffic disruption, efficient
labor utilization, ease of rehabilitation, optimization of load
transfer, and cost effectiveness. CFC-WVU laboratory has
been actively involved with FRP wrapping advanced
technology development, Including specific design methods,
material selection, field installation procedures, performance
requirements and subsequent inspection techniques.
Figure 12 is a group of photos showing how damaged piles
of 11 timber railroad bridges on South Branch Valley Railroad
(SBVR) lines in Moorefield, WV were rapidly rehabilitated
and restored in-situ without affecting the rail traffic, with the
use of Fiber Reinforced Polymer (FRP) composites (July
2010). These timber bridges consisted of total span lengths
varying from 75 ft. to 1200 ft. with timber pile bents spaced
15-20 ft apart. The deteriorated piles were cracked, heartrotted, and damaged to varying lengths. This rapid
rehabilitation technique can be used on various other structural
members including steel and reinforced concrete members in a
highly cost effective manner to extend the service life of
structural systems.
Furthermore, much of the existing building stock in
Europe, as well as in developing countries, has been designed
According to old standards and has little or no seismic
provision and Often suffers from poor materials and
construction practices. As a result, many existing buildings
have deficient lateral load resistance, insufficient energy
dissipation and can Rapidly lose during Earthquakes Reviews
their strength, leading to collapse. Retrofit of seismically
deficient structures before Earthquakes provides a feasible and
cost-effective approach to improving Reviews their load
carrying capacity and reducing Reviews their vulnerability.
Over the last decade, the use of externally bonded fiber
composite materials (FRPs) has offered engineers a new
solution for strengthening seismically deficient buildings
(Figure 13). The initial cost of FRP for strengthening is
usually higher than conventional structural materials.
However, they are much Easier to apply, and this is where
composites offer significant economic benefits.
Figure 13. Strengthening of a RC columns with FRPs
27
Proceeding of International Symposium on Smart Material and Mechatronics
III. RESEARCH MODEL
shown the maximum load difference of different variations of
the test specimen.
In this article the authors also present the results of
research on the use of CFRP materials are applied to the round
concrete columns that serve as external reinforcement in
concrete columns. The work is done through analysis and
experimental studies. The study analysis was conducted to
study the stress strain models that utilize concrete as a material
CFRP reinforcement in concrete columns.
Models stress strain of CFRP material is then summarized
and implemented in the form of a computer program Confined
Column v.1.0 (CC-v.1.0) that have been made to produce a
stress strain relationship chart. The program is used to validate
the results of experimental studies.
In this study the implementation of the test specimen used
in the column that is round with a diameter of 130 mm
diameter by 700 mm long round columns. In this study
conducted experiments on plain concrete columns (PS), which
uses concrete columns internally reinforced steel (BT), and
given a plain concrete column steel reinforcement internally
and externally CFRP material (B-1 LS). Concrete used is
normal strength concrete with a target compressive strength of
20.75 MPa. For longitudinal reinforcing steel used 6 ∅ 10 and
ridden spiral reinforcement is ∅ 8-50 mm. Furthermore, an
analysis and evaluation of the results of testing that has been
done to study the behavior of restrained concrete columns with
fiber polymer (CFRP) as well as models of effective restraint.
In addition, we will get the formulation / formulas stress strain
constitutive relations that occur due to the confined of the
fiber polymer (CFRP). The resulting formulation results will
be validated using the constitutive equations of the results of
other researchers with the help of Confined Column v.1.0
program (CC-v.1.0) that have been made.
TABLE 1. RESULTS OF TESTS ON CIRCULAR COLUMN
Specimen
Code
Maximum
Load
(kN)
1
2
3
4
5
6
7
8
9
PS-A
PS-B
PS-C
BT-A
BT-B
BT-C
B-1 LS-A
B-1 LS-B
B-1 LS-C
150
140
160
230
250
240
280
270
260
Maximum
Load
Average
(kN)
Increased
Maximum
Load
(%)
150
-
240
60
270
80
Based on the research that has been summarized in Table 1
show that the concrete column specimen without restraint (PS)
capable of withstanding a load of 150 kN, while the concrete
columns with transverse reinforcement confinement and
longitudinal reinforcement (BT) is able to withstand a load of
240 kN, and the concrete columns with transverse
reinforcement confinement and longitudinal reinforcement as
well as externally with CFRPconfinement (B-1 LS) capable of
withstanding a load of 270 kN. This suggests that an increase
in the strength of the concrete column specimen BT by 60%
when compared to columns that are not confined, while the
test specimen B-1 LS increase is as high as 80% when
compared to the concrete columns that are not confined. It
shows that the functioning of the confinement of transverse
and longitudinal steel reinforcement and confinement
externally with the use of CFRP materials.
Stress strain curves for all test objects can be seen in
Figure 14 as the comparative column specimens studied.
Concrete column specimens were observed without the use of
confinement, and concrete columns using CFRP restraint. In
figure 14 it can be seen that the presence of transverse and
longitudinal reinforcement (Specimen BT) can improve axial
compressive stress. The most influence on the value of
confinement is the specimen B-1 LS because in addition to
using the transverse and longitudinal reinforcement, also using
CFRP material as an external confinement. Increased
confinement posed CFRP compared to concrete specimen
(BT) is 12.5%. The results showed that with the use of CFRP
materials as an external confinement can increase the capacity
of the concrete column. It is appropriate that disclosed by Mac
Gregor (1997) which states loading triaxial strength of
concrete with concrete (confinement) is greater than the
compressive uniaxial loading.
Model collapse on circular concrete columns after testing
in the laboratory can be seen in figure 13 below.
(b) B T
No
Source: Research Results
IV. RESULT AND DISCUSSION
(a) PS
ISBN 978-602-71380-1-8
(c) B-1LS
Figure 13. Collapse of Specimen Model PS, BT,
and B-1 LS
Based on laboratory test results obtained from the
maximum load of each variation of the concrete columns were
tested, as shown in Table 1 in the column of concrete without
confinement (PS), concrete columns with transverse
reinforcement confinement and longitudinal reinforcement
(BT), and concrete columns with confinement transverse and
longitudinal steel reinforcement and external confinement
with CFRP 1 (one) layer spacing (B-1 LS). In Table 1 are
A. Validation of Experimental Results
Validation of Value Unconfined Concrete Strength
The result of the increase in strength of confined concrete
validation (K) to review the model formulation by previous
researchers using triaxial test results can be seen in Table 2.
The model being simulated is a model of Campione and
28
Proceeding of International Symposium on Smart Material and Mechatronics
Miragle (2003), the model of Li et al (2003), and the model of
Lam and Eng (2003).
on the results of laboratory testing through system testing
using Load Control technique with the speed of movement
(stroke) of 0.012 mm / sec. The results of testing this model
produces a stress strain curve the shape of the ascending
branch. The resulting model is then summarized and carefully
observed the movement of the model curve shape of the
experimental results. The resulting shape of the curve in
general form a parabolic curve with peak coordinates
(
). The results of the model formulation of the stress
strain curve of the experimental results with comparison of
some models of previous investigators are shown in Table 3,
while, for the model of confined concrete stress strain curve of
the experimental results with two models, namely the model in
terms of Li et al (2003) and Model Campione and Miragle
(2003) is shown in figure 15.
TABLE 3. CURVE MODEL OF STRESS STRAIN
FOR CONFINED CONCRETE
Figure 14. Stress strain curve of specimens
Researchers
Formulation of the model equations are then processed to
determine predicted for confined concrete strength (K) as a
validation of the experimental results of short column testing
(short column) with CFRP confined concrete is given a
concentric load. Validation is performed to determine the
accuracy of each equation in predicting an increase in
confined concrete strength (K) based on the experimental
results. All three models are reviewed each have a value of
COV (Coefficient of Variation) above 9%. Among the three
models is the model of Lam and Teng have COV higher value,
is 10.71%, which means closer to the experimental results
with a 11.13% COV value. Meanwhile, the model of Li et al
have COV values of 10:07%, and The Campione and Miragle
model has the value COV of 9.27% of the experimental
results. All three models are reviewed indicates that predicted
for confined concrete strength (K) to the experimental results
are considered quite good because it has the value COV
proximity to the experimental results.
Lam and
Teng Model
(2003)
COV (%) for (
Campione and Miragle
(2003)
9.27 %
Li et al (2003)
10.07 %
Lam and Teng (2003)
10.71 %
Experimental Result
11.13 %
Curve Model of Stress Strain for Confined Concrete
Ascending Branch
Descending branch
=
Model Li
et al (2003)
Campione
and Miragle
Model
(2003)
Proposed
Model
Experimental
Result
Source: Research Result of Lam and Teng, Li et al,
and Campione and Miragle Model
TABEL 2. COV VALUE OF PREDICTION RESULT VS
EXPERIMENTAL RESULT
Model
ISBN 978-602-71380-1-8
)
Source: Analysis Results
Curve model validation of stress-strain unconfined
concrete for experimental results
Stress strain curve modeling confined concrete (confined
concrete) transverse and longitudinal reinforcement and
externally CFRP layers are calculated based on the results of
experiments on 9 test specimens in the form of columns of
normal strength concrete (NSC), and tested with concentric
loading. Proposed stress strain curve is given one part based
Figure 15. Proposed Model of Confined
Stress Strain Curve with Li et al, and
Campione and Miragle Model
29
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
[3] C. M. J. Ongpeng, and C. W. A. Oreta, “Effect of Carbon FRP
in Confining Circular RC Columns Using Artificial Neural
Networks”.
[4] D. A. Moran, C. P. Pantelides, “Variable Strain Ductility Ratio
for FRP Confined Concrete. Journal of Composites for
Construction,” pp.224-232, November 2002.
[5] F. Braga, R. Gigliotti, M. Laterza, “Analytical Stress-Strain
Relationship for Concrete Confined by Steel Stirrups and/or
FRP Jackets,” Journal of Structural Engineering, ASCE, ,
pp.1402-1416, September 2006.
[6] G. Campione, N. Miraglia, “Strength and Strain Capacities of
Concrete Compression Members Reinforced with FRP”. Cement
and Concrete Composites, 25(1), 31-41, 2003.
[7] G. Elnabelsy, and M. Saatcioglu, “Design of FRP Jackets for
Seismic Retrofit of Circular Concrete Columns”. Emirates
Journal for Engineering Research, 9 (2), 65-69, 2004.
[8] G. Hota, and R. Liang, “Advanced Fiber Reinforced Polymer
Composites for Suistainable Civil Infrastructures”. International
Symposium on Innovation & Sustainability of Structures in
Civil Engineering Xiamen University, China, 2011.
[9] L. A. Bisby, A. J.S. Dent, M. F. Green, “Comparison of
Confinement Models for FRP Wrapped Concrete,” ACI
Structural Journal, pp.62-72, January-February 2005.
[10] L. Lam, and J. G. Teng, “Design-Oriented Stress-Strain Model
for FRP-Confined Concrete. Construction and Building
Materials,” 17, 471-489, 2003.
[11] R. Benzaid, and A. H. Mesbah, “Circular and Square Concrete
Columns Externally Confined by CFRP Composite:
Experimental Investigation and Effective Strength Models”.
Fiber Reinforced Polymer-The Technology Applied for
Concrete Repair, 2013.
[12] R. Eid, A. N. Dancygier, and P. Paultre, “Elastoplastic
Confinement Model for Circular Concrete Columns,” Journal of
Structural Engineering, ASCE, pp.1821-1831, december 2007.
[13] R. García, I. Hajirasouliha, K. Pilakoutas,. and M. Guadagnini,
“Seismic behaviour of EBR FRP retrofitted frames,” Advanced
Composites in Construction, Edinburgh, 2009.
[14] R. Garcia, I. Hajirasouliha, K. Pilakoutas, and M. Guadagnini,
"Seismic Strengthening of RC Buildings Using CFRP," 9th US
National and 10th Canadian Conference on Earthquake
Engineering (EERI), Toronto, Canada, 2010.
[15] R. Garcia, I. Hajirasouliha, and K. Pilakoutas, "Seismic
Behaviour of deficient RC Frames Strengthened with CFRP
Composites”, Engineering Structures, In press. 2010.
[16] S. A. Carey, K. A. Harries, “Axial Behavior and Modeling of
Confined Small-Medium-, and Large-Scale Circular Sections
with Carbon FRP Jackets,” ACI Structural Journal, pp.596-604,
July-August 2005.
[17] S. Khan, I. Hajirasouliha, Pilakoutas, and K.. M. Guadagnini,
“A Framework for Earthquake Risk Assessment for Developing
Countries,” 9th US National and 10th Canadian Conference on
Earthquake Engineering (EERI), Toronto, Canada, 2010.
[18] U. Tamon, “FRP for Construction In Japan”. Hokkaido
University, JAPAN 060-8628.
[19] S. Samdani, and A. S. Sheikh, “ Analytical Study of FRP
Confined Concrete Columns”, 2003.
[20] Y. K. Yong, M. G. Nour, and E. G. Nawy, “Behavior of
Laterally Confined High Strength Concrete Under Axial Load.
Journal of Structural Engineering,” ASCE, V.114, No.2, pp.332351, February 1988.
[21] Y. Y. Li, S. H. Chen, K. C. Chang, and K. Y. Liu, “A
Constitutive Model of Concrete Confined by Steel
Reinforcements and Steel Jackets,” Canadian Journal Civil
Engineering, pp.279-288, 2005.
V. CONCLUSION
Based on the results of experimental studies that have been
done, it can be concluded as follows:
 Calculation results of experiments on the effectiveness
of the confinement of a plain column (PS), reinforced
column (BT), as well as external confinement CFRP
reinforced column (B-1 LS) with a COV value of
11:13% is considered good enough to see the result of
validation of the value of K generated by Lam and
Teng model, Li et al model, and Campione and Miragle
model and the experimental result which each have a
COV value of 10.71%, 10:07%, and 9:27%.
 The capacity of strength that occurred in plain concrete
column (PS) is 150 kN. If given additional concrete
column internally reinforced steel, the strength
increased capacity is 240 kN. Effect capacity is the
greatest force if given the confinement of steel concrete
columns internally with CFRP material externally is
equal to 270 kN. Thus, the addition of transverse and
longitudinal reinforcement confinement (BT) has
increased the strength by 60% when compared with
plain column without confinement (PS), and an
increase in capacity of the column concrete were
confined by transverse and longitudinal reinforcement
(BT) to concrete column were confined with transverse
and longitudinal reinforcement or external confinement
1 layer CFRP spacing (B-1 LS) of 12.5%. With the
results of these experiment that utilize CFRP material
as an external confinement can provide increased
strength to the concrete column and can be an
alternative material for building construction and other
building as reinforcement.
 Model of constitutive formulation proposed for stress
strain can predict the stress strain curve of CFRP
confined to the accuracy of the model is not much
different from the model of Li et al, as well as the
Miragle and Campione model.
ACKNOWLEDGMENT
The author would like to thank for department of civil
engineering, faculty of Engineering, Gorontalo State
University for conducting the research. Lastly, the author
would also like to thanks the International Conference of
ISSMM 2014 for this publication.
REFERENCES
[1] A. Mirmiran, M. Shahawy, M. Samaan, H. Echary, J. C.
Mastrapa, and O. Pico, “Effect of Column Parameters on FRP
Confined Concrete,” Journal of Composite for Construction,
pp.175-185, November 1998.
[2] C. M. J. Ongpeng, “Retrofitting RC Circular Columns Using
CFRP Sheets As Confinement”. Symposium on Infrastructure
Development and the Environment, SEAMEO-INNOTECH
University of the Philippines, Diliman, Quezon City,
PHILIPPINES, 2006.
30
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Material properties of various light metals produced
by heated mold continuous casting
Yuta Miyamoto
Mitsuhiro Okayasu
Dept. Materials Science and Engineering
Ehime University
3 Bunkyo-cho, Matsuyama, Ehime, 790-8577, Japan
[email protected]
Dept. Materials Science and Engineering
Ehime University
Matsuyama, Ehime, Japan
[email protected]
Abstract— In the present work, an attempt was made to
develop high quality cast aluminum alloys via a new casting
technology, e.g., the heated mold continuous casting (HMC)
with ultrasonic vibration (UV) process. With the UV process
in the continuous casting process, fine and spherical grains
were obtained, where the lattice structure is formed similarly
before the UV process while dislocation density increases. The
mechanical properties of the UV-HMC Al alloys are higher
than those for the related cast Al alloys without UV although
still high material ductility is obtained. The lattice and
dislocation characteristics of the continuous cast samples
made with and without the UV processes were analyzed
systematically by the EBSD observations to interrupt clearly
their mechanical properties.
subjected to ultrasonic waves of different power levels for 5
min under frequency of about 20 kHz and the maximum
power of 600 W, in which strong effect on the size and
sphericity of alpha dendrites is obvious. Moreover, high
applied ultrasonic power resulted in small, more rounded and
uniformly distributed –grain and eutectic particles. Feng et al.
[3] have attempted to treat UV into the melt hypereutectic Al–
23%Si alloy in a horn crucible.
From the above previous works, it appears that a number
of experimental works have been conducted to make high
quality cast Al alloys by the UV process [3]. However, the
authors believe that there would have still chance to apply the
UV technology in casting process. This is because, in the
previous studies, the ultrasonic vibration is conducted only to
the melt in crucibles and molds, i.e., simple approach.
Moreover, there is apparently lack of the investigations to
understand clearly the detailed vibration effect on the material
properties. This is because the previous examination has been
executed with the limited vibration conditions, e.g., a few
vibration amplitudes and frequencies.
Thus, in the present study, an attempt was made to propose
a new casting system of a heated mold continuous casting
method with ultrasonic vibration in advance. With this casting
system, mechanical properties of several Al alloys have been
investigated. To understand clearly the effects of the UV
process on the material properties of the cast Al alloys, the
lattice and dislocation characteristics were scientifically
analyzed.
Index Terms: aluminum alloy; ultrasonic vibration; continuous
casting; mechanical property; microstructural characteristic
I. INTRODUCTION
In recent years, high fuel efficiency of automotive is
required in our society, because of environmental issue. To
make this, the reduction in exhaust gases from the automotive,
such as carbon dioxide and nitrogen oxide, would be required,
as the number of automotive has been increasing to be more
than 1 billion in the world. The automotive consists of a
number of the related parts, and many of them have been
made by cast irons and steels. It has been expected to replace
Fe–based automotive parts with more lightweight metals, e.g.,
aluminum alloys. The specific weight of Fe is about 7.8,
which is more than 2.8 higher than that for Al. Recently, the
production amount of automotive parts, made of Al alloy, has
been increasing gradually.
It is general consideration that small grains with spherical
shape are significantly important to make excellent
mechanical properties. To obtain such microstructural
characteristics, some practical techniques of rapid
solidification, high casting flow and adding fine nucleating
elements are employed. Furthermore, new technologies have
been proposed with mechanical modification, including
electromagnetic vibration [1], mechanical vibration and
mechanical shearing processes.
Aghayani and Niroumand [2] have examined the effects of
ultrasonic vibration (UV) treatment on microstructural
features and tensile strength. The melt alloy in sand molds was
II. EXPERIMETNAL PROCEDURE
II-1. Material preparation
In the present study, two aluminum alloys (AC4CH and
ADC6) and pure aluminum (99.9%Al) were used. In order to
create the high mechanical properties of cast aluminum alloys,
a hybrid casting system was originally proposed, where an
ultrasonic vibration (UV) device was added to our original
heated mold continuous casting system, see Fig. 1. In this case,
a small UV device (PEF-L25A, Sanki Corp.) was employed.
The specification of this device is as follows: electric voltage:
0–240V and frequency: 40–400Hz. Such vibration is applied
directly to the cast sample during the casting process. The
HMC arrangement consists of a graphite crucible in a furnace,
a graphite mold of 5 mm in diameter, a cooling device and a
dummy rod for withdrawal of the cast sample. The graphite
mold is jointed with the graphite crucible. The cooling system
31
Proceeding of International Symposium on Smart Material and Mechatronics
was set just out of the mold. The ultrasonic vibration system
was attached near the cooling system, and the vibrations were
executed directly to the casting rod during the casting process.
The melts in the crucible were fed continuously into the mold
at 1.9 mm/s.
In the gravity casting (GC) process, the melt was poured
directly into a metal mold. Note, the GC process would not be
a represent conventional gravity casting process, as our gravity
casting system does not include sprue, runner and gate.
ISBN 978-602-71380-1-8
EDX analysis was carried out to investigate the
microstructural characteristics with an acceleration voltage of
20 kV a scanning electron microscope. The EBSD analysis
was conducted to observe the crystal orientation
characteristics with an acceleration voltage of 15 kV, beam
current 5 nA and step size 0.5–20 m. The samples were
prepared with sectioning to less than 5 mm thick and with
mirror flatness. This EBSD analysis was executed with HKL
Channel 5 software.
Cooling device
Ultrasonic vibration device
Cast rod
Casting
φ5
direction
φ50
Molten
metal
Heated
mold
40
Dummy rod
(Stainless steel)
40
Furnace
Graphite crucible
Fig. 1 Schematic illustration for the heated mold continuous casting device
with ultrasonic vibration system.
II-2. Experimental
Microstructure, lattice structure and strain characteristics
were investigated by various approaches including energydispersive X–ray spectroscopy (EDX), electron backscatter
diffraction (EBSD).
GC
III. RESULTS
III-1. microstructural characteristics
Fig. 2 depicts the optical micrographs for the pure
aluminum, AC4CH and ADC6 alloys produced by GC and
HMC processes. In this case, the HMC process was carried
out with and without ultrasonic vibration. It can be seen that
fine -Al phase and tiny eutectic structures are observed in the
HMC samples compared to their GC ones. In addition, those
grains seem to be altered slightly to more fine spherical shape
of -Al grains with the ultrasonic vibration. Interestingly,
core-like structures can be characterized in the middle of their
grain for the UV pure-aluminum. From the EDX analysis,
such core-like structure is related with the iron element (Fig.
3).
HMC
HMC with UV
Pure Al
20μm
AC4CH
Mg2Si
α-Al phase
Si
ADC6
Mg2Si
Al6(Fe, Mn)
Fig. 2 The optical micrographs for the pure aluminum, AC4CH and ADC6 alloys produced by GC and HMC processes with and without ultrasonic vibration.
32
Proceeding of International Symposium on Smart Material and Mechatronics
Fig. 4 presents the crystal orientation maps (IPF) for pure Al
and ADC6, obtained by the EBSD analysis. It is obvious that a
relatively uniform lattice structure is obtained over a large area
in the HMC samples without ultrasonic vibration, where
almost perfectly orientated crystal structure, i.e., single
crystal-like formation.
It is interest to mention that even if the UV process conducted
strongly, the crystal orientations are still relatively organized.
However, their lattice structures, i.e., misorientation angle, are
slightly disordered.
HMC–pure Al
Al-Kα
SEM image
Fe-Kα
Fe
10μm
α-Al
Fig. 3 EDX analysis for HMC samples of pure aluminum without UV.
SEM image
Misorientation angle
(2°-5°)
IPF map
Pure Al
without UV
Pure Al
with UV
ADC6
without UV
ADC6
with UV
111
20μm
001
101
Fig. 4 The crystal orientation maps for pure aluminum and ADC6 by HMC with and without ultrasonic vibration.
33
Proceeding of International Symposium on Smart Material and Mechatronics
Fig. 5 shows the electric conductivity (EC) of the cast
samples. Note, in this case, the EC values were measured
using the same specimen of 1 mm  100 mm. To understand
EC characteristics clearly, this was also carried out for
commercial wrought pure Cu, wrought pure Al and
continuous cast Al alloys. The data obtained in Fig. 5 is
indicated with the rate of the EC value based upon the copper
wire. It is clear that the electric conductivity for the
commercial wrought pure Al wire is about 60% of the Cu wire
one. Interestingly, slight improvement of the electric
conductivity for the HMC–pure Al is obvious, which is
approximately 15% higher than that for the wrought pure Al
wire. This may be attributed to the uniformly organized
crystal orientation, as mentioned in Fig. 4. Furthermore, it is
obvious that the EC values for the cast Al alloys without UV
(AC4CH and ADC6) are about 25 % higher than those for UV.
This is also influenced by the different crystal orientation
characteristics.
100
80
60
2
1. Pure Al wire
2. Pure Al
3. AC4CH
4. AC4CH (UV)
5. ADC6
6. ADC6 (UV)
7. Cu wire
350
300
1
40
3
5
4
20
0
Pure Al
Tensile stress, MPa
Electric conductivity, %
7
Fig. 7 depicts the representative tensile stress-versus-strain
curves for HMC–ADC6 alloys with and without UV process.
It is clear that there are different trends of the tensile
properties depending on the UV process. Based upon the
stress–strain curves obtained, ultimate tensile strength and
fracture strain are summarized in Fig. 8. The tensile properties
slightly increase for ADC6 with the vibration process. Such
increment of the tensile strength would be caused by the
change of the microstructural and lattice structures, as
mentioned above. On the other hand, slight high ductility for
the UV samples is attributed to the grain refinement and
spherical structure. Fig. 9 represents the relationship between
stress amplitude and cyclic number to final fracture (S-N
curve) for ADC6 with and without the UV process. It is
obvious that, like the tensile properties, the S-N curve for
ADC6-UV is located to the higher level compared to the
without UV one, namely the higher fatigue strength for
ADC6-UV. On the basis of the above experimental results, it
could be briefly summarized that the UV process is useful to
improve the mechanical properties of the cast aluminum
alloys.
AC4CH
6
ADC6
Cu
250
With UV
200
Without UV
150
100
50
Fig. 5 Rate of the electric conductivity for various metals on the basis of
the copper wire one.
0
0
79
Amp.=50%
77
Amp.=99%
10
20
Strain, %
30
Fig. 7 Representative tensile stress vs. tensile strain curves for ADC6
produced by the HMC process with and without ultrasonic vibration process.
(a)
300
Ultimate tensile strength, MPa
Vickers hardness, Hv
Fig. 6 displays the Vickers hardness of HMC–ADC6 alloy
as a function of the vibration frequency. As seen, the hardness
level of the Al alloy does not change significantly even if the
frequency is altered. However, it is obvious that high hardness
value is obtained for the sample with higher vibration
amplitude. The highest hardness by UV is about 7% high
compared to that for the cast samples without UV.
75
73
71
69
290
280
270
260
67
Without
UV
65
0
100
200
300
Frequency, Hz
400
Fig. 6 Vickers hardness of HMC-ADC6 alloy as a function of the vibration
frequency and vibration amplitude.
34
With
UV
Proceeding of International Symposium on Smart Material and Mechatronics
(b)
ACKNOWLEDGMENTS
24
This work was supported by a grant (Grant-in-Aid for
Scientific Research (C), 2014) from the Japanese
Government (Ministry of Education, Science, Sports and
Culture).
Fracture strain, %
22
20
18
REFERENCES
16
1 C Vivès, Effects of forced electromagnetic vibrations
during the solidification of aluminum alloys: Part II.
Solidification in the presence of collinear variable and
stationary magnetic fields, Metall. Mater. Trans. B,
27B(1996)457–464.
14
12
10
Without
UV
With
UV
2 M.K. Aghayani, B. Niroumand, Effects of ultrasonic
treatment on microstructure and tensile strength of AZ91
magnesium alloy, J. Alloys Compd. 509(2011)114–122.
Fig. 8 Tensile properties of the HMC–ADC6 with and without UV process:(a)
ultimate tensile strength and (b) fracture strain.
3 H.K. Feng, S.R. Yu, Y.L. Li, L.Y. Gong, Effect of
ultrasonic treatment on microstructures of hypereutectic
Al-Si alloy, J. Mater. Process. Technol. 208(2008)330–
335.
Stress amplitude, MPa
350
300
250
200
150
100
Without UV
50
0
With UV
3
10
1000
104
105
106 10000000
107
10000
100000
1000000
Number of cycles to failure
Fig. 9 S-N curves for the HMC–ADC6 with and without UV process.
IV. CONCLUSIONS
1) Electric conductivity for the HMC–pure aluminum is about
15% higher than that for the wrought pure Al wire. This is
attributed to the uniformly organized crystal orientation.
With the UV process, the EC levels for HMC-Al alloys
decrease about 25% compared to those without UV, which
is affected by the randomly distributed lattice structure
arising from the UV process.
2) The hardness level of the ADC6 alloy is not changed
significantly with increasing the UV frequency. In contrast,
the high hardness was obtained as loaded at the high
vibration amplitude. The highest hardness by UV is about
7% high compared to the mean hardness of the cast
samples without UV.
3) The tensile strength and fatigue strength increase for the
ADC6 alloy with the UV process. In addition, similar to the
mechanical strength, the material ductility is also relatively
increased with the UV process. Such increments of the
strength are attributed to the change of the microstructural
and lattice structures.
35
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Microstructure and Mechanical Properties of Al10Zn-4.5Mg-xCu Turbine Impeller Produced by
Investment Casting
Muhammad Syahid 1,2 a
Bondan T. Sofyan1,b
Insani Mukhlisa 1
1
Department of Metallurgy and Materials,
University of Indonesia, Indonesia
2
Department of Mechanical Engineering,
Hasanuddin University,Indonesia
a
[email protected]
Abstract--- Aluminium alloy can be applied for a turbine
impeller of Organic Rankine Cycle power plant that operates at
temperatures below 150 oC. Aluminum turbine impeller can
enhance efficiency of turbine due to light weight material. Al
7xxxx commonly use for impeller due to good mechanical
properties. Turbine impeller commonly produced by machining
but which is time-consuming and less efficient because of material
removal. . It can be replace by Investment casting to produce
impeller turbine due to their complex geometry and precision.
This study identifies effect Cu content of Al-9Zn-4Mg-xCu on the
microstructure and mechanical properties of turbine impeller
produce by investment casting. The study also identifies casting
defect of turbine impeller. The structures consisted of α-Al,
MgZn2, CuMgAl2 and CuAl2. Higher Cu content is the higher
hardness value due to CuAl2 phase. Visual examination showed
that the turbine impeller was free of macro defects and misruns.
Keywords: investment casting, aluminum alloys, turbine
impeller, organic rankine cycle
I.
1
Department of Metallurgy and Materials,
University of Indonesia, Indonesia
b
[email protected]
addition of Zn, Mg and Cu considerably increase the strength
of Al alloys, due to precipitation of a much dispersed MgZn2
and CuAl2 phase during ageing.
The aims of this study is to produce turbine impeller for an
ORC power plant by investment casting. The turbine impeller
was made from Al-9Zn-4Mg with varied additions of Cu
content. Microstructure and mechanical properties were
identified to show the effect of additional Cu content.
II EXPERIMENTAL METHOD
The gating system and ceramic mold is used refer to
previous research [7]. An Al ingot, Mg and Cu ingots were
used as master alloys. Mg content was 4 wt. % and Cu content
varied at 1, 3, and 5 wt %. The alloys were melted in a graphite
crucible. Degassing was conducted by using argon gases to
avoid gas porosity. The nominal composition of the cast alloys
is presented in Table 1.
INTRODUCTION
Process technology to produce turbine impeller is
commonly by machining and investment casting because the
geometry of turbine impeller is more complex, need highly
precise, and the tips of the blades are very thin. Fabrication by
machining produces is high precision but the cost is high due to
lengthy process and material removal. Alternatively,
investment casting can produce a highly precise parts
characterized with more complex geometry. It is cheaper than
machining since the material removal is not required [1].
However the microstructure and mechanical properties of cast
products highly depend on alloys, ceramic shell, gating system
solidification process. The failure rate in production of the
impeller by investment casting has been 30-40% [2]. Some
defects commonly appear in products of investment casting:
these include misruns, inclusion, macro and micro porosity and
hot cracks[3,4].
Al 7xxxx commonly use for impeller due to lightweight,
high strength to weight ratio, high corrosion resistance and
excellent mechanical properties. Compressor turbocharger that
operates on conditions similar to ORC turbine impeller, used
Al-(1-2) Mg-(2-3) Cu (wt %) with addition of titanium [5].
Wallace et al [6], produced turbocharger impeller that operates
at temperature of 90 ° C by using thixocasting method. The
composition of the alloy is Al-6Si-3Cu-0.35M (wt. %) with
tensile strength at 400 MPa and the elongation of 7.7 %. The
Alloys
I
II
III
Table 1: Composition of cast alloys
Composition (wt%)
Zn
Mg Cu
Fe
Mn
10.73 4.48 1.02 0.21
0.006
11.21 4.45 3.18 0.25
0.009
9.6
4.39 4.94 0.22
0.011
Al
Balance
Balance
Balance
The melting alloys was pouring into ceramic mold at
temperature 750 oC while preheating temperature for the
ceramic shell mold was 730 oC. After solidification, the
ceramic shell mold was broken. The turbine impeller casting
product was characterized by visual examination.
Microstructural characterization by Optical microscopy and
scanning electron microscopy (SEM). Mechanical properties
was identified by hardness testing were cut from the hub of the
impeller. The hardness based on the Rockwell B scale
III RESULTS AND DISCUSSION
A. Visual inspection.
Turbine impeller casts were free of macro defects, such as
misruns, macro porosity, and surface cracks. Smooth surfaces
were obtained for all cast specimens. The successful
elimination of defects resulted from a good design of gating
system, ceramic shell molds, and casting parameters (i.e.,
36
Proceeding of International Symposium on Smart Material and Mechatronics
pouring temperature and preheating temperature for shell
molds)[2].
The poor design gating system will lead to misruns and
shrinkage. The smooth surface indicated that no reaction
occurred between the alloys and the first layer of the ceramic
mold. The absence of misruns and shrinkages in the cast
showed that the permeability of the molds was adequate to
expand the gases within [4].
ISBN 978-602-71380-1-8
.
30 mm
b
a
Figure 2 Microstructures of the hub of turbine impeller
of Al-10Zn-4,5Mg-xCu with (a, b) 1 (c, d) 3 and
(e, f) 5 wt. % Cu produced by investment casting.
30 mm
Figure 1 (a) cast product of investment casting b)Turbine
impeller that made of Al-9Zn-4Mg-5Cu produced by
investment casting,
B. Effect of Cu on Microstructures
Figure 2 shows microstructures of the hub of impeller with
varied Cu content. Microstuctures of the Al-10.7Zn-4.48Mg1.02Cu alloy are shown in Figures 2a and 2b. It can be seen the
dendritic grain structucture with the second phase around the
grain boundaries. The second phase commonly found in Al-ZnMg-Cu alloys are MgZn2, Mg3Zn3Al2 , Al-Mg3Zn3Al2 dan AlMgZn2. The addition of Cu form CuMgAl2 and CuAl2 phase.
The microstructures of the Al-11.21Zn-4.45Mg-3.18Cu alloys
are shown in Figures 2c and 2d. The structure is relatively the
same as that of the Al-10.7Zn-4.48Mg-1.02Cu. But it has a
finer grain size and the second phase in the grain bondary look
thicker. The microstructures of the Al-9.6Zn-4.39Mg-4.94Cu
alloy are shown in Figures 2e and 2f. The grain size more finer
than the others. The second phase in the grain bondary also
thicker than the others. The CuAl2 and CuMgAl2 formed in this
alloys due to higher Cu content. Different amount of in CuAl2
and CuMgAl2 phase caused significant differences in
mechanical properties.
Figure 3 Backscettered SEM of a tip of impeller
11.21Zn-4.45Mg-3.18Cu alloy.
of Al-
Table 2 Elemental composition on the structure in the alloy Al11.21Zn-4.45Mg-3.18Cu at position shown in
Figure 3
No
37
Rata-rata unsur (wt.%)
Zn
Mg
Cu
Al
Phase may form*
1
4.48
2.70
1.52
91.29
Al (matriks)
2
3
4
3.80
5.49
4.15
1.91
2.91
2.49
2.09
0.62
9.75
92.20
90.98
83.61
CuMgAl2
Mg3Zn3Al2
CuAl2
Proceeding of International Symposium on Smart Material and Mechatronics
Figure 3 shows the FE SEM of Al-11.21Zn-4.48Mg-1.02Cu
alloy. The α-Al matrix can be seen (position 1), and the
CuMgAl2 phases (position 2), the thicker phase around
boundary is MgZn2, Mg3Zn3Al2 , Al-Mg3Zn3Al2 or Al-MgZn2
and it is clear that the white phases are CuAl2 (position 4)),
The alloys and the corresponding elemental analysis is provide
in Table 2. Phase formations on as-cast 1 Cu and 5 Cu
containing alloys were similar except for quantities. The
presence of a second phase MgZn2 and CuAl2 will increase the
hardness as cast and are expected to be precipitates after heat
treatment [7].
ISBN 978-602-71380-1-8
higher the Cu content the higher the hardness value due to finer
grain size and promote more CuAl2 and CuMgAl2 phases
which is hard and tough.
ACKNOWLEDGMENTS
The research was funded by the PUPT 2014. The authors
would like to thanks also to PT. Metinca Prima for facilitating
investment casting process. MS is grateful for the provision of
scholarship by BPP-DN Dikti .
REFERENCES
Hardness Value
(HRB)
78
[1] Jones, S. and Yuan, C., Advances in shell moulding for
investment casting, J. Mat. Proc. Tech., Vol. 135, 2003,
pp. 258–265.
[2] Lia, D. Z., Campbell, J., Li, Y. Y., Filling system for
investment cast Ni-base turbine blades, J. Mat. Proc.
Tech., Vol. 148, 2004, pp. 310–316.
[3] Gunasegaram, D. R., Farnsworth, D. J., Nguyen, T. T.,
Identification of critical factors affecting shrinkage
porosity in permanent mold casting using numerical
simulations based on design of experiments, J. Mat. Proc.
Tech., Vol. 209, pp.1209-1219
[4] Jovanovic, M.T, Dimcic, B., Bobic, I., Zec, S.,
Maksimovic, V., Microstructure and mechanical
properties of precision cast TiAl turbocharger, J. Mat.
Proc. Tech., Vol. 167, 2005, pp.14-21
[5] Furukawa-Sky Aluminum Corporation Chiyoda-Ku
Tokyo (Jp), Cast Aluminum Alloy Compressor Wheel
For A Turbocharger, European Patent Application, 2005
[6] Wallace.G, Jackson.A.P, Midson. S. P. and Zhu.Q,
“High-Quality Aluminum Turbocharger Impellers
Produced By Thixocasting “, Trans.Nonferrous
Met.Soc.China, vol 20, 2010, pp 1786 – 1791
[7] Muhammad Syahid, Bondan T.Sofyan, Singgih Giri
Basuki, Bayu Adam, “Characterization of Al-7Si-Mg-Cu
Alloy Radial Inflow Turbine Blade produced by
Investment Casting, Advances Material Research,
vol.789, 2013, pp 324-329
[8] Zang jin-xin, Zhang kun, Dai Sheng-long, Precipitation
behavior and properties of a new high strength Al-Zn-MgCu alloy, Tran. Nonferrous Met. Soc. China, Vol. 22,
2012, pp. 2638-2644
76
74
72
70
1
3
5
Cu(wt. %)
Figure 4 Effect of Cu on hardness of the Al-10Zn-4.5MgxCu alloys
C. Effects of Cu on hardness value
Figure 4 shows the effect of Cu on hardness of the Al10Zn-4.5Mg-xCu alloys. The addition of Cu significantly
increases the hardness value. The highest hardness value was
obtained by 5Cu of 77 HRB. The alloy contain 5 wt. % Cu has
finer grain than 1 and 3 wt. % Cu caused highest hardness
value, also the higher Cu content lead to more phase CuAl2 and
CuMgAl2 which is hard and though. The presence of phase
CuAl2 and CuMgAl2 leads to significantly higher hardness.
MgZn2 phase also contribute to increase mechanical properties
of the alloys [8].
IV CONCLUSION
The turbine impeller produce by investment casting show
free from macro defect such as misrun, macro porosity,
shrinkage and others surface defects. It is indicate that gating
system design, casting parameters and ceramic shell mold work
optimally.
Microstructures of turbine impeller made of Al-10Zn4.5Mg-xCu alloys fabricated by investment casting mainly
consisted of α-Al, MgZn2, CuAl2, and CuMgAl2. Alloy
containing 5 wt. % Cu achieved the hardness of 77 HRB. The
38
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Study of Performance Improvement of Various
Stoves with Waste Biomass Briquettes Fuel
Effendy Arif
Sallolo Suluh
Dept. of Mechanical Engineering
Faculty of Engineering
Hasanuddin University
Makassar, south Sulawesi, Indonesia
[email protected]
Dept. of Mechanical Engineering
Faculty of Engineering
Akademi Teknologi Industri Dewantara
Palopo, South Sulawesi, Indonesia
[email protected]
Abstract—The utilization of effective and efficient biomass
briquettes is strongly influenced by the type of stoves used and
how to use them. This study aims to make coconut shell charcoal
briquettes as stove fuel, to conduct the proximate, heating value
and physical property tests, and then to test the performance of
temperature, water boiling ability and the efficiency of three types
of stoves before and after modification. The method used is an
experimental method by utilizing waste coconut shell charcoal
briquettes as stove fuel on three different types before and after
modification. The results of the analysis of the chemical
composition test with average proximate analysis were 1.67 % of
water content, 13.03 % of ash content, 28.69 % of volatile matters,
and 56.69 % of fixed carbon. Heating value was obtained in 4949
kcal/kg, compressive strength 0.489 gr/cm2 and density 0.705
gr/cm3. The results of combustion tests on three different types of
stoves before and after modification indicated that the stove
burners K’1 and K’3 were two types (codes) of the most superior
briquette stove in terms of the boiling ability and modifications to
suit the combustion efficiency (K’1 for 71.30 % and K’3 for 70.73
%). Both of these stoves have their respective advantages. K’3
stove, in particular, can be locally produced (from clay) and
affordable in price. The three stoves had significant increase in
performance improvement after modification.
Key words: coconut shell charcoal briquettes, chemical
composition, calorific value, physical property, and stove efficiency
I. INTRODUCTION
Indonesia was formerly known as one of the OPEC countries, a
world oil-producing organization. However, since 2003
Indonesia has turned into an oil-importing country. In 2005,
Indonesia’s energy consumption was approximately 700 barrels
of oil equivalent (BOE) per year. This amount, approximately
57% of energy comes from oil, 24 % gas, 13 % coal, and the
remainders are from hydroelectric, geothermal, and so on.
Due to the impact of the prolonged economic crisis, the
conditions change drastically when the fuel subsidies were
gradually being phased out. Several layers of society, not only
the lower class and the middle class but also the domestic
industry, began to feel the weight of the fuel subsidy removal.
Facts and data show that the use of fossil fuel was reaching to
the end, because the amount of oil reserves were running low.
High oil prices are unstable and continue to rise. The issues that
fossil fuels cause environmental damage already started to
prove. Along with the growth of the world population which
continues to increase, people are encouraged to find alternative
sources of new energy by utilizing renewable energy sources.
Some types of energy sources which can be renewed and
developed are solar energy, ocean thermal energy (OTEC) and
biomass energy. These biomass or organic materials can be
processed as alternative fuel, such as briquette.
Coconut shell charcoal is the product obtained from the
incomplete combustion of coconut shell. Charcoal gives a
higher heat and less smoke and can be smoothed and then
compressed into briquettes in a variety of forms, in which the
use of briquettes will be more practical, efficient and
economical as well as easy to get than fire wood.
The studies carried in connection with coconut shell
briquettes that have more heating value than other biomass
briquettes i.e. : Siti Jamilatun (2008 and 2011), found that
coconut shell briquette was the most optimum and economic
alternative fuel, which was quite high in heating value 5779.11
kkal/kg. Herotje Siwi (2010) obtained the heating value of
coconut shell of 4569.22 kcal/kg and Meli and Muslimin (2010)
obtained 5410.77 kkal/kg. The heating value difference in some
previous research were probably because of the briquette’s
different manufacturing process and material composition.
Esmar Budi (2011) found that coconut shell charcoal has
carbon content of 76.32 % which was potentially good as fuel.
Based on the above considerations, biomass energy in the
form of coconut shell briquette used as fuel in various briquette
stoves, and their modification to improve performances in
increasing the effectiveness and efficiency of alternative fuels
to ease the burden of government, especially to the people who
had a hard time finding kerosene. Therefore, it is necessary to
study performance improvement possibility of various stoves
with waste biomass fuel briquettes to reduce the dependence on
petroleum, especially kerosene, and to look for a more
economic alternative energy.
II.
THEORETICAL BACKGROUND
Briquette stove a cooking appliance that uses fuel from
briquette, which was a solid material that has been processed
either with or without carbonization process derived from coal
biomass or the like. Nowadays, the use of briquette is not
unfamiliar anymore, because of the recommendation of
government to diversify energy. Moreover, Indonesia’s coal
reserves were very abundant, as well as biomass. Materials used
in producing the stove affect the appearance, durability, and
quality of heat utilization. The types consist of:
39
Proceeding of International Symposium on Smart Material and Mechatronics
Britubara Stove (briquette-coal) is one type of stoves which
are coated with flame-retardant materials and heat resistant.
However, if it is not carefully used, it will be easily broken
and it can not be used anymore. Hereinafter referred to as
K1 stove.
2. KM stove is a stove briquette made of durable metal
material, but is not stainless so the appearance changes
along with the duration of use. Hereinafter referred to as
K2 stove.
3. Clay oven or is commonly called brazier, is made of
pottery raw materials, such as burnt clay, is widely
available in the community and is generally used mostly in
rural communities. Hereinafter referred to as K3 stove[1].
Coconut (Cocos nucifera) is a plant that grows in tropical
regions and lowlands which now has become industrial
plantation crops. Plants of this palmae tribe has a straight trunk
and the only species in genus cocos. This plant is believed to
have come from the shores of Indian Ocean on the Asian side,
but has since spread throughout many world tropical beaches.
Coconut is also a multipurpose tree in the tropical community[2].
Fuel briquettes are defined as fuels produced from organic
material through compaction, external charcoaling, full
carbonation or combined. The others means briquetting
according is basically a densification of compacting process
which aims to improve the physical properties of a material so
as to facilitate its handling[3].
Biomass materials used to make the briquettes are from:
1. Wood processing wastes such as : logging residues, bark,
saw dusk, shavinos, waste timber.
2. Agricultural wastes such as : straw, bagasse, dried leaves.
3. Fibrous material wastes such as : cotton fiber, jute, coconut
coir.
4. Food processing wastes such as: nut skin, fruit seeds, fruit
peel.
5. Cellulose such as : paper waste, cardboard.
Based on its shape, briquette shape can be divided into two
types, namely bee nest and egg.
1) Cylinder-type(bee nest), for household use. This type of
briquette is more known and popular, cylindrical shape
with a large hole in the middle and several small holes.
ISBN 978-602-71380-1-8
1.
Figure1. Cylinder-type briquette (wasp nest)
2) Egg-type, for domestic industry. This type of briquette is
usually used for burning lime, brick, tile, pottery, and
blacksmith. It is oval-shaped with customized size[4]
Figure 2. Egg-type briquette
Several factors used as the standard of charcoal briquettes
are:[5]
a. Water content (moisture)
The water content in the fuel, the water contained in the
wood or wood products is defined as moisture content.
b. Ash content (Ash)
Ash or mineral contained in the solid fuel is a fireproof
material after combustion process. Ash is burnt material when
it is solidified
c. Volatile matters
Volatile matters are one of characteristics contained in
briquette. The more content of volatile matters in the biobriquette, the easier it is to burn and lit so the combustion rate is
faster.
d. Fixed carbon
The content of fixed carbonor also called fixed carbon
content (FCC) which is contained in fuels such as charcoal
(char), is a component which does not form a gas when it
burns.
e. Heating value
Heating value of the fuel is the amount of heat generated
from and caused by a gram of fuel to raise the temperature of
one gram of water from 3.5 oC – 4.5 oC, with unit of calories.
Loam or clay soil is a soil with a very fine grain, is plastic
(malleable) and has adhesive power. Clay soil is divided into
two types, primary and secondary clay. Type of clay soil used
in this research is secondary clay. This is because of its
physical form, which after burning, the clay remains with the
color of darkish light brown. This is the characteristic of
secondary-type clay[5]
Cassava flour (tapioca flour) is starch obtained from
cassava root tubers. Tapioca has physical properties similar to
sago starch, so the use of both can be interchanged. It is
widely-used in food industry, such as in pudding making, soup,
etc.
Combustion is rapid reaction between the fuels from the
air. This process is the release of thermal energy from the fuel.
This thermal energy is released during the combustion reaction,
where the oxygen, CO2, water and other substances contained
in the combustion gases through the release of heat.
III.
RESEARCH METHOD
This study was conducted between February and April 2014
with a range of activities including : measuring the dimensional
of the three stoves that would be modified, making coconut
shell charcoal briquettes in the shape of a bee nest, proximate
and heating value testing, physical property testing as well as
water boiling testing and briquette combustion on three
different stove before and after modification. The modification
efforts intended to improve the stove cooking time and their
40
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
effiency by getting a high temperature and to minimize heat 2. KM stove (K2) was distributed by UD Barokah at Jalan
loss. The process of making briquettes and combustion test was
Ketilang number 9 Makassar, South Sulawesi. This stove
carried out in Production Process Engineering Laboratory
was made from steel plate. This stove would be modified
UNHAS Makassar, proximate and heating value tests were
on the exterior walls of the air hole stove. By adding 8
carried out in Animal Feed Chemical Laboratory, Faculty of
pieces of air hole which were originally 11 mm in diameter.
Animal Husbandry UNHAS Makassar and physical test was
And with asbestos insulation and zinc bound with wire. The
carried out in Laboratory of Metal Science, Faculty of
distance between each air hole to others was 31 mm. With
Engineering UKIP Makassar. All data including combustion
this addition and insulations, more air was expected to enter,
time, flame temperature, water temperature, and ambience
so the combustion could produce higher temperature and
temperature displayed on this study derived from measurements
shorter cooking time.
Photo K2 stove and photo
on experiments conducted in the laboratory, while the formula
modification of K2 stove can be seen in Figure 4 below.
used to calculate the thermal efficiency was obtained from
several reference books.
A. Materials and Equipment
Materials and equipment used in this study were as follows:
1. Material:
a. Coconut shell
b. Tapioca starch
c. Clay
d. Water
2. Equipment:
- Briquette press machine
- Coffee grinder
a.
K2 stove before modification
b. Modifield of K2 stove
- Carbonization drum
Gambar 4. Photos of K2 stove
- Sieves (40-60 mesh)
- Briquette stove
3. K3 stove was made by Takalar’s pottery center with wood
charcoal as fuel and as a substitute for kerosene stove. This
- Thermocouple
stove was made of clay (pottery) which was very easy to
- Scales
find in traditional markets. This stove would be modified by
- Water
adding aluminum plate cylinder with a thickness of 0,9 mm,
- Bomb calorimeter
a diameter of 90 mm,a high of 140 mm by making one row
- Aluminum pot
of air holes surrounding the cylinder. The diameter of each
- Beaker glass
air holes 10 mm and the distance between the air holes was
B. Modified Stoves
20 mm. The purpose was to add insulation to reduce the
The method used in this study was an experimental method,
heat loss to the walls in radial direction. Photo K3 stove and
which was to modify 3 different types of stoves.
photo additing cylinder could be seen in Figure 5 below.
1. Britubara Stove (K1), made by PT Britubara Indoraya
Indonesia, made of porcelain-coated steel plate, heatresistant (1500 oC), which was a surface heat-resistant
cylinders. This stove modified by adding aluminum plate
cylinder with a thickness of 0,9 mm, a diameter of 90 mm, a
high of 140 mm by making one rows of air holes surrounding
the cylinder. The diameter of each air holes 10 mm and the
distance between the air holes was 20 mm. The purpose was
to add insulation to reduce the heat loss to the walls in radial
direction. Photo K1 stove and photo additing cylinder
modification can be seen in Figure 3 below.
c.
K3stove before modification
b. Additional cylinder for K3 stove
Figure 5. Photos of K3 stove
Stove code testing which was used before and after
modification in various briquette stove could be seen in Table 1
below:
a.
K1 stove before modification
b. Additional cylinder for K1 stove
Figure 3. Photos K1 stove
41
Proceeding of International Symposium on Smart Material and Mechatronics
Table 1. Stove’s codes which are observed
No
Stove’s
Codes
1.
K1
2.
K’1
3.
K2
4.
K’2
5.
K3
6.
K’3
Stand-for
Britubara
(charcoalbriquette) stove
Modification by additing
with one cylinder .
KM stove
Modified K2 stove by
additing 8 new air holes
and insulation
Stove/clay oven
Clay oven by additing
one cylinder
ISBN 978-602-71380-1-8
Briquette volume calculation:
V = Total volume – hole volume – 4 x Side holes volume
= (3.14 x 3.252 x 4,5)-(3.14 x 0.752 x 4.5) – (4 x(3.14x
0.42 x4.5))
= 132.2568 cm3
2. Proximate and heating value testing results shown below in
table 2
Table 2. Result recapitulation of Proximate and heating
value
Composition (%)
Heating
No
value
Volatile Fixed
Water
Ash
(Kcal/kg)
matters Carbon
1.
1.67
13.03
28.61
56.59
4949
Notes
Britubara
stove
KM stove
Stove/clay
oven
3. Briquette physical test (Compressive Strength and density)
Physical examination consists of 2 parts:
1) Compressive Strength
C.Research Procedure
The result of compressive force was obtained as 14.39 kgf.
1. Preparation of coconut shell charcoal briquettes in the shape
Maximum pressure that could be accepted by the briquette
of bee nest.
was obtained by the equation:
2. Proximate analysis testing to obtain water content (moisture),
F
ash content (ash), volatile matters, fixed carbon and heating
Pmax 
A
value.
where:
3. Physical testing to obtain compressive strength and density.
F = compressive force = 14.39 kgf
A = compressive area = 29.39 cm2
4. Briquette combustion and water boiling testing on the three
The compressive strength found :
different types of stove before and after modification.
14.39kg
The testing consisted of two parts : briquette burning/water
Pmax 
 0.489kg / cm 2
boiling and calculations efficiency (ηth) :
29.39cm 2
ηth = Qw  Qp ...(1)
2) Density
LHV x Mbb
Density of briquette was obtained using the following equation:
 th 
( Ma x Cp air x (Ta  Tb ) )  ( M p xCp al x(Tc  Tb ))  ( Mu x Hl)
LHV x Mbb
where :
 th
Ma
Mbb
Mu
HL
Cpair
Cpal
LHV
Tb
Ta
IV.
m
Vtotal
where: m
= mass of briquette (gr) = 93.3 gr
Vtotal = briquette total volume (cm3) = 132.2568 cm3
So that:
: thermal efficiency of briquette burning (%).
: initial water mass (kg),
: remaining briquette mass in the stove
(kg).
: mass of water vapor (kg).
: vapor latent heat (kJ/kg).
: water specific heat 4.1769(kJ/kg 0C).
: aluminum/pot material specific heat (kJ/kg 0C).
: briquette lower heating value (kJ/kg).
: water’s ambient temperature
: water vapor temperature (100 0C)
TC

...(2)
 
93.3gr
 0.705 gr / cm 3
132.2568cm 3
4. Combustion / boiling testing and thermal efficiency
a. Briquette combustion
Briquette combustion data test by using water boiling
method for three types of stove before and after modification
were collected every 5 minutes, i. e : flame temperature, water
temperature, water mass, briquette mass before, after
combustion, and vapor mass. The data are presented in figure 6
to 11.
: POT TEMPERATURE (0C)
RESULTS AND DISCUSSION
A. Results of Research
1. Produced briquetees with specification :
a) Cylinder-shape briquette (bee nest)
Briquettes produced had average diameter dimension of (d) =
65 mm, high (t) = 45 mm, hole in the center hole (d1) = 15 mm
and around (d2) = 8 mm (four holes).
b) Briquettes mass and volume
The briquette mass was 93.3 gr
42
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Figure 6. Graph of temperature vs combustion time at K1
stove
Figure 10. Graph of temperature vs combustion time at K3
stove
Figure 7. Graph of temperature vs combustion time at K’1
stove
Figure 11. Graph of temperature vs combustion time at K’3
stove
Figure 6-7 was a graph of relationship between temperature
and briquette combustion time of K1 stove before and after
modification. It could be seen in the Figure 6 that water boiling
ability of K1 before modification only two times at the 48 th
minute and 123rd minute (blue line). The last, the water could
be only heated reached 96 oC (red line), where the maximum
flame temperature was 309 oC. And compared with Figure 7 the
K1 stove after modification (K’1) could boil water 5 times at
the 24th, 51st, 91st, 142nd, and 229th minute (blue line), could
only heat the water until 96oC (red line), where the maximum
flame temperature was 393oC.
Figure 8. Graph of temperature vs combustion time at K2 Figure 8-9 was a graph of relationship between temperature
stove
and briquette combustion time of K2 stove before and after
modification. It could be seen in the Figure 8 that water boiling
ability of K2 before modification only three times at the 31 st,
67th, and 111th minute (blue line). The last, the water only
heated reached 94oC (red line), where the maximum flame
temperature was 451oC. And compared in Figure 9 the K2
stove after modification (K’2) could boil water 4 times at the
31st, 67th, 107th, and 165th minute (blue line), could only heat
the water until 83oC (red line), where the maximum flame
temperature was 398oC.
Figure 10-11 was a graph of relationship between temperature
and briquette combustion time of K3 stove before and after
modification. It could be seen in the Figure 10 that water
boiling ability of K2 before modification only one time at the
rd
Figure 9. Graph of temperature vs combustion time at K’2 83 minute (blue line), the last the water could be only heated
o
reached 86
C (red line), where the maximum flame
Stove
43
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
temperature was 255 oC. And compared in Figure 11 the K3
stove after modification (K’3) could boil water 5 times at the
27th, 58th, 93rd, 157th, and 227th minute (blue line), could only
heat the water until 87oC (red line), where the maximum flame
temperature was 373oC.
b. Combustion efficiency
Example for thermal efficiency calculation was for K1 stove
boiling water for 2 times with maximum flame temperature of
309 oC with briquette burning time for 240 minutes (4 hours),
and spending briquette 0.22 kg. Furthermore, the data could be
seen as follows:
Calculation efficiency was :
K1 stove before modification only two times water boiling
with maximum temperature of 309oC and efficiency 24.92%.
And after the modification, K’1 stove’s water boiling had
suffered almost as many as 5 times with temperature of 393 oC,
efficiency was 71.30%. This was because the addition of a
modified aluminum cylinder plate hole above the line that was
able to maintain the briquette’s fire heatfor almost more than 5
hours.
K2 stove before modification produces 3 times water boiling
with temperature of 451oC and efficiency of 36.69%. K2 was
also the best before modification because the preparation of the
briquettes in the combustion chamber is arranged horizontally.
And after the modification of 8 holes variation with asbestos
insulation and zinc bound with wire, that had K’2 stove. So, the
performance improvement could be obtained twice better than
previous modification with the process of water boiling to 4
times.
K3 stove before modification only boiling the water one
time with flame temperature of 255 oC and efficiency 15.70 %.
After modification of K’1 stove, water boiling ability was
The comparison of the three types of stoves data testing obtained 5 times with not too high temperature of 373 oC and
before and after
modification in water boiling ability, efficiency of 70.73%.
This was due to the additional
maximum flame temperature, and thermal efficiency of the modification of one top row aluminum cylinder plate which was
burnt-out briquettes presented in the Table 3 below.
also able to sustain the flame briquette for 335 minutes.
Table 3. Tabulation of stove performance improvement
V.
K1 stove before modification had maximum flame temperature,
water boiling ability, thermal efficiency and burnt-out briquette
mass respectively as : 309 oC, 2 times within 240 minutes, and
24.96%. After modification of K’1, the maximum flame
temperature was 393 oC, 5 times boiling within 345 minutes
and efficiency 71.30%. K2 stove before modification was
respectively 451oC, 3 times within 215 minutes and efficiency
36.69 %. And after modification of K’2 stove was 4 times
within 240 minutes, 398 oC and 54.17 %. K3 stove before
modification produced 255oC, 5 times within 180 minutes, and
15.07%. K’3 stove after modification was 5 times within 335
minutes, 373oC, and 70.73 %.
B. Discussion
Briquettes was burned in a bee nest shape with diameter
dimension of 65 mm, height of 45 meter, one 15 mm hole and
4 small holes with diameter 8 mm. The average proximate,
heating value and physical property test results were water
content of 1.67 %, ash content 13.03 %, volatile matters 28.61
%, fixed carbon 56.69 %, heating value 4949 kkal /kg, density
0.705 gr/cm3and compressive strength 0.489 gr/cm2. The result
of these tests indicate that they did not meet with the briquette
standard, except for the water content and volatile matters.
Although quality improvement efforts by improving the
charcoaling process (reducing the ash content) and drying
process (reducing the water content) as well as the optimal use
of particles of 40-60 mesh sieve have been done.
CONCLUSION AND SUGGESTION
A. Conclusion
Briquettes had been made in the form of bee nest, with the
diameter of 65 mm, height of 45 mm, a 15 mm hole, and 4
holes with diameter of 8 mm. The average proximate test result
showed water content of 1.67 %, ash of 13.03 %, volatile
matters of 28.61 %, and fixed carbon of 56.69 %. Heating value
was obtained in 4949 kcal/kg, compressive strength 0.489
gr/cm2 with the density of 0.705 gr/cm3. The overall results of
proximate and physical property tests did not meet the existing
quality standard of briquettes, except for water content and
volatile matters.
The results of performance testing of 3 types of stove before
modification obtained : that K2 stove was superior in terms of
water boiling ability, maximum flame temperature, and thermal
efficiency of each of 3 times within 215 minutes, 451oC and
36.69%. Followed by K3 stove with water boiling 2 times
within 240 minutes, 309 oC and 24.92 %. And the lowest was
K3 stove which was to boil water as much as one time within
180 minutes, 255 oC and 15.70 %. Water boiling ability and the
best thermal efficiency after modification were produced by
K’1 and K’3 stoves which was 5 times, where K’3 stove was 10
minutes faster than K’1 stove with each efficiency 71.30% and
70.73 %. Of the overall testing both before and after
modification, the most superior are K’1 and K’3 stoves. Both of
these stoves had their respective advantages, which K’1 stove
was slightly more superior in terms of thermal efficiency and
maximum flame temperature. Nevertheless, K’3 stove excels in
ignition time and also had the additional advantage as it could
be produced locally (from clay) and affordable in price. The
best efficiency improvement was obtained by K1 stove for
46.34 % (from 24.96 %-71.30%), K2 stove for 17.485 (from
44
Proceeding of International Symposium on Smart Material and Mechatronics
36.69%-54.17%) and K3 stove for 55.03% (from 15.7070.73%)
B. Suggestions
Improved performance had been successfully performed
on three different types of stoves. But there were still
deficiencies found in this research that the process of water
boiling (cooking time) was quite long and less amount of
materials cooked. Therefore, it was necessary to conduct
further research, especially in terms of designing the optimal
stove with more fuel capacity and more ingredients that could
be cooked in relatively short time. Emission testing was also
necessary to conduct.
REFERENCES
[1] Kuncoro Heru dan Damanik Ladjiman (2005), ―Kompor Briket
Batubara Tanpa BBM Dan Hemat Biaya,‖ Penebar Swadaya,
Jakarta 2005.
[2] MM Faozi., 2008, ―Peluang Pasar Produk dari Kelapa Indonesia,
Analisa Dampak Menipisnya Cadangan Minyak Bumi Dan
Perubahan Iklim,‖ Http://www.mmfaozi.com/peluang_pasar,dll.
Diaskes 11 Februari 2009.
[3] Arif, E,.2008, ―Pemanfaatan Briket Limbah Biomassa Sebagai
Sumber Energy Alternatif,‖ Laporan Penelitian Fakultas Teknik
Universitas Hasanuddin Makassar
[4 Mahadir Sirman., 2013, ―Peningkatan Kualitas Briket Campuran
Limbah Ketam Kayu Merbabu, Sekam Padi dan Tongkol Jagung
Pada Berbagai Komposisi,‖ Laporan Penelitian Fakultas Teknik
Universitas Hasanuddin.
[5] Syahrir M., 2011, ―Limbah Batang Jagung Sebagai sumber Energi
Alternatif. Laporan,‖ Penelitian FakultasTeknik Universitas
Hasanuddin.
[6] Mangkau A (2011), ―Karakteristik Pembakaran Briket limbah
Tongkol Jagung dan Sekam Padi Dengan Berbagai Perbandingan
Tongkol Jagung Dan Sekam Padi,‖ Laporan Penelitian Fakultas
Teknik Universitas Hasanuddin.
[7] Jamilatun S., 2011, Kualitas sifat-sifat dari pembakaran tempurung
kelapa, briket serbuk gergaji kayu jati, briket sekam dan briket
batubara,‖ Prosiding seminar Nasional Teknik Kimia‖Kejuangan‖
Universitas Ahmad Dahlan Yogyakarta.
[8] Jamilatun S., 2008, ―Sifat-Sifat Penyalaan dan Pembakaran Briket
Biomassa, briket batu bara dan Arang Kayu,‖ Jurnal Rekaya
proses., Vol. 2, no. 2, 2008.
[9] Siwi H (2010), ―Pemanfaatan Limbah Tempurung Kelapa dan
Enceng Gondok Sebagai Sumber Alternatif,‖ Laporan Penelitian
Tesis Fakultas Teknik Universitas Hasanuddin
[10] Esmar Budi., 2011, ―Tinjauan Proses Pembentukan dan
Penggunaan Arang Tempurung Kelapa sebagai Bahan Bakar,”
Jurnal Penelitian Sains Vol. 4, No. 3 (B), Oktober 2011.
[11] Meli dan Muslimin (2010),
―Pengaruh Dimensi Arang
Tempurung Kelapa Terhadap Mutu Briket,‖ Skripsi S1 Teknik
Mesin Fakultas Teknik Universitas Hasanuddin.
[12] Arianto (2010), ―Daun Kering Kakao dan Daun Kering Kayu Jati
Dijadikan Sebagai Energi Alternatif,‖ Skripsi S1 Teknik Mesin
FakultasTeknik Universitas Hasanuddin.
45
ISBN 978-602-71380-1-8
ISBN: 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
Position Control of an X4-Flyer Using a Tether
Yusuke Ouchi, Keigo Watanabe, Keisuke Kinoshita, Isaku Nagai
Graduate School of Natural Science and Technology
Okayama Univercity
Okayama, Japan
[email protected]
Abstract—In Japan, aging of infrastructures, such as roads,
bridges, and water and sewer services, etc. poses a problem, and
it is required to extend the life-span of such infrastructures by
maintenance. Among infrastructures, especially bridges are
periodically inspected by short range visual observations, which
check the damage and deterioration of the surface. However,
since there are some cases where the short range visual
observation is difficult, an alternative method is required so as to
replace the short range visual observation with it. So, "X4-Flyer"
is very attractive because of realizing a movement at high altitude
easily. The objective of this study is to develop a tethered X4Flyer, so that the conventional short range visual observation of
bridges is replaced by it. In this paper, a method for the
measurement and control of the position is described by using a
tether for controlling the position of the X4-Flyer. In addition, it
is checked whether the tethered X4-Flyer can control the position
using the proposed method or not, letting it fly in a state in which
a tether is being attached.
Index Terms— Aerial Robotics, Unmanned Aerial Vehicles,
vehicle dynamics, Control.
I. INTRODUCTION
In Japan, there are about 700,000 bridges whose length is
2[m] and more. Aging of such bridges is a serious problem [1],
because those about 50[%] and more exceed 50 years in 2030,
which is the life of a bridge. As a general rule, an inspection
period of the bridge is determined to be five years and less by
the short range visual observation to cracks and corrosion [1].
The short range visual observation by a human is widely used
to evaluate the degree of damage and understand the damage
status of concrete structures such as bridges, because it is
possible to check the deterioration and damage of the surface
cracks, etc. However, there exists a case where the bridge
inspection vehicle cannot be used due to an insufficient space
under the bridge digit, and also exists a case where a close
visual inspection is difficult because of a large-scaled traffic
control, a necessity of installation of scaffolding, etc. For this
reason, it needs a substitutive method of the short range visual
observation. So, an aerial robot to move at high altitude easily
is very attractive. In particular, an "X4-Flyer", which is a kind
of VTOL type aerial robot, has high maneuverability,
compared to conventional VTOL aerial robots possessing other
rotor arrangements [2]. Therefore, it is expected to be used in
various applications, such as security, pipe inspection, etc. [3]
[4]. It needs to control the position and attitude of the X-4 Flyer,
if it is used for the inspection of infrastructures, such as bridges
etc. Although the position control using the GPS is common, it
is difficult to use such a control method in environments, such
46
as a tunnel or under a bridge, where the GPS signal does not
reach to or is weak. In addition, the manual operation by a
joystick etc. is difficult when affected by disturbances such as
wind etc. Lupashin and D’Andrea [5] have proposed a method
for controlling an X4-Flyer using a tether, not relying on the
operation of a joystick or the use of GPS. However, this
method only controls the tilt of the aircraft towards the tether,
so that it is impossible to vary independently the altitude and
position of the X4-Flyer, respectively. Therefore, it is thought
to be difficult to be used in an inspection of infrastructures,
such as bridges etc., as it is.
In this study, it aims at developing the X4-Flyer with tether
to replace the short range visual observation of infrastructure.
The position and attitude are controlled by the inertial sensor
and the altitude sensor that are attached on the airframe of the
X4-Flyer, and by a tether attached at the bottom of the airframe.
In this paper, we first describe the summary of the X4-Flyer
and a controller for the position and attitude. Then, a method is
explained for measuring the position of the X4-Flyer by
applying a tether. In addition, it is checked whether the tethered
X4-Flyer can control the position using the proposed method or
not, letting it fly in a state in which a tether is being attached.
II. OVERVIEW OF AN X4-FLYER
f1
f2
X
I
f3
{B}
f4
\
Z {E}
Ex
T
Y
Ey
Ez
Fig. 1. Definition of the coordinate system for the X4-Flyer.
Fig. 1 shows the coordinate systems and the appearance of
X4-Flyer, respectively. The body coordinate system of the X4Flyer is denoted by B and the inertial coordinate system is E ,
ISBN: 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
IV. POSITION MEASUREMENT
where a right-handed coordinate system is adopted in each
coordinate system. x, y, z denotes the coordinate of center of
gravity of the aircraft in the inertial coordinate system, and I ,
T , and \ are the rotational angles around the X , Y , and Z axis, respectively. Furthermore, the X4-Flyer mounts a circuit
and a battery near the center of the airframe, and has a total of
four rotors around these. While carrying out the flight by the
thrust generated by each rotor, the attitude control of the
airframe is also performed by adjusting the number of
revolutions of each rotor.
In this study, the airframe position of the X - and Y -axis
directions is determined by measuring the airframe height and
the slope of the tether is attached to the X4-Flyer. In this
section, a mechanism is described for measuring the slope of
the tether, and it is applied to measuring the airframe position.
A. Mechanism for Position Measurement
Potentiometer
III. CONTROLLER OF THE X4-FLYER
A. Controller of the Attitude Angle
In this article, the attitude of the X4-Flyer is controlled
using a PD control method developed in Bouabdallah’s [6].
When defining the P gains of the controller as k1 , k 3 , and k 5 ,
the D gains of the controller as k 2 , k 4 , and k 6 , the target
value of the attitude of the aircraft as I d , T d , and \ d , control
inputs as U 1 , U 2 , U 3 , and U 4 , the PD controllers for
postures are given by
U2
k1 (I Id ) k 2I
U3
k3 (T T d ) k 4T U4
k5 (\ \ d ) k6\
Fig. 2. A device for measuring the inclination of the tether attached to the
airframe.
Fig.2 shows a situation where a device for measuring the
inclination of the tether is attached to the airframe. This device
consists of a gimbal mechanism equipped with potentiometers.
This gimbal mechanism is composed of orthogonal two axes,
which can incline in any direction respectively. The inclination
of the X4-Flyer can be known by measuring the slope of each
axis, because two axes move in any direction.
B. Controller of the Position
The position control of the X4-Flyer is performed by
changing the attitude of the airframe. It is found from Fig.1 that
the X4-Flyer can move X -direction and Y -direction by tilting
the airframe to the direction T and I , respectively.
Therefore, the X4-Flyer in this paper is controlled to X -axis
and Y -axis directions by changing the target value T d and Id
in Eq. (4), respectively. That is, a feedback-roop is constructed
to generate and change the target values of attitude angles of
the airframe, by using the errors from the current position to
the target position of the airframe. Here about the position
control, a PD controller is assumed to be used as the control at
the attitude angles. When defining the P gains of the controller
as k 7 and k 9 , the D gains of the controller as k 8 and k10 ,
and the target values of attitude of the airframe as x d and y d ,
the PD position controllers are given by
Td
k 7 ( x x d ) k 8 x .
Id
k 9 ( y y d ) k10 y .
B. Position Measurement Using the Tether
za
J
E
D
Ex
l
xa
c
ya
Ey
Ez
The X4-Flyer is equipped with a tether, maintaining in the
state where it is stretched. Then, the control input U 1 is set to
be constant so as to generate a constant thrust to the height of
Z -axis direction.
Fig. 3. The tether tilt and the airframe positions.
Fig.3 shows the relationship between the inclination of the
tether and the airframe position. Let the E x -, E y -, and E z axis positions of the airframe be x a , y a , and z a . c denotes
the distance from the origin of the coordinates to a point at
which a perpendicular line given from the center of the
47
ISBN: 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
performing the attitude control are set to k1 4.5 , k 2 1.5 ,
k 3 4.5 , k 4 1.5 , k 5 1.2 , and k 6 0.4 . The constant
gains in the PD controller for performing the position control
are set to k 7 0.12 , k 8 0.8 , k9 0.18 , and k10 0.2 .
airframe intersects the E x - E y plane, and l is the length of
the tether. The slopes of the tether against the perpendicular
line directed to E x -axis and E y -axis are defined by D and
E , respectively, and the J is a slope of the tether. Then, the
X4-Flyer
airframe position in the E x -axis is given by
xa
z a tan D PC
Fixed point
Furthermore, the airframe position in the E y -axis is reduced to
ya
z a tan E The height za is required to measure the position of the X4Flyer using Eq. (3) and Eq. (4). Now, the height z a is fixed to
the height at which the tether is extended up to the maximum
length, satisfying the condition that the slope of the tether to
the airframe becomes 0 [deg] .
Controller
Wi-Fi
Tether
Fig. 5. Experimental setup.
V. EXPERIMENTS THE POSITION CONTROL USING THE TETHER
B. Results and Consideration
The experimental results are shown in Fig. 6 to Fig.8. It is
seen from Fig. 6 that the error in X -axis direction is in the
range of r0.2 [m] . It is seen from Fig. 7 that the error in Y axis direction is in the range of at most r0.25[m] . However, it
is found from Fig.8 that the airframe position in the X -axis
direction deviates about -0.2 [m] . Furthermore, this graph
shows that the flight range of the airframe is in the range of at
most 0.4 [m] from -0.6 [m] .
From these results, it is considered that the airframe
position can be measured and controlled by using the tether.
However, it is considered that the constant gain in the position
controller can be tuned more suitably to reduce the deviation in
the X -axis direction as shown in Fig. 6. In addition, it is
effective to consider that a PID controller is introduced to the
position control so as to stabilize the flight range of the
airframe in a narrower space, as shown in Fig. 8. However, it is
considered that since the position of the airframe shown in
these graphs are affected by the inclination of the aircraft when
measuring the inclination of the tether, it is a larger or smaller
value than the actual position in some cases.
The position of the X4-Flyer is measured and controlled by
using the position measurement method that applied the tether,
shown in the previous section. In this paper, the proposed
method is verified by mounting the position measuring device
by the tether on the X4-Flyer, and measuring and controlling
its position.
A. Experimental Conditions
Fig. 4. Overview of the X4-Flyer used for experiment.
Fig.4 shows the X4-Flyer used in the experiment. The
center of gravity of the airframe is approximately consistent
with the center of the airframe, by collecting heavy loads, such
as electronic circuits, batteries, etc., near the center of the
airframe. Also, a brushless DC motor is used for rotating the
rotor. The experimental setup is shown in Fig.5. A Wi-Fi
module mounted on the X4-Flyer can realize wireless
communication with a PC, so that it can be operated by a
controller (i.e., a gamepad) via the PC and obtain the log data.
Assume that the length of the tether is l 1[m] and the
experiments are conducted by fixing the other end of the tether
on the ground. The target positions and attitudes in flight are
set to (I T \ I T \ x y z) T (0 0 0 0 0 0 0 0 1) T .
The values provided from the gamepad are used for the control
input U 1 in the height direction to perform an operation such
as takeoff etc. The constant gains in the PD controller for
0.8
䢢
Position X [m]
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
0
䢢
10
20
30
Time [s]
40
50
Fig. 6. Position in the X-axis direction of the airframe.
48
Proceeding of International Symposium on Smart Material and Mechatronics
0.8
VI. CONCLUSION
䢢
In this paper, a method for measuring and controlling the
position of an X4-Flyer has been described by using a tether.
Furthermore, the proposed method was verified using a real
system. It was concluded that although the airframe position
was able to be measured, the accuracy of the position control
was to be not too high because the airframe position in X -axis
direction deviated. For future work, the introduction of a PID
controller as the position controller is considered to improve
the accuracy of the position control. In addition, the flight
experiment of the X4-Flyer is being fixed to the ground tether,
so that, we are going to have a flight experiment that the tether
will be handled by a human so that in the future.
Position Y [m]
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
0
䢢
10
20
30
Time [s]
40
50
REFERENCES
Fig. 7. Position in the Y-axis direction of the airframe.
0.8
Position Y [m]
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-0.5
0
Position X [m]
ISBN: 978-602-71380-1-8
0.5
Fig. 8. Position in the Y-axis direction relative to the position of the X-axis
direction of the airframe.
49
[1] K. Yamada, H. Adachi, M. Itami, S. Kato, “Research of flying
robot of bridge inspection -Implementation of control to avoid
contact with structure-,” Proceedings of the 2014 JSME
Conference on Robotics and Mechatronics, No. 14-2, pp.2A1B02(1)- 2A1-B02(4), May, 2014.
[2] K. Watanabe, K. Izumi, “On the Controllability of Rotor-type
Flying Robots: Why is the Drive of Four Rotors Effective for
X4-Flyer?,” Proceedings of The Society of Instrument and
Control Engineers System Integration (SICE SI 7th), pp. 820821, December, 2006.
[3] Y. Totshuka, K. Ohno, E. Takeuchi, S. Tadokoro, “Quad-rotor
Flying along the Pipe” Proceedings of the 2012 JSME
Conference on Robotics and Mechatronics, No. 12-3, pp.2A2L02(1)- 2A2-L02(4), May, 2012.
[4] K. Yokota, A. Ohya: “Children Watching System Using a Small
UAV -Position Estimation And Following Control of a Target
Person-”, Proceedings of the 2013 JSME Conference on
Robotics and Mechatronics,, no.13-2, pp.1A1-F04(1)-1A1F04(4), May, 2013.
[5] S. Lupashin and R. D’Andrea: “Stabilization of a flying vehicle
on a taut tether using inertial sensing”, Intelligent Robots and
Systems (IROS), pp.2432-2438, 2013.
[6] S. Bouabdallah and R. Siegwart: “Towards Intelligent Miniature
Flying Robots”, Springer Tracts in Advanced Robotics, vol.25,
pp.429-440, 2006.
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Development of a Mobile Robot as a Test Bed
for Tele-Presentation
Diogenes Armando D. Pascua
Sherwin A. Guirnaldo
School of Computer Studies
Mindanao State University-Iligan Institute of Technology
Iligan City,Philippines
[email protected]
Dept. of Mechanical Engg., College of Engineering
Mindanao State University
Marawi City,Philippines
[email protected]
Abstract— In this paper a human-sized tracked wheel robot
with a large payload capacity for tele-presentation is presented.
The robot is equipped with different sensors for obstacle
avoidance and localization. A high definition web camera
installed atop a pan and tilt assembly was in place as a remote
environment feedback for users. An LCD monitor provides the
visual display of the operator in the remote environment using
the standard Skype teleconferencing software. Remote control
was done via the internet through the free Teamviewer VNC
remote desktop software. Moreover, this paper presents the
design details, fabrication and evaluation of individual
components. Core mobile robot movement and navigational
controls were developed and tested. The effectiveness of the
mobile robot as a test bed for tele-presentation were evaluated
and analyzed by way of its real time response and time delay
effects of the network.
by seating in a control station-a computer with a joystick and a
webcam. The wide angle display of the robot cameras presents
the doctor with the view of the robot’s environment for
navigation and to examine patients and converse with the
hospital staff. Blackwell’s QB on the other hand is a general
purpose remote telepresence robot. It is a Wi-Fi enabled,
vaguely body-shaped wheeled robot with an ET-looking head
that has cameras for eyes and a display in its chest that shows
an image of the person it's standing in for. You can slap on
virtual-reality goggles, sensor gloves, and a backpack of
electronics to link to it over the Internet for an immersive
telepresence experience. Or you can just connect to the robot
through your laptop's browser.
Telerobots, teleoperators, and remotely operated
vehicles belong to a class of machines used to accomplish a
task remotely, without the need for human presence on site.
They are typically used in situations that are too hazardous to
human health or survival, like deep water, outer space, or toxic
environments. A growing number of telerobots is used for
applications where it would be too expensive or too timeconsuming to send humans, for example in telemedicine or
tele maintenance that requires highly trained individuals with
special skills. Sheridan [11] defines a telerobot as a machine
with sensors of the environment and devices to perform
mechanical work. The human operator supervises the telerobot
through a computer intermediary. The operator communicates
to computer information about goals, plans, and orders relative
to a remote task, getting back integrated information about
accomplishments, difficulties and sensory data. The telerobot
executes a task based on the information received from the
human operator plus its own artificial sensing and intelligence.
The cost of a telerobotic system can be considerably
reduced by using personal computers and prevailing standard
software for most of the computing tasks[1]. Another way to
reduce costs for some applications is to use the free Internet
for communication between the computer that the operator
interacts with and the computer that controls the robot.
Telepresence robots on the other hand are specialized types of
networked telerobots that offer the operator some form of both
visual and tactile feedback giving him a sense of as if he is at
the actual site of the robot.
Compared to plain robotic systems, in which a robot
executes a motion or other program without further
consultation of a user or operator, telerobotic systems provide
Index Terms—— telepresence, teleoperation, remote sensing,
Skype, Teamviewer.
I. INTRODUCTION
A.BACKGROUND
Telepresence has been the hype of the modern day
fictional films like the Surrogate, Avatar, Sleep Dealer and
Gamer. In Surrogate for example, a person will be assigned to
a robot replica of oneself which one has to control via his
mind in a recliner seat at home. The much younger, stronger,
better looking replica will then do your biddings in the world.
The real technology behind such fantastical fiction is
grounded both in far-out research and practical robotics. In the
present day real world telerobots is presently limited to
physical interfaces-through wireless internet connections,
video cameras, joysticks, and sometimes audio. Humans move
robots around at the office, in the operating room, underwater,
on the battlefield, and on Mars. Examples of advances in
today’s telepresence robots are the RP-7 of InTouch Health
Systems and
QB of Blackwell's Anybots [5]. RP-7 is a
mobile robotic platform that enables the physician to be
remotely present. Through the integration of key technologies,
RP-7 can remove time and distance barriers and effectively
extend the physician's reach to manage patient care. The
Robot’s visualization system consists of a camera,
microphone, and a speaker. Mobility and navigation is
possible via a holonomic drive system and an array of infrared
sensors. Physicians teleoperate the robots through the internet
50
Proceeding of International Symposium on Smart Material and Mechatronics
information to and require commands from the user. Their
control architectures can be described by the style and level of
autonomy.
Using standard internet technology for telerobotic
applications offer the advantage of low cost deployment.
There is no longer a requirement for expensive purpose built
equipment at each operator’s location. Almost every computer
connected to the internet can be used to control a teleoperable
device. The downside is the limitation of the varying
bandwidth and the time delays.
The internet offers the infrastructure for
communication but still the operator requires software that
displays the user interface and communicates with the
telerobot over the internet. Powerful browsers are freely
available and often updated with increased functionality. User
interfaces can be developed using a web browser only can be
achieved. Operator visual feedback can also be realized using
readily and freely available Skype software, freeing the
developer from developing its own visual feedback.
Existing commercial telepresence robots are way far
expensive ranging from 6000 to 15000 dollars. Some have
their own dedicated web servers to manage control, and visual
feedback commands thus adding to the cost overhead by
charging users a monthly subscription fee[5]. This research
was conceptualized bearing in mind the cost savings when an
elementary telepresence robot will be developed from off the
shelf components and doing away with a dedicated web
server.
Having a dedicated web control infrastructure
however leads to a much effective telepresence robot by
ensuring an almost zero communication downtime, thus a
significant reduction in delays, by efficiently rerouting IP
packets to other routes. Since the robot to be designed do
away with this type of infrastructure, a controller/robot
communication system must be developed with the ultimate
goal of minimizing the effects of transmission delays.
Telepresence robots are useful in various areas such as
remote presentation, teleconferencing, telemedicine: remote
diagnosis, remote treatment of patients with the aid of a
medical staff and remote consultation, military applications,
remote surveillance, teleoperation , advertising and remote
education Due to these benefits, a low cost telepresence robot
from off the shelf components was developed. A heightened
perception of the presence of the speaker in the remote area
was sought and for safety purposes, a control scheme with
minimal delay effects was developed.
Most of the existing telepresence robots are either
private projects or commercial ones. Each of the robots have
their own advantages and disadvantages and this study would
like to fill in the gaps left by these telepresence robots. In this
study, a robot was developed that somewhat mimics the
properties of these established telepresence robots but has
some inherent characteristics that somehow complements the
shortcomings of these previously reviewed robots. First of all
a robot made up of locally available materials was built and
thus leads to a low cost platform. The blueprint for the robot
was laid out as simple as possible so that a generalized model
can be easily duplicated. The design of the hardware was open
sourced and the sources of the components were
easily
obtained locally. The electronic controls components of the
ISBN 978-602-71380-1-8
robot come in kits that can be obtained in local stores. So in
general, a telepresence robot that can be built using modular
components was developed. In terms of robot control
software, most of the previously reviewed robots have their
own proprietary software and most of them are closed source.
In this particular telepresence robot, it was decided to use free
and open source software. This decision was based on the
notion that open source and free software can contribute to the
robot’s low price and a generalized robot can be synthesized
by just anyone with the right means. The previously reviewed
telepresence robot are considered complex in terms of control
and software and the designed robot however was not.
Moreover, the teleconferencing component of the robot came
from the freely available Skype software. Overall, a
telepresence robot designed with modular components and
free software was developed. This means that anyone with
sufficient knowledge in electronics and computer
programming can develop their own telepresence robot since
the components can be obtained readily.
To solve the problem stated thus we have done the
following:
1.Designed
and constructed
a web controlled
telepresence robot platform with the following
features
a. A tracked wheel differential drive motion
capable robot with motion control system
incorporating two sets of optical wheel encoders
for dead reckoning linear displacement
measurements and speed control using pulse
width modulation.
b. Incorporated a webcam, microphone sets and an
LCD display for two way audio and video
transmission between the robot and remote
controller.
c. Employed three sets of ultrasonic distance sensor
for the robots obstacle avoidance system.
d. Incorporated on the robot a digital compass for
direction sensing based on the 4 cardinal
directions.
e. Incorporated on the robot a 3 axis accelerometer
for inclination measurement on the three
cartesian axis.
f. Use of the Skype teleconference software for the
audio and video information transmission
between the robot and the remote controller.
g. Low cost and sourced from Commercial off the
Shelf (COTS) materials.
2.
3.
4.
51
Developed a purely local robot control using a
processing GUI whose control is ported to the
operator via TEAMVIEWER VNC.
Tested the effectiveness of the robot as a telepresence
agent based on the evaluation of the time delay
between the transmission and execution of control
commands.
Developed and optimized control algorithms for the
robots navigational control.
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
II.SYSTEM DESIGN
To reduce complexity, the telepresence robot was a passive
terminal with minimal autonomy. A little intelligence however
was incorporated through the sensors in a way wherein the
robot can have control in case sensors detect obstacles.
Mobility was limited to positioning controls, although the
prototype can be designed for much general and complex
teleoperation task, modelling, calibration and testing for the
prototype would be done for remote tele-presentation only. The
software used for teleconferencing was limited to Skype and
robot control was achieved through forwarding the remote
desktop to the controlling computers screen using Teamviewer.
The Processing programming language was adopted in
developing the robot controllers GUI.
The choice of the type of robot was based on some existing
commercial telerobot designs. The robot first and foremost
must be able to move around in the remote environment thus a
suitable mobile platform was designed. Since the robots task
was to interact with other people on the remote site as
naturally as possible, a human sized robot was designed.
The robot can be remotely controlled by a remote PC. Fig.
1 shows the overall scheme of the system. Basically, the
system consist of two computers, one top of the robot and
other will be the remote controller station. These PC’s were
both connected to the net and communicates with each other
via Skype and Team viewer VNC package.
Figure 2.The robot Graphical User Interface
A. Hardware Design
The hardware of the robot platform is shown in Fig. 3 and
Fig. 4. The remote robots skeletal system was made up of a
combination of 1x1 square aluminum bars comprising the robot
frame and a main vertical post made up of a 15/8 by 1 5/8 14
gauge slotted angle bar. The post acts as a backbone to support
the visual feedback system consisting of the 14 inch LCD
monitor and the camera system. Nuts and bolts of various sizes
were utilized as fasteners. Acrylic plastic sheets with thickness
of 2.5 mm and 3 mm was used as coverings. A tracked
industrial platform acts as the robot base. The main controller
of the robot was a microcomputer with an Intel-based
processor. This computer processes the video feed from the
robot mounted camera, processes the audio feed from the
microphone , outputs the received video information from the
controlling PC to a robot mounted LCD, outputs the received
audio signal from the controlling PC to the robot speakers,
control and monitor the communication link between the
controller and the robot, and perform motion and tactile
commands to the robot as well as process the robots sensors
relaying feedback signals to the controlling PC. The robot
moves via a differentially driven track wheels which are in turn
webcam
Pan and Tilt Assembly
Figure 1.The telerobotic system showing the telepresence robot and its
remote controller station
LCD
Video, audio data were relayed from the remote PC to the
controlling PC in the robot and vice versa. In this setup, the
computer in the remote robot runs a java controller program
whose inputs were controller commands and sensor data. The
controller commands were implemented by normal inputs
consisting of buttons, sliders and textbox in a Graphical User
Interface (GUI). Outputs were implemented as text labels and
graphical elements like gauges and arrows. This GUI were
virtually transported on the remote controller station, appearing
on the controllers monitor and thus controlled remotely. Fig. 2
shows the Graphical User Interface (GUI) in the remote
controllers end.
Speakers
Digital Compass
Robot Frame Made of Slotted Angle Bars
12v Inverter
Robot controller Laptop
Ultrasonic Sensors
Wifi to serial module
motors
12v Lead Acid Battery
tracked wheels
Figure 3.The remote Robot System
52
Control Electronics
Proceeding of International Symposium on Smart Material and Mechatronics
connected to the shaft of the motor was coupled to an optical
encoder wheel. Two sets of optical incremental wheel encoder
system were developed for the two motors using infrared
reflective sensor combined with the wheel encoder mounted on
the motor shaft. These absolute encoders determine the relative
distance travelled by the robot at a certain time difference. A
Phillips KMZ52 based electronic compass acts as a sensor for
azimuth position. The output of this sensor was used as an
input to the motor controller system.
A proportional integral control scheme was employed
for the mobile robot to control its displacement and heading.
The output signal of the PI controller will be a PWM signal
which will be the input to the motor controllers. Two sets of
PWM controllers were employed, one for the left motor and
one for the right motor. Fig.6 shows the PI controller
implementations for the robot.
The controller station consists of a laptop PC
equipped with a web cam, a microphone and speakers.
Broadband Internet access must be provided for this PC using
any of the existing services in the Philippines like DSL, 3G,
4G or Wimax .This controller PC must have Skype,
Teamviewer , Java Runtime installed. The hardware of the
controller station is shown in Fig.7.
The controller PC will be the one responsible
controlling the remote PC mounted on the robot. The two
computers must communicate via the internet using Skype and
The VNC software Teamviewer.
Figure 4. The remote robot system dimensions
controlled by an arduino microcontroller via motor controller
boards. Slave microcontrollers were employed to process
sensor data and a pan and tilt mechanism for the web camera
system. Robot navigation and tactile commands were
processed by the main microcontroller. Obstacle avoidance was
achieved by the use of three ultrasonic sensor arrays spaced
symmetrically around the body of the robot. Connectivity on
the robot site was provided through 3G internet service via a
3G wireless router installed in the remote site. The overall
block diagram of the system is described in Fig 5.
Two Nissan mt3-12 high torque DC geared motors
were employed to drive two threaded drive chains. The drive
chains are configured for a differential drive robot base. The
two motors were controlled by two 6.0 Ampere H-bridge
motor driver kits from E-gizmo Mechatronix Central. These
kits control the direction and speed of both motors. The axel
B. System Software Design
The video and audio transmission between the robot
and the controller was handled by the Skype Software. Both
the Robot PC and The Controller PC have Skype installed,
each with their corresponding Skype accounts registered.
Banking on the popularity of this software, it was assumed
that this part of the design was already been taken cared off
US-100
ultrasonic
sensors
x3
Wifi to
Serial
Module
webcam
Microphone
ISBN 978-602-71380-1-8
Digital
Compass
Intel based
Laptop
Computer
3-Axis
Accelerometers
slave mcu 1
master mcu
UART
soft serial
Wireless
adapter
Optical
Encoder
x2
12V
inverter
digital inputs
digital input
interrupt
analog inputs soft serial
LCD Screen
Buzzer
digital input digital output
Atmega328
digital output
Atmega 8l
slave mcu 2
Atmega 8l
soft serial
digital output
2 nissan
High power
geared
Motors
speakers
Figure 5. Overall block diagram of remote robot
53
6A Motor
Controllers\
x2
Tilt Servo
Motor
Pan Servo
Motor
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Figure 8. The overall robot system hardware: (a)overall robot system;(b)LCD
video system and
speakers;(c)Web camera mounted on pan
and tilt
servo mechanism; (d)Control electronics
consisting
of microcontrollers, motor controllers
and
power
conditioning circuits.
Figure 6. Two PI controller implementation. (a) for position (b) for
angular heading
delay on the internet and it was investigated through a series
of experiments.
The Graphical User interface permits the control of
the movement of the robot. Two methods of robot control
movements were allowed, manual direct control and
semiautomatic. In the manual control, 5 directional buttons
corresponding to forward, left, right, backward, stop were
used. Clicking on the desired a particular button effects the
desired movement of the robot . Eight bit values ranging from
0 to 255 can be used to implement PWM speed values to the
motors. While the robot is executing a particular robot
movement, the clicking of another movement button will stop
the robot. One has to click the button again to effect the
desired movement. Once the robot is moving, the ultrasonic
sensors are active. When an obstacle is directly in front of any
of the three ultrasonic sensors, the robot will stop its
movement and sends the “obstacle detected” robot status on
the robot GUI. One can in anytime stop the movement of the
robot by sending any commands to the robots GUI. In the
manual method, the robot implements the command and wait
strategy wherein the robot executes the sent command and
then waits for the next command. Some commands have a
definite duration in terms of execution and once the control
routine is finished, the robot stops and waits for the next
command. Examples for this are the pan and tilt commands for
the camera.
In the semiautomatic mode of control, sliders and
buttons are implemented. Two possible movements are
implemented in this method:
Figure 7. Controller hardware setup
since the software has already been proven in the internet in
terms of affectivity and efficiency. Albeit to say, the system
therefore was highly dependent on Skype in terms of
reliability of connection and transmission delays of the video
and audio information.
On the remote robot computer, a Graphical User
Interface was implemented using the java based Processing
language. This interface was actually a TCP client which
connects with the wifi server module of the robot. These client
sends control signals to the server and at the same time
receives feedback data from the server. This GUI was then
virtually transported on the controller monitor through the
Teamviewer software. That is, the remote controller screen
has a display of the remote robots PC desktop.Through
Teamviewer, the remote controller PC can control the GUI on
the robot PC. The effectivity of sensing and executing control
commands via this setup was dependent on the transmission
54
Proceeding of International Symposium on Smart Material and Mechatronics
1.
distance traversal
2.
angular heading seeking
There were three arduino microcontroller board
utilized in the implementation of the control of the robot. One
Atmega 328 based master microcontroller board acts as the
master controller and two Atmega8l based microcontroller
board acts as slaves. One of the slave module acts as the
controller for the pan and tilt servo motor assembly. This slave
microcontroller waits for serial command coming from the
master microcontroller. These commands are the pan and tilt
angle for the servos. The other slave microcontroller controls
the three US -100 ultrasonic sensors constantly getting its
distance reading. When the distance of the obstacle on any of
the three sensors was below the threshold level, this
microcontroller sends a low level signal on an output pin. This
pin was connected to an interrupt pin of the master
microcontroller. The master microcontroller therefore had an
interrupt whenever a low level signal was present on this pin.
The master microcontroller performs numerous task.
These tasks were:
1. Communication with the WIFI serial module.
2. Controlling the two 6 A h-bridge motor controller.
3. Receives encoder counts from the two optical
encoders.
4. Receives serial data from the digital compass.
5. Reads the output of the three axis accelerometer.
6. Executes the overall control loop for the control of the
robot.
Fig. 9 shows the overall flow chart for the main controller
loop of the master microcontroller firmware.
In the distance traversal, the user uses the distance slider to
enter the required distance of travel. One then clicks the
navigate button and the robot moves forward to the required
travel distance. This movement is implemented using a
Proportional Integral controller algorithm where the inputs are
the actual travel distance and desired travel distance. The
output of the controller will be the PWM signals on the two
motors driving the track wheels. The actual travelled distance
was measured using an optical encoder.
In the angular heading seeking movement, one uses
the heading slider to choose the desired heading. The heading
angle correspond to the earth magnetic pole directions with 0
degrees as north, 90 degrees as east, 180 degrees as south and
270 degrees as west. To lock to the desired heading, one clicks
the FIND button and the robot turns towards the desired angle.
The robot turns towards the desired angle using a PI controller
where the inputs are the desired angle and the actual heading
angle. The output of this controller will then be the PWM
signals that drives the robot motors. The implementation of
the PI controller in this system is complemented with a logical
system wherein the choice of turning direction is dependent on
the amount of angle to be traversed. The system is
programmed wherein the turning angle to be traversed will be
the one which entails lesser turning distance. For example if
the present heading angle is 90 degrees and the desired angle
is 180 degrees, there are two possible scenarios for
implementing this heading seeking:
1.
Clockwise at an angular displacement of 90 degrees
2.
Counterclockwise at angular displacement of 270
degrees (360-90).
ISBN 978-602-71380-1-8
Start
For this example, the robot selects the clockwise movement
for it entails lesser angular displacement to be covered.
The actual heading angle was measured by a digital
compass sensor whose output is a serial data sent to the robot
controller. The actual desired angular headings of the robot are
displayed on the GUI using a circular gauge with two dials.
The red dial is for the actual heading and the blue dial is the
desired heading. Whenever the robot moves, this actual
heading dials location is updated. One can also manually
update the actual angular position by clicking the
READCOMPASS button. The Robots Pan and Tilt Camera
can be controlled via the Tilt Angle and Pan Angle sliders in
the GUI. The Pan angle slider value corresponds to the amount
of angular degrees the camera pans. These angular values can
range from 0 to 180 degrees. The tilt angle slider values
correspond to the amount of degrees the camera tilts. These
angular values can range from 0 to 180 degrees. By clicking
the neutral button, one can reset the camera to point to a
default viewing position which is the usual orientation of a
webcam. Neutral position corresponds to a pan angle of 85
degrees and a tilt angle of 90 degrees. An overall GUI for the
robot controller was employed on the robot. This GUI serves
as a frontend to a telnet session between the robot PC and The
Wi-Fi to serial converter server. The Wi-Fi to serial converter
System
Initialization
Routine
Wait for Serial
Commands
Command
Recieved?
no
yes
Process Commands
Figure 9. Flowchart for the main controller loop
55
Proceeding of International Symposium on Smart Material and Mechatronics
The initial step in the main microcontroller code was the
initialization routine. Here every piece of hardware interface
was initialized, I/O pins data direction are set depending on
the peripheral connected to it. Figure 10. is the flowchart for
the initialization routine. The initialization is implemented on
the setup( ) function of the main microcontroller code.
ISBN 978-602-71380-1-8
III.TESTING AND EVALUATION
The main goal of this study was to develop a basic
telepresence robot whose affectivity was dependent on how
accurately it performs the tasks assigned by the controller. The
study implements the objective method of evaluating the
performance of the telerobot through a series of tasks that it
has to complete at the shortest time possible, thus task
completion time and reaction time to a remote stimuli were the
metrics used. Task completion time was measured and
averaged while the reaction time was measured by getting the
time difference between the completion time in the local
station and the time of completion as perceived in the remote
station. Two tasks have been designed, one is the traversal of a
square path and the other one is the traversal of a straight path.
The traversal of the square path employs the shared
continuous means control. In this method a “command and
wait strategy” was implemented by which the robot was sent
commands one at a time. One command must be finished first
before another command was sent. The robot on the other hand
monitors obstacles along the path, and once an obstacle was
detected, it stops the present maneuver of the robot. The next
step for the robot was then to wait for the next command and
execute it. The path was traversed 10 times and the time of
completion was recorded through Wireshark, an open source
network sniffing software. The perceived time of completion
was also measured on the controlling computer through
Wireshark by looking at the timestamp on the TCP
transmission of the last command coming from the remote
machine. The testing area would be a tiled surface with an
outline drawing of the path to be traversed. Two paths have
been defined for testing, one is a square path with a side
dimension of 1.2 m and the other is a straight line with a
length of 1.6m.
In the experiment, the robot performs its usual startup
routine whereby it connects to the controller station via
internet. Skype and Teamviewer were run on both the robot
and the controlling station. The robots controller GUI was
then forwarded to the controller screen so that it can remotely
control the robot. Skype provides the visual feedback to the
controller PC so that it can effectively maneuver the robot.
Once connection was established, the robot then traverses the
paths. While the robot was traversing the paths, Wireshark
monitors the wireless data packets leaving and entering the
robot PC. These data packets are then sent remotely to the
controller stations PC for monitoring. Wireshark in the
controller station also monitor its wireless data packets
particularly the timestamps on each TCP packet. Through the
timestamps in the TCP packets, the time of completion and
time delays was calculated and recorded. In calculating the
time delay, the tasks starting time and ending time were
extracted from the TCP stream trace and packet display list of
the TCP transmission between the robot PC and the wifi
module. Fig. 11. shows software modules and the data flow
during the experiments.
Figure 10. . Flowchart for the initialization routine
The bulk of the main loop of the master controller’s firmware
was a blocking wait loop. The program waits for serial data
coming from the wifi serial module. The data coming from the
wifi module were actually commands coming from the main
graphical user interface. As mentioned previously, the
graphical user interface processes commands entered by the
user as relayed through the internet via Teamviewer. The
commands received by the GUI were then encapsulated into a
TCP packet which was then relayed to the serial wifi module.
The serial wifi module then decapsulates the commands from
the TCP packet into a stream of serial data. These serial data
will now be the serial commands the main controller loop
processes. Commands are formatted as a series of 4 bytes The
wait loop in the main loop waits for this and upon reception,
the first byte then was used as an argument in a switch case
block in the code. Different commands then are executed
based on the decoded first byte. Likewise if a particular
command needs a response from the robot, the said response
was sent to the wifi serial module as a serial character stream.
This serial stream was then converted to a TCP packet which
will then be transmitted to the GUI client on the controlling
PC. An example of this was the string “ready” which was
transmitted every time a command was successfully executed
by the robot.
56
Proceeding of International Symposium on Smart Material and Mechatronics
teamviewer control
signals
TCP and UDP
ISBN 978-602-71380-1-8
Internet
Local Area Network
Robot PC
Controller PC
Putty is used to signal the
robot that the last manuever
was observed to have been
accomplished
Teamviewer
Teamviewer
tcp packets
Putty
Wifi-Serial
Module
acts as server
Teamviewer control
commands passed on
the Java GUI
tcp packets
Java GUI acts
as client
Wireshark monitors tcp
packet via the PCs wireless
network adapter
WinPCap monitors tcp packet
and sends them to Wireshark via
local network
Wireshark
WinPCap
Figure 11. Software modules and the data flow during the experiments
resulted to an average time delay of execution of less than 2
ms in both tests. This test was done not taking into account the
quality of service for the internet connection. Overall, the
design of the robot leads to a low cost system; as such
materials were coming from off the shelf items.
A Series of time of task completion test and time of
execution delay was conducted through the use of Wireshark
network sniffing software. The researcher suggests therefore
that more test be done on this aspects taking into account the
different quality of service at different times of the day. As for
the robot control, the researcher recommends further
investigation by implementing control software that is not
dependent on the commercial Teamviewer software. A pure
client and server program for robot control is therefore
recommended for further study.
As for mobile robot mobility, wheel slippage has
been observed due to the nature of the wheels employed on the
project. This resulted to severe errors on the optical encoder
system which limits its effectivity on smaller distances. Due to
this, the researchers recommend the use of other means of
robot locomotion such as wheels. Odometry using optical
encoders deemed erroneous when large distances was
traversed, thus the use of MicroElectroMechanical Modules
(MEMS) such as accelerometers as position sensors are
recommended.
IV. CONCLUSIONS AND RECOMMENDATIONS
The goal of the study was to develop a human sized
tracked mobile robot suitable for telepresence applications. A
mobile robot equipped with sensors and tele-presentation
capabilities were developed. The robot was built from off the
shelf components so that the overall cost of the system is as
minimal as possible. The robot design process was
implemented through the manufacture and testing of
individual components. Core mobile robot movement and
navigational programs were developed and tested. The
effectiveness of the mobile robot as a test bed for telepresentation were evaluated and analyzed by way of its real
time response and time delay effects of the network. Iterative
process of software creation, testing and debugging were done
to come up with an optimized code suitable for both mobility
and remote robot control. The physical hardware components
of the robot were developed first through an iterative process
of mixing and matching. Once the hardware components were
developed, an iterative process of software development was
implemented in order to fit the desired function of the robot.
The mobility of the robot has been proven effective through a
series of mobility test scenario where the robot has been
controlled. The use of Skype has also been proven effective in
sending the audio and video information between the robot
and its controller. The use of Teamviewer as the controller
medium for control signals deemed effective provided a good
quality of service is expected from the internet provider. The
use of Wireshark deemed effective in capturing the time of
execution of task as well as time delays with accurate
resolutions of up to 10 ms. A test was conducted which
REFERENCES
Cobs, “The human role in
telerobotics”. Springer Tracts In Advanced Telerobotics,
2007, vol 31, pp. 11-22.
[1] R. Aracil, M. Buss, and S.
57
Proceeding of International Symposium on Smart Material and Mechatronics
[2]
[3]
[4]
[5].
[6]
[7]
[8].
[9].
[10]
[11]
[12].
[13].
J. Borestine, and H.R. Everet, Where am I, Sensors and
Methods for Robot Positioning. Michigan:UM Press, 1996.
T. Bräunl, Embedded Robotics. Berlin:Spinger Science, 2006.
M. Buss, and G. Schmidt, “Control problems in multimodal
telepresence systems,” Advanced
Control Highlights
Europe Control Conference, 1991, pp. 65–101.
Caroll, T. “Then and now:telepresence,” Servo, Vol 9, pp. 7680, March 2011.
J. FENG, Introductory Robotics. NewYork:Prentice Hall,
1194.
D. Geer, “Remote presence robots put
physicians within
reach,” Servo,vol 8, pp. 76-88,June 2009.
G. McComb,Robot builders bonanza. New York:Mc Graw
Hill, 2002.
G. Niemeyer, C. Preusche, and G. Hirzinger, “Telerobotics,”
Springer Handbook of Robotics, 2008, pp.741-754.
W. Schloerb,Handbook of Robotics.New York:Prentice Hall,
2002.
T.B. Sheridan, Telerobotics, Automation and Human
Supervisory Control. New York: The MIT Press,1992.
R. Siegwart, and I.R. Nourbakhsh,
Introduction to
Autonomous Mobile Robots. London : MIT Press,2007.
D. Song, K. Goldberg, and N.K.Chong, “Networked
Telerobotics,” Springer Handbook of
Robotics, 2008,pp.
759-769.
58
ISBN 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Intelligent Machine Vision for Automated Fence
Intruder Detection Using Self-organizing Map
Veldin A. Talorete, Jr.
Sherwin A. Guirnaldo
MSU–Iligan Institute of Technology
Iligan City, Philippines
[email protected]
MSU–Main Campus
Marawi City, Philippines
[email protected] gmail.com
building but not really effective in preventing break-ins. The
recorded events can be firm evidence as enforcement in
catching and prosecuting the burglar. The Hidden Cameras are
an improvement, but still no power to stop a crime. Observing
the visual display or video stream of surveillance cameras (i.e.
Close-Circuit Television (CCTV) cameras) can be time
consuming because the activity captured by each of the
installed camera is shown in one monitor at the same time. If
there are 9 cameras installed in a building, the person who is
assigned to observe the situation surely have difficulties in
watching the video stream at the same time, which also makes
the system ineffective in preventing burglars. This type of
system will be much vulnerable from intruders when the
assigned person falls asleep.
Accordingly, the study focuses on the development of an
intelligent machine vision fence intruder detection using Selforganizing map, aimed to identify the intrusion level within
the set perimeter of a building or a residence, promising a
reliable high detection rate and a low false alarm rate. It also
aimed to make the job of the assigned person much easier in
supervising the system’s visual display by providing simple
but accurate visual display based on the intruder position. The
system will trigger the alarm in case the assigned person fall
asleep
Abstract – This paper presents an intelligent machine vision for
automated fence intruder detection. A series of still captured
images that contain fence events using Internet Protocol cameras
was used as input data to the system. Two classifiers were used;
the first is to classify human posture and the second one will
classify intruder location. The system classifiers were
implemented
using
Self-Organizing
Map
after
the
implementation of several image segmentation processes. The
human posture classifier is in charge of classifying the detected
subject’s posture patterns from subject’s silhouette. Moreover,
the Intruder Localization Classifier is in charge of classifying the
detected pattern’s location classifier will estimate the location of
the intruder with respect to the fence using geometric feature
from images as inputs. The system is capable of activating the
alarm, display the actual image and depict the location of the
intruder when an intruder is detected. In detecting intruder
posture, the system’s success rate of 88%. Overall system
accuracy for day-time intruder localization is 83% and an
accuracy of 88% for night-time intruder localization.
Index Terms - Intelligent Machine Vision, Self-Organizing Map, and
Image Segmentation, Classifier
I. INTRODUCTION
Anyone wants to have a safe home, properties or offices.
The Self-Organizing Map Classifier for Vision Based Auto
Intruder Detection embracing the method of Artificial Neural
Network (ANN) will offer a remarkable result in defending the
residence, stuffs or properties from thieves, intruders or
burglars. Crime is increasing day by day, thus the demand of
reliable, fast and accurate security system is rapidly increasing.
It was stated that whether or not a person is a victim of crime,
the mere thought of an unwanted visitor lurking around his
house can make him cringe. Many home security systems have
proven effective in preventing home burglaries. It was grouped
differently depending on the usage of the system and on the
technical features it offers. There are magnetic, electric circuit
and motion detecting systems, infrared systems and wireless
security systems. There are systems with security cameras,
electric fences and guard dog perimeters. Some other systems
can communicate via household electric wires called the X10
security systems and those operated remotely via the Internet.
Though the mention security system offer a
remarkable result but, none of them has an ability to identify
and analyze the current situation. It simply gives a warning
when there is a disturbance on an emitted signal or on an
established circuit. Usually this type of security system has
high percentage of error or high false alarm rate.
Establishments like banks, airports, casinos, convenience
stores, and military installation use surveillance cameras to
monitor and record activities both inside and outside the
II. SYSTEM ARCHITECTURE
The system classifies the acquired still images from data
acquisition unit as Not Intruder, Potential intruder, Intruder
Level-1 (L-1), or Intruder Level-2, using two system’s
classifiers: Human Pose Intruder Classifier and Intruder
Localization Classifier after applying several image
segmentation. The system will trigger the alarm and display
the actual location and position of the intruder.
Figure 1: General system architecture
59
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
system’s classification technique, it is classified as may or may
not be an intruder. Similarly, Level one (L-1) intruder and
Level two (L-2) intruder is an object classified as human and
specifically located in a certain point of a fence, refer to Figure
5.
The designed details of Classifiers development is
described in Figure 2.
a
c
b
Figure 5: Relative intruder position (a) Potential
Intruder, (b) L-1, (c) L-2
Unlike human recognition ability, computer recognition
ability is pretty much limited to numbers. With this, the
developed system viewed the input data in a form of still
images, as a series of random numbers that represented each
and every color, lines and shapes of the picture. Part of the
system goal was to extract the human and the fence from the
given data and make it available for further processing. Static
method of object extraction was used to extract the fence from
input image. Fence was modeled using four line segments
which depicted four corner points. Based from four corner
points the system then calculated the image pixel at position
(xi ,yi) that was known to be a part of fence model using the
equation of a line:
Figure 2: Classifier development block diagram
Human data samples were taken randomly with varying
distance based from the installed effective position of Internet
Protocol Camera (IP Cam). Intruder Clothing color was also
considered during data gathering. This was to anticipate the
possible error caused by dress color. The study included only
four clothing color: Blue, Green, Red, and Black. 160 images
of human data samples were used experimentally per testing
distance in both day time and night time. Sample images of
human intruder are shown in Figure 3 and Figure 4.
( )
(
.
/
)
(1)
Whereslopem = (y2i – y1i) / (x2i – x1i)and the value of pixel ycoordinate ( )was defined by the relation:
( )
using constant interval of 1.
Foreground extractor was to eliminate unwanted data and
to extract the target object from the given image. Most likely
Human intruder fell to the category of a moving object; it
constantly changes its position throughout the time. Given a
single video frame as input data, the value of image pixel was
considered at position (
)takenfor a certain period of time.
( )and treated as a random
This value was referred to as
process of variable
( )
(2)
The current pixel value was modeled as a mixture of K
Gaussian distribution. The weight of this value was
determined by a distribution, π
Figure 3: Day-time human sample
Figure 4: Night-time human sample
( )
All data or images that could possibly appear in testing area
were all included during the data gathering of non-human data
sample. It included different types of leaves and different types
of animals such as dog, cat, horse, elephant, bird, butterfly, pig
and etc. Non-human data come in different sizes, different
angles, different shapes, and different positions. Basic
geometrical shapes such as: circle, diamond, square, rectangle,
triangle, ellipse, stars, and etc. were also included as nonhuman data sample.
Intruder localization data was subdivided into three: The
Potential intruder data, The Level one intruder data, and The
Level two intruder data. Potential intruder is an object
identified to have human properties and with the use of
(
)
(
)
(
)
∑
( )
(
) (3)
where πi is an estimated weight of the ith Gaussian, and N is
the evaluation of a standard Gaussian with mean µi, t and
covariance matrix ∑i, t :
N(
60
)=(
(
)
)
(
)
(
)
(4)
Proceeding of International Symposium on Smart Material and Mechatronics
Based on the calculated Gaussian value, individual pixel of
an input data was being group into background pixel or
foreground pixel. The background subtraction process was
implemented.
.
Below was the procedural approach used in foreground
detection:
For each pixel in a video frame:
1. For every N values taken from the pixel
2. Find the K Gaussians and weights that best fit to
sample N values using Expectation Maximization
(EM) algorithm
3. Find the Gaussian with the largest weight and
store its mean as the value of the background
image for that pixel.
4. Subtract the background image from the frame.
ISBN 978-602-71380-1-8
that have a pixel value of 0, surrounded by pixels that have a
pixel value of 1. Unwanted blob was identified by counting
the number of ones within the detected blob or by considering
the minimum blob area. Human blob had greater numbers of
ones or greater area compared to unwanted blob which made it
easier to identify and eventually removed from target object.
Object contour extraction technique involved data conversion
or transformation. The method was used to extract the object
contour in preparation to the object classification. The object’s
outline details were the only considered data, instead of using
the entire object blob as object representation, shown in Figure
7(b). Furthermore, silhouettes or object contour was the
refined object representation taken from object blob.
Object Silhouettes in Figure 7 (b) is defined by pixel
location: P(1,2), P(1,3), P(1,4), P(1,5), P(2,3), P(2,4), P(3,2),
P(3,5), P(4,2), P(4,5), P(5,2) P(5,5), P(6,2), P(6,5), P(7,3)
and P(7,4).
In the resulting difference image, any value larger than
three standard deviations from the mean was considered
foreground, and any other value was considered background.
Feature vector extraction and formulation was the final
stage of eliminating unwanted data, refining and finalizing the
target object using several image segmentation techniques. The
technique used in this stage is depicted in Figure 6.
a
b
Figure 7 Object Silhouettes extraction
Feature Vector was the data structure that contained all the
unique details of the target object, needed in object
classification. Feature Vector dimension was formulated using
the equation 5 and equation 6.
(
)
(5)
(6)
(
)
Where:
To human Pose
Intruder Classifier
Figure 6: Feature vector extractions
For simplification, the input data was converted into binary
image, 0 and 1. Zero value was used to represent the unwanted
pixel and 1 for a pixel of interest that composed the structure
of the target objects. Blob detection was applied in input
images to detect points or region that differ in properties like
brightness or color compared to the surrounding. In this study,
the value of individual pixels and the minimum blob area,
which was 250 unit pixels were the properties used as a basis
for region detection. Blob area analyses were realized by
counting the number of touching pixels or pixels with adjacent
side and have a pixel value of 1. Image re-composition aimed
to recreate a whole new image representation concerning only
the detected region of interest which generated form blob
detection stage. At this point the processed image still
contained the unwanted data and still considered as noisy
image but much refined compared to the data produced after
applying region filtration. The image morphology correction
technique used in the project development was a collection of
non-linear operations related to the shape of target image. The
goal of this technique was to correct the target object
imperfections or distorted pixels. Even though, input image
already undergone pixel correction, still there were parts of the
image that needed for further refinements that were not meet
in pass pixel distortion correction technique. Filling of holes
was one of the concerns. Image hole was referred to pixels
Furthermore, the Feature Vector dimension can be reduced
by half of its size, using the distance formula from the equation
of a line (Equation 7) to calculate the unique attribute of the
object.
(7)
*
+, where aiwas the
individual value represented the object details, shown in
Figure 8 (f) (page 47) the red doted color pixels.
*
+, where bi was the
individual value taken from the point of reference, refer to
Figure 3.18 (f) the blue doted color pixels.
The number of raw a from Feature Vector dimension
(
) was taken from the total number of object details
from Equation 7.
Figure 8, illustrated the logical representation of
object details reduction, from originally acquired image to
process or converted image.
Three classifiers were used in the designed project
development such as Human Pose Intruder Classifier, Day-time
Intruder Localization Classifier and Night-time Intruder
Localization Classifier.
61
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8

3,800 total number of Day-time and Night-time
localization data sample. 1,400 in Level-1 and 1,400
in Level-2 data sample. 1000 total number of Daytime and Night-time localization potential intruder
data sample
The accuracy of newly created classifiers was tested using
the data specified below:
For Human Pose Intruder Classifier:
5. 1000 samples for intruder human pose.
6. 1000 samples for non – human
For Intruder Localization Classifier:
Both daytime and nighttime
7. 100 samples per meter in level-1 intrusion, total
of 700 samples
8. 100 samples per meter in level-1 intrusion, total
of 700 samples
9. 100 samples per meter in level-1 intrusion, total
of 700 samples
To test Intruder Localization Classifier accuracy,
total of 2100 data samples were used. In Human Pose
Intruder Classifier, 2000 total of samples were used to test its
accuracy.
III TESTING RESULTS
Considering the constant change of light intensity and the
image background stability, eliminating the unwanted data was
never been easier. During noise reduction, the system could
possibly hurt the desired object pixels The object detected was
classified as human intruder only if it was positively identified
as human and its localization belong to intruder localization.
Furthermore, classifying the detected object was possible using
only a single classifier, merging the object details as well as the
object localization as object feature vector, but the time
required to train the network was very expensive. Training
period to develop a single classifier merging the object details
and object localization did reached to approximately one month
using 10 thousand epoch of network training. That was
exhausting and time consuming network training. The
downside of a single classifier was when there was an update
for intruder sample that required re-training of network. The
update would take another one month approximately. In
contrast with the single classifier, by using two classifiers
(HPIC and ILC) the network training period was effectively
reduced to approximately 2.5 hours. Approximately 2 hours for
HPI training using 10 thousand network training epochs.
Approximately 0.5 hours for ILC training using 30 thousand
training epochs.
a
d
e
c
f
Figure 8: Object details extraction
The ability or accuracy of the Map to classify was
adjusted experimentally by manipulating the parameter;
number of epochs which defines the number of iterations used
to train the system classifier; size of SOM defined the finite
number of different classification type ( )of an input data;
and number of data sample.
Furthermore, the focused of SOM was to convert high
dimensional presentation into two dimensional presentations
, ( )
( )- by grouping the input data based on its
likelihood. This understanding conveyed that a map with a
dimension of 20 X 20 will have 400 different types of data
classifications, arranged according to its likelihood.
( )
( ) ( )
(8)
The number of data sample can affect directly to the
performance of the Classifiers, as mentioned above that
classification variation of a classifier was modeled in the data
sample details. In the field of Artificial Neural Network
having a greater number of data samples used in network
training depicted a better performance.
The development of Human Pose Intruder Classifier
(HPIC) which designed to classify human pose in both
daytime and nighttime used the following parameters during
Map training:
 10,000 training epochs
 Map dimension of 10 x 10
 4,480 total no. of Human pose data sample
 3200 total no. of Non-human data sample
Intruder Localization Classifier (ILC) was use to classify
the input data that was labeled as positive human intruder pose
by HPIC according to its level of intrusion such as Potential
Intruder (PI), Level-1 (L-1) Intruder
and Level-2 (L-2)
Intruder.
There were two types of ILCs: Day-time ILCs and Nighttime ILCs were developed separately using separate data
samples. This was to address changing properties of input data
during daytime and nighttime. The development of 2 ILC(s)
used the following parameters during Map training:
 30,000 training epochs
 Map dimension of 10 x 10
a
b
Figure 9: Sample data hit (a) Not Human Pose (b) Human
Pose
62
Proceeding of International Symposium on Smart Material and Mechatronics
The data that were used in mapping were the same
data that were used during map development. The result below
was depicted according to its level as well as to its horizontal
distance from the camera.
The intruder hit sample result depicted a clear boundary
between three intruder level classifications. As expected, level
1 intruder classification would be closer to potential intruder
level and level 2 intruders. Level 2 intruders and potential
intruder level as shown in Figure 10 and 11, was isolated by a
clear boundary distinction. The result implies that a potential
intruder classification could hardly be identified as intruder
level 2 and vice versa, but it could possibly be classified as
intruder level 1. However, there were few occurrences of
classification fluctuations as shown in Figure 10 (c), this was
usually happened when the noise reduction process fails to
preserved some important object details.
The first experiment was done using 1000 image for
both human and not human data samples. Human sample was
acquired using the same acquisition device used in actual
experiments and all of the representations contain the pose of
human intruder in different level and different distance. The
data used to represent not human pose were taken from
internet. The data were composed of different basic
geometrical shapes and different animal shapes that were
already in masked format. Human Pose classifier that was
trained using 10 thousand epochs, painted a hit rate of 0.884
and a false rate of 0.116. Hit rate value (α) depict a result of
0.8942857 and false hit rate of 0.105714286 (β) for
Introduction Localization Classifier using 30 thousand epoch.
Hit rate has a value ranges between 0 – 1 (0<=α<=1). Value
close to 1 means high hit ratings, 0.5 values was the neutral hit
rating of the system, implies that system has accuracy of 50
percent (50%). Figure 13 presented the sample images that
were false identified as not human and Figure 14 were the
sample image of falsely identified as human from a set of not
human data samples.
Figure 12: ID as not human
b
a
c
Figure 10: Night-time (a) L-1 (b) L-2 (c) Potential Intruder
Figure 13: ID as human
Similar to Night-time hit sample results, level 2 intruder
classification was expected to be the neighbor of intruder level
1 classification and intruder level 1 classification to potential
intruder classification. Level 2 intruder could possibly be
classified as intruder level 1 or vice versa. Additionally,
intruder level rarely could be classified as potential level
intruder, based from Figure 11, hit sample result.
a
ISBN 978-602-71380-1-8
Figure 12 and Figure 13 were the result sample during
simulated experiment using 10 x 10 map dimension that
trained 10 thousand times (10K epoch). To aid the deficiency
of Human Pose Intruder Classifier, Intruder Localization
Classifier was used to confirm the position of the identified
human. In connection with this, mostly non – human objects
have the position located below or distance from the fence and
the human object usually located near or climbing in the fence.
In intruder localization classification, potential identifier
and level one intruder were the false classification possibly
occurs, because of the location which was near to identical.
The reason stated was also true to Level – 1 Intruders and
Level – 2 Intruders. Classifier test result summary using 30
thousand epoch and 10x10 map dimension was shown in
Table 1 using Human Pose Classifier, Table 2 for Intruder
Localization Classifier – Night-time and Table 3 for Intruder
Localization Classifier – Day-time.
b
TABLE 1: HUMAN POSE INTRUDER CLASSIFIER
Result
Human samples
Not Human samples
1000
1000
c
Figure 11: Day-time (a) L-1 (b) L-2 (c) Potential Intruder
63
Proceeding of International Symposium on Smart Material and Mechatronics
No. of
Epoch
Positive
ID
Negative
ID
3K
5K
10K
1519
1525
1768
481
475
232
Hit
Rate
(α)
0.7595
0.7625
0.884
False
Rate
(β)
0.2405
0.2375
0.116
device. It was because; the acquisition device could not have a
clear view or good image representation of the given potential
intruder level, due to the physical setup of the acquisition
device, as shown in Figure 6. The system can only have a
clear view to the intruder when the intruder starts to climb on
the fence, and nearly the situation could be possibly identified
as Level – 1 Intrusion.
Red line was the target system performance, blue curve was
the approximation of the actual performance of the system,
and green curve was the average system performance.
Accuracy
(%)
75.95
76.25
88.4
TABLE 2: INTRUDER LOCALIZATION CLASSIFIER–NIGHTTIME
Result
TABLE 3: USING INTRUDER LOCALIZATION CLASSIFIER –
Used 700 data
sample per Level
DAY-TIME
No. of
Epoch
Accur
acy
Used 700 data sample per Level (β)
(%)
Epoch
10K Positive
1905Negative
195 Hit Rate
0.9071 False0.09 Accuracy
90.71
No.30K
10K
30K
Positive
ID
ID 1878
1828
1806
Negative
Hit
Result
ID
Rate (α)
ISBN 978-602-71380-1-8
False
Rate
(α)
Rate (β)
ID 222
0.8942
0.10 (%)
89.42
272
0.8704 0.1295 87.0476
294
0.86
0.14
86
The result in Table 1 depicted map accuracy with respect to
the number of training epochs. The experiment evidently
showed the result different from the three tested training
epochs: 3k, 5k and 10k. The experiment result met the
expectation that the greater the number of training epochs the
better the classifier performance will be. For Table 3and Table
2 the classifier was tested and trained using 10k training
epochs and 30k training epochs. Based from the result,
expectation was a little bit of. The accuracy of system
classifier using 30K epochs was lower by one unit compared
to the classifier with 10k training epochs, but the result does
not show any significant difference in terms of classifier
performance. The only difference was the training time used
to develop the classifier, it took 25.55 minutes to train a
classifier using 30k training epochs. It was 3 fold higher than
the training time of a classifier using 10k training epochs.
The second experiment was done around 10:00 am to 2:00
pm. Because of Foreground Detection and Background
Subtraction technique, the time of experiment conducted was
not a much of a concerned of the said experiment, as long as
the sun still up high, then still it’s good to go. Acquisition
device was installed approximately 2.35 meter, measured from
the ground and aiming to test the system accuracy using the
distance of 2 meters, 4.5 meters, 5.5 meters, 7 meters, 8.5
meters, and 10.5 meters as the horizontal distance of the
intruder from the camera. Recall, that the horizontal distance
of the sample data (intruder) used to train both network
(Human Pose Intruder and Intruder Localization Classifier)
were taken using the specified distance from the camera: 3 to
2 meters, 4 meter, 6 meters, 7 meters, 9 meters, 10 meters, and
12 meters for Day-time data sample. The distance of used in
the second experiment was a little bit off, this was done to test
if the system still able to classify the object given the distance
odd.
The experiment was using 10 intruder samples per meters as
shown below. Given a single intruder the accuracy of the
system was tested to detect the non-intruder level, the
potential intruder level, the level-1 intruder level and the level2 intruder level. As a result, the accuracy of the system was a
little bit low for the intruders which have a horizontal
distance: 2 meters and 4.5 meters, closer to the acquisition
Figure 14: Potential Intruder Detection in Experiment 2
Figure 15: Level-1 Intruder Detection in Experiment 2
Figure 16: Level-2 Intruder Detection in Experiment 2
The result of experiment 2 shows that the system
performance was much more accurate if the sample intruder
was far from the acquisition device (the further the sample, the
accurate the system can get), given that the horizontal distance
of the sample intruder does not exist to 12 meters as the set
distance limit. The average detection rate of intruder level 1
was low compare to potential intruder detection rate and
intruder level 2 intruder rate; this was due to the system
turnaround time. The dilemma usually happened when an
intruder climb a little bit faster. During second experiment, the
system was also tested using non-intruder samples. The
sample had just walk along the fence, sometime extended their
hand to hold the side of the fence. The experiment was done
using ten (10) samples passing back and forth from 1 meter
64
Proceeding of International Symposium on Smart Material and Mechatronics
distance from the camera up to 12 meter distance. As a result,
the detector system did not positively identify the given
sample as fence intruder, probably because of the system filter
that filtered the location of non-intruder samples. The system
was also tested using multiple intruders designated in random
horizontal distance from the camera and climbed concurrently.
As a result, the system still detected the intruder, if not all of
them, at least the system detected the first climber or the
intruders whose clearly emphasized from the perception of the
acquisition device.
The third experiment was done around 10:00 pm to 1:00 am
without the moon’s presence. With our without the presence
of the moon during testing, the system can still be operational.
Foreground Detection and Background subtraction technique
was very much suitable in slight changes of image background
illumination. The image background light, shown in Figure
4.21 was the light emitted by the acquisition device itself. The
same with the first experiment acquisition device was installed
or was setup 2.35 meters from the ground. Intruder horizontal
distances used in third experiment were measure using the
designated distances from the acquisition device: 2.5 meters, 5
meters, 7 meters, 9 meters, 10 meters and 11 meters. The
designated distances were slightly different from the sample
data used to train the network. The same reason with second
experiment, the designated distances omitted purposely to test
the classification accuracy of the system, which was one of the
reason why SOM technique was developed.
This third experiment tested the system accuracy by
providing 10 intruders sample per designated distances. Each
of the intruders was identified on its potential level, on its
level-one position and on its level-two position, with respect
to the fence of interest. The same with experiment number one
and experiment number two, the size of the map used was
10x10 that was train 10 thousand times for Human Pose
Intruder Classifier and 30 thousand times for Intruder
Localization Classifier.
ISBN 978-602-71380-1-8
Figure19: Intruder L-2 Detection Result of Experiment 3
The presented intruder sample in 6 different designated
positions in both experiment 2 and experiment 3 were all
detected by the system. Some Levels of intrusion were not
identified or did not trigger the classifier due to the processing
time or period required to detect or classify the intruder.
The system classified an intruder and gave a visual display
approximately in less than 3 seconds. For intruder
classification alone, the system classified the acquired image
in less than 1.5 second. But since the system used wireless
data transmission from acquisition unit to image classification
unit and to monitoring unit, the time to complete single image
classification reaches 3 seconds approximately, in both
experiments 2 and 3.
However, the system accuracy in both experiment 2 and 3
was altered when the intruder or intruders climb faster than 1
second, since the time required in classifying single image was
less than 1.5 seconds. Furthermore, when the intruder climbs
the fence in less than 1 second, the system can still detect or
can still classify the intruder in Level – 1 Intrusion category,
given that the system did not detect the intruder in potential
level position. However, if the said intruder was detected in
potential level position, then there will be a big possibility that
the intruder could not be detected anymore in Level-1
intrusion, but can still be detected in level 2 position.
III. CONCLUSIONS
In spite of all the factors that affected the system’s accuracy
such as: sudden change of light illumination, movement of
undesirable object, noises, and the speed of the intruders; the
developed detector system had successfully detected the
presented intruder samples that were located not beyond 12
meters horizontal distance from the location of the acquisition
device. With the use of classifiers or identifiers developed
using Self-Organizing Map, the detector system was able to
identify the pose of the extracted object as well as the current
position of the object with respect to the fence of interest. The
developed intruder detector system was also able to provide a
simple yet accurate visual presentation of the detected
intruders and triggered the system alarm when the image was
positive from intruder(s).
Figure 17: Potential Intruder Detection Result in
Experiment 3
References
[1] T. Kohonen, “The Self-Organizing Map”, Institute of
Electrical and Elictronics Engineers, Vol. 78, No. 9, 1990
[2] K. Bapat and S. Mantri, “Neural Network Based Face
Recognition Using Matlab”, International Journal of Soft
Computing and Engineering,Vol. 1, No. 1, Pp 6-9, 2011.
[3] N. Gupta et al., “To Reduce the False Alarm in Intrusion
Detection
System
Using
Self
Organizing
Figure 18: Intruder L-1 Detection Result of Experiment 3
65
Proceeding of International Symposium on Smart Material and Mechatronics
Ma”.International Journal of Soft Computing and
Engineering,Vol. 1, No. 2, 2011.
[4] J. T.Jimenez et al., “ Analysis of Performance Metrics
from a Database Management System Using Kohonen’s
Self Organizing Map”, WSEAS, Vol. 2, No. 3. 2003.
[5] P. A. Aguilera et al., “Assessment of Groundwater
Quality by Means of Self-Organizing Maps: Application
in a Semiarid Area”, Environmental Management,Vol.
30, No. 5, 2002.
ISBN 978-602-71380-1-8
[20] X. Bai et al., “Skeleton Pruning by Contour Partitioning
with Discrete Curve Evolution”, M.S. theses, China:
Huazhang University of Science and Technology, 2005.
[21] P. Kumar et al., “Study of Robust and Intelligent
Surveillance in Visible and Multimodal Framework” M.S.
theses, India: Indian Institute of Technology, 2006.
[22] L. Zhao, Liang, “Dressed Human Modeling, Detection,
and Parts Localization”, Ph.D. dissertation, United States:
The Robotics Institute Carnegie Mellon University
Pittsburgh, PA 15213, 2001.
[6] A. Iroala. (2009). “Skeleton based visual pattern
recognition: application to table top interaction, Ph.D.
dissertations”, Donostia – San Sebastian: University of
the Basque, 2009.
[7] A. Mitrokotsa et al., “Detecting Denial of Service Attacks
Using Emergent Self-Organizing Maps”, M.S. theses,
Greece: University of Piraeus, 2005.
[8] A. Balaz et al., “Intrusion Detection System Using Self
Organizing Map”, M.S. theses, Slovak Republic:
Technical University of Kosice, 2006.
[9] C. Charnsripinyo et al., “Real-time Intrusion Detection
and Classification”, M.S. theses, Thailand: University of
Technology Thonburi.
[10] S. Kaski, Sami, “The Self-Organizing Map (SOM)”, M.S.
theses, Finland: Helsinki University of Technology,
(1999).
[11] E. Weippl and K. I. Ismail, “Content-based Management
of Document Access Control”, M.S. theses,
Autria:Software Competence Center Hagenberg, 1999.
[12] M. Kumar et al., “Feature-Based Alert Correlation in
Security Systems Using Self organizing Maps”, Ph.D,
Karachi Pakistan: NED University of Engineering and
Technology,2009
[13] S. Kaski and Kohonen, “Exploratory Data Analysis by
the Self-Organizing Map: Structures of Welfare and
Poverty in the World” Research Project, Finland:
University of Technology Rakentajanaukio, 1995.
[14] I. Yoo, “Visualizing Windows Executable Viruses Using
Self-Organizing Maps”. M.S. theses,Switzerland:
University of Fribourg, 2004.
[15] M. Kolemainen et al., “Reducing energy consumption by
using self-organizing maps to create more personalized
electricity use information”, M.S. theses, Finland:
University of Kuopio, 2008.
[16] Moore, David, “A real-world system for human motion
detection and tracking”, M.S. thesis,California: California
Institute of Technology, 2003.
[17] S. V. Ioannou et al., “Emotion reconition through facial
expression analysis based on a neurofuzzy network”, M.S.
theses,Greece: National Technical University of Athens,
2005.
[18] N. Aggarwal et al., “Vision Based Hand Gesture
Recognition”, M.S. theses, World Academy of Science,
Engineering and Technology, 2009.
[19] E. BosKurt and O. S. Mehmet, “Ambient Intelligence in
Home
Environments”,
M.S.
theses,
Turkey:
Telecommunications Engineering Sabanci University,
2004
66
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Simulation and Experimental Works of
Quadcopter Model for Simple Maneuver
Rafiuddin Syam
Mechanical Engineering Department, Hasnuddin University,
Jl. P. Kemerdekaan Km 10 Makassar - Indonesia
(corresponding author phone: +6285827881000; e-mail:
rafiuddinsyam @gmail.com)
Mustari
Mechanical Engineering Department,
Dayanu Iksanuddin university, Baubau-Indonesia
(phone: +6282292850988; e-mail: [email protected]).
Abstract—This study aims to create a simulated and experimental
of aircraft movements for multirotor quadcopter. The research
method is theoretical and experimental methods. For theoretical
method consists of calculating the dynamics and kinematics.
While the experimental method consists of the aircra
II. DYNAMICS MODEL
In general, a quadcopter described simply as four rotors
that are in a cross configuration. Vertical movement is
obtained by adding or reducing the speed of the rotor with all
the same value. This movement produces a vertical force U1
(N) to the body frame which will be up or down the
quadcopter. Roll motion is obtained by increasing or reducing
the rotor speed of the left and at the same time reducing or
increase the right rotor speed. Pitch motion is obtained in the
same way on both the other motors.
Front and rear motors turn to the the anti-clockwise
direction while the two other motor turn to the clockwise, so
that the direction of the yaw motion/ anti-clockwise obtained if
the front-rear speed propeller ˗ increase/ decrease and the left
and right speed of propeller decrease/ increase.
1.751 m/s2, y = 2.038 m /s2 = 1.6371 m dan
the maximum error between the theoretical and the actual
movement is ex = 0.682 m; ey and ez = 0.353 m = 0.546 m.
Theoretical movement pattern already resembles the actual
movement..
Keywords: model
kinematics control
simulation,
quadcopter,
dynamic
control,
I. INTRODUCTION
One type robots attract much attention is a mini unmanned
aerial aircraft UMAVs (“Unmanned Mini Aerial Vehicles”),
because of its ability to perform rescue tasks in hazardous
locations and difficult to reach. This type of helicopter flying
robot has the advantage over other flying vehicle that can
maneuver in cramped areas and perform takeoff and vertical
landing that it is called vertical take˗off landing (VTOL).
Figure 2. Framework configuration of Quadcopter
To design a dynamic quadcopter model, first, defined two
frames (coordinate axes) as shown in Figure 2, condition of an
aircraft that explains the position and a vector velocity defined
in a state of XH. Each vector consists of linear and angular
position and linear velocity and angle.
Figure 1. Quadcopter with 4 rotors
Adapun penelitian ini akan membahas suatu jenis pesawat
mini udara tanpa awak dengan tipe sayap berputar RUMAV
(“Rotary˗wing Unmanned Mini Aerial Vehicle”) yang
dinamakan quadcopter. Shown at Figure 1, the quadcopter
which is a flying robot that has four blades ˗ independent rotor
propeller mounted at each end of a cross frame.
If
is linear position vector consisting of the
components in the direction of the x,y,z (x,y,z) and A
angular position vector consisting of the components of the
angular position x,y,z (ϕ,θ,ψ) then Quadcopter position vector
XH [1] is
67
Proceeding of International Symposium on Smart Material and Mechatronics
XH = [
]T = [x y z ϕ θ ψ] T
A
(1)
If the action of the movement vector UH, OH propeller
gyroscopic matrix and propeller angular velocity Ω, then the
action vector is a general movement
Λ = UH + OH. Ω
(6)
If ̇
is the linear velocity vector consisting of components
of velocity of the x,y,z (ẋ ,ẏ ,ż ) and ̇ A is the angular velocity
vector consisting of the components of the angular velocity on
the x, y, z (p, q, r) then velocity vector,
Both equation above can be written in the following form
MH . ̈
̇
= [ ̇
̇A
] T = [ẋ ẏ ż
(2)
p q r] T
XH0 = [x0 y0 z0 ϕ0 θ0 ψ0]
̇ 0 = [ẋ 0 ẏ ż 0 p0 q0 r0] T
0
F4
ω4
ω3
..
z
..
y
0
M =[
(3)
(4)
0
3x3
0
]= 0
0
[0
3x3
3x3
..
Φ
..
ψ
F3
+ GH + OH. Ω + UH
..
x
3
0
0
0
0
1
F2
ω2
C =[
4
3x3
3x3
3x3
˗S( . ̇ A
..
θ
0
0
0
]= 0
)
0
[0
0
0
0
0
0
0
Gravitational vector,
0
0
0
0
0
0
0
0
0
0
xx
0
0
yy
0
0
0
0
0
0
zz ]
(8)
Figure 3, shows the dynamics of the movement of a
quadcopter. If the aircraft is given an acceleration of the
aircraft will change its position and velocity. If ̈
is a
ai n
nsis ing
n n s (ẍ ,ÿ , z̈ , ̇ , ̇ , ̇)
, the new velocity vector at time t is obtained by the
acceleration vector mengintegral quadcopter against t. While
the new position vector at time t is obtained by integration of
the velocity vector respect to time t. Percepatan yang bekerja
pada pesawat diakibatkan oleh gaya yang diberikan pada
pesawat.
Then, the equilibrium of the dynamics model of
quadcopter will be shown in this part. Then it is important for
analyzing the dynamic equilibrium in the plane so it will know
the forces and moments that can work on the plane.
Furthermore, of the forces and moments acting on the aircraft
will be known acceleration occurs.
If the moment of inertia matrix MH, ̈ acceleration
matrix ̈ , S n i a ˗C i is matrix CH, velocities matrix
̇ , GB gravitational vector and the action vector Λ of the
general movement of an aircraft dynamics model of
quadcopter can define in the following matrix form [1],[2]:
GH = Λ
0
0
0
0
0
0
0
0
0 ˗ zz
0 yy
0
0
0
zz
yy
˗
xx
0
0
0
˗
yy
(9)
xx
0 ]
0
0
Figure 3. Freebody diagram of quadcopter with four rotors.
+ CH . ̇
0
0
It appears that MH is a diagonal matrix and a is a
constant. C n i a ˗C i is matrix,
Fg
MH . ̈
(7)
F1
ω1
2
= CH . ̇
Subscript H indicates that the variable corresponding relative
to the H frame.
If m [kg] is mass of quadcopter and I [N m s2] is moment
inertia matrix, the the inertia system could be described as,
Subscript E, B and H indicates that the variable relative to
the frame-E, B-frame and H-frame.
It will be distinguished initial state vector and the state
vector after acceleration in a given time t, respectively notated
XH0, ̇ 0 , XHt dan ̇ 0 ,
T
ISBN 978-602-71380-1-8
=[
]=
3x1
- g
0
0
[ 0 ]
(10)
If JTP [N m s2] is the total rotational moment of inertia
h
axis and Ω [ ad / s] is h angu a
iy
the propeller whole, then the propeller gyroscopic matrix,
Ω=
0
0
0
˗
[0
0
0
0
˗
0
0
0
0
˗
0
0
0
0
Ω
˗
(11)
0]
If lift force factor b [N s2] and d [N m s2] is the drag
factor, then the movement matrix can be shown below,
(5)
68
Proceeding of International Symposium on Smart Material and Mechatronics
0
0
=
0
0
0
0
0
˗
˗
[ ˗d
0
0
̇=
yy
˗
̇=
zz
˗
Ω +
xx
Ω +
̇=
xx
˗
yy
yy
+
zz
If U1 [N], U2 [N m], U3 [N m] and U4 [N m] are the lift
force, roll torque, torque pitch and yaw torque respectively,
then lift factor b [N s2] and the drag factor d [N m s2 ] then
action vector relative to frame-B,
0
0
0
0
2
2
2
2
(Ω
+
Ω
1
2 + Ω3 + Ω4 )
U1
2
Ω=
=
(Ω24 ˗ Ω22 )
U2
U3
(Ω23 ˗ Ω21 )
[U4 ] d (Ω2 + Ω2 ˗ Ω2 ˗ Ω2 )
[
]
U =
2
4
1
(13)
Tabel 1.Spesifikasi Pesawat
Parameters
L
b
d
m
Ixx
Iyy
Izz
JTP
3x3
(14)
θ=
sψ
θ
˗sψ ϕ + ψ sθ sϕ
sψ sϕ + ψ sθ
ϕ
ψ ϕ +sψ sθ sϕ
˗ ψ sϕ +sψ sθ
ϕ
[ ˗sθ
θ sϕ
If tk = tan k then transfer matrix,
1 sϕ θ
θ
= 0
ϕ
[0
sϕ /
θ
(15)
|
|
|
= |
|
|
(16)
= (˗ CH ̇
= [ẍ ÿ z̈ ̇ ̇ ̇]
+ GB + OH Ω + UH) M˗1
|
|
|
= |
|
|
(17)
sψ sinθ
sϕ)
ÿ = (- s ψ sinϕ + sinψ sinθ
sϕ)
U1
z̈ =
g + ( sθ
sϕ)
0.0000
1.5760
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
1.5760
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
|
|
|
Gravitasional Vector G_H = |
|
|
In the simple equation is described as (17):
ẍ = (sin ψ sinϕ +
1.5760
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0153
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0001
0.0000
0.0000|
0.0000|
0.0000|
0.0000|
0.0000 |
0.0317 |
Sentripetal-Coriolis Matrix C_H (.10^-4)
ϕ/ θ]
Acceleration vector of quadcopter relative to frame-B
could be described as,
̈
arm
lift coefficient
drag coefficient
mass of quadcopter
body inertia moment for axis-x
body inertia moment for axis-y
body inertia moment for axis-z
rotational moment inertia
Inertia Matrix of the System , M_H,
ϕ θ
˗sϕ
Nilai
| 0.1243|
| 0.1397|
| 26.0946|
Vektor Aksi Pergerakan, U_H =| -0.0484|
| 0.0003|
| -0.0002|
]
θ ϕ
Deskripsi
240 mm
72,081.10-6Ns²
3,433.10-6Nms²
1,576 kg
15,293.10-3Nms²
15,293.10-3Nms²
20,307. 10-3Nms²
0,103. 10-6Nms²
An example of output program using matlab program:
If ck = cos k, sk = sin k then matriks matrix,
ψ θ
zz
This section will be calculated dynamics of the movement of
aircraft with specifications as shown in Table 1 The calculation
process will be assisted with Matlab program.
(sψ sϕ + ψ sθ ϕ )U1
3x3
yy
III. DYNAMICS MODEL OF QUADCOPTER
If
rotation matrix then relative action vector aksi relatif
respect to frame-H,
U =[
U4
U3
On equation (18) shows the mathematical model of
quadcopter.
3
(˗ ψ sϕ +sψ sθ ϕ )U1
( θ ϕ )U1
]U =
U2
3x3
U3
[
]
U4
U2
xx
xx
yy
0
d]
˗d
zz
xx
(12)
0
0
d
ISBN 978-602-71380-1-8
U1
U1
0.0000
0.0000
0.0000
-0.0000
0.1189
0.2275
0.0000 0.0000|
0.0000 0.0000|
0.0000 0.0000|
-0.1189 32.7469 |
-0.0000 -32.7469 |
32.7469 -0.0000|
0.0000
0.0000
-15.4448
0.0000
0.0000
0.0000
Gyroscopic Propeller Matrix O_H (.10^-4)
| 0.0000 0.0000 0.0000 0.0000|
| 0.0000 0.0000 0.0000 0.0000|
| 0.0000 0.0000 0.0000 0.0000|
(18)
69
|
|
|
|
|
|
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
= | 0.1965 -0.1965 0.1965 -0.1965 |
| 0.2208 -0.2208 0.2208 -0.2208 |
| 0.0000 -0.0000 0.0000 -0.0000 |
IV. QUADCOPTER SIMULATION
This section will show a simulation of movement
quadcopter both theoretically and experimentally. Movement
of the experimental results obtained from the GPS data
attached to Quadcopter. Furthermore, the results of
calculations using the model dynamics and kinematics will be
demonstrated in this section. For illustration, the movement in
the x direction is planned to reach a distance of 5 m. As for the
motion in the y direction is planned to reach a distance of 4.5
m. For motion in the z direction is planned and maintained at a
height of 5 m.
Angular velocities vector of each propeller,
Omega
|
|
= |
|
300.8505|
302.9173|
300.8799|
298.7114|
Angular velocities of quadcopter,
= -0.1017
Acceleration Vector for aircraft,
| 0.0789|
| 0.0886|
| 6.7575|
XDDot_H = | -3.1632|
| 2.8624|
| -0.0252|
In Figure 5, shows the graph of the trajectory of aircraft
movement in three dimensions obtained from the calculation of
dynamics and kinematics using matlab program
6
Translation B-F
Translation L-R
Vertical Translation
L-R
5
Actual
Aktual
Teoritik
Theoritical
x (m)
4
3
2
1
time (s)
time (s)
time (s)
Rotation B-F
Rotation Bjj-Sjj
0
Rotation L-R
-1
0
0.5
1
1.5
2
2.5
3
3.5
2.5
3
3.5
2.5
3
3.5
waktu (s)
time (s)
4.5
Aktual
Actual
Teoritik
Theoritical
4
3.5
time (s)
3
time (s)
2.5
y (m)
time (s)
Figure 4. Acceleration and Time for vertical and horizontal
movements
2
1.5
1
0.5
Figure 4 shows the acceleration (m/s2) and angular accelration
(rad/s2) respect to time.
0
-0.5
0
0.5
1
1.5
2
waktu (s)
time (s)
6
5
Aktual
Actual
Teoritik
Theoritical
4
z (m)
a
x
i
s
z
3
2
1
0
-1
0
ax
i
0.5
1
1.5
2
waktu (s)
time (s)
s
y
axi s
Figure 6. Position (m) vs time t(s)
x
In Figure 6, shows the graph plane distance (m) and time (s) in
the x, y and z both theoretical and experiments using matlab
program.
Figure 5.Simulation of aircraft movement
70
Proceeding of International Symposium on Smart Material and Mechatronics
V. CONCLUSIONS
A
a i n du
h h us an is a g n ugh ha x =
1.750
/ s2, y = 2.0377 / s2 and z = 1.6371 / s2
each mileage is x = 0.5565 m , y = 0.6307 m and z = 1.9700 m.
Thus indicating that the aircraft is quite agile in maneuvering.
Quadcopter movement simulation shown that the maximum
error position between the theoretical and the actual movement
is ex = 0.683 m; ey and ez = 0.353 m = 0.546 m. Movement
patterns and the theoretical value of the error is influenced by
the control limit membership function of the fuzzy logic
control design. Theoretical movement pattern similar to the
actual movement.
REFERENCES
[1] Bresciani,T. 2008.Modelling, Indentification and Control of a
Quadcopter Helicopter. Department of Automatic Control, Lund
University.
[2]
dić A ksanda , M s
yu a. 2011.The Modeling and
Simulation of an Autonomous Quad-Rotor Microcopter in a
Virtual Outdoor Scenario, University of Belgrade, Institute
Mihajlo Pupin, Robotics Laboratory, Belgrade.
[3] Matilde Santos, Victoria Lopez 2010. Intelligent Fuzzy
Controller o\f a Quadcopter.
Facultad de Informatica
Universidad Complutense, 28040 Madrid, Spain.
[4] E. Abbasi1, M. J. Mahjoob, R. Yazdanpanah.Controlling of
Quadcopter UAV Using a Fuzzy System for Tuning the PID
Gains in Hovering ModeCenter for Mechatronics and
Automation, School of Mechanical Engineering College of
engineering, University of Tehran Iran.
[5] Birkan Tunç, K. y un Ya ı ı.Fuzzy Logic Control Of A Four
Rotor Unmanned Air Vehicle, Quadcopter Istanbul Technical
University Mechanical Engineering Department, System
Dynamics & Control Unit
71
ISBN 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Design of Wheeled Mobile Robot with Tri-Star
Wheel as Rescue Robot
Paisal
Mechanical Engineering Department
State Polytechnic of Ambon
Jln. Ir. M. Putuhena Wailela Rumahtiga AmbonIndonesia-97234, phone: +6281342233140; email:
[email protected]
Rafiuddin Syam, Wahyu H. Piarah
Mechanical Engineering Department
Engineering Faculty, Hasanuddin University
Jl. P. Kemerdekaan Km 10 Makassar, Indonesia,
90245 e-mail: [email protected],
[email protected]
Abstract— This study aims to design, and analyze a mobile
robot that can handle some of the obstacles, they are uneven
surfaces, slopes, can also climb stairs. WMR in this study is Tristar wheel that is containing three wheels for each set. On
average surface only two wheels in contact with the surface, if
there is an uneven surface or obstacle then the third wheel will
rotate with the rotation center of the wheel in contact with the
leading obstacle then only one wheel in contact with the surface.
This study uses the C language program. Furthermore, the
minimum thrust to be generated torque of the motor and
transmission is 9.56 kg. The results obtained by calculation and
analysis of DC motors used must have a torque greater than
14.67 kg.cm. Minimum thrust to be generated motor torque and
the transmission is 9.56 kg. The experimental results give good
results for robot to moving forward, backward, turn left, turn
right and climbing the stairs.
Figure 1. Design of robot
II. THE CONCEPT OF TRI-STAR
A. Stages of creating robot
In the Figure 1, shows stages of the manufacturing. There
are three stages of manufacture of the robot, the first is
planning, Including the selection of hardware and design, the
next stage of manufacture, Including the manufacture of
mechanical, electronic, and programs, and the last stage is the
testing [3].
Index Terms—WMR, tri-star, kinematics, dynamics model.
I. INTRODUCTION
Wheeled mobile robots (WMR) are generally used for flat
areas or smooth trajectory, then the region with a significant
height difference then the robot will stop or be programmed to
turn look for another way. One of the unique and
multifunctional robots are made to address the issues above are
a robot with wheels arranged in a triangular shape, so there are
three sets of wheels each wheel. This model is commonly
called tri-star wheels.
On average surface only two wheels in contact with the
ground, if there is an uneven surface or obstacle then the third
wheel will rotate with the rotation center of the wheel in
contact with the leading obstacle then only one wheel in
contact with the surface. This kind of robot that uses wheels
contacted to the ground, commonly used on uneven road
surfaces. And also, these types of robots are very well used to
climb the stairs.
Several previous studies which can be used as a reference,
i.e. design rescue robot [1], and design wheel chair that can be
climbing the stairs [2]. Meanwhile, this research aims to design
and analyze the performance of a mobile robot using a tri-star
wheels.
B. Wheeled Mobile Robot
Some types of wheels are often used on WMR among other
types of Bi-wheeled form of a pair of wheels that can be moved
with a soft, very suitable for modeling, but still prone to the
risk of slippage. Caterpillar type, form two pairs of wheels
(each wheel connected to each other) with the appropriate
characteristics of a straight movement, not to risk a skid, but
cannot accurately model the turning movement. Type
omnidirectional, has characteristics can move freely in all
directions throughout the complex structure of the wheel, but
still has the disadvantage of frame [4].
(a)
(b)
Figure 2, Wheel type that is often used in the WMR
(a) Bi-Wheel ; (b) Caterpillar ; (c) omnidirectional
72
(c)
Proceeding of International Symposium on Smart Material and Mechatronics
C. Electrical Motor
Wheel of WMR uses an electric motor. For high torque
motors and torque motors larger then it is better to use a DC
motor. The most important in the selection of a motor is the
torque necessary. Note that, Figure 3 shows how the tri-star
wheel work. Then, to calculate the minimum torque on the axis
robot with the following equation [4]:
A.
Kinematics of WMR
Technically, WMR robot has two main wheels are each driven
by an independent drive, and the other that there is a wheel
with one or two castor wheels. Those wheels are placed at the
back of the robot that serves as a counterweight.
Figure 4 shows the architecture of the robot viewed from the
top. If both drive wheels are spinning at the same speed, the
robot will move straight direction, whereas if one of the wheel
speeds is slower, then the robot will move with a curved
trajectory toward the direction of one of the wheels that move
more slowly [5].
For the wheel radius r, and the rotation speed of the right
wheel, and left respectively ωR and ωL then, the linear speed of
the right and left wheels can be calculated by the following
equation:
Figure 3. How to calculate torque of motor
(5)
(6)
Static friction force Fr on wheels, with a coefficient of
static friction between the wheels k and the robot mass mserta
landasanμ and g is the acceleration due to gravity.
Fr = μ × m × g
ISBN 978-602-71380-1-8
When the robot motion play when time t with the
length of the radius R is measured from the center of rotation
and the center point of the two wheels, the angular velocities
can be calculated as:
(7)
(1)
Friction Force F, with radius of wheels R dan radius of axis is
R0,
(2)
(8)
Torque at the axis can be shown,
(3)
Torque of motor is,
Robot linear velocity v(t) and the angular velociti of
the robot ω(t) can be determined based on both the linear
speed of the wheel. In detail, can be shown in matrix presented
in equation below,
(4)
(9)
where hG is an efficiency of gears, , Tstall is stall torque, it
shown at specification of motor and N is gear ratio.
This equation shows the relation between the direct
kinematics linear velocities of the wheels, and linear and
angular velocities of the robot, while this equation below
shows the opposite relationship.
(10)
Terms of absolute position control of the mobile robot is
a robot knows the position and orientation of each moment.
Furthermore, one solution is to calculate the distance the
wheel every time. Mileage left wheel (SL), the right wheel
(SR), and average distance (S) in successive time zones as
follows:
Figure 4. Position and orientation of the mobile robot
at Cartesian frame
(11)
(12)
73
Proceeding of International Symposium on Smart Material and Mechatronics
(13)
ISBN 978-602-71380-1-8
by using the method of complex numbers [7] as follows, note
Figure 7.b.
Formula approach to orientation, position coordinates x,
and y coordinates respectively as follows.
(14)
(15)
(16)
D. Tri-star model of WMR
.
(a)
(b)
Figure 7. (a) Motion of gears (b) Freebody diagram for A wheel respect to the
center A.
position vector of point A is expressed as
equation form
Figure 5. Ilustration the movement of Tri-Star Wheel uneven surface
Note, Figure 6 below, gears gear A is connected to the drive.
B gears numbering three, connected by gears roda.Dan C is a
liaison and steering. All three gears A, B and C are connected
to the arm. There are two models of the movement of the
WMR by tri-star wheel that is currently running on a flat
foundation and up the stairs at the time.
the following
(20)
If 2 a function of time t, the velocity of point A is obtained by
derivate the A position respect to the time t.
B
(21)
A
Lengan
Acceleration of point A obtained by derivate velocity point
A respect to time t.
C
(a)
(b)
Figure 6. (a) Model of gears of Tri-Star Wheel (b) Moving at flat surface
When moving on a flat surface then, there are two wheels in
contact with the runway, each wheel just spins on its axis and
each arm is not moving, see figure 6.B. Then, comparison of
the angular velocity of the wheels of their gears are
(22)
For dynamic analysis of tri-star wheel by using complex
numbers, the first step is to describe the free-body diagram.
The second step is to create equilibrium equation based on free
body diagrams. Furthermore, outlining all relevant vectors in
the form of complex numbers.
Consider the free body diagram in Figure 7 the equilibrium
equation is given as follows
(17)
(18)
Velocity ratio (N) of this model is
(23)
(19)
Where the vectors described above as follows
When climbing stairs, there is only one wheel in contact
with the ground, and the other spinning wheels on the axle in
contact with landasan.akibat rotation arm, see Figure 7.a.
Kinematics equations under these conditions can be explained
(24)
(25)
(26)
74
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
0,098 m/s, |AA | = |AB | = 0,216 m/s2, and d |AG | is 0,113
m/s2.
Then the equilibrium equation can be rewritten
(27)
Components described in the real and imaginary components,
generating the equation below
(28)
(29)
Figure 9. WMR Tri-Star
3) The results of the analysis of the dynamics of the wheel TriStar, is given in the graph in Figure 11. dynamic
calculations to obtain the value of the torque to turn the
wheels when climbing stairs.
Figure 8. Tri-star wheel in equilibrium position.
In order to balance the system, then
, where
(30)
It can be shown as,
Figure 10. Tri-Star Wheels
In the Figure 10, WMR Robot Products Tri-Star has a
length of 110 cm, width 75 cm and height 60 cm. And for the
tri-star wheel has a small gap between the wheel axis 18 cm
and smaller wheelbase to the center tri-star wheels 10 cm.
(31)
Torque
Angle vs Torque
III. ANALYSIS MODEL AND DISCUSSION
A. Result of Model Analysis
From this research generated several things, among others:
1) Wheeled Mobile Robot (WMR) with Tri-Star Wheels are
made visible in the image 11. This robot consists of a
frame, tri-star wheel just as seen in Figure 12, the motor
and transmission torque amplifier (Figure 13), and
electronic devices.
2) Kinematic analysis of results of Tri-Star Wheels with
kinematics equations by the method of complex numbers,
give the values of position, velocity and acceleration of
each points on the tri-star wheel as follows |rA | and |rB | =
0,18 m, |rG | = 0,09 m, |VA | = |VB| = 0,195 m/s, |VG | =
Angle
Figure 11. Relation between angle and torque
WMR Tri-Star robot is controlled by an open loop control
system, wherein the command signal is given via an infra-red
remote control system that is subsequently processed in the
microcontroller and then proceed to the next motor controllers
75
Proceeding of International Symposium on Smart Material and Mechatronics
motor will rotate according the applied voltage. Overall the
controls are stored in a control box. From the observation, the
command would be well received if the distance between the
signal receiver remote with 1-10 meters.
From the test results, the performance of the robot WMR
acquired Tri-Star is moving forward with a speed of 0.10 m / s,
either moving backward with a speed of 0.10 m / s, turn left
and right both with velocity 0.0024 rad / s, ladder that can be
climbed up to a height of 20 cm and a width of at least 25 cm
IV. CONCLUSION
WMR robot by tri-star wheel can perform experiments for
uneven surfaces. In addition, the tri-star robot successfully run
straight, turn and climbs stairs. Meanwhile, the average of
velocityof the Tri-star robot is 0.1 m / s. Furthermore, the
controller of this system is done by open loop method using
infrared control system.
ACKNOWLEDGMENT
Thanks to DIKTI (DGHE, Directorate General Higher
Education) Government of Indonesia (GOI) for the
Competitive Grant Universities 2014 and LP2M of Hasanuddin
University.
REFERENCES
[1] Jufri, Sialana , Design Rescue robot for searching, Hasanuddin
University, Makassar, 2010. (Original in Bahasa)
[2] Hudan, Nuzula Alfian, Design Build Wheelchairs Can Be Up Stairs,
ITS- Surabaya, 2010. (Original in Bahasa)
[3] http://www.tutorialgratis.net, tutorial to create intelligent robots, 2010
(Original in Bahasa)
[4] Sakti, Indra et. al, Driving force mechanism for soccer robot, Journal
Information Technology (Journal in Bahasa), Vol. 5 No. 1, 2004
[5] Santoso, Junaidi, “Design of Mobile Robot finder and obstacle
avoidance using fuzzy logic control, Diponegoro University, Indonesia,
2009.
[6] Stan, Gibilisco, Concise Encyclopedia of Robotics, McGraw-Hill, New
York, Amerika Serikat, 2003.
[7] Hutahaean, Y. Ramses, “Mechanism and Dinamics for Machine”,
Revision edition, ANDI Publishing, Yogyakarta,. 2010 (Original in
Bahasa)
76
ISBN 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Application of Genetic Algorithm for Determining
the Optimum Ship Route
Faisal Mahmuddin
Rahmad Patarru
Department of Naval Architecture, Engineering Faculty
Hasanuddin University
Makassar, Indonesia
[email protected]
Department of Naval Architecture, Engineering Faculty
Hasanuddin University
Makassar, Indonesia
[email protected]
Rahimuddin Samad
Department of Naval Architecture, Engineering Faculty
Hasanuddin University
Makassar, Indonesia
[email protected]
Abstract—Determining the most optimum route is a crucial
factor in ship operation in terms of technical and economical
points of view. In the present study, genetic algorithm (GA) is
applied to determine the most optimum route of a ship with
several destinations to visit. The problem and solution are
considered to be similar with the travelling salesman problem
(TSP). In GA implementation, the fitness of each individual is
measured from the distance of its origin and final destination
points. Moreover, several standard genetic operators are also
implemented in the present study such as selection, mutation and
crossover. An interactive program is developed with visual basic
programming language to demonstrate and simulate the
practicability and effectiveness of the method to solve the
problem. Three cases are simulated in order to analyze the
performance of the solution method which are simulations with
varying number of population, probability crossover and
mutation crossover. From the computation results, it can be
shown that even though the average fitness of the population is
fluctuating, the most optimum route still can be obtained.
ships, dangerous coral reefs, or bad weathers in certain area of
the ocean.
In order to determine the most optimum route of a ship with
several destinations, genetic algorithm is adopted as the main
computation method in the present study. The method is one of
popular methods in optimization research field. The flexibility
of the method makes it convenient to be adopted in numerous
research fields. For the purpose of finding the shortest path, it
was also implemented by several researchers to the vehicles
routing problem [1, 2, 3].
As described previously, in the present study, it is assumed
that the ship has several destinations to visit. The ship needs to
visit each destination location at least once. Therefore, the
problem and solution method in the present study are similar to
the travelling salesman problem (TSP). The implementation
and solution of TSP with GA can also be found in Larranaga et
al. [4] and Grefenstette et al. [5].
In order to define and solve the problem, ship sailing area is
divided into a grid to make it easier to define the location and
distance of the ship inside sailing area. The fitness of an
individual is measured from the distance the ship sails starting
from origin to the final destination points. In the present study,
even though it is assumed that the origin point can be from any
position inside the grid, it would be easy to extend the problem
with fixed origin point.
Moreover, a computer program is developed to simulate the
process of determining the most optimum route using Visual
Basic programming language. With this program, the user can
define the number and position of all locations that ship needs
to visit. For this purpose, the user can determine them by
manually clicking inside the grid or requesting the application
to randomly determine certain numbers of locations to visit.
In order to demonstrate the applicability and performance
of the solution method and program, simulations are performed
by varying 3 (three) variables, which are:
 Case 1: varying initial population
 Case 2: varying crossover probability
 Case 3: varying mutation probability
Index Terms—Genetic algorithm, optimum ship route,
travelling salesman problem, visual basic simulation.
I. INTRODUCTION
As fuel price increases, the ship sailing operation costs also
get higher. This is not a favorable condition either for ship
operators or for ship passengers in terms of ticket prices and
sailing time. Therefore it is an important and crucial task to
make the ship operation effective and efficient.
There are several ways to address the problem such as
installing a fuel-efficient engine, utilizing cheap and renewable
energies, designing ship hull forms with less resistance, etc.
Another way to resolve the problem is to sail following the
shortest route. As a result, the engine fuel and sailing time can
be saved.
In the present study, the shortest route is assumed to be the
most optimum route. It is also assumed that the ship needs to
visit or reach several locations before the ship arriving at a final
destination. The reason of the ship to visit or reach several
locations can be to load/unload passengers and cargos at other
ports or simply to avoid static obstacles such as anchoring
77
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
With these 3 cases, the effect of varying those variables to
the computation results can be analyzed and discussed.
Because of the flexibility of the solution method, the program
simulation can be easily further developed and implemented to
several other applications such as real time ship autopilot,
remote underwater vehicle (RUV) obstacles avoidance,
optimum ports development, etc.
II. GENETIC ALGORITHM
Genetic algorithm is one of popular methods used for
optimization purpose. It is a method inspired by the evolution
theory and uses the survival of the fittest strategy to find the
optimal solution in a defined search space. It is an iteration
process which starts by generating a certain number of random
individuals. After that, genetic operators such as reproduction,
crossover and mutation are applied to form the next generation
which satisfies a certain fitness function. More detail
explanation can be found in Coley [6] and Sivanandam and
Deepa [7].
A. Outline of GA
Before starting the GA analysis, the chromosome
representation of each individual or so-called encoding must be
defined. There are some types of encoding such as binary
encoding, integer encoding, permutation encoding, and tree
encoding. In this study, integer encoding will be used. The
integer will represent the number of locations that ships need to
visit. Besides encoding, the genetic operators should also be
defined. The following genetic operators are applied in this
study.
 Selection (reproduction) is the process of choosing
parents for mating. There are several methods which
can be used in parents selection process for example
roulette-wheel, tournament, stochastic universal
damping, and truncation selections. The present study
will adopt tournament selection.
 Crossover is used to interchange limited parts of
parents. The parents will be decided to undergo
crossover or not based on crossover probability (Pc).
 Mutation is used to flip the value of each chromosome
of an individual. It is decided to apply mutation based
on mutation probability (Pm).
 Elitism is copying the fittest member of previous
population if the maximum fitness of the new
population is lower than this fittest member.
The main flow of GA is shown in Fig. 1. In this figure, an
initial population is generated as the first step. An individual in
the population represents the distances of the ship route to sail.
At this step, some random numbers are thrown for some
number of genes which constitute a chromosome. After
obtaining the values of each gene, the fitness of each individual
is computed. The fitness is measured to determine the
probability of an individual to be selected as parents which will
be mated using genetic operators to obtain offsprings. These
offsprings are then used to replace the old population and to be
computed again to obtain their fitness. The process will
continue for certain number of generations until the
computation result converges. The computation is considered
to be converged when there is no more increase in the best
fitness for more than 100 generations.
Fig. 1. GA workflow
B. Individual Representation
In order to solve the problem, several locations that ship
needs to visit are represented by filled circles with certain
numbers on them as shown in the figure below.
Fig. 2. Problem definition
Each individual in the present problem is represented by a
possible route that ship can sail. The chromosome of the
individual is represented by the number of locations that the
ship needs to visit. Therefore, each individual will have a
chromosome consisting of random numbers of integer from 0
to the numbers of locations. 3 (three) examples of individuals
with number of location equals to 10 are:
Individual 1 = 8 2 0 5 9 6 3 7 1 4
Individual 2 = 3 1 8 5 3 7 1 9 4 6
Individual 3 = 5 0 4 9 7 4 2 8 1 3
The first integer of chromosome represents the origin
point, the last integer represents the final destination point, and
all other integers represent the locations or positions that need
to be visited or reached at least once.
C. Fitness Definition
In order to evaluate the performance of an individual and
convergence of the calculation, the fitness of the individual
78
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
needs to be defined. In this study, the fitness is defined to be
the distance that ship needs to sail from origin to final
destination. Therefore, lower value of fitness means shorter
distance which represents higher fitness.
The fitness represents the probability of an individual to be
chosen as parents for mating. Therefore an individual with
higher fitness will have a bigger chance to be selected for
mating. In implementation, the ship sailing area will be divided
into grids, so the distance of two positions is measured from
the number of grids which separate the two positions.
Because the fitness is measured by the distance of the grid,
it can be determined using the following simple formula
d12 
 x2  x1 
2
  y2  y1 
2
where
(x1, y1) = ordinate of the first position
(x2, y2) = ordinate of the second position
Fig. 4. Position definition window
III. SIMULATION PROGRAM
The window shown in Fig. 4 consists of an empty panel,
three buttons and an input text fields. In order to define the
position, the user can directly click any point inside the panel
or fill a certain number on input text field and then click
“Random Position” button. The program will randomly place
several positions on the panel. After the users defines the
position, they can click button “OK” to quit or button “reset” to
remove all the defined positions and start again the process
from beginning.
After the positions definition processes are completed. The
window can be closed so the program will return to the start
window again. However, this time, the start window will
inform the user that the positions have been defined by
changing the message to be “positions defined”. At this stage,
the simulation can be started by filling all input fields and then
pressing “Start Simulation” button.
Simulation program is developed with Visual Basic
programming language. The program consists of two windows
with several controls on them. The first window shows the
introduction about program and several input fields that needs
to be filled by user which are number of population (PopSize),
maximum generation (MaxGen), crossover probability (Pc),
and mutation probability (Pm).
The start window of the program is shown in the following
figure.
IV. RESULTS AND DISCUSSION
With the problem definition defined above, simulations are
performed. The number of locations are randomly chosen to
demonstrate the simulation as follows
Fig. 3. Program start window
Besides the introduction text and input fields, there are also
3 buttons. One button is to define the position labelled
“Position Definition”, another one is to start the simulation
labelled “Start Simulation” and third button is in the lower left
corner of the window which is used to quit the program. In
order to start the simulation, the user firstly needs to define the
position by clicking “Position Definition” button. The program
also shows a message below the “Positions Definition” button
says “Positions not defined” when the user is still not defined
the positions.
When the user clicks the “Positions Definition” button, the
following window will be shown.
Fig. 5. Example of random locations to demonstrate the program.
79
Proceeding of International Symposium on Smart Material and Mechatronics
As shown in the figure above, the number of positions are
chosen to be 20. The positions are created randomly by
clicking “Random Position” button.
In order to analyze the effect of genetic variables and
performance of the method, 3 (three) cases are analyzed in the
present study which are varying the number population,
crossover probability and mutation probability.
In Fig. 7, it can be observed that the average fitness is
fluctuating in all 4 values of PopSize. This could mean that not
all individuals in the population change to be the optimum
ones. The results might be caused by the type of encoding used
in the present study. The integer encoding limits the number of
chromosome to be modified, hence limits the identities to be
introduced in the population as well.
Moreover, from Fig. 7, it can also be observed that the
average fitness of the population has the same trend with best
fitness where higher value of PopSize will converge faster than
lower one.
1) Varying Number of Population
The first case is simulated by computing 4 different
numbers of population which are 5, 20, 50 and 70. The
crossover probability and mutation probability are fixed to be
75% and 30%, respectively. The number of maximum
generation is chosen to be 1000 generation. The computation
results are shown below
160
2) Crossover Probability effect
The second case is simulated by varying the crossover
probability to be 50%, 80%, 90% and 100%. In this case, the
number of population and mutation probability are fixed to be
50 and 30%, respectively. The computation results in terms of
best fitness are shown in the following figure
Best Fitness of Case 1
Pc = 75 %
Pm = 30 %
140
120
160
150
140
130
120
PopSize = 5
PopSize = 20
PopSize = 50
PopSize = 75
100
Fitness
ISBN 978-602-71380-1-8
Best Fitness of Case 2
PopSize = 50
Pm = 30 %
Pc = 50 %
Pc = 80 %
Pc = 90 %
Pc = 100 %
110
80
Fitness
100
60
90
80
70
60
40
0
200
400
600
800
1000
Generation
50
Fig. 6. Best fitness with varying population size
40
It can be seen from the figure above that optimal results can
be obtained when the number of population is 5, 20 and 50.
Moreover, it can also be noted that for these 3 values of
PopSize, the solution method can obtain the most optimal
solution faster as the number of population higher.
The average fitness for the first case are shown in the
following figure
160
Fitness
80
60
40
0
200
400
600
800
600
800
1000
From the figure above, a different trend with the previous
case can be observed. In this case, the most optimum solution
is obtained with crossover probability equals to 80% while
higher crossover probability (Pc=90%) obtains the optimum
solution slightly with more number of generations. Moreover,
the highest crossover probability (Pc=100%) converges with
optimal solution lower than results from other 3 values of Pc.
The results show that in genetic algorithm, the higher crossover
probability will not mean the best choice to use in the
implementation. The best value of crossover probability will
depend on the problem we are solving.
The average fitness of the population for 1000 generations
of the second case are shown in the figure below
PopSize = 5
PopSize = 20
PopSize = 50
PopSize = 75
100
400
Fig. 8. Best fitness with varying crossover probability
Pc = 75 %
Pm = 30 %
120
200
Generation
Average Fitness of Case 1
140
0
1000
Generation
Fig. 7. Average fitness with varying population size
80
Proceeding of International Symposium on Smart Material and Mechatronics
160
Average Fitness of Case 2
160
PopSize = 50
Pm = 30 %
140
Fitness
Fitness
100
80
60
0
200
400
600
800
1000
40
0
200
400
600
800
1000
Generation
Fig. 9. Average fitness with varying crossover probability
Fig. 11. Average fitness with varying mutation probability.
Fig. 9 above shows that the average fitness have the same
trend with the best fitness one as shown in the previous figure.
It is also interesting to see that a relatively high peak occurs
when the crossover probability equals to 5% and 100%. The
peak can be attributed to the randomization implemented in the
present study.
From the figure above, it can be noted that the average
fitness for the value of Pm=90%, 75%, and 50% have lower
average fitness compared to their best fitness. The results
suggest that many individuals in the population which have
poorer fitness even after many generations. The trend is
different to all other cases previously. The results support the
previous statement that the genetic operator of mutation has
made the convergence of the solution method slower.
From all computations, the most optimum solution is found
to be 51,344. However, there are several routes which have the
same fitness value. One of the route is shown in the following
figure.
3) Mutation Probability effect
The final case is simulated by varying the mutation
probability to be 10%, 50%, 75% and 90%. The computation
results in terms of best fitness are shown in the following figure
Best Fitness of Case 3
PopSize = 50
Pc = 75 %
Pm = 10 %
Pm = 50 %
Pm = 75 %
Pm = 90 %
110
100
Fitness
80
60
Generation
160
150
140
130
120
Pm = 10 %
Pm = 50 %
Pm = 75 %
Pm = 90 %
PopSize = 50
Pc = 75 %
120
100
40
Average Fitness of Case 3
140
Pc = 50 %
Pc = 80 %
Pc = 90 %
Pc = 100 %
120
ISBN 978-602-71380-1-8
90
80
70
60
50
Fig. 12. One of the most optimum routes
40
0
200
400
600
800
In the figure above, the optimum route has origin on
position 9 while the final destination position is on 0.
Moreover, by observing the route shown in Fig. 12, it can be
suggested that the route is the optimal one. One reason of this
judgment is because there is no arrow crossing other arrows.
Even though the demonstrated example seems to be simple
with not so many number of positions, the simulation with
large number of position can be performed with the program
developed in the present study.
1000
Generation
Fig. 10. Best fitness with varying mutation probability
From Fig. 10. It can be observed that the fastest optimum
solution is obtained with lowest value of crossover probability
which is when Pc=10%. Moreover, the figure also shows that
computation results with the highest mutation probability
converges slowest. The results could give us a suggestion that
the genetic operator of mutation gives a negative contribution
to the convergence of the solution method.
The computation results in terms of average fitness for
1000 generations for the third case are shown below
V. CONCLUSION
In the present study, genetic algorithm is applied to
determine the most optimal route of a ship for sailing. The
implemented method is also applied in a Visual Basic program.
One example is demonstrated in the paper. From the
81
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
[3] C. W. Ahn and R. S. Ramakrishna, “A Genetic Algorithm for
Shortest Path Problem and the Sizing of Populations,” IEEE
Transaction on Evolutionary Computation, vol. 6, no. 6 pp. 566579, December 2002.
[4] P. Larranaga, C. M. H. Kuijpers, R. H. Murga, I. Inza, and S.
Dizdarevic, “Genetic Algorithms for the Travelling Salesman
Problem: A Review of Representation and Operator,” Artificial
Intelligence Review, vol. 13, pp. 129-170, April 1999.
[5] J. Grefenstette, R. Gopal, B. Rosmaita, and D. V. Gucht,
“Genetic Algorithms for the Travelling Salesman Problem,”
Proc. of the First Int. Conf. on Genetic Algorithms and Their
Applications, pp. 160-168, New Jersey, 1985.
[6] D. A. Coley, An Introduction to GA for Scientists and
Engineers, World Scientific Publishing, 1998.
[7] S. N. Sivanandam and S. N. Deepa, Introduction to Genetic
Algorithms, Springer, 2008.
computation results, it can noted the genetic algorithm method
is viable to be implemented in this problem. Moreover, it is
also shown that even though the solution can be obtained, the
average fitness of a population is still fluctuating which can be
contributed to type of chromosome type of the individual.
Another important point in the present study that mutation
probability gives a negative effect to the convergence of the
solution method.
REFERENCES
[1] B. M. Baker and M. A. Ayechew, “ A Genetic Algorithm for
Vehicle Routing Problem,” Computers and Operations
Research, vol. 30, no. 5, pp.787-800, April 2003.
[2] S. B. Pattnaik, S. Mohan, and V. M. Tom, “Urban Bus Transit
Route Network Design Using Genetic Algorithm,” J. Transp.
Eng., vol. 124, no. 4, pp. 368-375, July 1998.
82
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Fatigue Life Prediction In Journal Bearing
Irsyadi Yani Hasan Basri and Hafizd Ibrahim Marsil
Dept. of Mechanical Engineering
Faculty of Engineering, Sriwijaya University
Palembang, Indonesia
[email protected], [email protected] [email protected]
Abstract— Failure of fatigue is damaged materials where caused
frequent load. Fatigue owing to some factors, which is Stress
concentration on fatigue, Stress life, Effect size and surface, and
Change properties of surface. The fatigue failure of a material is
dependent on the interaction of a large stress with a critical flow. In
essence, fatigue is controlled by the weakest link of the material, with
the probability of a weak link increasing with material volume. This
phenomenon is evident in the fatigue test results of a material using
specimens of varying diameters. From this research we can get effect
of concentration stress on strength fatigue with S-N method. On this
method only count fatigue life or endurance limit from Journal
bearing housing. By Finite Element Analysis, it is not so easy to
determine fatigue life. When we find the first yield point, it means
this point is in the highest stress state. Then we can refer S-N curve.
In this paper, the effect of bearing and housing elasticity on the stress
field, which could result in surface fatigue in journal bearing, has
been investigated. This condition is proved with occurred slip lines
on surface of specimen. These slip lines are caused on some
thousands stress cycles. Additional crack is happened immediately
and finally long enough crack. So that formed unstable crack that
caused fracture of brittleness or fracture of toughness because section
of specimen cannot keep
an iterative design cycle is centered on the construction of
real prototype components. This inhibits the ability to
develop new concepts and reduces confidence in the final
product due to a low statistical sample of tests. It is also
common to find early products released with „known‟ defects
or product release dates being delayed whilst durability issues
were addressed.
A more desirable approach is to conduct more testing
based on computer simulations. Computational analysis can
be performed relatively quickly and much earlier in the design
cycle.
Confidence in the product is therefore improved because
more usage scenarios can be simulated. It is not
recommended, however, that these simulations completely
replace prototype testing. It will always remain desirable to
have prototype signoff tests to validate the analysis performed
and improve our future modeling techniques. However, the
number of prototype stages, and hence the total
development time, can be reduced.
The following subsections are including: bearing in
general, journal bearings, thrust bearing, other types of
bearings, rotor-bearing system. Coupled thermomechanical
non-linear finite element models have been developed to
study 2D and 3D rolling, and rolling plus sliding contact
problems. The various less or more realistic material
constitutive models have been used to model behavior of
bearing materials. The contact stress fatigue is considered as
a primary wear mechanism. The damage process under
contact loading such as, for example, is the cracking,
spaling, and tribological reaction, can be study by the finite
element method. We can study the mechanics of the subsurface or near surface modes of rolling contact failure.
In this paper we overview the physical behavior
responsible for fatigue stress from initiation to final
component failure of journal bearing.
Index Terms— journal bearing, numerical, fatigue life
I. INTRODUCTION
While many parts may work well initially, they often fail in
service due to fatigue failure caused by repeated cyclic loading.
Characterizing the capability of a material to survive the many
cycles a component may experience during its lifetime is the
aim of fatigue analysis. In a general sense, Fatigue Analysis
has three main methods, Strain Life, Stress Life, and Fracture
Mechanics
According to independent studies carried out by the Battelle
group in 1982, between 80-90% of all structural failures occur
through a fatigue mechanism and the estimated annual cost of
fracture and fatigue to the US was 4.4% of GDP.
Furthermore the Battelle Study concluded that this could be
reduced by 29% by application of current fatigue analysis
technology.
In the past, fatigue analysis was largely the domain of the
development engineer, who used measurements taken from
prototype components to predict the fatigue behavior. This
gave rise to the traditional “Build it, Test It, Fix It” approach
to fatigue design. This approach is known to be very costly as
II. JOURNAL BEARING
The bearings are important enough to be studied because if
the shaft‟s orbit is not stable, or t he bear ing is no t wel l
des igned , cont act between the shaft and the bearing will
appear
83
Proceeding of International Symposium on Smart Material and Mechatronics
The plain journal bearings are fully used in hydraulics due
to their small size, low price, and its capability of carrying
load.
The journal bearing appears in the finite element equations
as a spring and damper. It appears at one node, linking the
shaft to a rigid structure. It has 4 degrees of freedom.
ISBN 978-602-71380-1-8
in the notch root. Different methods can be used for this
conversion, depending on the stress state in the notch root and
the applied loading. The most well-known approach is that due
to Neuber which relates nominal elastic values to notch root
stress and strain as εσ = K t ε n E = C, where C is a
constant and en is nominal elastic strain. Plane stress is
assumed in this analysis, which can be shown not to be the
case for a circumferentially notched bar. A method for general
stress states is outlined later.
The fatigue strength of a welded component is defined as
the stress range which fluctuating at constant amplitude
causes failure of the component after a specified number of
cycles (N). The stress range is the difference between the
maximum and minimum points in the cycle. The number
of cycles to failure is known as the endurance or fatigue life.
m
The expression linking N and
R can be plotted on a
logarithmic scale as a straight lineand is r efer red to as a n
S-N curve. The relationship holds for a wide range of
endurance. It is limited at the low endurance end by static
failure when the ultimate material strength is exceeded. At
endurances exceeding about 5-10 million cycles the stress
ranges are generally too small to permit propagation under
constant amplitude loading. This limit is called the nonpropagating Stress.
Finite element technique involves element-modeling
discretion, which is defined through a displacement function of
each node.
Fig 1. FE Equation of Journal Bearing
The first thing we need to do is determine the static
(mean) position of the shaft in the bearing
Fig 2. Static Bearing Calculation
Two important parameters are obtainedfrom the bearing
test: (i) bearing yield (sb,yield) and (ii) bearing ultimate
(sb,ult) of the material; where bearing stress is defined
with the following
relation:
sb=P/(Dt).
The
yield parameter is defined as the stress at a 2% permanent
hole deformation, which is a definition comparable to the
tensile yield. Bearing ultimate is defined as the first maximum
load peak, which generally was the maximum stress reached.
For the material model, it is assumed that it behaves as an
isotropic material with isotropic hardening. Uniaxial tensile
test data are simplified into a trilinear behavior, consisting of
(i) an elastic part, (ii) plastic part up to necking (15%
p l a s t i c s t r a i n ) w i t h s t i f f n e s s e q u a l t o 1467.67
MPa, which is followed by (iii) a description of the necking
behavior.
Fatigue is defined as „Failure under a repeated or otherwise
varying load which never reaches a level sufficient to cause
failure in a single application.‟ Fatigue cracks always develop
as a result of cyclic plastic deformation in a lo ca l iz ed area.
This plastic deformation might arise through the presence
of a small crack or pre-existing defect on the surface of a
component, for both cases it is practically undetectable
and unfeasible to model using traditional Finite Element
techniques.
The fatigue life prediction follows the strain life approach
used for notched geometries. Surface g r o o v e s a r e
t r e a t e d a s microscopic notches, where elastic stresses
and strains are converted to local plastic stresses and strains
{F}= [k]{D}
(1)
Handling Finite Element Analysis stress requires a good
understanding of the stress-concentration effect, quantified as
a factor Kt. The theoretical stress-concentration factor is
based on a theoretical elastic, homogeneous, isotropic
material and can be expressed as:
k t = σ max / σ nom
(2)
Handling FEA fatigue stresses correctly also requires
good understanding of fatigue stress-concentration factor
By Finite Element Analysis, it is not so easy to determine
fatigue life. When we find the first yield point, it means this
point is in the highest stress state. Then we can refer S-N
curve.
III. RESULT AND DISCUSSION
From this research we can get effect of concentration
stress or Kt on strength fatigue with S-N method. On this
method only count fatigue l i fe o r e n d u r a n c e l i m it
f r o m Journal Bearing. The lubricant was assumed to supply at
ambient pressure via a full width line groove in the upstream
groove. The Reynolds equation was solved using the GaussSeidel iterative method with over relaxation factor. The
84
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
boundary conditions required were at the bearing edges and at
the lubricant supply line. The lubricant was assumed to
capitated at ambient pressure. Thus; Pcavitation = Patmofer =
0. For use with the Reynolds equation and full-width film
applied, the following boundary conditions adapted were,
a) P = 0 at  = 0.
b) P = 0 at  = /2 - , and P = 0 at  = 3/2 c) P = dP/dX = dP/dZ = 0 at  = 2 = +α
d) P = 0 at z = + b/2.
Fig 5. Fatigue for Journal Bearing
Table 1. Fatigue Test for Journal Bearing
Stress (Mpa)
Fatigue life (cycle)
13,872
48671
21,047
20518
27,728
15191
35,312
10513
40,128
5732
From the data we can draw S-N curve for j ournal
bearing.
Fig 3. FEA Modelling for Journal Bearing
From this research we can get effect of
concentration stress or Kt on strength fatigue with SN method. On this method only count fatigue life or
endurance limit from Journal bearing.
Fig 6. S-N curve
IV. CONCLUSION
A general approach to modelling the durability of Journal
Bearing has been developed. The approach removes the
requirement of rebuilding FEM models in order to capture the
important stress raising features which significantly affect
fatigue life predictions.
The method is ideally suited for predicting data for fatigue
life calculations in Journal Bearing.
Fig 4. FStress Consentration for Journal Bearing
Example applications have been presented
demonstrating some of the capabilities of the method. The
most important point addressed in this work is that the method,
which uses, can be implemented easily to predict fatigue life of
journal bearing by Finite Element Analysis. Moreover, the
algorithm provides robust and fast results because the proposed
method avoids the extra computational burden for
85
Proceeding of International Symposium on Smart Material and Mechatronics
preprocessing since only stress concentration of the
components are used to predict the fatigue stress of the journal
bearing. The proposed method can identify stress
concentration, fatigue stress and calculate the fatigue life of the
journal bearing.
ISBN 978-602-71380-1-8
[4] Calculating and Displaying Fatigue Result, Raymond
Browell, Al Hancq, Development Engineer ANSYS, Inc,
March 29, 2006
[5] L. Roy and S. K. Kakoty, Optimum Groove Location of
Hydrodynamic Journal Bearing Using Genetic Algorithm,
Advances in Tribology, Article ID 580367, Hindawi
Publishing Corp, 2013.
[6] W.A. Gross, et. all., Fluid Film Lubrication, John Wiley
& Sons, New York, 1980.
[7] F.A. Martin, C.S. Lee, Feed Pressure Flow in Plain
Journal Bearings, Trans. ASLE. Vol. 26, No. 3 (1982)
381-392.
[8] E.S.D.U. Item No. 84031, Calculation Methods for
Steadily Loaded Axial Groove Hydrodynamic Journal
Bearings, E.S.D.U., London, 1984.
[9] F.P. Brito, J. Bouyer, M. Fillon, A.S. Miranda, Thermal
behaviour and performance characteristics of a twin axial
groove journal bearing as a function of applied load and
oil supply temperature, TRIBOLOGIA - Finnish Journal
of Tribology 3 vol 25 (2006) 24-33.
REFERENCES
[1] Fatigue Life and Crack Growth Prediction Using FEM
Data, Robert A Adey, John M. W. Baynham, Sharon
Mellings, Tom Curtin, Computational Mechanics BEASY,
Ashurst Lodge, Southampton, Hampshire, SO40 7AA, UK
[2] Fatique Life Prediction of machined components using
finite element analysis of surface topography, S. K. As, B
Skallerud, BW Tveiten, B Holme, Science Direct,
Elsevier, International Journal of Fatigue, 27 (2005) 15901596
[3] Finite element analysis of machine elements, A
bibliography (1977 – 1997), Jaroslav Mackerle,
Engineering Com, Vol. 16 No. 6, 1999, MCB University
Press.
86
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Development of 5-DOF Robot Arm Manipulator
Ismail Thamrin
Irsyadi Yani
Dept. of Mechanical Engineering
Faculty of Engineering, Sriwijaya University
Palembang, Indonesia
[email protected]
Dept. of Mechanical Engineering
Faculty of Engineering, Sriwijaya University
Palembang, Indonesia
[email protected]

Abstract— This research is designed and made prototype
articulated type robot arm with five degrees of freedom (5-DOF).
Where the end-effector is a gripper is used that has two claws. This
robot arm joint rotation has 5 pieces / revolute which serves to
connect the links that one with the other link. Each joint is restricted
1 degrees of freedom. It aims to facilitate the mechanical and control
system of the manipulator. As the actuator (driver) joint use stepper
motors, which required the driver to control (control circuit) with
input data in the form of computer commands communicated with the
system working port. Parallel communication or control system of
the manipulator is to adjust the angular position of each joint with
computer program created with Visual Basic 6.0 language, which is
working parallel to the driver circuit so that the driver circuit
generates pulses to drive the stepper motor.
Unilateral action gripper; contact with the surface
of the object. (Method: vacuum, magnetic and
principles of adhesion
Bilateral action gripper; contact with the surface of
the object 2. (Method: clamp)
Multilateral action gripper; more than 2 contacts
the surface of the object. for example, multijointed
fingers (method: gripping).


Mechanical system illustrates how the shape of the robot
hand and the type of components to be used.
Mechanical design is the principle that determines the
skill/dexterity of a robot arm. 3 basic aspects that must be
considered in designing a robot arm:
•
The number of rods (link)
•
Number of connections (joint linear / prismatic and
revolute joint)
•
The size and movement of each rod (degrees of
freedom / DOF)
Index Terms— degree of freedom, robot arm, manipulator
I. INTRODUCTION
Products rapidly evolving field of technology today is
robotics. The reason is because the robot can be designed to
work on the process of working with high accuracy (quality) as
well as having more job productivity (quantity) that can help
increase the volume of production.
One of the advantages of the robot is able to be placed for
the operations that are considered at risk (high-risk) to be done
by humans, such as in the area of high temperature, hazardous
chemicals, toxic gases, radioactive risks, including a mission in
outer space. In addition, the robot is able to do continuous
process repeatedly deemed too tedious to be done by humans.
The work process includes a series of robot control system that
enables a robot to take orders, run the program works, and
produces both interrupt and process repeatedly. One of the
robot controller is a computer that is often used.
The robotics system is being developed not to replace the entire
human role, but rather help people to more easily, quickly and
comfortably in his work.
End Effector
Joint
Links
Fig 1. Schematic Diagram of Robot Arm
Robot arm control system in general is a microprocessorbased electronic circuit that functions as a regulator of all
components
in
shaping
the
work
function.
The control system generally has three functions:
• Opening and closing movement of the manipulator
components at the desired stage and specific point.
• Storing data in memory positions and phases.
• Allow the robot to be able to interface with a particular
device.
Broadly speaking, the motion control system in robotics have
structured parts, which include:
II. ROBOT COMPONENT
The main components of the robot arm generally has three
important parts, namely manipulators, mechanical systems, and
control systems.
Manipulator is a mechanical part that can be used to move,
lift, and manipulate the work piece. Function depends on the
type of robot end-effector mounted on the manipulator. Endeffector can be a device welding, painting, machine tools or a
gripper which has a movable jaw opens and closes. Gripper can
be classified into 3 types:
Input
(Volt)
Possition endeffector
Processing
Driver
Fig 2. Control System of Robot Arm
87
Output
(Radian)
Proceeding of International Symposium on Smart Material and Mechatronics
Robots can be analyzed in two studies domain, namely the
analysis of kinematic and dynamic analysis. Kinematic
analysis related to robot movement regardless of inertial
effects / inertia that occurs when the robot movement, while
dynamic analysis related to the inertia effect of the physical
structure of the robot is the result of the movement generated
by the torque of the actuator when the robot is doing the
movement.
Robotic systems in general consist of a system controller,
electronic and mechanical robots.
kontroller
ri
e
+
G(s)
ISBN 978-602-71380-1-8
[4]
III. RESULT AND DISCUSSION
Flowchart for Program Control Robotic Arm in general can
be seen in Figure below:
Robot
(elektronik & mekanik)
u
y
H(s)
Fig 3 System Robot Arm
G (s) is the controller mathematical equation is H (s) is the
equation for a physical robotic systems including actuators and
physical systems. Component ri is the reference input in the
application can be either a reference position, velocity and
acceleration. E is the error component and the component u is
the output of the controller. The output y is a function of the
expected motion of the robot is always equal to the reference
(motion) which is defined in the input ri.
In a study of 3D movement known use of vector algebra
and matrix algebra to simplify the analysis.
p x 
p u 
 
p 
p

R
 y
 v
p 
p w 
 z
p x  r11
p  r
 y    21
p z  r31
  
 1   0
Fig 4 Flow Chart of System Robot Arm
In the movement of the robot arm are the coordinates of the
beginning of the robot arm. If described as follows
[1]
r12
r13
r22
r23
r32
r33
0
0
sx 
s y 
sz 

1 
p u 
p 
 v 
p w 


 1
[2]
Homogeneous transformation matrix (homogenous
transformation matrix) is a combination of the rotation matrix
and translation matrix. Homogeneous transformation matrix
can be considered to consist of 4 submatrik, namely:
Fig 5 DOF of System Robot Arm
Based on calculations of data for each link and gripper,
obtained the effective length of the center of rotation of the
joint (d & a) and the mass of the system at each link (m):
[3]
d0 = 3 cm = 30 mm
d1 = 6,5 cm = 65 mm
a2 = 10 cm = 100 mm
a3 = 10,9 cm = 109 mm
d4 = 0 mm
d5 = 13 cm = 130 mm
DH represents the only solid link depends on 4 parameters of
the geometry of each link (for revolute or prismatic joint).
These four parameters are defined as follows:
• θi is the angle of rotation on the axis Zi-1.
• αi is the angle of rotation on the axis Xi.
• The translation is on the axis Zi-1.
• ai is the translation in X¬i axis.
For joint rotation, the parameters ai, αi, and is constant, while
the θi parameters change when the link i moves (rotates) with
based on link i-1.
mbase = 480 gram = 0,48 Kg
m1 = 304,7 gram = 0,3047 Kg
m2 = 91,9 gram = 0,0919 Kg
m3 + mmotor bandul = (94,1 + 160) gram
= 254,1 gram = 0,2541 Kg
mgripper + mmotor gripper = (94,6 + 80) gram
= 174,6 gram = 0,1746 Kg
Maximum distance of movement of the arm is not free
because of the mechanical limitations on the arm, where the
magnitude of the maximum angle (θ) achieved by each arm,
namely: θ1 = 900, θ2 = 1200, 1500 = θ3, θ4 = 750, θ5 = 900, θ
6 = 900.
88
Proceeding of International Symposium on Smart Material and Mechatronics
Table 3 Results of Measurement Circuit Voltage Driver
condition
Voltage
Saturation
0V
Cut Off)
21 V
Voltage in
21,5 V
The magnitude of the maximum displacement distance of
each link:
In the same way it will be obtained:
Analysis: From the results of stress measurements obtained
parallel port voltage value logic 0 and logic 1 are the normal
limits, ie, logic 0 in the range 0 s / d 0.8 V and logic 1 in the
range of 2.4 s / d 5 V. therefore concluded parallel port can
work well.
Voltage standard deviation value of less than 5%, so that
the circuit can be concluded worked well.
Test Results: Form programs appear and function properly.
Analysis: Program under normal circumstances, because if it
detected an error in the program after the Run, then the form
and a visual basic program that has been made cannot be
displayed.
If R2 = 100 mm, then the S2 = 0.209 m
If R3 = 109 mm, then S3 = 0.285 m
If R4 = 130 mm, then S4 = 0.170 m
If R5 = 25 mm, then the S5 = 0.039 m
The amount of energy the maximum movement of each link:
E=FxS=mxgxS
E1 = 0,164 Joule
E2 = 0,188 Joule
E3 = 0,710 Joule
E4 = 0,291 Joule
E5 = 0,036 Joule
IV. CONCLUSION
Assuming gears used are ideal, then
d1 = 0,6 cm = 6 mm
N1 = 12
d3 = 0,9 cm = 9 mm
N3 = 17
d2 = 2,5cm = 25 mm
N2 = 60
d4 = 3,8 cm = 38 mm
N4 = 76
ISBN 978-602-71380-1-8
Based on the above can be taken several conclusions,
among others:
• Program Visual Basic 6.0 is a language program that
can be used to control the computer hardware whose
function is to control the robotic arm.
• In the hardware design of a robot arm using the openloop system (without feedback), so that no correction
to the angle error is generated.
• Design of mechanical influence on the performance
of the robot, such as limitation of motion, mass, arm
length, and the interaction force on each joint. That of
all that greatly affect the design of a robot arm
control system program
REFERENCES
[1] Fu.K.S, Gonzales.R.C, Lee.C.S.G. 1987 “Robotics:
Control, Sensing, Vision, and Intelegence” New York:
McGraw Hill Publishing Company.
[2] Tood DJ. 1986. “Fundamental of Robot Technology:
Introduction to industrial robot, teleoperators and robot
vehicles” London: Kogan Page LTD.
[3] Prasetya Retno, Widodo, Edi Catur. 2004 “ Interfacing
Port Parallel Dan Port Serial Komputer dengan Visual
Basic 6.0”. Yogyakarta: Andi Yogyakarta.
[4] Pitoworno Endra. 2006. “Robotika: Desain, Kontrol, dan
Kecerdasan Buatan”. Yogyakarta: C.V ANDI OFFSET
(Penerbit ANDI)
[5] Snyder Wesley E. 2005 “Industrial Robots: Computer
Interfacing and Control” Prentrice-Hall.
[6] Mico Pardosi. “Microsoft Visual Basic 6.0: Bahasa
Pemograman Windows dan Internet”. Selaras.
Additional load lifting capacity = 0,270 kg
From the above calculation is known that the ability to lift
maximum load is 0,270 kg robotic arm
Table 1 Results of Measuring Voltage Power Supply
Load condition
Voltage
Without Load
21,5 V
With load
20,5 V
Table 2 Results of Parallel Port Voltage Measurement
Load condition
Voltage
Logical “0” (low)
0V
Lo
4,5 V
gical“1” (high)
89
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN: 978-602-71380-1-8
Kinodynamic Motion Planning for an X4-Flyer
Using a 2-Dimentional Harmonic Potential Field
Kimiko Motonaka, Keigo Watanabe, and Shoichi Maeyama
Okayama University
Okayama, Japan, 700-8530
Email: [email protected]
gradient which is calculated from the HPF. The system input
U = [U1 U2 U3 U4 ]T , which is constructed by nonholonomic
control input uc and control input ∆u based on the gradient
of the HPF, is as follows:
Abstract—In this research, we present a control method using
kinodynamic motion planning based on a harmonic potential field
(HPF) for an X4-Flyer moving in a 3-dimensional space. In the
previous research, it was confirmed that a controller using two
HPFs generated on the X-Y and X-Z planes was able to guide the
X4-Flyer to the arbitrary target point in a 3-dimensional space.
In this paper, the previous method is extended to the case where
three HPFs generated on the X-Y, X-Z, and Y-Z planes are used,
and it is verified that the X4-Flyer can move efficiently by using
the proposed method through some simulations.
I.
U = uc + ∆u.
Here, U1 is a control input for acting on each translational
motion, and U2 , U3 and U4 are control inputs for acting on
roll angle ϕ, pitch angle θ and yaw angle ψ respectively. In
the following subsections, we describe the control input based
on nonholonomic control uc and the proposed control input
∆u based on the gradient of an HPF.
I NTRODUCTION
Recently, there are many researches on the autonomous
locomotion for an X4-Flyer[1]-[3]. An X4-Flyer is a vertical
takeoff and landing (VTOL) aerial robot with four rotors, and
it has received attention in recent years as search and rescue
robots because of its highly maneuverability and hovering
ability. In traditional motion planning, dynamic constraints and
kinematic constraints are generally solved separately. In that
case, dynamic constraints are mostly solved by designing the
control input according to the result of kinematic constraints.
There is “kinodynamic motion planning” that is aimed at
solving these two constraints simultaneously, for designing the
control input from the current state[4]. Kinodynamic motion
planning can design the control input in one-step, and therefore
it has an advantage of being able to decide the control input
simply, compared to existing motion planning. There are many
techniques for kinodynamic motion planning, and kinodynamic
motion planning based on using “Harmonic potential field
(HPF)” was proposed as one of them[5][6].
A. Nonholonomic Control Input
The control input uc = [uc1 uc2 uc3 uc4 ]T is added for z
direction and three attitude angle and given as follows[7]:

mÛ1

uc1 = cos mg

ϕ cos θ − cos ϕ cos θ


uc2 = − Ilx (ϕ − ϕT ) − k1 ϕ̇
(2)
I

uc3 = − ly (θ − θT ) − k2 θ̇



uc4 = −Iz (ψ − ψT ) − k3 ψ̇.
Here, Û is :
Û = k4 (z − zT ) + k5 ż.
(3)
In these equation, k1 -k3 are positive constant gains, and zT is
an arbitrary altitude and (ϕT , θT , ψT ) are the desired angles.
In this research, a control method is proposed for guiding
an X4-Flyer to an arbitrary target point in a 3-dimensional
space. The present controller guides an X4-Flyer by appropriately switching the HPF generated in a 2-dimensional plane.
It was already confirmed in the previous research that a
controller using two HPFs generated on the X-Y and X-Z
planes can guide the X4-Flyer to the arbitrary target point
in a 3-dimensional space. In this paper, the previous method
is extended to the case where three HPFs generated on the
X-Y, X-Z, and Y-Z planes are used. Moreover, it is verified
by simulations that the X4-Flyer can move to the arbitrary
target point efficiently by using the present method based on
2-dimensional HPFs.
II.
(1)
B. Added Control Input
In this subsection, an added control input ∆u is described
for the translational motion. Kinodynamic motion planning
which is proposed in the previous research can only guide an
X4-Flyer on the X-Y plane, because it uses an HPF generated
in the X-Y plane. The proposed method can control the Z
direction of an X4-Flyer by switching an HPF including the
Z direction and an HPF in the X-Y plane.
For the control in the X-Y plane, the HPF on the X-Y
plane including the current position of the X4-Flyer is used.
At that time, using the gradient of the HPF on the X-Y plane,
an added control input ∆u is designed by

−bc · ẋ − kv ∇V

√(x) − kc · FC (x, ẋ)

if σ < (x − xT )2 + (y − yT )2
(4)
∆u =
−b
·
h(x,
ẋ) − kv ∇V (x)

d

otherwise,
K INODYNAMIC M OTION P LANNING FOR AN
X4-F LYER
In the proposed method, kinodynamic motion planning is
achieved by combining nonholonomic control input and the
90
Proceeding of International Symposium on Smart Material and Mechatronics
h(x,
[ ẋ)
= nT ẋn+
(
ISBN: 978-602-71380-1-8
Z
V (x)T
∥∇V (x)∥
Current position of an X4-Flyer
]
)
V (x)T
· ẋ · Φ(∇V (x)T ẋ) ∥∇V
(x)∥ ,
An HPF on the X-Y plane
FC (x, ẋ)
= (xT − x) · Φ(σ − |xT − x|) · Φ(ẋT (xT − x)).
Y
where bc , bd , kv and kc denote a positive constant gain,
h(x, ẋ) = [0 h(y, ẏ) h(x, ẋ) 0]T , ẋ = [0 ẏ ẋ 0]T , ∇V (x) =
[0 fy fx 0]T , and FC (x, ẋ) = [0 FC (y, ẏ) FC (x, ẋ) 0]T .
Note that, fx and fy mean the gradient of HPF parallel to the
direction of x- and y-axis respectively.
X
Fig. 1.
An HPF on the Y-Z plane
An HPF on the X-Z plane
A schema of generating 3-directional gradient using two HPFs.
For the Z-directional motion control, the HPF on the X-Z
or the Y-Z plane including the current position of the X4-Flyer
is used. In this control, the X4-Flyer moves in the Z-direction
by subtracting the Z-directional gradient fz calculated with the
HPF from the current altitude z and regarding it as the target
altitude zT in Eq. (3), so it follows that
Obstacle
40
Z [m]
Û = k4 {z − (z − fz )} + k5 ż.
50
(5)
30
20
10
C. Switching of the HPFs
50
0
0
The proposed method uses 2-dimensional HPFs for guiding
the X4-Flyer to an arbitrary target point in a 3-dimensional
space. Only two HPFs generated on X-Y and X-Z planes are
applied in the previous research, whereas three HPFs are used
in this method. Fig. 1 shows a schema of generating the X, Y-, and Z-directional gradients using three HPFs. In this
method, the HPFs are generated on the X-Y, X-Z and Y-Z
planes including the current position of the X4-Flyer as shown
in Fig. 1. By giving the X-, Y-, and Z-directional gradients
to the X4-Flyer, the controller guides the X4-Flyer in the 3dimentional space. Here, the variables fx , fy , and fz in Fig.
1 are X-, Y-, and Z-directional gradients, respectively.
10
20
40
0
50
Y [m]
X [m]
Fig. 2.
The assumed field.
50
45
40
Current position
(45, 25, 10)
50
35
Y [m]
Z [m]
40
30
20
30
25
20
15
10
50
10
40
0
0
As an illustration, consider the case where the X4-Flyer
moves from the current position (x, y, z) = (45, 25, 10) to the
target position (xT , yT , zT ) = (5, 5, 10) in the environment
shown in Fig. 2. In this case, the HPF on the X-Y plane
including the current altitude z = 10 (see Fig. 3 (a)), the HPF
on the X-Z plane including the current y-position y = 25 (see
Fig. 4 (a)), and the HPF on the Y-Z plane including the current
x-position x = 45 (see Fig. 5 (a)) are generated. In order to
generate the HPFs, (xT , yT ) = (5, 5) is set as a target point
in Fig. 3, (xT , zT ) = (5, 10) is set as a target point in Fig. 4,
and (yT , zT ) = (5, 10) is set as a target point in Fig. 5. Then,
by using the gradient of the current position generated as in
Fig. 3 (b), Fig. 4 (b), and Fig. 5 (b), the controller guides the
X4-Flyer toward the target point.
30
10
5
20
20
X [m]
30
40
10
Y [m]
50 0
0
0
10
20
30
X [m]
40
50
(a) X-Y plane including the (b) The gradient of the HPF
on the X-Y plane
current position
Fig. 3.
The HPF on the X-Y plane.
50
45
40
50
Z [m]
40
30
Z [m]
35
Current position
(45, 25, 10)
30
25
20
20
15
10
0
0
III.
30
S IMULATIONS
40
10
5
20
10
20
30
X [m]
40
50 0
Y [m]
0
0
10
20
30
40
50
X [m]
(a) X-Z plane including the (b) The gradient of the HPF
on the X-Z plane.
current position.
In this section, it is confirmed that the proposed method
can guide the X4-Flyer 3-dimensionally in the environment
shown in Fig. 2, by simulations in MATLAB. Additionally,
the proposed method is compared with the previous method
to verify that the proposed method is more efficient than the
previous one.
Fig. 4.
91
The HPF on the X-Z plane.
ISBN: 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
50
45
40
50
35
Z [m]
Current position
(45, 25, 10)
40
Z [m]
30
30
25
20
20
15
10
0
0
50
10
5
10
20
X [m]
Y [m]
30
40
50
0
0
0
10
20
30
40
50
Y [m]
(a) Y-Z plane including the (b) The gradient of the HPF
on the Y-Z plane
current position
Fig. 6.
Fig. 5.
The X4-Flyer.
The HPF on the X-Y plane.
A. Conditions
TABLE I.
In this simulation, it is assumed that the X4-Flyer moves
from the initial position (x, y, z) = (45, 25, 10) to the target
position (xT , yT , zT ) = (5, 5, 10) in the environment including
an obstacle (shown in Fig. 2). The target attitude is always set
as (ϕT , θT , ψT ) = (0, 0, 0), and the following equation is used
as a dynamical model of the X4-Flyer[7]:

1
U1
ẍ = (cos ϕ sin θ cos ψ + sin ϕ sin ψ) m






1


U1
ÿ = (cos ϕ sin θ sin ψ + sin ϕ cos ψ) m






1

U1
 z̈ = −g + (cos ϕ cos θ) m


ϕ̈









θ̈







ϕ̈
= θ̇ψ̇(
Iy −Iz
Ix )
−
x
= ϕ̇ψ̇( IzI−I
)−
y
= ϕ̇θ̇(
Ix −Iy
Iz )
+
Jr
Ix θ̇Ω
+
l
Ix U2
Jr
Iy ϕ̇Ω
+
l
Iy U3
M ODEL PARAMETERS FOR
Parameter
g
m
l
Ix
Iy
Iz
Jr
b
d
Description
Gravity
Mass
Distance
Roll inertia
Pitch inertia
Yaw inertia
Rotor inertia
Thrust factor
Drag factor
THE
X4-F LYER .
Value
9.80665
1.3
0.248
0.01467
0.01467
0.02331
175.69×10−6
0.0000434
0.000002188
Unit
m/s2
kg
m
kg·m2
kg·m2
kg·m2
kg·m2
(6)
C ONTROL GAINS FOR THE X4-F LYER .
TABLE II.
Gain
k1
k3
k5
bc
kc
1
Iz U4 .
In this equation, let us define m [kg] as the mass of the
X4-Flyer, l [m] as the length from the center of airframe to
the center of rotor, g [m/s2 ] as the gravity acceleration, Ix ,
Iy and Iz [kg/m2 ] as the inertial moment around each axis
respectively, and Jr [kg/m2 ] as the inertial moment of a rotor.
Here, U1 is the control input for acting on each translational
motion, and U2 , U3 and U4 are the control inputs for acting on
roll motion, pitch motion and yaw motion respectively. Then,
Ω and the system’s inputs U1 , U2 , U3 , U4 can be written by
using the rotational speed ωi of the rotor i (i=1,...,4), i.e.,

U1 = b(ω12 + ω22 + ω32 + ω42 )



 U2 = b(ω42 − ω22 )
(7)
U3 = b(ω32 − ω12 )

2
2
2
2

U
=
d(ω
+
ω
−
ω
−
ω
)
 4
2
4
1
3

Ω = ω2 + ω4 − ω1 − ω3 .
Value
0.015
0.007
25.0
0.006
0.002
Gain
k2
k4
kv
bd
σ
Value
0.015
10.0
0.001
0.002
10
50
Z [m]
40
Here b is the thrust coefficient and d is the drag coefficient.
The parameters used for this simulation are shown in Table
I, and they are obtained from the X4-Flyer developed in our
laboratory (shown in Fig. 6). The gains are set as in Table II,
from a rule of thumb.
30
20
10
50
0
0
10
20
30
X [m]
B. Results
Fig. 7.
Figs. 7-10 show the simulation results. The solid red line
drawn in Fig. 7 denotes the trajectory of the X4-Flyer when
92
The trajectory of the X4-Flyer.
40
50
0
Y [m]
ISBN: 978-602-71380-1-8
Distance between the current position
and the target position [m]
Proceeding of International Symposium on Smart Material and Mechatronics
Fig. 8.
using the proposed method, and the dotted blue line denotes the
trajectory when using the previous method. Fig. 8 shows the
change of the Euclidean distance between the current position
and the target position, and Fig. 9 and Fig. 10 show the
change of the attitude angles (ϕ, θ, ψ) when using the proposed
method and the previous method respectively.
C. Discussions
25
proposed method
previous method
20
15
It is found from Fig. 7 that the X4-Flyer was able to
reach the target point from the initial state while avoiding
the obstacle in the 3-dimensional environment by using the
proposed method or the previous method. Moreover, the X4Flyer was able to move 3-dimensionally by using the gradient
of HPFs in spite of only using the 2-dimensional HPFs.
However, the trajectory when using the previous method takes
a long detour to upside of the obstacle. On the other hand,
the trajectory when using the proposed method takes a shorter
route from the side of the obstacle and it seems that an efficient
trajectory was chosen.
10
5
0
0
50
100
150
200
250
300
Time [s]
Distance from the target position.
As shown in Fig. 8, the X4-Flyer was able to keep its
attitude on the target point after reaching the target point
even if which controller was used. However, the proposed
method was able to guide the X4-Flyer to the target point
about two times faster than the previous one. Small steady-state
errors remained after the X4-Flyer reached the target point,
irrespective of the control method adopted.
0.05
φ
θ
ψ
0.04
Angles [rad]
0.03
All the attitude angles of the X4-Flyer always fell within ±
0.04 [rad] as shown in Figs. 9 and 10. This result is acceptable
realistically, because the X4-Flyer assumed in this simulation
was able to easily cope with the oscillation of about ± 0.2
[rad]. From these results, it is confirmed that the proposed
method can choose a more efficient route and guide the X4Flyer to the target point faster than the previous method.
0.02
0.01
0
−0.01
−0.02
−0.03
IV.
−0.04
−0.05
0
50
100
150
200
250
In this paper, a method for guiding the X4-Flyer to an
arbitrary target point in a 3-dimensional space is proposed by
switching the HPFs, which are generated on 2-dimensional
planes. The usability of the proposed method was confirmed
in simulations by comparing it with the previous method.
In future works, we are going to implement the proposed
method to an actual machine, and make experiments in actual
environments.
300
Time [s]
Fig. 9.
Angles of the X4-Flyer using the previous controller.
R EFERENCES
0.05
φ
θ
ψ
0.04
0.03
[1] Y. Bouktir, M. Haddad, T. Chettibi, “Trajectory planning for a quadrotor
helicopter,” in Proc. of 16th Mediterranean Conf. on Control and
Automation, pp. 1258–1263, Jun. 2008.
[2] P. Bristeau, F. Callou, D. Vissiere and N. Petit, “The navigation and
control technology inside the AR.Drone micro UAV,” in Preprints of the
18th IFAC World Congress, pp. 1477–1484, 2011.
[3] Y. Suh, “Robust control of a quad-rotor aerial vehicle,” Int. J. of Applied
Electromagnetics and Mechanics, vol. 18, pp. 103–114, 2003.
[4] B. Donald, P. Xavier, J. canny, and J. Reif, “Kinodynamic motion
planning,” J. of the ACM, vol. 40, no. 5, pp. 1048–1066, 1993.
[5] A. Masoud, “Kinodynamic motion planning,” IEEE Robotics & Automation Magazine, vol. 17, no. 1, pp. 85–99, 2010.
[6] K. Motonaka, K. Watanabe, and S. Maeyama, “Offline Optimization of
Gains for Kinodynamic Motion Planning for X4-Flyer,” SICE Annual
Conference, 2013.
[7] B. Samir, M. Pierpaolo and S. Roland, “Design and control of an indoor
micro quadrotor,” IEEE Robotics and Automation, Vol. 5, pp. 4393–4398,
2004.
Angle [rad]
0.02
0.01
0
−0.01
−0.02
−0.03
−0.04
−0.05
0
50
100
150
200
250
300
Time [s]
Fig. 10.
C ONCLUSIONS
Angles of the X4-Flyer using the proposed controller.
93
ISBN: 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
The Stabilization of Position and Attitude
for a Blimp by a Switching Controller
Yoshikazu Nakamura, Keigo Watanabe and Isaku Nagai
Department of Intelligent Mechanical Systems
Graduate School of Natural Science and Technology, Okayama University
Okayama, Japan
[email protected]; [email protected]; [email protected]
Abstract—In recent years, the development of unmanned air
vehicles aiming at vegetation observation, information gathering
of a disaster site, etc. is increasing. Among them, airships are
attractive because of good energy efficiency and it is possible to be
employed for a long time cruise. Especially, small airships called
“blimp” have been developing to make the management easy.
Although most of existing airships employ control methods by
combining propellers and rudders, such a control approach has
the problem that the maneuverability is deteriorated if their
traveling speed is slow because the airflow received by rudders is
weakened. In this research, “X4-Blimp” is proposed as a blimp
controlled by only four propellers without any rudders, and it is
controlled by a switching controller.
Index Terms—X4-Blimp, underactuated control, switching
control.
I. INTRODUCTION
In recent years, unmanned aircrafts are expected to play an
important role in observing vegetation and gathering
information on disaster sites etc.[1] where it is hard for human
to enter. Especially, airships that can float by its own buoyancy
are attractive for good energy efficiency to travel for long time.
However, a big airship requires a wide space and cost for
maintenance. Thus, small airships which are called “blimp”
have been developed, because it is easy to maintain and use it.
Most of existing airships have propellers and rudders for
controlling them. In this control method, the airframe is
controlled by the rudders, using the airflow flowing on its
surface. Such a method has a problem that if the traveling speed
is slowed down, then the operability is deteriorated because of
the weak airflow. Thus, it is desired to develop a blimp
controlled without using rudders.
In this research, a controller method is proposed for an “X4Blimp” where the airframe is controlled by only four propellers
without any rudders. Since the X4-Blimp can control the
positions and attitudes in three-dimensional space by regulating
the output of the propellers, it can realize high operability,
irrespective of its traveling speed. However, it is not easy to
control the X4-Blimp, because it is an underactuated system.
From an actual experiment, we have found that it was hard for
a conventional X4-Blimp [2], in which the envelope is placed
at the upper part of the airframe whereas the gondola is placed
at the lower part of the airframe to fly downward.
Fig. 1. Definition of the coordinates
When the airframe is inclined, the righting moment prevents
the X4-Blinp from being controlled, because the conventional
X4-Blimp has the center of gravity and the center of buoyancy
in different point. This paper proposes a new X4-Blimp, which
is symmetric in structure. The new X4-Blmip can fly stable
because the new X4-Blmip has the center of gravity and the
center of buoyancy in the same point. A method for controlling
the X4-Blimp by switching two controllers is adopted, one of
which is designed by using a model that includes nonlinear
parts or a model that only includes linear parts, where those are
separated from the derived dynamical model. The effectiveness
of the proposed method is verified by some simulations.
II. OVERVIEW OF THE X4-BLIMP
A. Structure of the X4-Blmip
The X4-Blimp proposed in this research is composed of
envelopes, a gondola and propellers as shown in Fig. 1. The
envelopes is filled with helium gas to balance airframe mass
with the buoyancy. The envelope form is a spheroid to decrease
air resistance for traveling direction. The gondola includes
batteries and controllers, and it is placed on the center of the
airframe. The gondola form is a rectangular solid to maintain the
space for the controllers etc. and simplify a calculation of the
moment of inertia. The four propellers are attached on up, down,
left and right sides of the gondola with the same distance from
the center of the gondola. This airframe is designed
symmetrically at a point C so as to be controlled easily.
94
ISBN: 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
B. Definition of the coordinates
A definition of coordinates is shown in Fig. 1, and the robot
coordinate C is difined such that the origin is the center of the
gondola, positive X-axis is set as the forward direction of the
airframe, positive Y-axis is set as the right direction of the
airframe, and Z-axis is set to be downward perpendicular to the
airframe. Similarly, the world coordinate E is a right-handed
coordinate where positive z-axis is set to be vertically downward.
The center position of the gondola is represented by 𝝃 =
[𝑥, 𝑦, 𝑧]𝑇 in the world coordinate, and the rotational angles for
roll, pitch, and yaw in the robot coordinate system are
represented as 𝜙, 𝜃 and 𝜓 respectively, then the sttitude of the
gondola is represented by 𝜼 = [𝜙, 𝜃, 𝜓]𝑇 . A rotation matrix 𝑹
to transform the robot coordinate to the world coordinate is
derived as follows:
𝑐𝜃𝑐𝜓 𝑠𝜙𝑠𝜃𝑐𝜓 − 𝑐𝜙𝑠𝜓 𝑐𝜙𝑠𝜃𝑐𝜓 + 𝑠𝜙𝑠𝜓
𝑹 = [𝑐𝜃𝑠𝜓 𝑠𝜙𝑠𝜃𝑠𝜓 + 𝑐𝜙𝑐𝜓 𝑐𝜙𝑠𝜃𝑠𝜓 − 𝑠𝜙𝑐𝜓]
(1)
−𝑠𝜃
𝑠𝜙𝑐𝜃
𝑠𝜙𝑐𝜃
where 𝑐𝐴 is cos 𝐴 and 𝑠𝐴 is sin 𝐴.
III. DERIVATION OF DYNAMICAL MODEL
A dynamical model of the X4-Blimp is derived by referring
to X4-AUV studied in Watanabe et al. [2], the dynamical model
of the X4-Blimp is derived as
𝑥̈ = (cos 𝜃 cos 𝜓 𝑢1 )/𝑚
𝑦̈ = (cos 𝜃 sin 𝜓𝑢1 )⁄𝑚
𝑧̈ = (− sin 𝜃𝑢1 ) ⁄𝑚
(2)
𝜙̈ = (𝜃̇𝜓̇(𝐼𝑌 − 𝐼𝑍 ) + 𝑢2 )⁄𝐼𝑋
𝜃̈ = (𝜙̇𝜓̇(𝐼𝑍 − 𝐼𝑋 ) − 𝐽𝑝 𝜓̇Ω + 𝑙𝑢3 )⁄𝐼𝑌
{𝜓̈ = (𝜙̇𝜃̇(𝐼𝑋 − 𝐼𝑌 ) + 𝐽𝑝 𝜃̇ Ω + 𝑙𝑢4 )⁄𝐼𝑍
Fig. 2. Concept of the proposed controller
𝑥̈ = 𝜔1
𝑦̈ = tan 𝜓 𝜔1
𝑧̈ = − tan 𝜃 sec 𝜓 𝜔1
𝜙̈ = 𝜔2
𝜃̈ = 𝜔3
𝜓̈ = 𝜔4
Then, the inputs are transformed as follows
𝜔1 = cos 𝜃 cos 𝜓 𝑢1⁄𝑚
𝜔2 = (𝜃̇𝜓̇(𝐼𝑌 − 𝐼𝑍 ) + 𝑢2 )⁄𝐼𝑋
𝜔3 = (𝜙̇𝜓̇(𝐼𝑍 − 𝐼𝑥 ) − 𝐽𝑝 𝜓̇Ω + 𝑙𝑢3 )⁄𝐼𝑌
𝜔4 = (𝜙̇𝜃̇(𝐼𝑋 − 𝐼𝑌 ) + 𝐽𝑝 𝜃̇Ω + 𝑙𝑢4 )⁄𝐼𝑍
(7)
(8)
(9)
(10)
(11)
The partial underactuated controller 1 is designed from a 2input/6-state partial model for x, 𝜓 and y, and from a 2-input/4state partial model for 𝜙 and 𝜃 . The partial underactuated
controller 2 is designed from a 2-input/6-state partial model for
x, 𝜃 and z, and from a 2-input/4-state partial model for 𝜙 and 𝜓.
When a chained form transformation in [4] is applied, the 2input/6-statepartial model for x, 𝜓 and y is denoted by
𝑧11 = ℎ1 = 𝑥
(12)
𝑧12 = 𝐿𝑓 ℎ1 = 𝑥̇
(13)
𝑧21 = 𝐿𝑔1 𝐿𝑓 ℎ2 = tan 𝜓
(14)
2
̇
⁄
𝑧22 = 𝐿𝑓 𝐿𝑔 𝐿𝑓 ℎ2 = 𝜓 cos 𝜓
(15)
where the mass of the airframe is 𝑚, the moment of inertia for
each axis is represented by 𝐼𝑋 , 𝐼𝑌 and 𝐼𝑍 respectively, the
moment of inertia of the propellers is 𝐽𝑝 and Ω = 𝜔2 + 𝜔4 −
𝜔1 − 𝜔3 . When four propellers are numbered from 1 to 4 in the
clockwise from the upper propeller and the direction of
rotational velocity of each propeller is positive if it is defined as
clockwise. And the input 𝑢1 of translational motion, the input 𝑢2
of roll motion, the input 𝑢3 of pitch motion and the input 𝑢4 of
yaw motion are represented by
(3)
u1 = 𝑏(𝜔12 + 𝜔22 + 𝜔32 + 𝜔42 )
u2 = 𝑑(−𝜔22 − 𝜔42 + 𝜔12 + 𝜔32 )
(4)
u3 = 𝑏(𝜔12 − 𝜔32 )
(5)
u4 = 𝑏(𝜔22 − 𝜔42 )
(6)
where the thrust coefficient is b and the resistance coefficient is
d.
1
𝑧31 = ℎ2 = 𝑦
(16)
𝑧32 = 𝐿𝑓 ℎ2 = 𝑦̇
(17)
Then, the inputs are transformed as follows
𝑣1 = 𝑤1
(18)
2 tan 𝜓 2
1
𝑣2 =
𝑤 +
𝜓̇
(19)
cos 2 𝜓 4 cos 2 𝜓
From the above results, a chained form is derived by
𝑧̈11 = 𝑣1
𝑧̈21 = 𝑣2
(20)
𝑧̈31 = 𝑧21 𝑣1
To apply a method in Xu and Ma [3] to Eq. (20), it is rewritten
for state variables such as
𝑥̇1 = 𝑥2 , 𝑥̇ 2 = 𝑣1
𝑥̇ 3 = 𝑥4 , 𝑥̇ 4 = 𝑣2
𝑥̇ 5 = 𝑥6 , 𝑥̇ 6 = 𝑥3 𝑣1
Then the control input 𝑣1 is denoted by
𝑣1 = −(𝑠1 + 𝑠2 )𝑥2 − 𝑠1 𝑠2 𝑥1
(21)
where s2 > 𝑠1 > 0 . To control the underactuated system, a
coordinate transformation is performed to design a controller
based on a discontinuous model:
IV. DESIGN OF PARTIAL UNDERACTUATED CONTROLLERS
Since the system of the X4-Blimp represented by the
dynamical model of Eq. (2) is an underactuated system with four
inputs and 12 states, it is different to realize underatuated
control .As shown in Fig. 2, tow partial underactuated
controllers for a model with 4 inputs 10 states are designed by
combining a controller for a 2-input/4-state partial model with a
controller for a 2-input/6-state partial model. The whole system
is controlled by switching these two partial underactuated
controllers. To perform a chained form transformation, the
dynamic model is partially linearized such that
95
Proceeding of International Symposium on Smart Material and Mechatronics
𝑧i = 𝑥i (𝑖 = 1,2,3,4),
𝑧𝑖 =
𝑥𝑖
(𝑖 = 5,6)
𝑥1
ISBN: 978-602-71380-1-8
(22)
The Eq. (22) is rewritten as follows
𝑧̇1 = 𝑧2
(23)
)𝑧
𝑧̇2 = −(𝑠1 + 𝑠2 2 − 𝑠1 𝑠2 𝑧1
(24)
𝑍̇3−6 = (𝐴1 + 𝐴2 (𝑡))𝑍3−6 + 𝐵𝑣2
(25)
where 𝑍3−6 = [𝑧3 , 𝑧4 , 𝑧5 , 𝑧6 ]𝑇 . Here, 𝐴1 , 𝐴2 (𝑡) and B are
denoted by
0 1 0 0
0
0 0 0 0
𝐴1 = [ 0
] , 𝐵 = [ 1]
0
0 𝑠1 1
𝑠12 0 0 𝑠1
0
0
0 0
0
0
0
0
0]
𝐴2 (𝑡) = 𝐶 [
0
0 −1 0
−(𝑠1 + 𝑠2 ) 0 0 −1
where 𝐶 = 𝑧𝑧21 + 𝑠1 . The controllability of [𝐴1 , 𝐵] is confirmed.
A controllable matrix is represented as [𝐵 𝐴1 𝐵 𝐴12 𝐵 𝐴13 𝐵]. It is
regular because 𝑠1 > 0 . Since 𝐴1 + 𝐵𝐿 is controllable, the
feedback gain 𝐿 = [𝑙1 , 𝑙2 , 𝑙3 , 𝑙4 ] is calculated to make matrix
𝐴1 + 𝐵𝐿 as the Hurwitz matrix by the pole placement method.
The control input 𝑣2 is denoted by
𝑣2 = 𝐿𝑍3−6 = 𝑙1 𝑧3 + 𝑙2 𝑧4 + 𝑙3 𝑧5 + 𝑙4 𝑧6
(26)
Thus, since it can be stabilized to the origin, the control input for
the chained form are derived as follows
𝑣1 = −(𝑠1 + 𝑠2 )𝑥̇ − 𝑠1 𝑠2 𝑥
(27)
̇
𝜓
𝑦
𝑦̇
𝑣2 = 𝑙1 tan 𝜓 + l2
+ 𝑙3 + 𝑙4
(28)
cos 2 𝜓
𝑥
𝑥
In this way, the controller for 2-input/6-state partial model for x,
𝜓 and y is designed. Next, the controller for the 2-input/6-state
partial model for 𝜙 and 𝜃 is designed by a linear feedback such
as
𝑤2 = −𝑘1 𝜙 − 𝑘2 𝜙̇ (𝑘1 , 𝑘2 > 0)
(29)
𝑤3 = −𝑘3 𝜃 − 𝑘4 𝜃̇ (𝑘3 , 𝑘4 > 0)
(30)
The partial underactuated controller 1 for a model with 4 input
and 10 state is designed by combining the controllers for x, 𝜓
and y with the controller for 𝜙 and 𝜃.
Similarly, the partial underactuated controller 2 is designed
by combining the controller for the 2-input/6-state partial model
for x, 𝜃 and z with the controller for the 2-input/4-state partial
model for 𝜙 and 𝜓. When the partial model for x, 𝜃 and z is
transformed to a chained form, the input transformation is
denoted by
𝑣1 = −(𝑠1 + 𝑠2 )𝑥̇ − 𝑠1 𝑠2 𝑥
tan 𝜃
𝜃̇
𝑧
𝑧̇
𝑣2 = 𝑙1 (−
) + 𝑙2 (−
) + 𝑙3 + 𝑙4
cos 𝜓
cos 𝜓 cos 2 𝜃
𝑥
𝑥
The control inputs based on the chained form transformation is
denoted by
𝑤1 = 𝑣1
(31)
2
2
̇
𝑤2 = − cos 𝜓 cos 𝜃 ∙ 𝑣2 − 2 tan 𝜃 ∙ 𝜃
(32)
The 2-input/4-state partial model for 𝜙 and 𝜓 is derived by a
linear feedback such as
𝑤2 = −𝑘1 𝜙 − 𝑘2 𝜙̇ (𝑘1 , 𝑘2 > 0)
(33)
𝑤4 = −𝑘3 𝜓 − 𝑘4 𝜓̇ (𝑘3 , 𝑘4 > 0)
(34)
The partial underactuated controller 2 for the model with 4
inputs and 10 states is designed by combining the controller for
x, 𝜃 and z with the controller for 𝜙 and 𝜓.
Fig. 3. Structure of energy regions
V. ENERGY REGION BASED SWITCHING METHOD
Switching the two partial underactuated controllers for 4
inputs 10 states is considered to control an underactuated system
with 4 inputs 12 states. However, if input chattering phenomena
occur when controllers are switched, an excessive burden is
placed on motors. Therefore, a switching method[5] that has
multiple boundary regions is used to prevent the chattering
phenomena.
The energy is defined from the errors of generalized
coordinates. Since the state x is doubly generated from the set of
(x, 𝜓, y) and (x, 𝜃, z), and similarly the corresponding attitude
angle 𝜙 is also doubly generated from the set of (𝜙, 𝜃) and (𝜙,
𝜓 ), the errors for the stabilization to the origin are directly
represented by 𝜓, y, 𝜃 and z because both partial underactuated
controllers always stabilize the state x and the angle 𝜙 to the
origin. Then, the energy based on the errors is defined as
follows:
𝐸1 = 𝜓 2 + 𝑦 2
(35)
𝐸2 = 𝜃 2 + 𝑧 2
(36)
In Fig. 3, a two–dimensional plane is represented by 𝐸1 and 𝐸2 ,
and hysteresis like boundary lines 𝜋1 and 𝜋2 to separate the
energy plane are represented respectively by
𝜋1 (𝐸1 ) = 1 − 𝑒 −√𝐸1
(37)
𝜋2 (𝐸1 ) = 2𝜋1
(38)
In Fig. 3, the partial underactuated controller 1 is used on the
region 𝑅1 , whereas the partial underactuated controller 2 is used
on the region 𝑅2 . Considering an overlapped region, switching
rules are decided as follows:
Rule 1:
If 0 < 𝐸2 ≤ 𝜋1 (𝐸1 ) then 𝑠𝑡 = 𝑦
Rule 2
If 𝜋1 (𝐸1 ) < 𝐸2 < 𝜋2 (𝐸1 ) and 𝑠𝑡−1 = 𝑦 then 𝑠𝑡 = 𝑦
Rule 3:
If 𝜋1 (𝐸1 ) < 𝐸2 < 𝜋2 (𝐸1 ) and 𝑠𝑡−1 = 𝑧 then 𝑠𝑡 = 𝑧
Rule 4:
If 𝜋2 (𝐸1 ) < 𝐸2 then 𝑠𝑡 = 𝑧
Where st represents the controller used for each rule. When
𝑠𝑡 = 𝑦, the partial underactuated controller 1 is used, whereas
when 𝑠𝑡 = 𝑧, the partial underactuated controller 2 is used. st−1
represents the controller used before one-sampling time.
According to this switching rule, the partial underactuated
controller 2 is used to control the state z. Similarly, the partial
underactuated controller1 is used to control the state y. It should
be noted that, in this switching rule, the chattering phenomena
96
ISBN: 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
TABLE I. PARAMETERS FOR THE X4-BLIMP
Parameter
m
l
𝐼𝑋
𝐼𝑌
𝐼𝑍
Description
Mass
Distance
Roll Inertia
Pitch Inertia
Yaw Inertia
Value
0.8
0.50
1.10
1.43
1.43
Unit
kg
m
kg ∙ m2
kg ∙ m2
kg ∙ m2
Fig. 5. Controlled attitudes
Fig. 4. Controlled positions
are unlikely to occur because an overlapped region between the
boundary lines 𝜋1 and 𝜋2 exists to switch the controllers.
VI. SIMULATION
This simulation is intended to verify that the state variables
related to the position and attitude of the airframe converge to
the origin by switching the two partial underactuated controllers
using the switching rules created in previous section. The initial
state of X4-Blimp is 𝒒𝟎 = [−10.0, 0.5, 1.0, 𝜋⁄18 , 𝜋⁄9 , 𝜋⁄4]𝑇 ,
and the goal state is 𝒒𝒓 = [0, 0, 0, 0, 0, 0]𝑇 . The physical
parameters used for simulation are shown in Table 1. The
feedback gains 𝑘1 = 0.8, 𝑘2 = 1.2, 𝑘3 = 0.6, 𝑘4 = 0.7, 𝑠1 =
1⁄100 , 𝑠2 = 0.45, 𝑙1 = −0.005, 𝑙2 = −0.37, 𝑙3 = −0.80, and
𝑙4 = −35.1 are for the partial underactuated controller 1,
whereas the feedback gains 𝑘1 = 0.8, 𝑘2 = 1.2, 𝑘3 = 0.6, 𝑘4 =
0.7, 𝑠1 = 1⁄100 , 𝑠2 = 0.45, 𝑙1 = −0.02, 𝑙2 = −0.25, 𝑙3 =
−0.14 and 𝑙4 = −10.08 are for the partial underactuated
controller 2.
It is found from Fig. 4 that the positions, i.e., the states x, y
and z converge from the initial positions to the goal positions.
Similarly, it is seen from Fig. 5 that all the attitudes 𝜙, 𝜃 and 𝜓
converge to the desired angles. Fig. 6 shows the energy
trajectory, where it starts from the point S. It is found that the
controller 2 was switched to the controller1 at the point P and
the energy finally converges to the origin at the point G.
Switching of controllers occurs at the point P and the state
variables are changed suddenly, if the energy trajectory exceeds
the boundary line 𝜋1 . Thus, it is confirmed that the positions and
attitudes of the X4-Blimp can be stabilized by switching the two
partial underactuated controllers.
Fig. 6. Energy trajectory
designed from the derived dynamic model, and switching rules
for switching two such controllers were constructed by
applying the conventional logical rules based on hysteresis-like
switching boundaries. The effectiveness of the proposed
method was checked by simulations. For future work, we will
apply this approach to a level flight for an X4 tail-sitter.
REFERENCES
[1] K. Kawabata, Y. Hada, and H. Asama, “Robotics research related
to lighter than air aircraft,” Journal of Robotics Society of Japan,
2004, pp. 901–905.
[2] Y. Nakamura, K. Watanabe, and I. Nagai, “Underactuated control
for a blimp with four-propellers by a logical switching method,”
Proc. of the 18th Int. Symposium on Artificial Life and Robotics
(AROB 18th '13), Daejeon, Korea, January 30-February 1, 2013,
pp. 69–72.
[3] K. Watanabe, K. Izumi, K. Okamura, and S. Rafiuddin,
“Discoutinuous underactuated control for lateral X4 autonomous
underwater vehicles,” Proc. of the 2nd International Conference
on Underwater System Technology: Theory and Applications,
2008, Paper ID 14.
[4] WL. Xu and BL. Ma, “Stabilization of second-order
nonholonomic system in canonical chained form,” Robotics and
Autonomous Systems 34, 2001, pp. 223–233.
[5] MC. Laiou and A. Astolfi, “Local transformations to generalized
chained forms,” Proc. of the 16th International Symposium on
Mathematical Theory of Networks and Systems, 2004.
[6] JP. Hespanha and A.S. Morse, “Stabilization of nonholonomic
intergrators via logic-based switching,” Automatica 35, 1999, pp.
385–393.
VII. CONCLUSION
In this paper, an underactuated controller has been proposed
for stabilizing an X4-Blimp whose structure is symmetric at a
point, where two partial underactuated controllers were
97
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
New Waste Beverage Cans Identification Method
Firmansyah Burlian, Yulia Resti*, Ihsan Budiman
Dept. of Mechanical Engineering, Faculty of Engineering, Sriwijaya University
Palembang, Indonesia
[email protected]
*Dept. of Mathematics , Faculty of Math. And Science, Sriwijaya University
Palembang, Indonesia
[email protected]
Abstract— The primary emphasis of this work is on the
development of a new waste beverage cans identification method
for automated beverage cans sorting systems known as the SVS
system. The method described involved window-based
subdivision of the image into X-cells, construction of X-candidate
template for N-cells, calculation of matching scores of reference
templates for the N-cells image, and application of matching
score to identify the grade of the object. The SVS system
performance for correct beverage cans grade identification is
95.17% with estimated throughput of 21,600 objects per hour
with a conveyor belt width of 18˝. The weight of the throughput
depends on the size and type of the objects.
Automated sorting systems are classified into mechanical
and optical systems. Since 1932 to 2009, different mechanical
and optical sorting methods have been developed to fill the
demand of object sorting. Mechanical sorting cannot achieve
commercially viable throughputs and accuracy. The popularity
of optical sorting systems has increased because of inadequate
throughput of mechanical sorting systems. The greatest
advantages of optical sorting systems include the following:
consistent and reliable production efficiency with a relatively
high hit rate and purity; and low operational cost because of
fewer manual workers on the production line.
The main objective of the research is to develop a smart
vision sensing (SVS) system for automatic recyclable waste
beverage cans sorting. More specifically, the aims is To select
the best features and classifier for the smart vision sensing
(SVS) system for automatic recyclable solid waste sorting.
Index Terms— Identification, SVS, beverage cans, sorting
system
I. INTRODUCTION
Computer vision (CV) deals with extracting meaningful
descriptions of physical objects from images (Ballard & Brown
1982, Brosnan & Da-WenSun 2003, and Kulkarni 2001). Due
to low cost powerful solutions, the applications of CV have
increased tremendously in diverse fields such as medical
diagnostic imaging, food industry for quality evaluation,
factory automation, robot vision, object identification, military
reconnaissance, remote sensing, mineral exploration,
cartography, and automated object grading and sorting. The
aim of this motivation is to realize the necessity of the
automated solid waste sorting system and justify the
development of a smart vision sensing (SVS) system for
automated recyclable waste beverage cans sorting using stateof-the-art of the CV.
The primary challenge in the recycling of beverage cans is
to obtain raw material with the highest purity. In recycling,
highly sorted stream facilitates high quality end product, and
save processing chemicals and energy because various grades
of beverage cans are subjected to different recycling processes.
In addition, the amount of sludge and rejects generated in
recycling processes is decreased for the utilization of sorted
object in recycling as well as reduces the amount of energy
needed to produce recycled cans. In this work, the type of a
beverage cans is based on weight, color, usage, raw material or
a combination of these factors.
II. THE SVS SYSTEM FOR SOLID WASTE SORTING
Figure 1 illustrates the block diagram of the proposed
intelligent computer vision system for automatic sorting of
recyclable beverage cans and a picture of the actual systems.
The computer vision process consists of three parts: perception,
cognition and action. The perception or image acquisition
portion of this vision system consists of a commercially
available webcam and a special lighting arrangement. The
main responsibility of the action component of the vision
system is to segregate waste beverage cans into different types
based on the command of the cognition part of the vision
system. Mechanical system are used to segregate and to pile
different type of object according to their respective waste bins.
In this research, we emphasize a beverage cans type
identification system, which covers the perception and
cognition components of the proposed system.
In this proposed system, 640480 RGB images are
captured from inspection zone on the conveyor belt by using
Logitech QuickCam Pro 4000 Web Camera [46], [47]. The
specifications of the webcam are CCD Optical sensor, color
Camera, CCD Image Sensor Lens Construction support
Manual Focus Adjustment, 4 pin USB Type A Interface for
98
Proceeding of International Symposium on Smart Material and Mechatronics
USB Expansion / Connectivity with Computer Interface. In
webcam properties setting, the brightness, contrast and
saturation are adjusted at 50%, 50% and 100% of their
respective scales.
.
ISBN 978-602-71380-1-8
provided any improvement over RGB. Thus, the RGB color
space is considered in this research. For color images, each of
the three color components – red, green and blue – are
considered separately. For gray scale image, standard grayscale
transformation is obtained from the original RGB image.
Identification is primarily based on the dominating color level
of the objects. In the feature selection process, special emphasis
is placed on those features, which provides significant
information regarding the dominant color level. Initially,
seventeen first order features, such as mean, standard deviation,
skewness, kurtosis, dispersion, lowest color level, highest color
level, mode of the color level, entropy, energy, lower quartile,
upper quartile, histogram tail length on dark side, histogram tail
length on light side, median color level, range of the color
level, and inter-quartile range, are extracted from the image
using equations to determine the significant features in
identification.
To calculate the first order features, the gray level
histogram of the image is calculated first. The histogram, h(x),
is a one dimensional array that represents the number of pixels
in the image with a gray level of x. The x parameter can take
any value between 0 and Z-1, where Z is the number of gray
levels in the image. For color images, three histograms are
calculated for the three color components: red, green and blue.
Figure 1 Block diagram of the intelligent computer vision
system for automatic sorting of Beverage Cans
In this experiment, it is observed that the performance of
the vision system is extremely influenced by the lighting
arrangement. For calibration and adjusting the lighting, three
different lighting techniques namely front lighting-directionalbright field illumination, front lighting-directional- dark field
illumination, and diffuse front lighting are considered in this
research as shown in Figure 2. In both front lightingdirectional-bright field illumination and diffuse front lighting,
the images from the inspection zone show some reflection
problems such that the object on the conveyor. Moreover, the
reflection from the surface of the object is not uniform. It is
important that the texture information of the objects is
analyzed. Even one object of the same color in whole body
showed different color combination in histogram analysis of
the segmented portions of the image due to non-uniform
lighting. In front lighting-directional-darkfield illumination,
image from inspection zone is distinctive for texture analysis
and the object surface is illuminated uniformly. Moreover,
front lighting-directional-darkfield illumination is widely used
in surface scratches or texture analysis (Pham D.T., & Alcock,
R. J., 2003; Burke M.W., 1996), thus, this illumination
technique is adopted for this experiment.
Z 1
 h( x )
M ean,  
[1]
x 0
Z
Z 1
Standard Deviation, σ 
 ( h( x )   )
2
x 0
[2]
Z
Z 1
Skewness 
 ( h( x )   )
x 0
3
[3]
Z 3
Z 1
Kurtosis 
 ( h( x )   )
x 0
4
[4]
Z 4
Z 1
Dispersion   h( x)  
[5]
x 0
lowest color level, c  x h(x)  0
where 0  x  Z and h(i)  0 i:0  i  x
highest color level, d  x h(x)  0
where 0  x  Z and h(i)  0 i:x  i  Z
Mode  x h(x)  h(i):i,0  i,x  Z,i  x
[6]
UpperLimit
Figure 2: Lighting Techniques: (a) brightfield illumination,
(b) darkfield illumination and (c) diffuse frontlighting
Entropy 
III. FEATURE EXTRACTION
 h(x).log
x  LowerLimit
2
(h(x))
[7]
tpixelsp
Where tpixelsp is the total number of pixels used to
calculate entropy
In the feature extraction phase, both color and gray scale
images are considered. Brunner et al. (Brunner, C. C.,
Maristany, A. G., Butler, D.A., Leeuwen D.V., & Funck, J.W.,
1992) converted the usual RGB color space into other
potentially more useful color spaces, but they found that none
UpperLimit
2
Energy 
99
 x .h( x)
x  LowerLimit
tpixelsg
[8]
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
where tpixelsg is the total number of pixels used to
calculate energy
IV. RESULT AND DISCUSSION
Since no databank was available for beverage cans
identification system following our method of image
extraction, we had to create a database of the objects. One of
the tasks to be studied for the enrollment process is the color
value of background that forms the ranges of different grades.
It is obvious that the bigger the number of samples used, the
more accurate range of color for respective grade of object will
be created. 20 samples are considered sufficient to create
accurate range of color for respective types of beverage cans.
We have collected 3 photographs with resolution 100 dpi (dot
per inch), 200 dpi, and 300 dpi for each of 160 objects.
In order to develop the proposed system, the software tools
Matlab 7.4 for front-end application, Microsoft Access 2000
for backend database support, and MS Excel 2000 for data
sheets and experimental results analysis are used.
The three types of waste beverage cans, Aluminum (ABC),
Non-Aluminum (NAC) and Non-Recyclable (NRC), were
considered in this experiment because of their dominating role
in waste object with 1500 samples. Different templates were
created for the same grade of object. Ten samples were
considered to create an accurate feature vector for the reference
template object grade.
In this section, a relative comparison is made based on the
outcomes of the proposed method for ABC, NAC and NRC.
The images P(a), Q(a) and R(a) represent the original images
of ABC, NAC and NRC with background noise; in addition,
the images P(b), Q(b) and R(b) represent the preprocessed
images of ABC, NAC and NRC, respectively. The calculated
first order features of the ABC, NAC and NRC are illustrated
in Figure 3. The discriminating capabilities of the significant
feature energy, mode, and histogram tail length on the dark
side, histogram tail length on the light side, lower quartile, and
upper quartile are illustrated in Figure 4.
The success rates of the object grade identification process
for absolute distance metrics at different values of K are
tabulated in Table 1. The correct identification rate is
calculated based on the percentage of the number of objects
that are classified into their respective object grades. Using the
absolute distance metric with KNN, the results are 90% and
93% for k=3 and k=5, respectively.
Figure 4. Energy
Table 1 Identification success rate for the two distance metrics
at different values of K
Method
K Value
Name of the Distance
Metrics
Correct Identification
Rate
K-nearest
neighbor
(KNN)
3
Absolute Distance
90%
5
Absolute Distance
93%
Finally, the best results of the SVS system is compared
with the results published in literature other methods shown in
Table 2. It is observed that the performance of the SVS system
is the best among all existing systems. The template matching
method showed the closest performance. The average
maximum classification success rate of the template matching
system is 94.67%, while the SVS system offered 95.17%
classification success rate. In real time implementation, the
SVS system is more effective and convenient than the template
matching technique with regards to computational time and
lighting consistency.
For instance, in template matching, significant time is
allocated for preprocessing, while in the SVS system
preprocessing is not required. Additionally, performance of the
template matching method depends on lighting consistency
during the enrollment and identification phases. With the SVS
system, the lighting dependency has been alleviated because
the system uses different reference templates for the same
beverage cans types which are taken in different lighting
conditions.
Thirdly, for template matching, a 5×5 template consists of
25 pixels; and for each pixel the RGB string length is 4 to 16.
The RGB string length for 5×5 template is thus 100 to 400. As
a consequence, there are 100 to 400 comparisons between one
reference template and one cell image template, which makes
the system inconvenient for real time implementation. For the
SVS system, the template consists of only two values, namely
mode and energy of the RGB components, which greatly
improves the speed of the matching process.
Figure 3 First order feature values
100
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
conditions, which overcame the need to maintain lighting
consistency during enrollment and identification phases.
The most important point addressed in this work is that the
method, which uses computer vision, can be implemented
easily to sort multiple types of beverage cans. Moreover, the
algorithm provides robust and fast results because the proposed
method avoids the extra computational burden for
preprocessing since only two features, mode and energy, of the
RGB components are used to identify the dominating color
value of the object image. The proposed method can identify
three major beverage cans types, ABC, NAC and NRC.
REFERENCES
[1] J. Petek and P. Glavic, An integral approach to waste
minimization
in
process
industries,
Resource,
Conservation and Recycling, Elesevier, 17: 169–188,
1996.
[2] WasteCap. (2008), WasteCap of Massachusetts, 68
Hopkinton Road, Westboro, MA 01581, March 2008,
available
at:[http://www.wastecap.org/wastecap/commodities/paper
/paper.htm]
[3] Paper
Grades
(2009),
available
at:
[http://www.paperonweb.com/ppmanf.htm],
(accessed
February 2009).
[4] M. K. Ramasubramanian, R. A.V enditti, C. M.
Ammineni and M. Mallapragada, Optical Sensor for
Noncontact Measurement of Lignin Content in HighSpeed Moving Paper Surfaces, IEEE Sensors Journal,
5(5), pp. 1132-1139, 2005.
[5] A. G. Doak, M. G. Roe, and G. R. Kenny, Multi-Grade
Object sorting system and method, US Patent
No.7173709, (2007).
[6] A. G. Doak, M. G. Roe, and G. R. Kenny, Multi-Grade
Object sorting system and method, US Patent No.
US2007/0002326, (2007).
V. CONCLUSION
The primary emphasis of this work is on the development
of a new waste beverage cans identification method for
automated beverage cans sorting systems known as the SVS
system.
Another important idea that has been implemented is the
adaptability to new subcategories of the primary Object grades.
The wide range of subcategories of object grades is used to
train the system to recognize new subcategories, and as a result
the system is scalable and able to provide robust decisions for
object identification tasks. Besides, the method was trained
with many reference templates using different lighting
101
International Journal of Smart Material and Mechatronics Vol.1 No.1 2014
ISSN: 9772356-531002
Experimental Test of the Thermoelectric Performance
on the Dispenser Cooler
Zuryati Djafar, Amrullah, Wahyu H. Piarah, Syukri Himran
Mechanical Engineering Department
Engineering Faculty Hasanuddin University
Makassar, Indonesia
[email protected], [email protected]
Abstract— This study aims to find out of the cooling performance
of thermoelectric coolers with single, double series, and double
parallel circuit. The experiment was conducted in the Cooling
and Heating laboratory of Mechanical Engineering Department,
Hasanuddin University, Makassar. The data taken were hot side
temperature, cold side temperature, water temperature, and
ambient temperature. Data analysis was carried out on water
temperature, temperature difference, absorbed heat, and COP
with some variations of thermoelectric circuit and DC electric
voltage in 360-minute period. The result reveal that the best
module was the double thermoelectric arranged with a series
circuit in the voltage of 10 V. This could be seen after 360
minutes with cold water temperature of 12 oC, temperature
difference of 28oC, absorbed heat of 19.52810 and COP of
1.25268.
Seebeck's discovery inspires Jean Charles Athanase
Peltier to examine the opposite of the phenomenon. He flows
the electric discharge on two metal pieces glued together in a
series. When electrical power is applied, the heat absorption
occurs at the junction of the two metals and heat release in
another connection. This heat release and absorption revert
each other when the current is reverted . The discovery which
occurred in 1934 then known as the Peltier effect [3]. Seebeck
and Peltier effect is then the basic for the development of
thermoelectric technology.
Simple mode of cooling is by using a thermoelectric
device. However, due to the limits of thermoelectric materials
performance, one degree of the thermoelectric cooler machine
can only be operated with a small temperature range. If the
temperature ratio between the heatsink and cooling is large,
then the coolant engine with one degree of thermoelectric will
lose its effectiveness. Thus, the application of thermoelectric
with two or more levels are combined in the coolant engine is
an important method to improve the performance of
thermoelectric [4].
Keywords: thermoelectric cooler , water temperature, DC electric
voltage.
I. INTRODUCTION
The national need for energy is increasing along with
the growth of national economy that needs the efforts to
ensure the continuous availability of energy in sufficient
quantity and quality at a reasonable price level. With the
decreasing amount of energy derived from fossil, humans are
trying to find new sources of alternative energy. One of the
solutions that can be used to generate energy and is
environmentally friendly is by using thermoelectric.
The selection of the thermoelectric module
specification is based on the heat load, the temperature
difference and the electrical parameters used. Thermoelectric
cooler has several advantages including no noise, easy
maintenance, environmentally friendly and does not require a
lot of additional components. In addition, another benefit of
thermoelectric cooler as the engine is able to reduce air
pollution and Ozone Depleting Substances (ODSs) because it
no longer uses Hydrochlorofluorocarbons (HCFCs) and
Chlorofluorocarbons (CFCs) known as Ozone Depleting
Substances (ODSs) [1].
Thermoelectric first discovered in 1821 by the
German scientist Thomas Johann Seebeck. He connected
copper and iron in a circuit. Between the two metals are then
placed compass needle. When the metal is heated, it turns the
compass needle move. Later known, it happens because
electricity that occurs in metals cause the magnetic field. This
magnetic field that moves a compass needle. This
phenomenon known as the Seebeck effect [2].
II. THEORETICAL FOUNDATION
A thermoelectric device works by converting heat
energy directly into electricity (thermoelectric generators), or
otherwise, the electricity generating cold (thermoelectric
coolers). Thermoelectric modul composed by semiconductor
material arrangement (usually Bismuth Telluride) which uses
three principles of thermodynamics, known as Seebeck effect,
Peltier and Thomson. Its construction consists of a pair of Ptype semiconductor material and N-type which forms
thermocouple like a sandwich between two thin ceramic
wafers [5].
Thermoelectric cooler (TEC), which is a
semiconductor circuit by utilizing the Peltier effect has been
used as a cooling device on some mini cooling system. In
which cooling has become a necessity in modern society that
has been proven to improve the quality in terms of taste and
hygiene of food and beverages [6].
Generally, thermoelectric module, has a measurement
of 40mmx40mm or smaller and has less than 4 mm thick. Age
of a thermoelectric module in accordance with the industry
standard is about 100,000-200,000 hours and more than 20
years when used as a coolant, and by the number and voltage
which is appropriate with the characteristics of each module
[5].
102
International Journal of Smart Material and Mechatronics Vol.1 No.1 2014
Test using thermoelectric cooler module is the
application of the Peltier effect to move heat. Thermoelectric
cooler which is used consists of a number of pairs of P-type
and N-type semiconductor connected in series and parallel
thermal electricity. Heat which is pumped directly can be
changed by changing the pole which is flowed by DC
electricity. The thermoelectric semiconductor material
composed of N-type made from a mixture of bismuthtelluride-selenium (BiTeSe) and P type made from a mixture
of bismuth-antimony-telluride (BiTeSb). The use of bismuth
telluride on thermoelectric cooler based on some studies that
suggest that bismuth telluride is a material that has the best
performance even though it has limitations on the heat
temperature [7].
In this study we want to know the performance of
cooler using single, double series, and double parallel
assembled thermoelectric cooler.
In analyzing the performance of thermoelectric
modules can be observed in Figure 1, the heat transfer occurs
from the heat load to the cold side of the thermoelectric
module can be determined from the amount of heat that is
pumped by the Peltier effect, heat moves from the hot side to
the cold side because the thermal conductivity of
thermoelectric materials, and partly of the total Joule heating
effect generated by the electric current to thermal resistance
[8].
ISSN: 9772356-531002
resistance (R) and assumed to be divided toward the cold side
and hot side.
(3)
Heat absorbed at the cold side of the thermoelectric module
(4)
Heat released at the hot side of the thermoelectric module
(5)
As described above, to determine the absorbed
calorific value (qc) and the released heat (qh) on thermoelectric
can be written in equation (4) and (5), where the first term is
given electrical energy, the second term is the heat energy
transmitted by conduction, and the third term is the loss of
heat due to electrical current.
Based on the type of thermoelectric modules used, TEC112706, number of connection elements (N) is 127 so that the
thermoelectric module is twice of the number of connection
elements (2N).
Seebeck coefficient value element (αm), thermal conductivity
element (Km), and the thermal resistance elements () usually
can be seen from the data vendors or from the corresponding
equations form a thermoelectric material, in this case the
material used is Bismuth Telluride.
Seebeck coefficient
Value of the Seebeck coefficient (α) is determined by
the value of the Seebeck coefficient element (αm) and the
number of elements on the thermoelectric modules.
(6)
Seebeck coefficient of the element
Figure 1. Heat transfer in thermoelectric
(7)
α0 = 2.2224 x 10-5; α1 = 9306 x 10-7; α2 = -9905 x 10-10
Heat pumped by the Peltier effect
Heat which is pumped by the Peltier effect (qp) is the
electrical energy which is supplied and can be known by
determining the value of the Seebeck coefficient (α), the cold
side temperature (Tc), and the electric current supplied to the
thermoelectric (I).
Thermal conductivity
The amount of the thermal conductivity (K) is
determined by the thermal conductivity of the element (Km),
the geometry factor (G), and the number of elements on the
thermoelectric modules.
(8)
(1)
The thermal conductivity of elements
Heat transfer because of thermal conductivity
The amount of heat move due to thermal conductivity
(qk) is influenced by the magnitude of the thermal conductivity
(K) and the value of the temperature difference (ΔT).
(9)
K0 = 6.2605 x 10-2; K1 = -2777 x 10-4; K2 = 4,131 x 10-7
Electrical resistivity
The amount of electrical resistance (R) is determined
by electrical resistance elements (), the geometry factor (G),
and the number of elements on the thermoelectric modules.
(2)
Joule heating effect generated by the electric current
Joule heating effect (qJ) is the heat loss that occurs as
a result of electrical current which can be determined from the
value of the square of the electric current (I) and electrical
(10)
103
International Journal of Smart Material and Mechatronics Vol.1 No.1 2014
Resistance of electric element
ISSN: 9772356-531002
Heat absorbed from the water
Heat absorbed from the water can be determined by
determining the value of the mass of water (m), specific heat
of water (Cp), and the difference between water temperature
(ΔTair) and time difference (Δt).
(11)
0 = 5.112 x 10-5 ; 1 = 1.634 x 10-6 ; 2 = 6.279 x 10-9
By substituting equation number (6), (8), (10) into
equation number (4) can be obtained calorific value which is
absorbed at the cold side of the thermoelectric module:
(19)
Average calor absorbed from the water up to 360 minutes
Average calor can be determined by determining the
total heat absorbed from water (qw) against the amount of
absorption of heat occurs (n).
(12)
By substituting equation number (6), (8), (10) to
equation number (5) can be obtained calorific value which is
released on the hot side of the thermoelectric module:
(20)
(13)
III. RESEARCH METHODOLOGY
The electric power supplied to the thermoelectric module
The amount of electrical power supplied to the
thermoelectric module influenced the size of the electric
current (I) and the amount of the electrical resistance (R).
The research method used is the experimental
method. Thermoelectric performance testing carried out by
variation of the DC power supply given,which is 8 V, 10 V, 12
V and variation of module by using single module, multiple
series module, and multiple parallel module with a 360-minute
long test as shown in Figure 1 and Figure 2 .
Data collection was performed by measuring the cold
side, hot side , water temperature, and ambient temperature
using a thermocouple and a temperature controller.
Determination of the value of the element geometry factor (G)
is using the AZTEC software; version 3.1 [10]. Data
processing done by calculating the calorific value absorbed,
heat removed, the electrical power used, figures of merit, and
COP.
(14)
Energy equilibrium
According to the working principle of thermoelectric
based on Peltier effect, heat is absorbed from the cold side by
qc and the heat released to the environment by qh. The
difference between the two is the amount of electrical power
required or Pin = qh-qc [9] so that the thermoelectric energy
equilibrium can be written in the following equation:
(15)
Figure of merit
Figure of merit (Z) is the default for determining the
efficiency of thermoelectric materials. If the value of Z
increases the capability of thermoelectric materials also
increased. Figure of merit value varies depending on the needs
of the thermoelectric material temperature [7].
Single Thermoelectric Testing Installation
(16)
Coefficient of Performance (COP)
COP is a measure of the efficiency of a
thermoelectric cooler that can be seen from the comparison of
the amount of heat absorbed at the cold side (qc) to the amount
of incoming power (Pin) [5].
Figure2. Single Thermoelectric Testing Installation
(17)
Average calor absorbed at the cold side of the thermoelectric
module up to 360 minutes
Average calor can be determined by determining the
total heat absorbed at the cold side (qc) the amount of heat
absorption occurs (n).
I. MODEL ANALYSIS AND DISCUSSION
From the results of data collection and the calculation in the
research, the thermoelectric performances are:
The temperature of the hot side of the thermoelectric module
In the initial conditions before supplying the voltage,
the hot side temperature is at room temperature and after the
supplying the voltage, the hot side temperature will increase
until it reaches a certain temperature. It shows that in
(18)
104
International Journal of Smart Material and Mechatronics Vol.1 No.1 2014
thermoelectric, hot side temperature will increase by the
addition of voltage.
ISSN: 9772356-531002
temperatures of 29°C and the water temperature can reach
28°C only.
Heat absorbed at the cold side of the thermoelectric module
The calculation of the absorbed heat associated with
the amount of the electric current generated from a given DC
voltage. The amount of electric current is influenced from the
resistance or thermoelectric module. The greater the voltage,
the electric current generated is also getting bigger and the
greater the electrical resistance, the current generated will be
smaller. Determination of the electrical resistance depends on
the number of constituent elements of the thermoelectric
module [4]. For double thermoelectric, the number of
elements of the module is two times of the number of elements
of single module. However, the resistance is also affected
from the series type. For double thermoelectric series, the
resistance and the variables that influenced the electric current
is twice bigger than of that of a single thermoelectric. While
for the double parallel thermoelectric, the amount of resistance
and the variables associated with the electric current is half of
a single thermoelectric. This can be proved by calculating the
energy balance which can be seen in appendix of calculation
table, where the amount of heat released (qh) is the amount of
electric power required (Pin) and the amount of heat absorbed
at the cold side (qc) [10].
In Figure 3 it can be seen the increase in the value of
the absorbed heat affected from the increase in the electrical
voltage to each circuit. For voltage of 8V, the highest qc value
is indicated by a series of parallel double that is equal to
17.44189 W. For voltageof 10V, the highest qc value
indicated by a single sequence that is equal to 20.61895 W.
For a voltage of 12 V, the highest qc values indicated by a
single sequence that is equal to 24.71738 W. I suggests that
the increase in the value of the absorbed heat is proportion to
the increase of the applied voltage but depends on the
variation of the thermoelectric circuit.
Double Thermoelectric Testing Installation
Figure 3. Double Thermoelectric Testing Installation
The temperature of the cold side of the thermoelectric module
At first, the cold side temperature is at room
temperature and then decreases until it reaches a certain
temperature. Cold side temperatures will continue to drop to
constant conditions. In single thermoelectric, giving 8 V and
10 V of voltage can reach temperatures lower than 12 V. At
double thermoelectric with series circuit, the voltage of 10 V
can reach the lowest cold side temperatures among the three
variations of voltage. In double thermoelectric arranged in
parallel, the voltage of 8 V can reach the lowest cold side
temperatures among the three variations of voltage.
Different temperature of thermoelectric module
In the initial condition, the temperature difference
value is zero because the temperature of the hot side and the
cold side is at the same temperature. In a single
thermoelectric, the greater the applied voltage, the value of the
temperature difference will be even greater. Likewise on
double thermoelectric series, the greater the applied voltage,
the value of the temperature difference will increase. But for
parallel thermoelectric double, on the voltage of 12 V the
value of the temperature difference is low because the value of
Th and Tc tend to be constant and not increased since the
beginning of cooling.
Figure 4. Graph of electrical voltage to the heat absorbed at the cold side on
360 minutes
The temperature of the cooled water
At the beginning of cooling, water temperature is
around 29°C and then will continue to decrease until a certain
temperature. In single thermoelectric, giving the voltage of 8
V and 10 V can reach lower water temperature is than giving
the voltage of 12 V. The addition of voltation to the double
thermoelectric which is assembled series can accelerate the
decrease in water temperature. In parallel double circuit with
the voltage of 12 V can be seen that very little heat is absorbed
from the water. This is because the temperature reaches 56°C
heat and heat can not be released properly into the air so that
the side of the thermoelectric cooler can only reach
The electric power supplied to the thermoelectric module
Figure 4 is a graph of relation between voltage
electricity to the electrical power supplied and variations of
sequence at 360 minutes. From the graph it can be seen that
the greater the voltage applied to each circuit, electrical power
used is also greater. When compared to the third variation of
the thermoelectric circuit, double circuit series shows the
value of the lowest power. This shows a double thermoelectric
series is a series that consumes the least power among the
three variations of the circuit.
105
International Journal of Smart Material and Mechatronics Vol.1 No.1 2014
3.
ISSN: 9772356-531002
After analyzing the performance of double
thermoelectric with paraller circuit, best performance
generated by giving voltage of 8 V.
Of the three variations of the circuit, the best
performance is the double thermoelectric modules
with series assembled on voltage of 10 V.
4.
Symbol
Cp
G
I
K
Km
m
N
Pin
= Specific heat of water [J/kgK]
= The geometry factor [cm]
= Electric current [A]
= Thermal conductivity [W/K]
= Thermal conductivity of elements [W/cmK]
= Mass of water [kg]
= Number of elements on the thermoelectric
= Electric power [W]
= Heat absorbed at the cold side of the thermoelectric [W]
= Heat released at the hot side of the thermoelectric [W]
= Heat absorbed from the water [W]
R = Electrical resistance []
Tc = Cold side temperature [K]
V = Electrical voltage [V]
Z = Figure of merit [K-1]
α = Seebeck coefficient [V/K]
αm = Seebeck coefficient of the element [V/K]
 = Resistance of electric element [cm]
∆T = Temperature difference [K]
∆Tw = Temperature difference of water [K]
∆t = Time difference [s]
Figure 5. Graph of electricity voltage to the electrical power supplied with the
circuit variation on 360 minutes.
Coefisien of Performance (COP)
COP value is a measure of the efficiency of a
thermoelectric cooler that can be seen from the comparison of
the amount of heat absorbed at the cold side (qc) to the
amount of incoming power (Pin). Now, thermoelectric cooling
still has low COP value that it can not compete with vapor
compression cooling system [9].
REFERENCES
Figure 6.Graphs of the relationship of voltage applied to the COP on 360
minutes
[1]
[2]
Figure 5 is a graph of the relationship between the
COP to the applied voltage and variation of circuit on 360
minutes. The amount of COP influenced by heat absorbed at
the cold side and the amount of electrical power used. For
voltage of 8V, the highest COP values shown in multiple
series, by 1.55046. For voltage of 10V, the highest COP
values shown in multiple series circuit is equal to 1.25268. For
voltage of 12V, the highest COP value is shown in a series of
multiple series that is equal to 1.09192. It shows that voltage
variation given, double series circuit showed the highest COP
value compared to other circuit variations.
From the three variations of the thermoelectric circuit, best
known performance is a thermoelectric module which is
arranged in double circuit with the voltage of 10 V as it can
achieve the lowest water temperature, the lowest power
consumption, and best cooling speed.
[3]
[4]
[5]
[6]
[7]
[8]
II. CONCLUSION
[[9]
From the calculation results and discussion can be concluded
as follows:
1. After analyzing the performance of single
thermoelectric, best known performance generated by
giving voltage of 8 V.
2. After analyzing the performance of double
thermoelectric with series assembled, the best
performance is produced by giving voltage of 10 V.
[10]
106
Tellurex. 2010. Introduction to Thermoelectric. 1462 International
Drive.Traverse Citi,MI. (http://www.tellurex.com).
California Institute of Technology.2013.Brief History of
Thermoelectrics.(Online).
(http://thermoelectrics.caltech.edu/thermoelectrics/history.html).
Chakib Alaoui. 2011. Peltier Thermoelectric Modules Modeling and
Evaluation. International Journal of Engineering (IJE), Volume (5) :
Issue (1).
Jincan Chen, Yinghui Zhou, Hongjie Wang, Jin T. Wang. 2002.
Comparison of the optimal performance of single- and two-stage
thermoelectric refrigeration systems.
Riffat, S.B; Ma X. 2003. Thermoelectrics: a review of present and
potential applications. Applied Thermal Engineering 23 913–
935.Pergamon-Elsevier Science Ltd.
Hendi Riyanto, Sigit Y. Martowibowo.2010.
Modeling and
Prototyping a Mini Portable Thermoelectric Beverage Cooling
Device. ICCHT2010 - 5th International Conference on Cooling and
Heating Technologies.
Christopher M. Jaworski. 2007.Opportunites for Thermoelectric
Energy Conversion in Hybrid Vehicles. The Ohio State University.
Department of Mechanical Engineering.
Rehab Noor Mohammed Al-Kaby. Study Of Thermal Performance of
Thermoelectric Cooling System. Mechanical department, Babylon
University-College of Engineering.
Yunus A. Çengel and M. A. Boles. 2006. Thermodynamics: An
Engineering Approach, 5th ed, McGraw-Hill.
Laird
Technologies.
2010.
Thermoelectric
Handbook.
(http://www.lairdtech.com, diakses pada 1 Agustus 2012).
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Potential Coir Fibre Composite for Small Wind
Turbine Blade Application
Bakri, S.Chandrabakty, R. Alfriansyah, A. Dahyar
Mechanical Engineering Department
Tadulako University
Palu, Indonesia
[email protected]
Abstract— Natural fibers have been developed as
reinforcement of composite to shift synthetic fibers. One of
potential natural fibers developed is coir fiber. This paper aims
to describe potential coir fiber as reinforcement of composite for
small wind turbine blade application. The research shows that
mechanical properties ( tensile, impact, shear, flexural and
compression strengths) of coir fiber composite have really similar
to wood properties for small wind turbine blade material, but
inferior to glass fiber composite properties. The effect of
weathering was also evaluated to coir fiber composite in this
paper.
Coir fiber composite have been developed to some
applications like automotive and structure. In this paper, the
potential application of this composite was presented for small
wind turbine blade and it was also evaluated effect of
weathering time to the mechanical properties.
Using of fiber composites in wind turbine blade was
applied with glass fiber as reinforcement [10]. Glass fiber
composite for small wind blade have been applied and
compared to flax fiber composite. Flax fiber as natural fiber
can replace possibly glass fiber for reinforcement composite
[11]. The designed small blades will be subjected to load when
operation, therefore they need good strength, stiffness and tip
deflection.
For application in wind blade, weathering will affect to the
materials of wind blade. Some literatures explained the effect
of weathering to the natural fiber composites. Kenaf high
density polyethylene (K-HDPE) composite has been tested for
durable behavior towards weather effect. The result shows that
composite obtained brittleness proportional to the amount of
weathering time [12]. Then, outdoor weathering affected
tensile and moduli of the banana/phenol formaldehyde
composite, and alkali treatment of fiber can improve tensile
strength if exposure to outdoor weathering [13]. Mechanical
properties (including impact, tensile and shear strengths) of
coir/epoxy composites were influenced by weathering when
composites were placed in outdoor for 10 days, 20 days and 30
days [14] [15]. These effect can be seen in Table 1 where we
were published.
Index Terms—Coir fibers, composites, wind turbine blade.
I. INTRODUCTION
Natural fibers have been applied in composite materials.
Related to this, natural fibers have beneficial properties over
synthetic fibers like high specific strength and modulus, low
density, low cost and abundant in many countries [1, 2, 3].
Some natural fibers used as reinforcement composite are coir,
flax, jute, and ramie fibers.
Coir fiber is a natural fiber which has been used for
reinforcement of composite. Coir fiber composite has been
developed in India and Brazil. Some researches of mechanical
properties of coir fiber composite were done. Flexural strength
was obtained for coir fiber/polyester composite really similar to
the conventional materials [4]. Meanwhile, impact strength of
coir fiber composites is higher than jute and kenaf composites
Alkali treatment of coir fiber increases its bonding with
polyester matrix. Coir fiber composites show tensile strength
improves when fibers is soaked in 2% alkali prior to mixing
polyester and flexural strength improves when 5% alkali [5].
This result was supported by another reseacher that states
tensile strength of coir fiber composites increased when fibers
are soaked with alkali prior to binding with matrix. This is
because good adhesive between fibers and matrix after alkali
treatment [6]. Tensile, impact and flexural strengths of
coir/epoxy composites were evaluated with the average values
of 17.86 MPa, 11.49 kJ/m2 and 31.08 MPa respectively. These
values have lower than glass reinforced composite laminate
[7]. The tensile strength of coir reinforced composites was also
tested and found lower its value. But, their impact strength
was found higher which have potential for application in
automotive that require impact resistance [8]. The impact
strength of coir fiber composites was also reported that its
value is higher than other natural fiber composites [9].
TABLE 1. THE EFFECT OF WEATHERING TIME ON
IMPACT,TENSILE AND SHEAR STRENGTH OF COIR FIBER
COMPOSITES
Specimens
Treatment Time
Tensile
Strength
(MPa)
Shear
Strength
(MPa)
Impact
Strength
(kJ/m2)
Without treatment
(WT)
17.56
14.57
384.99
17.37
14.42
328.13
16.38
13.83
307.22
16.40
13.63
296.00
10 days
20 days
30 days
107
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
II. METHODOLOGY
III. RESULTS AND DISCUSSION
Coir fibers were extracted from the husk of coconut shell.
Prior to mixing with matrix, fibers were soaked in 5% NaOH
and water during 24 hours. Epoxy resin was used as matrix.
Coir fiber composites were made with 17% volume fraction of
fiber by pressing molding for 24 hours. Molded composite is
shown in Fig.1. Then, specimens were divided into two types
including without treatment (WT) and treatment to weathering.
Specimens with treatment time to the environmental effect
(outdoor weathering) were 10 days, 20 days and 30 days.
Mechanical properties of coir fiber composites were tested in
this paper including compression and flexural strength with
specimens as shown in Fig. 1 and Fig.2. Flexural testing used
three point bending and compression used compressive testing.
For impact, tensile and shear strength have been published
[14],[15].
Coir fiber composites were tested the mechanical properties
including compression, flexural, tensile, shear and impact
strength. These properties can be seen in Table 2. Impact,
tensile and shear strength results were published previously. In
this paper, flexural and compression strengths of coir fiber
composite were described. Flexural and compression strengths
were presented in Table 2 with values of 44.89 MPa and 26.27
MPa respectively.
TABLE 2. MECHANICAL PROPERTIES OF COIR FIBER
COMPOSITES
Properties
Sources
Tensile Strength (MPa)
17.56
[15]
Shear strength (MPa)
14.57
[15]
Impact strength (kJ/m2)
384.99
[14]
Flexural strength (MPa)
44.89
Compression strength (MPa)
26.27
From mechanical properties of coir fiber composite as
explained before, it can be seen that for application of small
wind turbine blade can be compared with other composites and
wood in Table 2. Mechanical properties of wood for small
wind turbine blade have similar to the coir fiber composites.
This indicated that coir reinforced composites have potential
for using of wind blade material. But, mechanical properties
of coir fiber composite are lower than glass fiber composite.
For development of application in wind turbine blade therefore
coir fiber composites need hybridization to other fibers for
improving strength and stiffness.
Fig 1. Molded coir fiber composite
TABLE 2. MECHANICAL PROPERTIES OF GLASS FIBER
COMPOSITES AND WOOD
Materials
GFRP
Fig 2. Specimens of compression testing (standar ASTM D695)
Laminated
veneer
lumber
Timberstra
n wood
Tensile
Stre.
(MPa)
Shear
Stre.
(MPa
)
826
90.9
19.1
Flex.
Stre.
(MPa)
28.7
5.1
Compr
Stre.
(MPa)
Sources
14.04
[11][16]
[17]
[17]
Related to weathering time effect, the relationship between
compression strength and treatment time of specimens is
demonstrated in Fig.1. Compression strength was not change
significantly on the treatment (weathering) time. When testing
of coBut, flexural strength decreased when specimens were
exposed during 20 days and 30 days (Fig.2). The decreasing of
flexural strength are about 9.71% for 20 days and 20,4% for 30
Fig 2. Specimens of flexural testing (standar ASTM D 790 – 02)
108
Proceeding of International Symposium on Smart Material and Mechatronics
days. The possible cause of decreasing its strength was due to
solar radiation and high humidity.
REFERENCES
[1] S. V. Joshi, A. K. Mohanty, and S. Arora, “Are natural fiber
composites environmentally superior to glass fiber reinforced
composites?,” Compos. Part, vol. 35, pp. 371–376, 2004.
[2] Z. Li, X. Zhou, and C. Pei, “Effect of Sisal Fiber Surface
Treatment on Properties of Sisal Fiber Reinforced Polylactide
Composites,” Int. J. Polym. Sci., pp. 1–7, 2011.
[3] S. Mukhopadhyay, R. Fangueiro, and V. Shivankar, “Variability
of tensile properties of fibers from pseudostem of banana plant,”
Text. Res. J., vol. 79, pp. 387–393, 2009.
[4] S. N. Monteiro, L. A. H. Terrones, and J. R. M. D’Almeida,
“Mechanical performance of coir fiber / polyester composites,”
Polym. Test., vol. 27, pp. 591–595, 2008.
[5] J. Rout, M. Misra, S. S. Tripathy, S. K. Nayak, and A. K.
Mohanty, “The Influence of Fibre Treatment on the
Performance of Coir-Polyester Composites,” Compos. Sci.
Technol., vol. 61, pp. 13023–1310, 2001.
[6] H. Gu, “Tensile Behaviours of the Coir Fibre and Related
Composites after NaOH treatment,” Mater. Des., pp. 1–4, 2009.
[7] S. Harisha, D. P. Michael, A. Bensely, D. M. Lal, and A.
Rajadurai, “Mechanical property evaluation of natural fiber coir
composite,” Mater. Characterisation, vol. 60, pp. 44–49, 2009.
[8] A. Ticoalu, T. Aravinthan, and F. Cardona, “A review of current
development in natural fiber composites for structral and
infrastucture applications,” presented at the Southern Region
Engineering Conference, Toowoomba-Australia, 2010.
[9] T. Sen and H. N. J. Reddy, “Application of Sisal, Bamboo, Coir
and Jute Natural composites in Structural Upgradation,” Int. J.
Innov. Manag. Technol., vol. 2, no. 3, pp. 186–191, 2011.
[10] B. Eker, A. Akdogan, and A. Vardar, “Using of Composite
Material in Wind Turbine Balde,” J. Appl. Sci., vol. 6, no. 14,
pp. 2917–2921, 2006.
[11] D. U. Shah, P. J. Schubel, and M. J. Clifford, “Can flax replace
E-glass in structural composites? A small wind turbine blade
case study,” Compos. Part B, vol. 52, pp. 172–181, 2013.
[12] A. Umar, E. . Zainudin, and S. . Sapuan, “Effect of Accelarated
Weathering on Tensile Properties of Kenaf Reinforced HighDensity Polyethylene Composite,” J. Mech. Eng. Sci., vol. 2,
pp. 198–205, 2012.
[13] S. Joseph, Z. Oommen, and S. Thomas, “Environmental
Durability of Banana- Fiber-Reinforced Phenol Formadehyde
Composite,” J. Appl. Polym. Sci., vol. 100, pp. 2521–2531,
2006.
[14] Bakri, S. Chandrabakty, and A. Dahyar, “Analisis Kondisi
Lingkungan Komposit Serat Sabut Kelapa Terhadap Kekuatan
Impak Untuk Aplikasi Baling-baling Kincir Angin,” presented
at the Proceeding Seminar Nasional Tahunan Teknik Mesin XII
(SNTTM XII), Universitas Lampung, Bandar Lampung, 2013.
[15] Bakri, S. Chandrabakty, R. Alfriansyah, and M. Tahir,
“Pengaruh Lingkungan komposit serat sabut kelapa untuk
aplikasi baling-baling kincir angin,” J. Mek., vol. 5, no. 1, pp.
448–454, 2014.
[16] W. R. Broughton, M. Kumosa, and D. Hull, “Analysis of the
Iosispecu shear test as applied to unidirectional carbon-fibre
reinforced composites,” Compos. Sci. Technol., vol. 38, pp.
299–325, 1990.
[17] D. Retallack, “Engginereed Wood Wind Turbine Blades,”
Delhousie University, Final Reports, Design Project – MECH
4020, 2005.
Compression Strength (MPa)
35
30
25
20
15
10
5
0
WT
10 days
20 days
30 days
Specimen Treatment Time
Fig 1. Relation between compression strength and specimen treatment time
Flexural Strength (MPa)
50
40
30
20
10
0
WT
10 days
20 days
ISBN 978-602-71380-1-8
30 days
Specimen Treatment Time
Fig 2. Relation between flexural strength and specimen treatment time
IV. SUMMARY
Coir fiber composite is one of natural fiber composites
having potential for small wind turbine blade application
because their mechanical properties were found competitive
with wood properties for wind blade materials. Although, it
has inferior to glass fiber composites. Related to treatment
(weathering) time of specimens, mechanical properties of coir
fiber composites were tend to decreasing but it is not
significant.
ACKNOWLEDGMENT
The authors wish to thank the Directorate General of
Higher Education Indonesia – Ministry of Education and
Culture for funding this research.
109
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
A new development of thermosiphon solar hot water
with paralel-serpentine tube configuration
Mustofa
Yuli Asmi Rahman
Mechanical Engineering Department
Tadulako University
Palu, Indonesia
[email protected]
Electrical Engineering Department
Tadulako University
Palu, Indonesia
[email protected]
Basri
Mechanical Engineering Department
Tadulako University
Palu, Indonesia
[email protected]
Abstract— The new configuration of the heat pipe copper
absorber fluid in solar water heater (SHW) has been developed
in an aluminium collector box that aims to improve the efficiency
and especially the outlet temperature of the fluid prior to
entering the stainless steel hot water. Parallel-serpentine pipe
configuration is used in this study in which 5/8" copper tube
installing on the inlet header and header outlet of parallel
configuration. The header outlet is connected with 3/8"
serpentine configuration tube before entering water tank.
Meanwhile collector panels mounted on a north-facing slope of
20o to get a lot of sunlight. From the observations show that the
temperature of the water out of the collector is around 90o C with
thermosiphon SHW efficiency of approximately 25%.
within flat plate and sinusoidal wave configuration and with
serpentine-parallel fluid tubes above the plates. The water
output temperature in pipe parallel is high enough till about
80°C, while water temperature in the reservoir averaged less
50oC. The reservoir water temperature is not enough last long
due to insulating ability of fiberglass reservoir material. In
other words, it is needed to be modified the tank material and
tubes configuration.
Thus, this research has focus on modification of some
collectors materials such as fiberglass reservoir to be stainless
steel, wood collector box becomes aluminium ones. In
addition, to accelerate the circulation of hot fluid entered into
the reservoir, the configuration of the serpentine-parallel tubes
is reversed into a parallel-serpentine in serial connecting. Water
flows from the reservoir made by stainless steel to fistly
parallel tube configuration and then serpentine within hot fluid
back into the tank in means of natural convection
(thermosiphon) [3]. That happens due to the water mass
difference within solar collector box that is exposed to north at
20o slope angle. The slope angle (β) is defined as the angle
between the plane of the collector and the horizontal. The
azimuth angle (  ) is defined as the displacement angle
between the projection on a horizontal plane of the normal to
the collector surface and due north. The incidence angle,  , is
the angle between the direct radiation on a surface and the
normal to that surface. For maximum direct radiation, the
incidence angle should be a minimum [4]. Fig. 1 shows these
angles.
Key words: solar hot water, parallel-serpentine, tube configuration
I. INTRODUCTION
Solar hot water (SHW) is cheaper application of solar
collector energy use than air ones. This can be seen a plenty of
water heater technology are manufactured with various models
according to customer needs. Recently, one obstacle water
heating technology products available in the market is the
selling price that is not affordable all society levels. Therefore
this study is aimed at producing water heating collectors that
are less expensive, easy installed and environmentally friendly.
Thermosiphon solar hot water collectors become interesting
topic for research and even more manufactured with various
modification leading to increasing efficiency. Mostly, tubes
configuration of current SHW are parallel and/or serpentine
laying out independently. Both configurations have advantages
and disadvantages for certain solar radiation hours in one day.
For instance, parallel tube configuration is more useful form
midday till afternoon, whereas serpentine ones is between
morning and midday for water solar heating. Therefore,
Mustofa dkk. [1] and Mustofa et al., [2] has been designed and
made a prototype of a hybrid model both serpentine and
parallel tube configuration in one single collector box leading
to minimizing the disadvantages fo SHW. Thermosiphon
principle is applied for collector plate heat with fluid absorber
Figure 1. Major angles in solar applications
110
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN 978-602-71380-1-8
Thus, the instantaneous efficiency for each tube configuration
is belows:

 par 
m C p (Toutpar  Tinpar )
IApar
(4a)
(4b)
So, the total efficiency for paralel-serpentine solar hot water is
accumalated of  par   ser
(5)
III. METHOD
The research begins with searching solar hot water collector
materials and some tube configurations topics, their advantages
and disadvantages. Mustofa et al., [2] used serpentine-parallel
in ½” of the tubes diameter. It takes 3 till 4 hours to reach
maximum hot fluid in the tank, while high temperature within
the reservoir is not long lasting. Therefore, research method is
started by turning the configuration to parallel-serpentine with
3/8 and 5/8” tubes diameter and so on. Testing has been done
with 2 mass flow rate (0.035 and 0.189 ltr/s) on clear weather
condition. Data collected from 9.30am till 3pm each 30
minutes both fluid temperatures and solar intensity. The
intesity was recorded with Solarimeter Tenmars as shown in
Figure 3, while fluid temperature is noted by thermocouple
digitals in Figure 4.
Figure 2. Prototype SHW parallel-serpentine configuration
For equator line like in Palu, Central Sulawesi, the slope
angle is less influence on solar intensity. Thus, 20o is accepted.
II. COLLECTOR DESIGN
As can be seen from Figure 2, collector box is mainly made
from aluminium with dimensions 128 x 78 x 10 cm comprising
copper tubes on flat and wave sinusoidal plates. The tube and
plates were painted by double Arclic Lacquer Black Metallic
leads to increasing heat absorption from the Sun. There are two
differences tube diameters for parallel configuration. For
parallel configuration tubes, they are 3/8 and 5/8”. Header
parallel tube is 5/8” and the rest of tubes dimension is 3/8”
including serpentine configuration. Insulator sponge 3 cm thick
is under the flat and wave sinusoidal plates.
For stainless steel tank is located above the collector in
which cool and hot water are circulating through the tubes. The
tank size itself is 80 x 30 cm with copper taps on two sides as
for output and input water circulating.
Based on those design, instantaneous equations associated
with collector performance are given as follows:

Qu
IAc
Figure 3. Solarimeter Tenmars
(1)
While Qu is divided by 2 correspondences formula,
Q par  AparS  UApar (Tpar  Ta )
For ambient temperature is measured by mercury
thermometer. All testing had been conducted at outdoor
Mechanical Engineering laboratory in Tadulako University.
Data are tabulated and calculated based on formulas from (1) to
(5).
(2a)
For parallel configuration tubes.
Qser  Aser S  UAser (Tser  Ta )
(2b)
For serpentine configuration, where Ta is ambient
temperature and S is energy from the Sun that can be
absorbed by collector.
The useful gain given collector to fluid:

qu  m C p T
(3)

Which
m is equal

m  A( par ser ) / 2 v
and
T  (Toutser  Tinpar)
Figure 4. Thermocouple digitals type-K
(3a)
(3b)
111
Proceeding of International Symposium on Smart Material and Mechatronics
IV. RESULT AND DISCUSSION
ISBN 978-602-71380-1-8
rate of 0.035 is more stable in low efficiecny than that 0.189
kg/s.
A. Solar Intensity
C. Water Outlet Temperature For Parallel-Serpentine
Figure 7 depicts that water outlet temperature (ToutletParl) in
parallel tube heat absorber is lower than that of serpentine
ones (ToutletSer). The highest temperature of serpentine is about
90oC. This temperature will heat water in the tank. As a result,
Figure 5. Solar intensity during testing time
Based on data collected by solar radiation of 18 locations
in Indonesia, the solar radiation in Indonesia can be classified
as follows: for the western and eastern Indonesia with the
distribution of radiation in the Western Regions of Indonesia
(KBI) about 4.5 kWh/m2/day with monthly variation of
approximately 10%, and in Eastern Indonesia (KTI)
approximately 5.1 kWh/m2/day with approximately 9% of the
monthly variation (EMR 2008). Continued the measurement
data from 1991 to 1994, the Central Sulawesi categorized as
the third largest in Indonesia after the District. Sumbawa
(5,747 Wh/m2) and Jayapura (5,720 Wh/m2) in terms of the
intensity of solar radiation, which amounted to 5,512 Wh/m2
[5] & [6]. This data, in fact, support for developing SHW in
Palu as indicated in Figure 4.
From Figure 5 shows that the average solar intensity
increased significanly from 11.00 to 11.30 and stable a couple
of hours. Means that solar energy for heating water collector is
quite high and effective in mid day till afternoon.
Figure 7. Water outlet temperature for both parallel and
serpentine tubes collector during the test
V. CONCLUSION
SHW for parallel-serpentine tubes configuration with single
cover glazing has been designed and made it. After that, testing
set-up for passive collector (natural convection) has been done
for 2 mass flow rates. It indicates that the higher mass flow
rate, the better of water outlet temperature. This means that hot
water in the tank/reservoir will be soon hot. Proven that
parallel-serpentine is better than of serpentine-parallel tubes
configuration [1] &[2] in terms of outer temperature prior to
entering the tank. Future research is needed to apply an active
collector (forced convection).
B. Collector Efficiency
The test procedure and efficiency calculation specified in
equation (5) with 2 mass flow rates, namely 0.189 and 0.035
(kg/s).
ACKNOWLEDGMENT
The author greatly benifited from discussions with and
comments by coleage Rustan Hatib. All remainining errors are
mine. Special thanks BOPTN Tadulako University (UNTAD)
in year of 2014 for financial support for carrying out this on
going work.
NOMENCLATURE
Apar
Aser
Cp
I
Qu
Figure 6. Total SHW efficiency with 0.189 and 0.035 kg/s mass flow rates
parallel collector area
serpentine collector area
water specipic heat
solar intensity
useful heat gain from collector
m2
m2
J/kgoC
W/m2
W
mass flow rate
collector heat losses coeficient
kg/s
W/m2K

m
The graph indicates that collector efficiency for 0.189 kg/s
mass flow rates is higher than that of 0.035 kg/s. However, the
U
112
Proceeding of International Symposium on Smart Material and Mechatronics
[3]
REFERENCES
[1] Mustofa, Y. A. Rahman, Muchsin & R.C.G. Nugraha, “Hybrid
Plat Datar dan Gelombang Sinusoidal pada Kolektor Pemanas
Air Surya”, Prosiding Seminar Nasional Teknologi Industri I,
Makassar, November 2013, ISBN 978-602-14537-0-4, hal. 99105.
[2] Mustofa, Y.A. Rahman, Muchsin & R. Hatib, “A New Copper
Tube Configuration of Solar Water Heating Collector: Single
and Double”, Proceeding of Engineering International
[4]
[5]
[6]
113
ISBN 978-602-71380-1-8
Conference, Semarang, 21 November 2013, pp. 1-4, ISBN
97925-2784.
U.C.
Arunachaiola,
“Performance
Deterioration
of
Thermosiphon Solar Flat Plate Water Heater Due To Scaling”
Engineering Journal, 2011, pp. 115-129
S. Bari, “Optimum orientation of domestic solar water heaters
for low latitude countries”, Energy Conversion and
Management, 2001, Vol. 42, pp 1205-1214.
Anonymous, online available at http://www.kamase.org,
accessed on 10 of October 2013, 2008a. Unpublished.
Anonymous,online available at http://www.energyterbarukan.net
accessed on 10 of October 2013, 2008b. Unpublished.
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN: 978-602-71380-1-8
Optimal Design of V-shaped Absorber Plate to the
Performance of Solar Water Heater
Jalaluddin*), Effendi Arief, Rustan Tarakka, Hairul
Arsyad, Andi Mangkau
Labusab
Graduate School of Mechanical Engineering
Hasanuddin University
Makassar, Indonesia
Department of Mechanical Engineering
Hasanuddin University
Makassar, Indonesia
*)[email protected],com
energy is known as an environmentally
friendly energy source and wide range of applications. This
energy is utilized in various applications such as domestic and
industrial water heating, refrigeration, cooking, power
production and water pumping etc. The present study analyzes
absorptivity of flat-plate absorber and various V-shaped
absorber plates. Analytical investigation of absorptivity of the
various V-shaped absorber plates and comparison with that of
the flat-plate absorber was carried out. The result shows that Vshaped absorber plate with angle of  = 21 0 (V-shaped dimension
of t = 4 cm and l = 4 cm) has a better absorptivity compared with
that of the flat-plate absorber and others V-shaped absorber
plates. Improving the absoptivity of absorber plate enhances
thermal performance of solar water heater. Utilization of Vshaped absorber plate will increase the efficiency of solar water
heater due to increasing its absorptivity of absorber plate.
efficiency. Various techniques to enhance the thermal
efficiency in solar water heater have been reviewed by
Jaisankar et al. [5]. Solar water heating system is an effective
technology to convert solar energy into thermal energy. The
efficiency of solar thermal conversion is around 70% when
compared to solar electrical direct conversion system which
has an efficiency of only 17%. Summary on the development
of various system components that includes the collector,
storage tank and heat exchanger are discussed in ―[6]‖.
The present work analyzes absorptivity of flat-plate
absorber and various V-shaped absorber plates. Absorptivity
of flat-plate absorber and V-shaped absorber plates with
various angle and dimensions were investigated.
Key words—Solar water heater, V-shaped absorber plate,
Efficiency
Thermosyphon solar water heaters which its components
includes the collector, storage tank and heat exchanger are
characterized by converting solar energy into thermal energy.
The system performance is affected by absorber plate and its
design, selective coating, thermal insulation, tilt angle of the
collector and working fluid. The schematic layout of a typical
thermosyphon solar water heater is shown in Fig. 1.
Abstract—Solar
II. SOLAR WATER HEATER
I. INTRODUCTION
Solar energy is a renewable energy source with wide range
of applications such as domestic and industrial water heating,
refrigeration, cooking, power production and water pumping
etc. Utilization of solar water heaters is increasing in the world
due to their simplicity and reliability. Thermosyphon solar
water heating system is now widely used in domestic as well
as industrial sector due to its ease of operation and simple
maintenance. Solar water heating system proves to be an
effective technology for converting solar energy into thermal
energy. Development of various system components that
includes the collector, storage tank, and heat exchanger is
interest subject to enhance thermal performance of solar water
heater. Several investigations have been reported to address
these issues. Collector efficiency of air solar heater has been
investigated [1]. Effect of glass cover of the solar water heater
collector has been reported in ―[2]‖. Ayompe and Duffy [3]
have investigated the thermal performance of solar water
heating system with 4 m2 flat-plate collectors in Dublin,
Ireland. The results show that solar fraction was 32.2 %,
collector efficiency was 45.6 % and system efficiency was
37.8%. Optical analysis, experimental study and cost analysis
of the stationary V-through solar water heater system have
been studied by Chong et al. [4]. They proposed a stationary
V-through solar water heater with the maximum solar
concentration ratio of 1.8 suns to improve the thermal
Storage
tank
Inlet pipe
Outlet pipe
Solar
collector
Fig. 1. The schematic layout of a typical thermosyphon solar water heater
114
Proceeding of International Symposium on Smart Material and Mechatronics
Design and thermal properties of the absorber plate
contribute to the performance of solar collector. The cross
sectional view of a solar water heater with flat-plate absorber
is shown in Fig. 2 (a) and V-shaped absorber plate is shown in
Fig. 2 (b).
III. ABSORPTANCE OF V-SHAPED ABSORBER PLATE
Two types of absorber plates which are flat-plate and Vshaped plate are investigated to study the absorptivity of the
plates. Increasing the absoptivity of the absorber plate lead to
enhance the collector efficiency. The solar energy is converted
to usefull energy by absorbed it in the absorber plate and
transfered it to working fluid. Various types of V-shaped
absorber plates are V-shaped plate with angle of  = 21 0 (t = 4
cm), 32 0 (t = 4 cm), 41 0 (t = 4 cm), 27 0 (t = 3 cm), 40 0 (t = 3
cm), 49 0 (t = 3 cm). Analytical method is applied to calculate
the absorptivity of absorber plates. Cross-sectional view of raytracing result of flat-plate absorber and V-shaped absorber
plates are shown in Appendix B.
Solar radiations with incident angle of  = 0 0, 30 0, 60 0
reach the surface of absorber plates. Solar radiations reach the
absorber plate perpendicular ( = 0 0) in the daytime, with  =
60 0 in the morning and afternoon. Absorptivity of absorber
plate is calculated by absorbed solar radiation in the plate based
on its incident angle. The absorber plates are assumed as a
black surface. In the case of V-shaped plate, solar radiation is
reflected by several times. Solar radiation is absorbed by the
plate in each reflection based on its incident angle.
Absorptance of black flat-plate with various incident angles is
shown in Appendix A [7]. Absorptivity of flat-plate absorber
and V-shaped absorber plates are shown in table I.
a) Flat-plate absorber
Solar radiation
Glass cover
Water pipe
Plat plate absorber
b) V-shaped absorber plate
Solar radiation
Glass cover
V-shaped plate
absorber
Water pipe
t

ISBN: 978-602-71380-1-8
l
Fig. 2. The cross sectional view of a solar water heater
TABLE I.
No.
1.
2.
3.
4.
5.
6.
7.
Absorber plate
orientation
Flat-plate absorber
V-shaped  = 410
V-shaped  = 320
V-shaped  = 210
V-shaped  = 490
V-shaped  = 400
V-shaped  = 270
ABSORTIVITY OF FLAT-PLATE ABSORBER AND V-SHAPED ABSORBER PLATES
Angle of
incident ()
Absorptivity(
)
0
30
60
0
30
60
0
30
60
0
30
0.963
0.951
0.894
0.985
0.968
0.961
0.980
0.974
0.963
0.970
60
0
30
60
0
30
60
0
30
60
0.981
0.966
0.966
0.963
0.979
0.960
0.966
0.970
0.968
0.970
0.973
115
Average
absorptivity
(average)
V-shaped dimension
(t and l), cm
0.936
-
0.971
4 and 2
0.973
4 and 3
0.975
4 and 4
0.965
3 and 2
0.968
3 and 3
0.970
3 and 4
Proceeding of International Symposium on Smart Material and Mechatronics
Figure 3 shows absorptivity of flat-plate absorber and
various V-shaped absorber plates. Absorptivity of the flat-plate
absorber decreases in increasing the incident angle. However,
absorptivity of V-shaped absorber plates changes slightly with
increasing the incident angle. This fact indicates that V-shaped
absorber plates provide a better absorptivity compared with the
flat-plate absorber. Several reflections of the solar radiation on
the V-shaped absorber plates increase its absortivity.
ISBN: 978-602-71380-1-8
others V-shaped absorber plates. Utilization of V-shaped
absorber plate will increase the efficiency of solar water heater.
ACKNOWLEDGMENT
This work was supported and financed by LP2M
Hasanuddin
University
under
contract:
16187/UN4.42/PL.09/2014.
REFERENCES
1.000
[1]
0.975
0.950
Absorptivity ()
[2]
0.925
Plat plate
V-shaped =41o
V-shaped =32o
V-shaped =21o
V-shaped =49o
V-shaped =40o
V-shaped =27o
0.900
0.875
0.850
0.825
[3]
[4]
0.800
0
10
20
30
40
50
60
70
80
[5]
90
Angle of Incident ()
Fig. 3. Absorbtivity of flat-plate absorber and various V-shaped absorber
plates
[6]
Average absorptivity (average)
1.000
[7]
Better absorptivity
0.975
Plat plate
V-shaped =41o
V-shaped =32o
V-shaped =21o
V-shaped =49o
V-shaped =40o
V-shaped =27o
0.950
0.925
0.900
0
1
2
3
4
5
6
7
8
9
10
Absorber Plate Configuration
Fig. 4. Average absorptivity of absorber plates
Based on the average absorptivity of V-shaped absorber
plates, optimal design is found in the V-shaped absorber plate
with angle of  = 21 0 (V-shaped dimension of t = 4 cm and l
= 4 cm) as shown in the Fig. 4. Applying the V-shaped
absorber plate in the solar water heating system will increase
its performance due to increasing the absorptivity of its
absorber plate.
IV. CONCLUSIONS
Analytical investigations of absorptivity of various Vshaped absorber plates and comparison with that of flat-plate
absorber have been carried out. The result shows that the Vshaped absorber plate with angle of  = 21 0 has a better
absorptivity compared with that of the flat-plate absorber and
116
Jalaluddin and A.E.E. Putra, Studi Teoritis Tentang Efisiensi
Kolektor Surya Pemanas Udara dengan Pelat Absorber
Gelombang, Jurnal Penelitian Enjiniring Fakultas Teknik,
Universitas Hasanuddin, 2002.
Jalaluddin, Analisis Perbandingan Prestasi Kolektor Surya
Pemanas Air dengan 1 (satu) dan 2 (dua) Penutup di Kota
Makassar, Sulawesi Selatan, Laporan penelitian Fakultas
teknik Universitas Hasanuddin, 2008.
L.M. Ayompe and A. Duffy, Analysis of the thermal
performance of a solar water heating system with flat-plate
collectors in a temperate climate, Applied Thermal
Engineering 58, 2013, 447-454
K.K. Chong, K.G. Chay, and K.H. Chin, Study of a solar
water heater using stationary V-trough collector, Renewable
Energy 39, 2012, 207-215
S. Jaisankar, J. Ananth, S. Thulasi, S.T. Jayasuthakar, and
K.N. Sheeba, A comprehensive review on solar water heaters,
Renewable and Sustainable Energy Reviews 15, 2011, 3045–
3050
R. Shukla, K.Sumathy, P. Erickson, and J. Gong, Recent
advances in the solar water heating systems: A review,
Renewable and Sustainable Energy Reviews 19, 2013, 173–
190
ASHRAE, ASHRAE Handbook: HVAC Applications, SI
Edition, Solar Energy Use (Chapter 35), American Society of
Heating, Refrigerating and Air-Conditioning Engineers, 2011
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN: 978-602-71380-1-8
Appendix A
Source: ASHRAE, (2011). ASHRAE Handbook: HVAC Applications, SI
Edition, Solar Energy Use (Chapter 35), American Society of Heating,
Refrigerating and Air-Conditioning Engineers, Inc., 1791 Tullie Circle, N.E.,
Atlanta, GA 30329.
114
Proceeding of International Symposium on Smart Material and Mechatronics
ISBN: 978-602-71380-1-8
Appendix B
Cross-sectional view of ray-tracing result of plat and V-shaped absorber plate
Flat-plate absorber
115
Proceeding of International Symposium on Smart Material and Mechatronics
V-shaped absorber plate
116
ISBN: 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
117
ISBN: 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
118
ISBN: 978-602-71380-1-8
Proceeding of International Symposium on Smart Material and Mechatronics
119
ISBN: 978-602-71380-1-8
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement