Design and analysis of foldable human-powered vehicle · Design and analysis of foldable...
Transcript of Design and analysis of foldable human-powered vehicle · Design and analysis of foldable...
Proceedings of Mechanical Engineering Research Day 2017, pp. 128-129, May 2017
__________
© Centre for Advanced Research on Energy
Design and analysis of foldable human-powered vehicle Mohd Azman Abdullah1,2,*, Mohamad Alif Fayumi Ahmad1, Shafizal Mat1,2, Faiz Redza Ramli1,2
1) Faculty of Mechanical Engineering, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 2) Centre for Advanced Research on Energy, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
*Corresponding e-mail: [email protected]
Keywords: Human-powered vehicle; foldable mechanism; transformable chassis
ABSTRACT – It is common for 2-wheel human-
powered vehicle (HPV) available in foldable
mechanism i.e. foldable bicycle. However, for 3-wheel
HPV, the foldable mechanism is quite difficult to be
designed. In this paper, a 3-wheel recumbent type
foldable HPV is designed and analyzed. The objective
of this research is to reduce the envelop volume and
areas of the HPV for storage and easy transportation.
Using simple mechanism, the foldable HPV is
successfully designed and analyzed with reduction of 50
% in envelope volume of its original condition.
1. INTRODUCTION
Human powered vehicle (HPV) is an
environmentally friendly and affordable vehicle that can
help people to achieve the green technology concept.
HPV is a one type of vehicle that powered only by
muscular strength to move on the road [1]. It is a land
vehicle with seat position that is inclined backwards.
The current HPV design is not foldable and may be
difficult to be carried around and stored for future use.
This HPV is redesigned for foldable capability by using
fasteners at the joint. The new design makes the HPV
compact [2].
Generally, the current HPV is not easy to be
carried to everywhere in a car rear trunk due to the
space constraint [3]. HPV may be bigger in size
compared to the bicycle. With all the problems in
current HPV, the next possible solution is the ability of
the HPV to be folded to reduce the envelope volume. If
this HPV is foldable, or detachable, it may use less
space and easy to be stored [3].
There is a part of this HPV which can be folded.
The main body of this HPV can be folded while the
wheel and other components are remaining unchanged.
The new design of the HPV also uses a compact joint
which can provide overlap condition when the main
body of the HPV is folded. Then, the seat of this HPV
also can be folded and adjusted for user comfortable
purpose.
The frame of HPV is the important part in the
system. This is because the frame or body will support
the load and vibration [4-7] and also can be folded.
There are several option in term of frame geometry. The
design space defined by the conventional, forward
facing and recumbent rider position. The proper design
and analysis is very crucial. The material used for the
frame can affect the load and dynamic stability [8] of
the HPV.
2. METHODOLOGY
The drawing for foldable HPV was designed using
a computer aided design (CAD) software CATIA. The
design concept for this foldable HPV is based on the
ergonomic, aerodynamic [9], highly engineered, and
easy to be manufactured and fulfil the safety criteria.
The original chassis design of the HPV is shown by
Figure 1. The location of the fold point is at the centre
of the chassis in order to maximise the reduction in
chassis length.
Figure 1 Original chassis design of HPV.
3. DESIGN & ANALYSIS
The assembly design for foldable HPV is shown in
Figure 2. The main objective of this project is to reduce
the space used for the HPV. After the HPV was folded
the percentage of space reduction will be calculated.
The percentage reduction for HPV before and after
foldable are calculated after all parts are completely
folded. The design for HPV before and after foldable
can be shown on the Figure 3.
4. RESULTS AND DISCUSSION
The percentage of size reduction of the foldable
HPV is calculated and summarized in Table 1. The
envelope volume of the foldable HPV is reduced 50.6
%. This is half of the original volume of the HPV. The
envelope areas for the top, side and front are also
reduced to 60.4 %, 17.3 % and 26.9 % respectively.
These have proven that the main objective of the
foldable mechanism for the HPV is achieved. The
cubical volume of the HPV is represented in Figure 4.
Abdullah et al., 2017
129
Figure 2 Foldable HPV design.
Figure 3 Foldable analysis.
Table 1 Percentage of reduction.
Fix
HPV
Foldable
HPV
%
Reduction
Volume, m3 2.32 1.15 50.6
Top area, m2 2.84 1.23 60.4
Side area, m2 1.79 1.48 17.3
Front area, m2 1.08 0.79 26.9
Figure 4 Cubical volume of foldable HPV.
5. CONCLUSION
The design and analysis of foldable HPV is
successfully performed and presented in this paper. The
drawing for foldable HPV is using the CAD software
CATIA. The envelope volume of the HPV is reduced 50
% and the envelope areas are reduced significantly. The
main objective of the paper is achieved.
6. ACKNOWLEDGEMENT
The authors gratefully acknowledged the
Advanced Vehicle Technology (AcTiVe) research group
of Centre for Advanced Research on Energy (CARe),
the financial support from Universiti Teknikal Malaysia
Melaka and The ministry of Education, Malaysia under
Short Term Research Grant, Grant no.
PJP/2014/FKM(10A)/S01330.
REFERENCES
[1] M.A. Abdullah, S.A. Shamsudin, F.R. Ramli, M.H.
Harun, M. A. Yusuff, “Design and fabrication of a
recreational human-powered vehicle”,
International Journal of Engineering Science
Invention, vol. 5, no. 2, pp. 11-14, 2016.
[2] A.V.R.K. Teja, J. Sri Harsha, N.S.K.Teja, A.Teja
and P.S Afridi Khan, “Kinematic design and
fabrication of four bar mechanism to steer a human
powered vehicle”, International Journal for
Research in Emerging Science and Technology,
vol. 3, no. 4, pp. 102-107, 2016.
[3] R.U. Urunkar and P.P. Deshpande, “Study of drive
mechanisms of bicycle, tricycle or like vehicles to
optimize operating performance - a review”,
International. Journal of Engineering Research
and Applications, vol. 4, no. 1, pp. 214–219, 2014.
[4] M.A. Abdullah, M.R. Mansur, N. Tamaldin, and K.
Thanaraj, “Development of formula varsity race
car chassis”, IOP Conference Series: Materials
Science and Engineering, vol. 50, no. 1, pp.
012001, 2013.
[5] M.A. Abdullah, M.R. Mansor, M.Mohd Tahir, S.I.
Abdul Kudus, M.Z. Hassan and M.N. Ngadiman,
“Design, analysis and fabrication of chassis frame
for utem formula varsitytm race car”, International
Journal of Mining, Metallurgy & Mechanical
Engineering, vol. 1, no. 1, pp. 75-77, 2013.
[6] M.A. Abdullah, A.H. Mohamad and F.R. Ramli,
“Design, Analysis and Fabrication of Fixed-Base
Driving Simulator Frame”, The Journal of
Engineering and Technology, vol 4, no. 2, pp. 85-
102, 2013.
[7] M.A. Abdullah, N. Tamaldin, M.A. Mohamad, R.
S. Rosdi, and M.N.I. Ramlan, “Energy harvesting
and regeneration from the vibration of suspension
system”, Applied Mechanics and Materials, vol.
699, pp. 800-805, 2015.
[8] M.A. Abdullah, M.A. Salim, M.Z. Mohammad
Nasir, M.N. Sudin, F.R. Ramli, “Dynamics
performances of Malaysian passenger vehicle”,
ARPN Journal of Engineering and Applied
Sciences, vol. 10, no. 17, pp. 7759-7763, 2015.
[9] M.N. Sudin, M.A. Abdullah, F.R. Ramli, M.T.
Musthafah, S.A. Shamsudin, “Review of research
on vehicles aerodynamic drag reduction methods”,
International Journal of Mechanical and
Mechatronics Engineering, vol. 14, no. 2, pp. 35-
47, 2014.
Proceedings of Mechanical Engineering Research Day 2017, pp. 130-131, May 2017
__________
© Centre for Advanced Research on Energy
Car seat design using RULA analysis S. Mat1,2,*, M.A. Abdullah1,2, A.R. Dullah1,2, S.A. Shamsudin1,2, M.F. Hussin1
1) Faculty of Mechanical Engineering, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 2) Centre for Advanced Research on Energy, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
*Corresponding e-mail: [email protected]
Keywords: Car seat; optimised; RULA
ABSTRACT – This paper presents an ergonomics
study of car seat design using Rapid Upper Limb
Assessment (RULA) analysis. A car seat was designed
using CATIA and it is then analysed using RULA to
obtain the initial ergonomics data of the car seat. An
optimisation process is done subject to the initial data
obtained and it is then analysed using RULA analysis to
obtain the final score that is aimed to achieve action
level 1. As a result, the final design of the car seat is
optimized ergonomically based on the final score
obtained from 4 to 2 (Action level 1).
1. INTRODUCTION
Nowadays, there are so many types and designs of
car seat. The common research questions arise from the
car seat designs are such as; whether the seat is suitable
and comfortable enough for a driver to sit on, and
whether the seat meets customer’s requirements. The
design of car seat plays a big role for drivers to achieve
a better driving experience and also for their health.
Good seats should meet customer’s needs and consider
ergonomics criteria and also safety features especially
for a driver who frequently commutes from one place to
another. The seat also has to be firm and strong enough
to hold the person who sits on it.
Ergonomics is the applied science of designing and
developing equipment like a workstation layout and
work strategies that best suit and protect the human
body. The goal of ergonomics is to minimize fatigue,
discomfort, injury, and emotional stress. Derived from
the Greek words ergo (work) and nomos (natural laws),
ergonomics literally means the laws of work.
Ergonomics defined by Fernandez [1] as the design of
the workplace, equipment, machine, tool, product,
environment, and system, taking into consideration the
human’s physical, psychological, biomechanical, and
psychological capabilities, and optimising the
effectiveness and productivity of work systems while
assuring the safety, health, and well-being of the
workers. In a nutshell, ergonomics encompasses the
relationship between humans, machines systems, job
design and the work environment. By approaching work
practices (stretching, reaching, or sitting) from an
ergonomically correct point of view, a worker actually
becomes stronger, healthier and more productive. If
management does not address ergonomics discomfort, a
worker will act on a subconscious level, adapting
his/her behaviour to lighten the pain [2]. Engineering
ergonomics focuses on the fit between a person’s body
size and physical capabilities (also known as
“anthropometrics”) and design of job task and
workspace. Engineering ergonomics can provide
recommendations on how to set up a work space. While
this approach plays a significant role in determining
design and furnishings in the office, it is limited by its
exclusive focus on the physical mechanics of work [3].
Depending on what a driver needs for his or her
comfort; seats can come from a wide range of selection.
The popular types of driver seats are bucket seats, bench
seats, and racing seats. A bucket seat is a seat contoured
to hold one person, distinct from bench seats which are
flat platforms designed to seat multiple people. A bench
seat is a driver’s seat that extends to connect the front
seats, forming a long bench that runs the width of the
car. Racing seats resemble bucket seats, only with a
much deeper base and additional equipment. One way
to identify a racing seat is by the bolsters, which are
wide elements on the sides of racing seats that hold the
driver in place during sharp turns.
2. METHODOLOGY
The Rapid Upper Limb Assessment (RULA)
provides an analysis of the manikin's upper limbs based
on variables such as weight, distance and frequency.
RULA is a survey method developed for use in human
factor or ergonomic investigations of workplaces where
work related upper limb disorders are reported. RULA
is a screening tool that assesses biomechanical and
postural loading on the whole body with particular
attention to the neck, trunk and upper limbs. RULA is
intended to be used as part of a broader ergonomics
study [4].
The RULA was developed and particularly used to
evaluate the risk to workers engaged sedentary tasks
such as workers at video terminals or the risk from other
tasks in which the operators sits or the risks for
workplace in which the worker stands for a large part of
the time. The analysis input data are body posture (head,
trunk, and upper limbs), the force used, and the type of
movements or actions performed, repetitively [4].
RULA analysis is used to analyse many facets of
manikin posture based on a combination of
automatically detected variables and user data. For this
study, certain parts of manikin’s body are adjusted
according to seat design and not all parts of the body are
Mat et al., 2017
131
moved. This is because the body parts that undergo
evaluation are back spine, leg, neck and head.
Table 1 shows the RULA action level and its
description. The score is given as number scale from
lowest score 1 to the highest score 7. Each action level
has its own descriptions which explain about what are
the consequences if the analysis meets the allocated
score.
Table 1 RULA scoring and descriptions [4].
Action
Level
RULA
Score Description
1 1-2 The person is working in the best
posture with no risk of injury from
their work posture
2 3-4
The person is working in a posture
that could present some risk of
injury from their work posture,
and this score most likely is the
result of one part of the body
being in a deviated and awkward
position
3 5-6
The person is working in a poor
posture with a risk of injury from
their work posture, and the
reasons for this need to be
investigated and changed in the
near future to prevent an injury
4 7+
The person is working in the
worst posture with an immediate
risk of injury from their work
posture, and the reasons for this
need to be investigated and
changed immediately to prevent
an injury
3. RESULTS AND DISCUSSION
The final concept design consists of less complex
design was selected (concept 5 from 6 concepts) using
weighted rating method as shown in Table 2. All the
curves and edges are minimal and not too complex as it
is one of the best design. The backrest and the headrest
are separated and this means the headrest can be
independently adjusted up and down according the
driver’s comfort. Furthermore, this gives better support
for the driver’s back.
Table 2 Weighted rating method.
Based on the RULA analysis, the final score value
is 4 which trunk, neck and leg are the main contributors
for this score and this means the person is working in a
posture that could present some risk of injury from their
work posture. This score most likely is the result of one
part of the body being in a deviated and awkward
position. Hence, this should be investigated and
corrected [4]. It is found that the back rest is larger than
driver’s back.
Figure 1 RULA analysis after improvement.
After improvement and optimization, the final
score has achieved 2 that is lower than the previous
design as shown in Figure 1. It indicates that posture in
this design is acceptable where the person is working in
the best posture with no risk of injury from their work
posture [4]. Regarding to redesigned process
(optimisation), the height and width of seat back is
reduced so that the manikin’s body (based on Asian
anthropometric data) is placed perfectly on the seat and
the whole body is fully supported. In addition, some
portion of the side bolsters has also been removed to
reduce its breadth.
4. CONCLUSION
The proposed design was developed using CATIA
and then it is analysed ergonomically using the RULA
analysis to obtain the final score. The initial design is
then optimised with some changes in the seat back
height and width, as well as the shrinking of the
bolsters. By using RULA analysis, the ergonomics
design of car seat is proposed with the reduction of final
score from 4 to 2. This indicates that the driver is
working in the best posture with no risk of injury from
their work posture.
REFERENCES
[1] J.E., Fernandez, “Ergonomics in the workplace,"
Facilities, vol. 13, no. 4, pp. 20-27, 1995.
[2] Z.M., Makhbul, D., Idrus, M.R., Abdul Rani,
“Ergonomics design on the work stress outcomes,”
Jurnal Kemanusiaan, vol. 9, pp. 50-61, 2007.
[3] K., Brookhuis, A., Hedge, H., Hendrick, E., Salas,
and N., Stanton, “Handbook of Human Factors and
Ergonomics Models,” Florida: CRC Press.
[4] L., McAtamney, and E.N., Corlett, “RULA: A
survey method for the investigation of work-related
upper limb disorders,” Applied Ergonomics, 24(2),
pp. 91-99, 1993.
Proceedings of Mechanical Engineering Research Day 2017, pp. 132-133, May 2017
__________
© Centre for Advanced Research on Energy
Automated sorting system design in manufacturing industries S. Mat1,2,*, M.A. Abdullah1,2, F.R. Ramli1,2, S.I. Abdul Kudus1,2, A.A. Abdul Aziz1
1) Faculty of Mechanical Engineering, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 2) Centre for Advanced Research on Energy, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
*Corresponding e-mail: [email protected]
Keywords: Automated sorting; recycling; optimized
ABSTRACT – This paper presents the conceptual
design of the automated sorting system in
manufacturing industries for plastic recycling that aims
to optimize the dimensions and space requirement for
the proposed design. The design process begins by
concept generation and selection. Three conceptual
designs were produced and then one was selected by
using concept screening method. The final design was
developed so that it optimizes the space usage when
installation for low to medium size of operation and it is
also having a suitable sorting rate that required by low
to medium size recycling industries.
1. INTRODUCTION
Automated sorting system is a process to sort
mixed items, colours or sizes like plastics, boxes, letters,
and clothes automatically using selected system.
Automated sorting system in manufacturing industries is
used instead of manual sorting system because it is
more reliable, cost effective and efficient. Manual
sorting relies on plant personnel who visually identify
and pick plastic bottles as they travel along the
conveyor belt [1]. These bottles are then sorted into the
respective containers. Manual sorting may not be a
suitable option for recycling facilities of high
throughput. It has also been noted that the high turnover
among sorting line workers had caused difficulties in
achieving consistency in the plastic separation process
[2].
Automated sorting systems employ a detection
system or a combination of detection systems to identify
the different types of recyclable plastic. These detection
systems utilize state-of-the art technologies to
automatically sort plastics either by resin type, shape,
colour or both. Automated sorting system of plastics for
recycling can be the solution to sort the plastics for
recycles in a lot of amount. The automated sorting
system of plastics for recycle has already existed in the
industries as shown in Figure 1. Some systems may
have very high-tech sorting methods. While others use
simpler ones. In this case, the performance of the
automated sorting systems may vary to each other.
Meanwhile, some existing systems have a big structure.
This might cause the space uses to install the system
become larger. More space was used to locate
machineries of the system. Thus, the proposed design of
the system is focus on low to medium size of recycling
industry that need to be less spacious. The objectives of
this project are to design automated sorting system in
manufacturing industries of plastics for recycling and to
optimize the dimensions and space requirement for
proposed design.
Figure 1 Existing of plastic and colour sorting system
2. METHODOLOGY
The beginning step to design an automated sorting
system is by generating the conceptual design. A
concept usually expressed as a sketch or as a rough
three-dimensional model and is often accompanied by a
brief textual description [3].
Figure 2 shows conceptual design A that uses two
conveyors. First conveyor is to move recyclables from
unsorted container onto the conveyor belt. The second is
conveyor belt that transmits the recyclables to the
sorting area. The ejection method uses multiple
compressed air guns. There are two sensors mounting
above and crossing the conveyor belt. Conceptual
design B uses hopper placed at beginning of the
conveyor as shown in the Figure 3. The sensor placed at
the side and followed by the ejector. The non-PET
recyclables will be ejected into the non-PET bin located
opposite to the ejector and PET recyclables will
continue to flow until it drops into PE bin located at the
end of the conveyor.
For conceptual design C (Figure 4), sensors are
placed across the conveyor. It uses compressed air
ejector to eject non-PET recyclables away from
conveyor flow into the non-PET bin. The hopper is
place beside the conveyor. It has plates to control the
recyclables flow rate and slide to allow recyclables exit
onto the conveyor.
From the concept screening matrix shown in Table
1, conceptual design C has the highest score and is
ranked in the first place. Therefore, it will be chosen for
the detail design process and development.
Mat et al., 2017
133
Figure 2 Conceptual design A.
Figure 3 Conceptual design B.
Figure 4 Conceptual design C.
Table 1 Concept screening matrix.
Concept Existing
design Selection criteria A B C
Size - + + -
Ease of use 0 0 0 +
Sorting method + - - +
Ejection method - - + 0
Hopper position + + - -
Ergonomics 0 0 0 0
Safety 0 0 + +
Durability + 0 0 +
Sorted container position - + + 0
Sensor position + - + 0
Sum +’s 4 3 5 4
Sum 0’s 3 4 3 4
Sum –‘s 3 3 2 2
Net score 1 0 3 2
Rank 2 3 1 -
3. RESULTS AND DISCUSSION
The assembly design of the system consists of
combination of single part and sub assembly is
visualized as in Figure 5. There are some development
activities done in designing the system such as
designing the slider of the hopper to have the
mechanism to control the flow rate of the recyclables.
The vision detecting system (sensor) is used for plastic
detection method that capable to identify type of plastic
bottles. The proposed design has a compact size and
reduce the requirement for space usage. The proposed
design of the system is 2.5 m in length and 1.2 m width.
It will utilize the area for placing just about 3 m².
Compared to the existing design which has 10.5 m
length and 4.5 m width and utilize the area of 47.25 m².
Therefore, the proposed design has managed to reduce
the space usage requirement for placing the machine.
Figure 5 Final design.
When review on the sorting rate, the proposed
design has 900 kg/hour of sorting rate. It is relatively
low compare to the existing design which has 2000
kg/hour of sorting rate. This is because the sorting and
ejection method for both designs are different. The
existing design uses multiple ejectors and can have the
multiple flows of the recyclables. For the proposed
design, it just has single ejector and only can manage
single recyclables flow. However, the sorting rate of
proposed design is acceptable for the low to medium
recycling industry.
4. CONCLUSION
From the data and information gathered through
literature review, conceptual designs of the system were
generated. By using concept screening matrix, concept
C was selected and developed so that it optimizes the
space usage when installation for low to medium size of
operation. In addition, the proposed design of the
automated sorting system for plastic recycling has a
suitable sorting rate required by low to medium size
recycling industry.
REFERENCES
[1] D. A., Wahab, A., Hussain, E., Scavino, M.M.,
Mustafa, and H., Basri, “Development of a
prototype automated sorting system for plastic
recycling”, American Journal of Applied Science,
vol. 3, pp. 1924-1928, 2006.
[2] P., Dinger, “Automated microsorting for mixed
plastics”, BioCycle, vol. 33, pp. 79-82, 1992.
[3] K. T., Ulrich, and S. T. Eppinger, “Product Design
and Development”, 3rd ed. University of
California: Mc Grow Hill. pp 97-136; 2003.
Proceedings of Mechanical Engineering Research Day 2017, pp. 134-135, May 2017
__________
© Centre for Advanced Research on Energy
Flexible shield for impact resistant purpose: A conceptual design H.M.S. Firdaus1,2,*, M.Y. Halyani1,3, M.I.H.C. Abdullah1,2, O.M. Rafi1,2
1) Faculty of Engineering Technology, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 2) Centre for Advanced Research on Energy, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
3) Centre for Robotics and Industrial Automation, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
*Corresponding e-mail: [email protected]
Keywords: Protective shield; impact resistant; RULA-analysis
ABSTRACT – The aim of this research is to develop
conceptual design of a shield for impact resistant. The
methodology used are customer survey, house of
quality, morphological chart, Rula-analysis and Pugh
method. Slide-in shield has been selected as conceptual
design based on material suitability since it used
polycarbonate that has very high Izod impact strength
between 600-850 J/m. In addition, it is also ergonomics
with the final score of 3 in Rula-Analysis for kneeing
and standing position. The main advantage of this shield
is the adjustable total length which lies in between 60-
100 cm.
1. INTRODUCTION
Shields were often the trusted defence of
infantrymen and knights alike throughout history and
this certainly did not change during the middle Ages.
Strong shields were used to block both attacks and
projectiles from harm to the soldier. After that, the plate
armour became a commonplace form of protection. It
was basically layered with the paste and tanned to make
it tougher [1]. Moreover, other equipment that compiles
with the shield were short throwing spear, open-face
helmet, and a sword [2]. Nowadays, demonstrations,
public disorder, and riots always happen in daily life for
a number of reasons such as economic hardships, social
injustices, ethnic differences (leading to oppression),
objections to world organizations or certain
governments, political grievances, and terrorist acts [3].
In certain years, the lightweight shield earned high
demand in the industry to lead to the personal
protection. Materials such as non-metallic materials,
ceramics and composites, have been increasingly
incorporated into more efficient lightweight armours
[4]. There are many factors that contribute to the quality
of the shield which includes size, thickness, and
material. The size varies greatly because some are
intended to protect the whole body, and others are meant
to be easily moved to cover specific parts at any given
moment. The objective of this study is to generate a new
conceptual design of a shield that is flexible and
lightweight to be used in accordance with the body of
the users.
2. METHODOLOGY
The project started by conducting customers
survey in defining customers needs. The feedback
received was generally in terms of maintenance,
material, shape, weight, handling, and function. The
needs were compared to engineering characteristics in
producing product design specification. Concept
generation was started by defining function analysis in
the morphological chart. Then, the best concept was
chosen from the analysis using Pugh Method [5]. In the
end, the body posture of the user in standing and
kneeing position was analyzed using Rula-analysis.
Figure 1 shows the details flow for the conceptual
design.
Figure 1 Flowchart for flexible shield conceptual
design.
3. RESULTS AND DISCUSSION
Figure 2 shows the House of Quality for the shield.
There are 8 customers’ needs to be discussed which are
costs, lightweight shield, flexibility, ease to carry, ease
of storage, ease of use, comfortability, and durability.
All of the information are obtained from several ways
includes brainstorming, historical data, literature review,
and a deep discussion with experts [6].
Firdaus et al., 2017
135
Figure 2 House of quality.
Product features also consist 8 features which are
impact strength, ergonomic factor, size, thickness, mass,
added function, portability, and safety. According to
George and Linda [5], engineering characteristics can be
ranked by summing up all relationships of each
characteristic. The most importance engineering
characteristics led by impact strength and weight of
shield. They were followed by size and ergonomics
factor that shared the same rank. Weight is also
important because it is strongly related to ergonomics
factor in terms of shield handling by the users. There are
5 types of shield used for ergonomics analysis. Figure 3
shows mannequin in standing and kneeing positions.
Figure 3 Rula-analysis in standing and kneeing position.
Figure 4 Score for kneeing position.
For standing position, wrist and arm has a score of 3 for
slide in shield. The final score is also 3, which is less
compared to the other types of shield. While for kneeing
position, both posture and wrist and arm have a score of
4. However, the final score of 3 showed that the shield
weight and positioning are still acceptable for the users.
Figure 5 Score for standing position.
4. CONCLUSIONS
Impact strength is the most important criteria in
creating shield with high impact resistance.
Polycarbonate has been selected as material because it
possesses high impact strength. It also has considerably
low mass of approximately 1.44 kg. In addition, the
height of the shield is adjustable between 60 cm to 100
cm because slide in type has been chosen for the design.
The design has made it flexible for all type of users. In
terms of ergonomics based on Rula-analysis done in
standing and kneeing position, the final score for slide
in shield is 3. This average score is acceptable and can
still be improved.
5. ACKNOWLEDGEMENT
PJP/2015/FTK(34C)/S01459.
6. REFERENCES
[1] H. Adam, K. Jeremy, R. Daniel, and S. Alex, The
evolution of arms and armors during the crusades,
Thesis; 2013.
[2] K. Donald and F.V. Gregory, Men of bronze:
hoplite warfare in ancient Greece, Princeton
University Press; 2013.
[3] R. Joseph, Security Officer Study Guide, Lulu
Enterprises; 2014
[4] Z. Fawaz, W. Zheng, and K. Behdinan, “Numerical
simulation of normal and oblique ballistic impact
on ceramic composite armours,” Composite
Structures, vol. 63, no. 3–4, pp. 387–395, 2004.
[5] E. D. George and C. S. Linda, Engineering Design,
Fourth Edition, McGraw-Hill, 2009.
[6] S. Praveen, “House of Quality; An effective
approach to achieve customer satisfaction &
business growth in industries,” International
Journal of Science and Research, vol. 5, no. 9, pp.
1365-1371, 2016.
Proceedings of Mechanical Engineering Research Day 2017, pp. 136-137, May 2017
__________
© Centre for Advanced Research on Energy
The analysis of design product preferences using curve fitted profiling
H. Sihombing1,*, W.N. Hidayah1, S. Shamsuddin 2, M.Y. Yuhazri1
1) Faculty of Manufacturing Engineering, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 2) Faculty of Mechanical Engineering, Universiti Putra Malaysia,
43400 UPM Serdang, Selangor Darul Ehsan, Malaysia.
*Corresponding e-mail: [email protected]
Keywords: Kansei engineering; bezier curve; mathematical approach
ABSTRACT – This study discussed about the
characteristics of product design based on customer
emotional preferences related to the ‘Kansei
Engineering’ towards ‘the Curve Fitted Profiling' using
mathematial approach. The focus of this project is
identify and analyze the customer preferences
compared to the image manipulated of product into
Bezier curves. This study use 20 city cars design in the
developed questionnaires were distributed in Malaysia.
From the data gathered, this study found that the
customer characteristics can be identified through the
certain variables of the Bezier curves equation. In this
study, the mostly preferences of city car profile design
is VW car.
1. INTRODUCTION
In today's aggressive market, the company has to
carry out the relevant strategy in order to fulfill their
customers' needs [1]. All companies have to, therefore,
prioritize their 'real extraordinary' consciousness to
catch customer design preferences before to produce a
certain product. According to Hsiao et al., this is due to
the paradigm of product designs that have to move from
the production-oriented approach to a marketing-
oriented approach, and finally to a customer-oriented
approach [2].
In addition, since the customers are not only to
demand the quality of product (but also their
satisfaction in psychological emotion conditions about
the product to be purchased), it is therefore, important
for producer to employ the improvement approaches in
their product development in order to satisfy their
customer needs and feelings. The companies that are
able to develop a product to fulfill their customers’
needs (that meet their psychological needs) will be
benefited to a profitable return when the product goes to market [3]. In this perspective, the Kansei Engineering
methods have been developed to support the valuation
of customer’s satisfaction in order to understand the
customer’s needs and desire [4]. However, this is only
limited on how to interpret the customer emotional
articulation refers to design of product. This study,
therefore, suggest the interpretation of the design
product related to profile using mathematical approach
related to profile, that is Bezier curve.
2. METHODOLOGY
Figure 1 shows the car design profiling
identification and measurement through segmentation.
By using segmentation into 4 quadrants, this study
activates the fitted curve profiling using 2 Bezier curve
for every quadrant.
Figure 1 Profiling the segmented of product car design
(side view).
Every cars profile was measured using curve fitted
to determine each mathematical values of every Bezier
curve. In this study, every car design is interpreted into
8 cubic Bezier curve. The cubic Bezier equation as
follows:
Bn (t) = An(1 - t )3+3Bn(1 - t)2 t +3Cn(1 - t)t2+Dnt3 …. (1)
Where n=1, 2,..,8 represents of each x-axis and y-axis
based on Bezier (n) equation.
3. RESULTS AND DISCUSSION
Table 1 shows the numbers of survey respondents
towards the car design preferences in table 2. Based on
survey distributed to 168 respondents at 5 states in
Malaysia (which are Kota Bharu (KB), Temerloh (TM),
Kemaman (KMM), Kuala Terengganu (KT) and Melaka
(MLK)), the highest number of choices is VW Polo,
followed by Toyota Etios Liva, and Toyota Aygo (Table
3).
Sihombing et al., 2017
137
Table 1 Total sample of 168 respondents.
State KB TM KMM KT MLK
Total Sample 30 32 30 30 46
Table2 Name of 20 city car design.
No. Car brand No. Car brand
1. Perodua Axia 11. Suzuki Splash
2. Mitsubishi Mirage 12. Toyota Etios Liva
3. Toyota Aygo 13. Opel Adam
4. VW Polo 14. Ford KA
5. Hyundai i10 15. Fiat 500
6. Nissan Micra 16. Geely Panda
7. Aston Martin Cynet 17. Kia Picanto
8. Chevrolet Spark 18. Peugeot 108
9. Datsun Go 19. Fiat Punto
10. Honda Brio 20. Chery QQ
Table 3 Selection of city car (side view) at 5 states.
CAR NAME KB TM KMM KT MLK TOTAL
VW Polo 15 8 14 10 18 65
Toyota Etios Liva 20 10 9 10 49
Toyota Aygo 10 12 13 35
Mitsubishi Mirage 9 14 23
Honda Brio 11 11 22
Based on the curve variables from each of 8 Bezier
equations towards 20 cars design, the profiling is
constructed through the normalization values (based on
scale defined) towards the average values. Since the
range value is maximum value minus minimum value,
the normalization is conducted using equation as below:
∑N
∑
BnRange
=Min -Max (2)
Where Bn is the parameter values of Bezier, n is the
numbers (n=1,2,…8), while N is the total of values of
Bezier parameter based on x-axis and y-axis.
In this study, the segmentation of every variable
values based on single individual Bezier curve were
scaled in 5 distance ranges. Based on the most car
design choices (for an example), this study found as
shown in Table 4 in which the range of the car
preferences (using Bezier 1) is in the yellow color
marks (or 1). This meant that the specific parameter
values which make the VW Polo chosen rather than else
is due to the values of D1 in x-axis for Bezier 1, which is
in the range 2. Whiles, towards the values of C1 for x-
axis of Bezier 1 is related to the choices of VW Polo
and Mitsubishi Mirage as the car design preferences of
Melaka respondents which is influenced also by the
value of A1 in y-axis.
Table 5 showed that the choice of VW Polo is also
due to variable of A2 and B2 in x-axis. Whiles, C2 and
D2 in y-axis were related to design of Toyota Etios Liva
Melaka respondents chose none.
Table 4 Range for Bezier 1.
Car Name X - Axis Y - Axis
A1 B1 C1 D1 A1 B1 C1 D1
VW Polo 5 3 2 2 5 5 5 5
Toyota Etios Liva 2 1 1 1 3 2 1 1
Toyota Aygo 2 2 1 1 3 2 2 3
Honda Brio 1 1 1 1 3 5 4 5
Mitsubishi Mirage 4 2 2 1 4 2 2 4
Table 5 Range for Bezier 2.
Car Name X - Axis Y - Axis
A2 B2 C2 D2 A2 B2 C2 D2
VW Polo 4 4 3 3 4 4 2 2
Toyota Etios Liva 2 1 1 1 1 1 1 1
Toyota Aygo 2 1 2 2 2 2 2 2
Honda Brio 2 1 3 2 3 3 2 2
Mitsubishi Mirage 2 1 2 3 2 2 2 2
4. CONCLUSIONS
This study found that the manipulation process
towards the design profile into curve profile will help to
depict the certain values of curve parameters that
represent the uniqueness of respondents' choices based
on the profile design of product. Using car product as a
case study, the distributed survey to 5 states (in
Malaysia) said that VW Polo has the highest choices
based on Kansei Engineering related to side view. This
can be tracked refers to Bezier curve fitted towards the
product profiles. This study, however, need to be
extended to further investigate relate to design features
with many different views, such as front view, back-rear
views, etc.
ACKNOWLEDGEMENT
The authors would like to thank CRIM-UTeM and
RMC-UPM. This study is supported by Research Grant
RACE/F3/TK6/FKP/F00301.
REFERENCES
[1] N. Cross, Engineering Design Methods: Strategies
for Product Design, 3rd ed. Chichester, UK: Wiley;
2000.
[2] S.W. Hsiao, F.Y. Chiu, S.H. Lu, “Product-form
Design Model Based on Genetic Algortihms,”
International Journal of Industrial Ergonomics,
Vol. 40, No. 3, pp. 237-246, 2010.
[3] M.D. Shieh, T.H. Wang, C.C. Yang, “A Clustering
Approach to Affective Response Dimension
Selection for Product Design,” Journal of
Convergence Information Technology, Vol. 6, No.
2, pp. 197-206, 2011.
[4] A.M. Lokman, “Design & Emotion: The Kansei
Engineering Methodology,” Malaysian Journal of
Computing, Vol. 1, No. 1, pp. 1-11, 2010.
Proceedings of Mechanical Engineering Research Day 2017, pp. 138-139, May 2017
__________
© Centre for Advanced Research on Energy
Optimization of process parameters variation on ION and VTH in n-channel double gate FinFET device
N.K. Norddin, F. Salehuddin*, N.R. Mohamad, Norhisham Mansor, A.S.M. Zain, K.E. Kaharudin,
A.H. Afifah Maheran, A.R. Hanim, H. Hazura, S.K. Idris
Centre for Telecommunication Research and Innovation, Faculty of Electronics and Computer Engineering,
Universiti Teknikal Malaysia Melaka, Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia.
*Corresponding e-mail: [email protected]
Keywords: Threshold voltage; drive current; Taguchi method
ABSTRACT – In this research, we investigated the
process parameters variation on drive current (ION) and
threshold voltage (VTH) in n-channel Double Gate
FinFET device. There are several input process
parameters were investigated such as poly doping dose,
poly doping tilt, S/D doping dose, S/D doping tilt, VTH
doping dose and VTH doping tilt. The setting of process
parameters was determined by using L18 orthogonal
array in Taguchi Method. From n-channel Double gate
FinFET MOSFET device, the VTH and ION obtained after
the optimal approach is 0.479V and 1.31mA/µm
respectively. The results obtained are well within the
ITRS 2013 prediction for the electrical characteristics.
As a conclusion, the optimal solution for the robust
design recipe of the device was successfully achieved.
1. INTRODUCTION
The Planar MOSFET structure is reaching its
scaling limits and alternative devices are being
investigated to further increase the transistor density and
performance of the device. However, as the transistor
shrunk, the close proximity between the source and the
drain reduces the ability of the gate electrode to control
the potential distribution and the flow of the current in
the channel region and produce undesired effect called
the short channel effects (SCEs). In a continuous effort
to increase the current drive and reduce the SCEs, the
MOSFET has evolved from the classical planar single
gate device to multiple gate structure device either
double-gate, triple gate or quadruple gate [1]. Among
the double gate MOSFET, FinFET have emerged as the
most desirable alternatives to MOSFET due to its
simple structure and ease of fabrication. Process
variations have become a critical issue in today’s high-
performance circuit design as technology keeps scaling
down. As technology continues to scale down, process
variation has a more significant impact on circuit
performance. Small dimension devices operating at low
supply voltages show an in increased sensitivity to
parameter variations. It is, therefore, essential to
characterize and control the parameter fluctuations to
improve the performance and yield of integrated
circuits. The increased process parameter variation has
been recognized as one of the major roadblocks to
further technology scaling. According to Kaharuddin et.
al. [2], one of the most systematic and efficient ways to
achieve a robust design is to use an optimization method
of designing experiments based on Taguchi Methods.
Taguchi Methods provide the most efficient and viable
solution in such cases with minimal experimental trials
[3]. In this work, n-channel Double Gate FinFET device
is developed by using Silvaco TCAD simulation
software, where the effects of process parameter
variation are investigated using L18 orthogonal array in
Taguchi method. This method is utilized to reveal the
ideal level of process parameter for the higher ION value
of the Double Gate FinFET device. At the end of study,
it was found that the optimum solution closer to the
desired value as well as within the ITRS 2013 for the
year 2016 [4].
2. MATERIALS AND METHODS
The research is based on simulation and program
development and physical modelling of nano device
performance. A statistical method that is used in this
research needed to run several experiments with
different combination of the parameters to get the
desired output response that fulfills the statistical
method requirement. There are two noise factors were
varied for two levels to get four readings of the output
responses for every row of experiments. The values of
the noise factor at the different levels are listed in Table
1. The process parameters and their level that has been
identified for the experiment are shown in Table 2.
Table 1 Noise Factors and their level. Symbol Noise Factor Unit Level 1 Level 2
Y Gate Oxide Temperature °C 875 880
Z Poly Oxide Temperature °C 910 915
Table 2. Process Parameter and their level
Sym Process
parameter
Unit Level 1 Level 2 Level 3
A Poly Doping
Dose
Atom
cm-3
3.50e14 3.60e14 -
B Poly Doping
Tilt
Degree 15 16 17
C S/D Doping
Dose
Atom
cm-3
1.12e18 1.22e18 1.32e18
D S/D Doping
Tilt
Degree 73 74 75
E VTH Doping
Dose
Atom
cm-3
1.85e13 1.95e13 2.05e13
F VTH Doping
Tilt
Degree 6 7 8
Norddin et al., 2017
139
3. RESULTS AND DISCUSSION
Figure 1 shows the two dimensional (2D) double
gate structure of a n-channel FinFET MOSFET device.
In this research, VTH of the device belongs to the
nominal-the-best quality characteristics and ION belong
to the larger-the-best quality characteristics. The S/N
ratio (SNR) is selected for VTH value must be closer or
equal to a given target value, that is 0.453V. Meanwhile
the S/N ratio that is selected for ION value must be the
largest value as possible. The SNR (dB) for each level
of the process parameters and factor effects on SNR (%)
for VTH and ION is summarized in Table 3. According to
this table, Factor B (35%) and Factor E (86%) was
found to be the major factor affecting the VTH and ION
respectively. Basically, the larger the SNR, the quality
characteristic of VTH and ION are better [5].
Figure 1 2D double gate structure of a n-channel
FinFET MOSFET device.
Table 3. SNR for threshold voltage and drive current.
Ou
tpu
tt
Res
po
nse
ss
Pro
cess
Para
met
er SNR (dB)
Fact
or
Eff
ects
on
SN
R (
%)
Level 1 Level 2 Level 3
Th
resh
old
Vo
ltage
(Nom
ina
l-th
e-b
est)
A 26.12 25.33 - 20
B 25.06 25.76 26.34 35
C 25.69 25.65 25.83 0
D 25.39 25.73 26.04 9
E 25.14 26.21 25.80 25
F 25.83 25.67 25.66 0
Dri
ve
Cu
rren
t
(La
rger
-th
e-b
est)
A -26.89 -25.39 - 0
B -26.92 -27.56 -23.94 14
C -25.32 -27.75 -25.35 0
D -25.42 -27.68 -25.31 0
E -31.75 -25.10 -21.57 86
F -25.35 -25.33 -27.75 0
The individual optimum condition for VTH and ION
were compared to choose the best optimization value for
each of the process parameters. The optimal value is
chosen due to the high percentage contribution of SNR
for each process parameter [4]. The best set of the
process parameters for device that had an effect on VTH
and ION is A1,B3,C3,D2,E2,F2. Once of the optimal
level of the process parameter is selected, the
confirmation test is performed to verify the accuracy of
the statistical method. The SNRs of VTH and ION for a
device after optimization approaches are 27.70dB
(28.47 to 26.78dB) and 2.31dB (2.41 to 2.21dB)
respectively. The values are within the predicted range.
Table 4 shows the comparison between the estimated,
simulation and the ITRS 2013 prediction values. This
indicates that the simulation value is closer to the
estimated value. Therefore, this verifies that the
simulated value is highly correlated with the estimated
result. This value of VTH is closer to ITRS prediction
and still in range ±12.7% of the nominal (target) value,
0.453V. Meanwhile, ION of the device exceeds the
minimum value which is 681µA/µm as specified in
ITRS 2013 for the year 2015 [4].
Table 4 Comparison between the estimated, simulation
and the ITRS 2013 prediction values. Output
Response
Estimation Simulation ITRS 2013 for
the year 2015
VTH (V) 0.482 0.479 0.45312.7%
ION (mA/µm) 1.29 1.31 > 0.681
4. CONCLUSIONS
Through this study, the factors that most affect the
output response of 18nm n-channel Double gate FinFET
MOSFET devices have been identified and the optimum
factor levels were also determined. Polysilicon Doping
Tilt and Threshold Doping Dose were found to be the
most significant factor that affects the threshold voltage
and drive current of device respectively. The threshold
voltage and drive current for a device after optimization
approaches are about 0.479V and 1.31mA/µm
respectively. These values have reached the target.
ACKNOWLEDGEMENT
The authors would like to thank the Ministry of
Higher Education (MOHE) and CeTRI, Universiti
Teknikal Malaysia Melaka (UTeM) for sponsoring this
study under the research grant
(RAGS/1/2014/TK03/FKEKK/B00064).
REFERENCES
[1] Rajiv Sharma, “Analytical modelling of volume
inversion and channel length modulation in fully
depleted double gate nanoscale SOI MOSFETs,”
Journal of Electron Devices, vol. 18, pp. 1553-
1563, 2013.
[2] K.E. Kaharudin, A.H. Hamidon, F. Salehuddin,
“Design and optimization approaches in double
gate device architecture,” International Journal of
Engineering and Technology (IJET), vol. 6, no. 5,
pp. 2040-2079, 2014.
[3] M.S. Phadke, Quality Engineering Using Robust
Design, Pearson Education, Inc. and Dorling
Kindersley Publishing, Inc; 2001.
[4] ITRS 2013 report, http://www.itrs.net
[5] H.A. Elgomati, B.Y. Majlis, I. Ahmad, F.
Salehuddin, F.A. Hamid, A. Zaharim, P.R. Apte,
“Application of Taguchi method in the
optimization of process variation for 32nm CMOS
technology,” Australian Journal of Basic and
Applied Sciences, vol. 5, no. 7, pp. 346-355, 2011.
Proceedings of Mechanical Engineering Research Day 2017, pp. 140-141, May 2017
__________
© Centre for Advanced Research on Energy
Reliability testing of inertial measurement units in the analysis of physiological variables in archery
Z. Taha1, R.M. Musa1,2,*, M.R. Abdullah2, M.H.A. Hassan1, M.A.M. Razman1, A.P.P. Abdul Majeed1
1) Innovative Manufacturing Mechatronics and Sports Lab, Faculty of Manufacturing Engineering,
Universiti Malaysia Pahang, Pekan Campus 26600, Malaysia 2) Faculty of Applied Social Sciences, Universiti Sultan Zainal Abidin, 21300, Terengganu, Malaysia
*Corresponding e-mail: [email protected]
Keywords: Archery; inertial measurements units; physiological indicators
ABSTRACT – Archery is a static sport. Detection of
any movement is beneficial in ensuring shooting
accuracy. Thus, determining the reliability of any
detection instrument is paramount. This study aims to
ascertain the postural balance, hand movement,
muscular activation as well as heart rate of an archer
using Shimmer sensors. An archer was observed over
two different tests (A&B). Test A with free movement
while B with restricted movement.
Kolmogorov/Smirnov test was utilized to measure the
reliability of the sensors over test re-test in two different
tests. The Kolmogorov/Smirnov test re-test reveals a
significant difference between all the indicators in both
tests A and B, p < 0.001. IMUs sensors appear to be
reliable in measuring some physiological indicators in
the sport of archery.
1. INTRODUCTION
Current innovative advances reinforce the
utilization of inertial measurement units (IMUs) as a
practical alternative for the appraisal and measurement
of exercise performance beyond the motion analysis
laboratory [1]. These IMUs offer various potential
points of interest over conventional marker-based
frameworks; they are little, cost-effective, simple to set-
up and empower the appraisal of human movement in
an unconstrained situation [2]. This implies that these
universal advancements may have the capacity to
possibly quantify human movement and give feedback
with respect to the nature of the movement performed
[3].
IMUs have been utilized in various ways from
evaluating energy expenditure [4]; to gait analysis [5] to
medical observation [6]. These sensors have
additionally been utilized in the athletic field and sports
such as skiing [7] and golf [8]. Recently, the usage of
IMUs as a technique for tracking gym and rehabilitation
exercises have been examined. Lin and Kulić assessed
data collected from IMUs at the hip, knee and ankle
during various lower limb works out. Information from
the IMUs were utilized to estimate joint angles; with the
authors comparing the IMU-derived joint angles to
those quantified via a marker-based motion analysis
capture system [9]. Despite, the aforementioned
development, however, the reliability of such sensors
are often neglected or not reported.The purpose of the
present study is to test the reliability of IMUs sensors in
measuring the postural sway, hand movement, muscular
activation as well as heart during execution of archery
related techniques.
2. METHODOLOGY
A total of 4 IMUs Shimmer sensors were used in
the present study to determine the postural balance,
movement of the bow, muscular activations of the
muscle flexor digitorum and extensor digitorum as well
as the heart rate of the archer. The experimental
protocol was implemented in two parts. In first part
(Test A), the archer was instructed to sway from the
center of his gravity while holding the bow meanwhile,
in the second part (Test B), the archer was permitted to
limit the movement so as to enable the researcher to
discover whether the sensors have the ability to
differentiate the selected physiological indicators in the
two types of the postural positions. Two shimmer
sensors were attached to the left muscle extensor
digitorum and the right muscle flexor digitorum to
obtain Electromyography (EMG) signals during the
performances of the archery related movements
described previously. All the data were streamed in real
time at a sampling rate of 51.2Hz using an Android
phone and transmitted via Bluetooth for further analysis.
The areas of all the sensors attachments on the archer’s
body are shown in Figure 1.
Figure 1 IMUs sensors’ location attachments on the
archer’s body.
Taha et al., 2017
141
2.1 Data analysis
The Kolmogorov/Smirnov test was applied to
measure the reliability of the application over test re-test
between tests A and B of all the movements measured at
a confidence level of p ≤ 0.05. The data for the total of
five body actions were analyzed and evaluated using
MATLAB 2016a and XLSTART add in version 2014
USA for Windows.
3. RESULTS AND DISCUSSION
Table 1 indicates the inferential statistics of the
Kolmogorov/Smirnov test. Two periods of testing,
observations, D-statistics as well as p values are shown.
It can be extracted from the table that the p-value of all
the tests (1 - 5) were < 0.001 which explains that there
is statistically significant difference between tests A and
B of all measured actions. This confirmed the reliability
of the sensors in evaluating as well as discriminating the
two types of movement executed by the archer.
Table 1 Inferential statistics of the
Kolmogorov/Smirnov test.
Test-retest Obs. D P value
1. Bow Movement Test A 0.24 0.001*
Test B
2. Postural Sway Test A 0.49 0.001*
Test B
3. Muscle Ex.Activation Test A 0.3 0.001*
Test B
4. Muscle Flex.Activation Test A 0.27 0.001*
Test B
5 Heart Rate Test A 0.47 0.001*
Test B
*Significant at p < 0.001
The finding of the current study is in agreement
with previous researchers who reported that IMUs had
been used in various ways such as evaluating energy
expenditure, gait analysis and medical observation [4-
6]. These sensors have additionally been employed in
the athletic field and sports such as skiing and golf [7-8].
Moreover, the findings from the present study are also
congruent with the study conducted by other researchers
who assessed data collected from IMUs at the hip, knee
and ankle during various lower limb works out.
Information from the IMUs were utilized to estimate
joint angles; with the authors comparing the IMU-
derived joint angles to those quantified via a marker-
based motion analysis capture system. The researchers
presumed that these joint angles were precise when
contrasted with those acquired by means of the most
conventional methodology [9].
4. CONCLUSION
The current study has successfully evaluated
postural sway, hand movement, muscular activation as
well as heart rate attributed to sport of archery in two
different analyses. The sensors used in the study have
demonstrated high sensitivity in detection of any
movement executed by the archer which is beneficial in
analyzing any form of movement during both aiming
and releasing of arrow. The study has indicated that
inertial measurement units can be utilized to evaluate
movements employed by the archers. The sensors are
capable of providing information on every action
executed which can go a long way in helping the archers
to be aware of his/her movements and any incorrect
techniques to help improve performance.
ACKNOWLEDGEMENT
The researchers wish to thank the National Sports
Institute of Malaysia for providing the grant for the
current study (ISNRG: 8/2014-12/2014). The
researchers have no conflict of interest to declare.
REFERENCES
[1] A. Ahmadi, E. Mitchell, F. Destelle, M. Gowing,
N.E OConnor, C. Richter, K. Moran. “Automatic
activity classification and movement assessment
during a sports training session using wearable
inertial sensors,” 11th IEEE International
Conference on Wearable and Implantable Body
Sensor Networks, pp. 98-103, 2014.
[2] O. Giggins, D. Kelly, B. Caulfield. “Evaluating
rehabilitation exercise performance using a single
inertial measurement unit,” in Proceedings of the
7th International Conference on Pervasive
Computing Technologies for Healthcare. Institute
for Computer Sciences, Social-Informatics and
Telecommunications Engineering, pp. 49-56, 2013.
[3] I. Pernek, K.A. Hummel, P. Kokol. “Exercise
repetition detection for resistance training based on
smartphones,” Personal and Ubiquitous
Computing, vol. 17, no. 4, pp.771-782, 2013.
[4] E.S. Rawson, T.M. Walsh. “Estimation of
resistance exercise energy expenditure using
accelerometry,” Med Sci Sports Exerc, vol. 42, no.
3, pp. 622-8, 2010.
[5] J.J. Kavanagh, H.B. Menz. “Accelerometry: a
technique for quantifying movement patterns
during walking,” Gait Posture, vol. 28 no. pp. 1-
15, 2008.
[6] M. Zhang, A.A. Sawchuk. “A customizable
framework of body area sensor network for
rehabilitation,” in 2009 2nd International
Symposium on Applied Sciences in Biomedical and
Communication Technologi. Nov, 2009.
[7] F. Michahelles, B. Schiele. “Sensing and
monitoring professional skiers,” IEEE Pervasive
Computing, vol. 4, no. 3, pp. 40-45, 2005.
[8] H. Ghasemzadeh, V. Loseu, R. Jafari. “Wearable
coach for sport training: A quantitative model to
evaluate wrist-rotation in golf,” Journal of
Ambient Intelligence and Smart Environments, vol.
1, no. 2, pp. 173-184, 2009.
[9] J.F. Lin, D. Kulić. “Human pose recovery using
wireless inertial measurement units,” Physiol
Meas, vol. 33, no. 12, pp. 2099-115, 2012.
Proceedings of Mechanical Engineering Research Day 2017, pp. 142-143, May 2017
__________
© Centre for Advanced Research on Energy
Wing design for blended-wing-body aircraft M.H. Mat Yazik*, M.T.H. Sultan, A. Hamdan
Aerospace Manufacturing Research Centre, Faculty of Engineering, Universiti Putra Malaysia,
Selangor Darul Ehsan, Malaysia
*Corresponding e-mail: [email protected]
Keywords: Distribution; blended-wing-body aircraft; pro-verse yaw
ABSTRACT – Blended-Wing-Body (BWB) aircraft is
a new design pitched as the future of air transport.
However, due to the nature of the design that is
relatively new, more research has to be made to exploit
its advantages. This article studies an initial scaled
blended wing body and the effect of different lift
distribution on a BWB design. The initial design is
adjusted via twist to obtain desired wing loading
distribution. Aerodynamic optimization is performed on
Selig S5010 aerofoil to further improve the
aerodynamic performance of the BWB design.
1. INTRODUCTION
Reservoir and effect on global warming is the main
concern for both industry and environmentalist.
Although significant improvement in efficiency with the
introduction of winglets and high performance engine,
the conventional tube-wing configuration has reached
its limits. A step change in fuel efficiency may be
realised through un-conventional configuration. The
demand for better and more efficient aircraft calls for
blended-wing-body (BWB) design as next generation
aircraft design to tackle the problem. Compared to tube-
wing design, BWB has advantage aerodynamically
through lower wetted area to volume ratio and lower
interference drag. BWB also have better lift-to-drag
ratio compared to conventional design credit to the
fuselage that also contribute to the total lift of the
aircraft. Having low radar cross sectional area makes
them fit into military aircraft which require stealth
aircraft i.e. reconnaissance and bomber mission. For
example, the Northrop Grumman B-2 Spirit which
employ the flying wing design. However, the
advantages in aerodynamic performance can only be
realised through detailed shape design of the BWB
shape.
Despite having advantage over cruise efficiency,
BWB design has poor stability control due to the lack of
tail surface control. Problem arise when BWB is require
to perform a roll. The effect of induced drag over a
finite wing aircraft will produce adverse yaw when
aircraft perform a rolling motion. This is caused by
increase in induced drag on one wing which cause the
aircraft to yaw in opposite direction then intended roll
motion. The objective of this paper is to study the effect
of span wise distribution on aerodynamic efficiency and
overcoming adverse yaw problem. Research has been
made to enable control of flight dynamics through
implementation of surface deformation to produce
different in lift between the wings. This eliminate the
need for vertical control surface which would introduce
additional trim drag [1]. Qin et al. investigate a series of
aerodynamic study on span wise lift distribution of a
blended wing body aircraft within a European
Commission funded project [2-3].
2. METHODOLOGY
The initial BWB design is divided into three
sections which are
a. The center body: from 0 to 0.12m
b. A pair of mid-wing: from 0.12 to 0.25m
c. A pair of outer-wing: from 0.25 to 0.55m
The baseline design has root chord is 0.39m and
tip chord of 0.052m. Figure 1 shows the planform of the
design BWB, showing dimension of the section location
at each span location. The cruise operating condition is
stated in Table 1. All aerodynamic analysis is analyzed
based on cruise operating condition.
Figure 1 BWB Baseline geometric planform.
Table 1 Cruise design condition.
The baseline geometry has no twist and dihedral
applied to it. Selig S5010 aerofoil has been chosen
along the span because it is designed for low Reynolds
number and it has near zero pitching moment coefficient
which is suitable for longitudinal stability of BWB
design. The airframe is analyzed in XFLR5 using 3D
panel method to obtain its aerodynamic properties.
Different load distribution is applied to the baseline
BWB through geometric twist across the span to
observe the effect of different lift distribution on the
Cruise Condition Unit Value
Altitude m 11500
Speed kmh-1 100
Reynolds number 440 090
Design lift coefficient 0.203
Yazik et al., 2017
143
aerodynamic performance. The aerofoil profile were
further optimized for improvement of aerodynamic
performance. The aerofoil optimizer program XOptFoil
is used to optimize aerofoil shape at cruise operating
condition. All of the design are analyzed using Athena
Vortex Lattice (AVL) program to obtain the stability
derivatives yawing moment with respect to roll, 𝐶𝑛𝑝.
The analysis was made at 25ᵒ bank angle to obtain a
standard aircraft banking rate of 3ᵒ per second.
3. RESULTS AND DISCUSSION
Figure 2 shows the distribution local lift along
then
Half-span for cruise condition. The bell shape has a
slightly longer span compared to elliptical distribution
as proposed by Jones and Bower et al. for a solution to
achieve optimum structural efficiency with fixed total
lift and root bending moment [4]. The load distribution
is achieved through geometric twist throughout the
span. As we can see the bell-shaped distribution have
maximum local lift shifted toward the root section. The
bell-shape distribution has its max local lift coefficient
around 0.25m. The movement of max local lift
coefficient towards the root reduce potential for tip
stalling and reduce wave drag at high speed flight.
Figure 2 Local lift distribution across half-span.
Table 2 compares the aerodynamic performance of
each design at cruise condition. The elliptical perform
the best among the designs in term of induced drag due
and total drag as expected according to research made
by Prandtl in [5]. Aerofoil optimization prove to be
beneficial to the aerodynamic performance of BWB
design as we can see a decrease in drag coefficient and
induced drag coefficient.
Table 2 Aerodynamic performance of BWB designs.
BWB Design C L C D C Di
Baseline 0.20291 0.009781 0.002055
Elliptical 0.20294 0.009677 0.001995
Optimized
Elliptical 0.20296 0.008814 0.001987
Bell-Shaped 0.20301 0.011308 0.002793
Optimized Bell-
Shaped 0.20290 0.010161 0.002833
Contradict to the aerodynamic performance, the
pitching moment and pro-verse yaw performance in
elliptically loaded BWB is worse than the bell shaped
loaded BWB. The pitching moment obtained from
elliptical loaded wing is further from zero compared to
the others. This require surface control to trim the
aircraft which introduce additional trim drag to the
BWB. The design also has negative yaw-roll
derivatives. This indicate that the BWB will yaw in
opposite direction of rolling motion. This is usually
compensating by vertical tail. With bell-shape wing
loading, the stability derivative appears to be positive
thus does not require vertical tail.
Table 3 Moment coefficient and Stability derivatives of
BWB design.
BWB Design C m C np
Baseline -0.010615 -0.004785
Elliptical -0.016271 -0.006549
Optimized
Elliptical -0.020069 -0.002903
Bell-Shaped 0.010201 0.012558
Optimized Bell-
Shaped 0.008097 0.017857
4. CONCLUSION
From this experiment, it can be concluded that
both load distribution has advantage for BWB design.
BWB with elliptical load distribution have higher
aerodynamic performance and lower induced drag
compared to bell-shaped load distribution. This can
translate to lower fuel consumption, higher range and
more efficient cruise flight. On the other hand, bell
shaped wing loading have advantage over the elliptical
shaped loading as it provides pro-verse yaw to the BWB
design. This is beneficial for a BWB design which lack
vertical tail. A more efficient design is one that can
employ both distributions packed in a single BWB
design. As future works, a morphing concept for BWB
aircraft can be developed to realize the design in which
currently under investigation by the author.
ACKNOWLEDGEMENT
GP-IPB grant no. 9490602.
REFERENCES
[1] W. Wu, D. Chen, N. Qin, X. Peng, X. & X. Tang,
“A new efficient control method for blended wing
body,” International Journal of Modern Physics:
Conference Series, vol. 19, pp. 396-405, 2012.
[2] N. Qin, A. Vavalle, A. Le Moigne, M. Laban, K.
Hackett & P. Weinerfelt, “Aerodynamic
considerations of blended wing body aircraft,”
Progress in Aerospace Sciences, vol. 40, no. 6, pp.
321-343, 2004.
[3] N. Qin, A. Vavalle, & A.L. Moigne, “Spanwise lift
distribution for blended wing body aircraft,”
Journal of Aircraft, vol. 42, no. 2, pp. 356-365,
2005.
[4] R.T. Jones, “The spanwise distribution of lift for
minimum induced drag of wings having a given
lift and a given bending moment,” 1950.
[5] L. Prandtl, “Uber tragflugel des kleinsten
induzierten widerstandes,” Zeitshrift fur
Flugtechnik und Motorluftschiffahrt, 1933.
Proceedings of Mechanical Engineering Research Day 2017, pp. 144-145, May 2017
Improving manufacturing facilities design layout for a coffee production company
D. Ibrahim*, S.M. Hashim, A.I.S. Yusof, Y. Yaakob, N. Hussin, A.A. Azahari
Faculty of Mechanical Engineering, Universiti Teknologi MARA, Kampus Permatang Pauh, Pulau Pinang, Malaysia
*Corresponding e-mail: [email protected]
Keywords: Layout; system layout planning (SLP); machine utilization
ABSTRACT – This study is to assist a local coffee
production company to increase their production due to
higher demands of Robusta coffee. There is high flow
intensity between workstations which have high
interrelationship leading to high traveling time. Two
alternative layouts are proposed using the 11 steps in
Systematic Layout Planning (SLP). The proposed
layouts involve re-position the Cooking, Breaking,
Grinding and Sifting workstations for both layouts. The
layouts are then evaluated using DELMIA QUEST
simulation. The best alternative is chosen based on the
most significant improvement in term of machines
utilization and total output produced.
1. INTRODUCTION
A well-designed facility layout must provide great
connections between raw materials, labors, equipment
and finished goods. Other than that, it must be at lowest
possible costs and in safe working environments.
Systematic implementation can result a reduction of
10% to 30% of the company’s operating cost [1]. This
study intends to propose to a small local coffee making
company an alternative layout that is possible to provide
better machines utilization and increase the number of
output produced to meet their customer’s demand.
2. METHODOLOGY
Systematic Layout Planning (SLP) developed by
Richard Muther consists of eleven steps to develop
several solutions, that can be divided into six basics
steps which is chart of relationships, establish space
requirements, diagram activity relationships, draw space
relationship layouts, evaluate alternative arrangements
and detail the selected layout plan [2, 3]. To generate a
layout that meets the requirements, Delmia Quest is
used as a tool in this study to improve elements such as
bottlenecks and idle time that the SLP method failed to
identify and could not solve. Delmia Quest contains
material element for modeling such as machines, buffer,
process, failure rate, maintenance, labor, path and
material export, which can help users simulate and
analyze the process flow in 3D facilities environment
[4].
2.1 Systematic layout planning
The first step of SLP is gathering the input data
required for the case study. The input variables are
product (2 in 1 Robusta Coffee); output per day (200
boxes); routing in the company’s existing production
layout; service available including utilities, equipment,
restrooms, lockers, cafeteria, entrances, and exits; and
working time (8 hours/day). The results are tabulated in
an Activity Relationship Chart (ACR). The relationship
chart displayed which department are related to others
and it also rates the importance of the closeness between
them. The degree of closeness between departments was
rated by using Total Closeness Rating (TCR) as shown
in Table 1.
Table 1 Total closeness rating.
Value Closeness
A Absolutely Essential
E Especially Important
I Important
O Ordinary
U Unimportant
X Not Desirable
ACR in Figure 1 is generated where proximity and
relationships are visually evident. After constructing the
ACR, the space requirements needed for each process
can be determined for the actual layout. Systematic
Layout Planning (SLP) is best employed when creating
a new facility starting from scratch and the design was
not yet finalized. In this case study, the existing facilities
were established hence there was limited ability to
expand the area for extra space.
Figure 1 Activity relationship chart (ACR).
Ibrahim et al., 2017
145
2.2 Layout simulation
The effectiveness of the improved layout is
evaluated from total output and machine utilization.
The simulation is based on ten processes involved to
produce the product that shares with 22 machines which
required different process time. The simulation run time
is to reflect the actual scenario of one working shift
which of 420 minutes including one 60-minute lunch
break. In this study, three models are constructed and
simulates using the DELMIA QUEST V5. One is the
existing layout while the other two are the proposed
alternatives. The existing production layout of the
company was modeled to get the benchmark of the
machine utilization and the number of output produced
per day.
3. RESULTS AND DISCUSSION
For the first alternative layout, the cooking and
breaking departments are placed close to each other. The
grinding machines are placed in a straight line to reduce
material handling time after mixing process as well as
enable to place the sifter machines closer to it. The
roasting and packaging departments remain at the same
location. In the second alternative layout, the cooking
and breaking departments are rotated for 90ᵒ counter
clockwise to reduce traveling time for worker to load
the materials to the cooking machines. The grinding and
sifting departments remain the same as in Alternative 1.
The roasting and packaging departments remain the
same as the existing layout.
The simulation results comparisons in Table 2 for
existing layout, Alternative 1 and Alternative 2 for
Robusta coffee, shows that the machine utilization of
the cooking process for Alternative 2 is more by about
1.5% compared to Alternative 1 and 3% increase
compared to the existing layout. This is due to the less
distance travel for the worker to load the materials to the
machines. The work-in-process (WIP) product produced
is still the same but the material handling time is
reduced. There is no significant change for breaking
process due to waiting time increased for breaking
process to finish before loading the next batch.
Table 2 Machine utilization for robusta coffee.
Process Element Existing
Layout
Alternative
1
Alternative
2
Cooking Cooker 1 55.96% 57.14% 58.87%
Cooker 2 55.87% 57.14% 58.66%
Cooker 3 52.93% 54.76% 56.02%
Breaking Breaker 1 68.77% 68.77% 68.83%
Breaker 2 68.56% 68.57% 68.41%
Breaker 3 68.04% 68.09% 68.25%
Grinding Grinder 1 60.71% 65.47% 70.23%
Grinder 2 59.52% 65.47% 69.04%
Grinder 3 59.52% 64.28% 69.04%
Sifting Sifter 1 35.24% 40.71% 44.00%
However, for the grinding process, machine
utilization is approximately 4% more for Alternative 2
compared to the Alternative 1 and approximately 9.5%
more than the existing layout. This is due to the closer
machines position and less material handling time
causing the worker to run the process faster.
Consequently, sifting process machine utilization also
increased due to the closer and better position between
grinding and sifting machines. The result shows 5.47%
increased for Alternative 1 compared to the existing
layout and 8.76% increased for Alternative 2 compared
to the existing layout. Positioning breaking machines
closer to the mixer might contribute to this increment.
Hence, the material handling time is reduced.
Table 3 indicates that Alternative 2 produced more
than Alternative 1 and the existing layout.
Table 3 Total Output of Robusta Coffee.
Layout Simulation (Boxes)
Existing Layout 175
Alternative 1 180
Alternative 2 197
4. CONCLUSIONS
In relation to the research objectives, the proposed
layout indicated that the performance of the company
can be improved with a well-designed facility layout.
Significant result might be achieved if flexibility for
future design changes is incorporated and interactions
between facilities and material handling system is
considered when designing layouts.
REFERENCES
[1] F. De Carlo, M. A. Arleo, O. Borgia, and M. Tucci,
“Layout Design for a Low Capacity Manufacturing
Line : A Case Study Regular Paper,” pp. 1–10,
2013.
[2] C. R. Shah, “Increased Productivity in Factory
Layout by Using Systematic Layout Planning
( SLP ),” Int. J. Adv. Eng. Technol., 2013.
[3] R. Muther, Systematic Layout Planning, Second
ed. MA, Boston: Cahners Books, 1973.
[4] E. G. Boteanu, “Improving layout and workload of
manufacturing system using Delmia Quest
simulation and inventory approach,” Int. J. Innov.
Res. Adv. Eng., vol. 1, no. 6, pp. 52–61, 2014.
Proceedings of Mechanical Engineering Research Day 2017, pp. 146-147, May2017
__________
© Centre for Advanced Research on Energy
Particles influence on breakdown voltage of liquid insulating medium M.H.S. Zainoddin*, H. Zainudin, N. Abu Bakar
1) Research Laboratory of High Voltage Engineering, Faculty of Electrical Engineering,
UniversitiTeknikal Malaysia Melaka, Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
*Corresponding e-mail: [email protected]
Keywords: Metallic particles; non-metallic particles; breakdown voltage
ABSTRACT – This paper presents the experimental
results of a comparative study of mineral oils, synthetic
and natural ester oils through breakdown voltage (BdV).
These dielectric liquids were contaminated with two
types of particles, metallic and non-metallic particles and
experimental results indicate that the AC breakdown
strength of insulating liquid are higher when it
contaminated with non-metallic particles. The influence
of metallic particles on the AC breakdown strength of
insulating liquid with lower viscosities are less sensitive
and the difference between mineral insulating oil and
vegetable insulating oil were obtained and analyzed to
find the prominent type of particles that strongly affected
the insulating liquid performance.
1. INTRODUCTION
Nowadays, the demand for vegetable insulating
liquid is increasingly as dielectric liquid in transformers,
however it is vulnerable to particles and special attention
need to be paid to the influence of particles on its
dielectric strength.
As a coolant and insulating medium that widely
used in transformers, the dielectric liquid must maintain
and maximize its performance to ensure the safe
operation of power transformer. According to CIGRE
working group WG 12.17, a significant number of high
voltage transformer failures were attribute to the present
of contaminant. It is known that dielectric liquids will be
in direct contact with the internal components of the
transformers such as insulation papers, iron cores and
metals [1]. This leads to the presence of impurities in the
dielectric liquids such as cellulose pressboard and iron
particles due to thermal and electrical stresses present
when the HVAC transformers are in service [2].
Besides that, previous studies found that the
breakdown voltage of ester are influenced by metallic
particles to a less degree than mineral oil [3]. Due to
different in molecular structures and viscosity, the effect
of metallic and non-metallic particles is different and it
should be attached a great importance. A comparative
study has been done to investigate the prominent type of
particles between metallic and non-metallic particles.
2. EXPERIMENTAL ARRANGEMENT
AC BdV of dielectric liquids were measured in an
electrode configuration which according to the ASTM
D1816-04 standard. All the tests were conducted under
room temperature and at atmospheric pressure. Two
spherically-capped electrodes with 36 mm diameter each
with 1 mm gap between electrodes is used throughout the
experiment. As discussed in previous section,
contaminant or impurity is an important factor that
influences the BdV of dielectric liquids. The type, size,
concentration level of contaminant in dielectric liquid
will affect it BdV performance and strength. So, two type
of contaminant with similar size (500 μm) and similar
concentration level (0.001g) were added after the
dielectric liquid was dried and degassed to indicate the
particle presence in dielectric liquid. The contaminated
dielectric liquids were stirred for 2 minute at room
temperature. Then, AC BdV test of contaminated
dielectric liquids were measured by using MEGGER
OTS60PB throughout the experiment. For the purpose
of statistical and visualization analysis, 50 breakdown
data are determining for each sample.
3. RESULTS AND DISCUSSION
The objective of this experiment is to evaluate the
effect of contaminant on the dielectric liquid
performance. Figure 1,2 and 3shows AC BdV results for
Gemini X, Midel 7131 and MideleN. It can be seen that
from Figure 1, at the first place, when the oil samples
without contaminant, the level of BdV are higher for all
three types of dielectric liquid compared when the
insulating liquid tested with metallic and non-metallic
particles as shown in Figure 2 and 3. Previous research
has drawn that vegetable oil has comparable dielectric
strength as mineral oil. However, finding from
experimental results clearly shown that BdV of two type
of ester oil are lower compared with traditional mineral
oil. This situation is believed due to the BdV properties
of the dielectric liquid itself which is different for every
dielectric type.
Figure 1 Distribution of breakdown voltage of tested
liquids without contaminant.
Zainoddin et al., 2017
147
Figure 2: AC breakdown voltages of dielectric liquids
contaminated with carbon particles.
Figure 3: AC breakdown voltages of dielectric liquids
contaminated with copper particles.
4. CONCLUSION
The present work focuses on the influence of
metallic and non-metallic particles on the breakdown
strengths of insulating liquid and its difference between
ester insulating oil and mineral insulating oil. Carbon
particles and copper particles with similar size and
concentration level were added into insulating oils and
then the AC breakdown voltages were measured. The AC
breakdown voltage of ester oils are higher and reduce
significantly when the non-metallic particles than those
of mineral oil. Finding from the experimental results
shows that ester oil is more sensitive to the metallic and
non-metallic particles than mineral oil.
REFERENCES
[1] M.G. Danikas, “Breakdown of transformer oil,”
IEEE Electr. Insul. Mag., vol. 6, no. 5, pp. 27–34,
1990.
[2] X. Wang and Z.D. Wang, “Particle effect on
breakdown voltage of mineral and ester based
transformer oils,” Annu. Rep. - Conf. Electr. Insul.
Dielectr. Phenomena, CEIDP, pp. 598–602, 2008.
[3] X. Wang and Z.D. Wang, “Motion of Conductive
Particles and the Effect on AC Breakdown
Strengths of Esters,” 2011.
Proceedings of Mechanical Engineering Research Day 2017, pp. 148-149, May 2017
__________
© Centre for Advanced Research on Energy
Optimization of FDM process parameters for ABS spur gear build time S.M.M. Nor1,2,*, M.N. Sudin1,2, S.H.S.M. Fadzullah1,2, M.R. Alkahari1,2, F.R. Ramli1,2
1) Faculty of Mechanical Engineering, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 2) Centre for Advanced Research on Energy, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
*Corresponding e-mail: [email protected]
Keywords: Optimization parameters; build time; fused deposition modeling
ABSTRACT –This paper presents optimization of
Fused Deposition Modeling (FDM) process parameters
such as layer thickness, infill density and fill angle to
achieve optimum build time for ABS spur gear by using
orthogonal array of Taguchi Method and signal-to-noise
(S/N) ratio analysis. From the results, it was found that
FDM parameters; layer thickness, fill angle and infill
density contribute to the build time performance of ABS
spur gear. It can be concluded that the optimized
parameter obtained in this study is 0.31mm of layer
thickness, 30% infill density 90° of fill angle in
minimizing the build time of ABS spur gear.
1. INTRODUCTION
Fused Deposition Modelling (FDM) enables quick
fabrication of 3D physical models directly from 2D
CAD data using layered manufacturing (LM) technique
by laying semi-molten plastic on a platform from
bottom to the top [1]. Besides, LM technique is widely
used by researchers in helping them to achieve the
required quality characteristic or the process
performance based on selection and optimal process
parameters [2]. Many researchers [2-4] had conducted
parameters optimization to obtain desired response
parameters such as build time, flexibility, surface
roughness and tensile strength respectively.
In Rapid Prototyping (RP) technology, build time
is seen as one of the key aspects of the quantitative
characteristic which is very important in calculating
production cost and also manufacturing durations. FDM
or RP has the ability in reducing product development
time [5]. Hence, this could be done by optimizing the
FDM parameters in order to obtain minimum build time
of a product. Therefore, investigations on FDM process
parameters to the build time of a spur gear are carried
out to achieve optimum build time.
2. METHODOLOGY
In this state, the 3D model was sliced into layers
with required thickness and other process parameters
were determined such as infill density and fill angle.
Semi-molten plastic such as Acrylonitrile Butadiene
Styrene (ABS) with diameter of 1.75mm was used as
FDM feedstock and extruded by the heated nozzle
which can be deposited on the bed as the nozzle travels
along the platform bed. 9 samples of ABS spur gear
were produced by using FDM while build time of 9 spur
gears were recorded and analysed by using analytical
software Minitab 17 for the S/N ratio analysis and
ANOVA. Confirmation test was carried out to verify
estimated result against experimental result by using
optimum parameter obtained from those analyses.
Another spur gear was produced by using the optimum
parameter and build time was recorded. This result was
used to verify estimated result whether it is strongly
correlated with the experimental result.
Figure 1 ABS spur gear.
In this experiment, layer thickness, infill density
and fill angle parameters are variable process
parameters while humidity and temperature were kept
constant. Layer thickness represents the height of the
sliced 3D model whereas fill density means the amount
of material deposited between the FDM parts and fill
angle is the method where roads can be deposited to fill
the interior part. In this experiment, the layer thickness
for level 1, level 2 and level 3 are 0.18mm, 0.25mm and
0.31mm respectively, were used whereas 30°, 45° and
90° were used for each level of the fill angle
respectively. Besides, previous studies had shown that
the infill density between 30%-90% were compliable
since the toughness of a part increases as the infill
density increases [6]. This experiment used Taguchi
method which proposed 𝐿9 Orthogonal Array of three
rows and nine columns with three process parameters at
three different levels.
3. RESULTS AND DISCUSSION
The signal-to-noise ratio measures the sensitivity
of the quality investigated to those uncontrollable
factors in the experiment. The objective of this
experimental plan is to minimize the build time of ABS
Spur Gear. Hence, the smaller the better quality
Nor et al., 2017
149
characteristic was implemented [7]. The optimization
parameters can be further studied and analyzed in
Figure 2.
Table 1: Obtained S/N Ratio Values Experiment
No
Parameters Responses
Layer
Thickness
(mm)
Infill
Density
(%)
Fill
Angle
(˚)
Build
Time
(s)
SNRA
for
Build Time
1 0.18 30 30 11940 -81.540
2 0.18 60 45 25500 -88.131
3 0.18 90 90 19620 -85.854
4 0.25 30 45 12960 -82.252
5 0.25 60 90 19020 -85.584
6 0.25 90 30 25020 -87.966
7 0.31 30 90 10980 -80.812
8 0.31 60 30 15720 -83.929
9 0.31 90 45 20340 -86.167
As depicted in Figure 2, the combination of
parameters and their level’s, 𝐴3𝐵1𝐶3 has produced
maximum value of S/N ratio and it yields optimum
quality characteristics.
Figure 2 Graph of S/N effects of process parameters on
build time.
Based on the comparison data on both calculated
from experimental and equation model, the results
shown that, the average percentage error was 0.05%
with the reliability of 99.9%. This step is very
important and highly recommended by the Taguchi to
verify the test outputs.
4. CONCLUSION
It can be concluded that the optimized parameter
obtained in this study is 𝐴3𝐵1𝐶3 while infill density is
the most significant parameter in minimizing the
processing time of ABS spur gear.
ACKNOWLEDGEMENT
Grant no.: (FRGS/1/2015/TK03/FKM/02/F00628).
REFERENCES
[1] F. Ning, W. Cong, Y. Hu and H. Wang, “Additive
manufacturing of carbon fiber-reinforced plastic
composites using fused deposition modeling:
Effects of process parameters on tensile
properties,” Journal of Composite Material, pp. 1-
12, 2016
[2] M. Srivastava, S. Maheshwari and T.K. Kundra,
“Optimization of Build Time and Model Volume
for A FDM Maxum Modeler Using Response
Surface Methodology,” International Journal for
Technology Research in Engineering, vol. 2, no. 7,
pp. 1050-1057, 2015.
[3] B.H. Lee, J. Abdullah and Z.A. Khan,
“Optimization of rapid prototyping parameters for
production of flexible ABS object,” Journal of
Materials Processing Technology, vol. 169, no 1,
pp. 54-61, 2005.
[4] M. Alhubail, D. Alenezi and B. Aldousiri,
“Taguchi-based optimisation of process parameters
of fused deposition modelling for improved part
quality,” International Journal of Engineering
Research & Technology, vol. 2, no. 12, pp 2505-
2519, 2013.
[5] R. Anitha, S. Arunachalam and P. Radhakrishnan,
“Critical parameters influencing the quality of
prototypes in fused deposition modeling,” Journal
of Materials Processing Technology, vol 118, no.
1-3, pp. 385-388, 2001.
[6] C. Mendonsa, K.V. Naveen, P. Upadhyaya and
V.D. Shenoy, “Influence of FDM process
parameters on build time using Taguchi and
ANOVA approach,” International Journal of
Science & Research., vol. 4, no. 2, pp. 2013-2016,
2013.
[7] N. Raghunath and M. Pandey, “Improving
accuracy through shrinkage modeling by using
Taguchi method in selective laser sintering,”
International Journal of Machine Tools &
Manufacture, vol. 47, no. 6, pp. 985-995, 2007.
Proceedings of Mechanical Engineering Research Day 2017, pp. 150-152, May 2017
__________
© Centre for Advanced Research on Energy
Behavior analysis of human walking and robot movement for person following robot
A.I. Tarmizi1,2,*, A.Z.H. Shukor1,2, N.M.M. Sobran1,2, M.H. Jamaluddin1,2
1) Faculty of Electrical Engineering, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 2) Centre for Robotics and Industrial Automation, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
*Corresponding e-mail: [email protected]
Keywords: IMU sensor; human walking; person following robot
ABSTRACT – Person Following Robot (PFR) is
generally used to accompany or assist human while
following the human’s movement. This paper presents
the analysis on the behavior of human walking
movements and robot movements. This experiment is
done for designing the tracking algorithm of PFR. The
analysis is done by extracting data from human walking
and mobile robot movement by using IMU sensor.
Human walking data is compared to the robot movements
data and analyzed. The speed of the normal human
walking is faster than the robot used in this research.
1. INTRODUCTION
PFR has three main important parts which are robot
hardware, tracking and following algorithm. The
hardware that often used for PFR is a mobile robot [1-2].
Whereas sensor fusion is often used in PFR for tracking
the human. Other than that, there are a lot of following
algorithm that used in PFR. For example, behavior based
algorithm and also fuzzy controlled algorithm.
This paper will be focusing on behavior analysis on
human walking for tracking part of the PFR. To be able
to track the human walking, the human walking behavior
is needed to be analysed. The human walking behavior is
analysed by its movement pattern of turnings on different
paths and also the speed. This analysis of human walking
and robot movements on different paths are important so
that it can be used to further applying fuzzy decision
making. Fuzzy decision making will classify each data to
be able to decide on different tasks to be done. This paper
aims to analyse the behavioral movement of human
walking and to compare with the robot movement on a
path for PFR.
2. PFR TRACKING EXPERIMENTAL SETUP
The robot used in this research is Pioneer AmigoBot
mobile robot. The max speed of this robot according to
the specification is 1m/s. However, the max speed of this
robot by experiment is 0.5m/s. In this research, the speed
of 0.5m/s is used for the straight-line path, whereas the
other that have turns, 0.4m/s speed is used. This is
because the turning radius increases when the robot
moves at its maximum speed resulting the robot not able
to go along the path. Contrarily, the normal human walk
speed is around 0.7m/s to 0.8m/s. PFR should be able to
follow a person either the robot need to be faster, same
speed, or a slightly slower than the human.
3. DATA EXTRACTION
To collect and analyze human walking movements
using IMU sensor different path in this research,
MPU6050 and Arduino UNO are used. Arduino UNO is
set up with coding that will extract yaw value in degree
from the IMU sensor. The Arduino and MPU6050 can be
connected to ZigBee for wireless connection. The
MPU6050 is put on the waistline of the subject. The
subject is required to walk normally on different paths of
walking straight, right turn, left turn, U-turn to the right
and lastly U-turn to the left. The data from the human
walking movements are recorded.
As for extracting the data of robot movements, IMU
sensor is put on the center of AmigoBot robot. AmigoBot
is controlled by a remote host from a computer to move
across the same path as the human walking. The
MPU6050 displays the yaw values of the movements and
the data are recorded. The human walking and the robot
movements data are collected and analyzed in two
separate experiments. Lastly, the recorded human and
robot data are then compared and analyzed.
4. DATA ANALYSIS
The data are analyzed by plotting the data extracted
from the human walking and robot movement on a same
graph. The data from the graphs plotted are the degree of
yaw turning and the time taken in milliseconds. From the
yaw data, the human and robot turning movements are
analyzed by observing the pattern of the graphs. Other
than that, the graphs are also analyzed in terms of speed
and time taken for the human and robot to complete one
path. The turning movement and time taken for the robot
and human to finish one path are discussed and compared
with each other. Lastly, feature extraction of standard
deviation and mean absolute value is applied for further
process of data classification by using fuzzy decision
making.
5. RESULTS AND DISCUSSION
The data from the IMU sensor are plotted in a graph
comparing the robot and human movement (Figure 1-5).
The distance of the path is kept constant with the same
path. The graphs are in linear graph so that the data can
Tarmizi et al., 2017
151
be observed and analyzed clearly. Moreover, the data is
more stable in a linear graph.
Figure 1 Human-robot moving straight.
Figure 2 Human-robot turning right.
Figure 3 Human-robot turning left.
Figure 4 Human-robot U-turn to the right.
Figure 5 Human-robot U-turn to the left.
The graphs above show different results on the
human walk and the robot movement on path. According
to the results, the turning movements of the human-robot
is quite similar with each other. However, the robot has a
sharper turning movement than human walking. From the
graph, human walking has a smoother turning
movements than the AmigoBot because the AmigoBot
was remotely controlled from the computer. The
smoothness of turning movement can be reduced by
programming the AmigoBot to move accordingly to the
path instead of remotely controlled. On the other hand,
the time taken of the human and robot to complete one
path has a slightly time difference. From the graphs, it
shows that the human walking is faster than the robot
movements. This can be seen on the graph where the time
taken for the human to complete on path is shorter than
the robot.
Table 1 and table 2 above are the feature extraction
applied to the data extracted from the IMU sensor. Table
1 shows the results for the standard deviation calculation
for the data extracted. Besides that, table 2 resulted by
calculating the mean absolute value of the data extracted.
From the data, mean absolute value calculation gives a
better value to be used for further data classification by
using fuzzy decision making.
Table 1 Feature extraction using standard deviation.
Path Human Robot
Straight 15.19 26.54
Turn right 54.15 35.27
Turn left 29.81 14.81
U-turn (right) 89.67 69.39
U-turn (left) 71.58 51.78
Table 2 Feature extraction using mean absolute value.
Path Human Robot
Straight 22.91 46.03
Turn right 62.23 61.69
Turn left -22.17 9.081
U-turn (right) -29.06 -13.42
U-turn (left) -61.71 -24.11
Tarmizi et al., 2017
152
6. CONCLUSION
As a conclusion, the movement of turning and
maneuverability of the robot is the same human walking.
This can be seen by analyzing the pattern of the graphs
that looks similar to each other. However, according to
the graph, we can see that the speed of the robot is slightly
slower than the human walking speed.
ACKNOWLEDGEMENT
This research is supported by the grant
RAGS/1/2014/TK03/FKE/B00056. Thanks to the
Robotics and Industrial Automation group, CeRIA.
REFERENCES
[1] C.A. Cifuentes, A. Frizera, R. Carelli, T. Bastos,
“Human-robot interaction based on wearable IMU
sensor and laser range finder,” Robotics and
Autonomous Systems, vol. 62, no 10, pp. 1425-
1439, 2014.
[2] S. Jia, L. Wang, S. Wang and C. Bai, “Fuzzy-based
intelligent control strategy for a person following
robot,” In Robotics and Biomimetics (ROBIO),
2013 IEEE International Conference on, pp. 2408-
2413, 2013.
Proceedings of Mechanical Engineering Research Day 2017, pp. 153-154, May 2017
__________
© Centre for Advanced Research on Energy
Roughness of 3D printed surface under influence of chemical post processing
A.R. Zolkaply1, M.R. Alkahari1,2,*, F.R. Ramli1, 2, N.S. Hamdan3, M.N. Sudin1, 2
1) Faculty of Mechanical Engineering, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 2) Centre for Advanced Research on Energy, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 3) Faculty of Engineering Technology, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
*Corresponding e-mail: [email protected]
Keywords: 3D printing; additive manufacturing (AM); surface roughness
ABSTRACT - Fused deposition modeling (FDM) is a
popular technology in 3D printing. This is due to its
capability to fabricate 3D part faster, effectively,
economically and more user-friendly. However, the
yield of the printed part is poor in surface finish and it
produces staircase effect resulted from layer by layer
deposition process. Hence, this paper studies the effect
of methyl ethyl ketone (MEK) used as post-processing
treatment for acrylonitrile butadiene styrene (ABS) 3D
printed part. During the experiment, controlled amount
of MEK was applied on 3D printed surface after
printing process. This research discovers that the surface
finish of 3D printed ABS can be improved significantly
with the use of MEK.
1. INTRODUCTION
The ability of fused deposition modeling (FDM) in
producing 3D part is undeniable and this technology has
brought a breakthrough in industries. However, FDM
fabricated part is less accurate due to its staircase issue
that affects the final quality of a prototype [1].
Performance of FDM process is also not comparable to
conventional manufacturing process [2]. This problem
can be reduced with several approaches such as slicing
strategy, parameters optimization and post-treatment [3].
According to Galantucci et. al., chemical treatment can
be used to reduce surface roughness on 3D printing
fabricated part and increasing the compactness of
structure [4]. The usage of chemical like dimethyl
ketone (Aceton) and methyl ethyl ketone (MEK) solvent
can be used to react with acrylonitrile butadiene styrene
(ABS) to improve the surface finish [5]. This method is
less expensive and easier.
One of the methods for post-processing techniques
is vapor polishing that is used as surface treatment by
coating the printed part evenly without affecting the
dimensional accuracy significantly. The treatment
makes the solvent react with ABS layers, which fused
together and reform a new structure that produces a
smooth surface finish. This also can increase cohesion
between layers and affects the mechanical properties of
ABS [6]. In this research, MEK was applied on 3D
printed surface as post processing and its influence on
surface roughness is studied.
2. METHODOLOGY
The experiment was implemented on Mendel Max
3D printer. Specimens were fabricated using ABS
thermoplastic and were post-processed with MEK. A
roller was designed to roll the injected solvent on the
ABS surface evenly. The purpose of using roller was to
reduce ABS dimensional changes during post-treatment
process. Chemical MEK was used to dissolve the ABS
filament. MEK is an organic solvent that has an ability
to break chains of amorphous thermoplastic like ABS
which is low in chemical resistance. The volume of
MEK was varied in order to investigate the effect of the
chemical amount on the surface roughness of 3D printed
surface. The initial reading of surface roughness for an
untreated sample was taken and the value was recorded
at 9.98 μm. Seven samples were fabricated on Mendel
and the volume of applied MEK was varied from 1 ml
to 7 ml. The top and cross section view was taken by
using Nikon Measuring Microscopes MM-800. The
parameter setting was set up as in Table 1.
Table 1 Parameter setting for 3D printing.
Volume, ml 1 2 3 4 5 6 7
Layer thickness, mm 0.2
Fill angle,˚ 90
Infill pattern Line
Infill density, % 30
3. RESULTS AND DISCUSSION
Based on the result, using chemical treatment as
post-processing improves the roughness at certain
volume of MEK. The amount of MEK plays important
role in determining the roughness of the ABS. Figure 1
shows that the roughness reduces from Ra= 9.98 μm to
Ra= 0.861 μm when 1 ml of MEK applied on ABS.
91.4% of improvement is achieved and glossy effect to
the sample is produced. The MEK dissolves the ABS
surface and forms a strong intermolecular bond as the
MEK evaporates and it creates cohesive bonding on
ABS structure. As a result, the distance of printing line
become closer and the material bond as shown in Figure
2(b). The surface roughness worsens when the amount
of MEK increases. This shows that the high amount of
Zolkaply et al., 2017
154
MEK can increase the surface roughness where Ra
9.979 μm was recorded for 7 ml MEK. Figure 2 and
Figure 3 show inverted microscope photographs of
treated and untreated specimens that have a substantial
impact in their surface finish.
Figure 1 Treated and untreated surface roughness of
ABS part.
(a) (b)
Figure 2 Top view of ABS (a) sample before treated by
MEK (b) Sample after treated by MEK.
(a) (b)
Figure 3 Cross sectional view of ABS (a) sample before
treated by 1ml of MEK and (b) sample after treated by
1ml of MEK.
Figure 3 shows the sectional view of ABS where
reaction between MEK and ABS plastic produced a
reactant that fills in the air gap of the sample. This
makes the molecular chain to slide over each other and
reforms a new secondary bond with the presence of
MEK solvent. As a result, the new form of ABS
structure is produced with glossy effect on its surface.
MEK is used due to its properties that slightly high in
boiling point which can slow the evaporation process
during post-processing treatment. This ensures the
bonding process is completely done by the MEK to
dissolve the ABS filament before it forms a new
structure. The low chemical resistance property of ABS
has made its random structure to easily absorb chemical
and destroy its secondary bond if the usage amount of
the chemical is not controlled. For the MEK treatment,
it was observed that the amount of the chemical was
very important in order to produce high quality of
surface finish. The interaction of ABS and MEK
material shows drastic improvement in surface
roughness. This has proven that MEK is a solvent that
can be used to enhance the surface finish at the suitable
amount.
4. CONCLUSION
FDM has a high potential in various application
but its quality of printed parts needs to be improved.
The drawbacks of FDM surface roughness has
encourage introduction of various methods in improving
FDM parts. This research indicates that application of
MEK as post-processing on 3D printed surface enable
the surface roughness to be improved significantly. This
process is much easier and economical. However, the
amount of MEK applied on the surface need to be
controlled so that required roughness can be obtained.
ACKNOWLEDGEMENT
Grant no.: RAGS/1/2015/TK0/FTK/03/B00123.
REFERENCES
[1] N.V. Reddy and S.G. Dhande, “Slicing procedures
in layered manufacturing: A review,”
Rapid Prototyping Journal, vol. 9, no. 5, pp. 274-
288, 2003.
[2] N. S. A. Bakar, M. R. Alkahari, and H. Boejang,
“Analysis on fused deposition modeling
performance”, J. Zhejiang Univ. Sci. A, vol. 11, no.
12, pp. 972–977, 2010.
[3] L.M. Galantucci, F. Lavecchia, and G. Percoco,
“Experimental study aiming to enhance the surface
finish of fused deposition modeled parts,” CIRP
Annals - Manufacturing Technology, vol. 58, pp.
189-192, 2009.
[4] L.M. Galantucci, F. Lavecchia, and G. Percoco,
“Quantitative analysis of a chemical treatment to
reduce roughness of parts fabricated using fused
deposition modeling,”
CIRP Annals - Manufacturing Technology, vol. 59,
no. 1, pp. 247-250, 2010.
[5] R.A. Sambasiva, A.M. Dharap, J.V.L. Venkatesh,
and O. Deepesh, “Investigation of post processing
techniques to reduce the surface roughness of
fused deposition modeled parts,” International
Journal of Mechanical Engineering and
Technology, vol. 3, no. 3, pp. 531-544, 2012.
[6] H. Gao, D.V. Kaweesa, J. Moore, and N.A. Meisel,
“Investigating the impact of acetone vapor
smoothing on the strength and elongation of
printed abs parts,” Journal of the Minerals, Metals
and Minerals Society, vol. 22, no. 4, pp. 1-6, 2016.
0
2
4
6
8
10
12
1 2 3 4 5 6 7
Surf
ace
rou
ghn
ess,
Ra
(u
m)
Volume, V (ml)
Treated…
Untreated surface roughness, Ra=9.98 um
Proceedings of Mechanical Engineering Research Day 2017, pp. 155-157, May 2017
__________
© Centre for Advanced Research on Energy
An investigation on applying different types of adhesive to reduce warping deformation in open source 3D printer
M.A. Nazan1,*, F.R. Ramli1, 2, M.R. Alkahari1, 2, M.A. Abdullah1, 2, M.N. Sudin1, 2
1) Faculty of Mechanical Engineering, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 2) Centre for Advanced Research on Energy, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
* Corresponding e-mail: [email protected]
Keywords: Open-source 3D printer; synthetic and natural based adhesive; warping deformation
ABSTRACT – The purpose of this paper is to
investigate the warping deformation problem by
applying adhesive material instead of applying heat onto
the open source 3D printing bed. The research processes
involved investigating the power consumptions of open
source 3D printers, preparation of the 3D printing model
and adhesive material, tensile strength of adhesive
samples and warping deformation measurement. Based
on the result, the power consumption has decrease to
23W in average when heat is not applied to printing bed.
Epoxy resin based adhesive gives better warping
deformation with 0.8% and higher bonding strength of
1.2233 MPa compare to cassava and soy adhesive.
1. INTRODUCTION
One of the problems in the open source 3D
printing especially filament deposition modeling (FDM)
process is that the plastic filament which is extruded
from the machine nozzle tends to shrink and warp from
the printing bed platform. This problem can be reduced
by applying heat to the printing bed where, the higher
the bed temperature is, the less the deformed shape will
be [1]. Moreover, the warping deformation has also
showing warping reduction when applying epoxy resin
based adhesive onto the printing bed [2-3]. Other type
of adhesive such as cassava and soy based adhesive has
massive potential to be used in the same ways to reduce
or to eliminate the warping problem. . Hence, the
purpose of this paper is to study how warping
deformation problem can be reduced by applying
adhesive material on the printing platform instead of
heat.
2. METHODOLOGY
2.1 Power consumption
Three types of open source 3D printer models i.e.
Kossel Mini, Fab Gear and Mendel Max was used in
this experiment to identify total power used when the
machine use heat bed and how much when it is not.
Digital Power Reader 2.0 was attached to the printer’s
power source to record the power used. It was measured
while beginning from the first material deposit from
nozzle until the end of fabrication of a 30x100x5mm³
solid model using Polylactic Acid (PLA) material such
as in Figure 1. Two different setting has been used
which one with a heat bed and without heat bed. The 3D
printing parameters is set based on [3] where the infill
density is 13%, 192˚C of printing temperature, and
0.2mm of layer height.
(a) (b)
Figure 1 Rectangular shaped of (a) sketched in solid
model and (b) printed sample.
2.2 Warping deformation preparation
Next, the same PLA model in Figure 1 with the
same parameters is built by applying adhesive onto the
3D printer bed. Three types of adhesive were used
where epoxy resin based adhesive from [3], cassava and
soy based adhesives are prepared based on [4-5]. The
adhesive was spread diagonally across the printing bed
is with angle of 45˚ and in the opposite direction to span
it all over the surface.
Figure 2 is shows the method of measuring the
warping deformation by using vernier height gauge and
Equation (1) to calculate the percentage number that
come out while measuring the warping deformations for
each sample. By referring to the Eq.(1), the value of
warping deformation, y is obtained by the difference
value of 𝑦1 , value of total height and 𝑦2, the deflected
total height.
Figure 2 Method of measurement at each sample’s
corner.
𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑜𝑓 𝑤𝑎𝑟𝑝𝑖𝑛𝑔
𝑑𝑒𝑓𝑜𝑟𝑚𝑎𝑡𝑖𝑜𝑛, 𝑦= |
𝑦1−𝑦2
𝑦1| 𝑥 100% (1)
Nazan et al., 2017
156
2.3 Mechanical testing
Standard test method for Tensile Strength of
Adhesive by Means Bar and Rod Specimens (ASTM
D2095) was used to obtain the bonding strength of the
adhesive. The specimens were prepared such as in
Figure 3 where a set of 10mm x 10mm x 40mm is
printed onto glass that was spread with different type of
adhesive. The specimens were undergoing tensile
strength test and the max stress and max strain were
recorded.
Figure 3 Tensile test sample arrangement.
3. RESULTS AND DISCUSSION
3.1 Power consumption result
Figure 4 shows the power consumption of the
tested open-source 3D printers. All 3D printers took
about 43-44 minutes to complete their task. Mendel
Max 3D printer that use heat bed has the highest power
consumption with P = 59 W. In general, the power
consumption of the printers that using heat bed is
double compares to printing without heat bed. Thus, for
complex model that takes longer time and higher power
consumption, removing heat bed can benefit the user
cost and the effect towards the environment.
Figure 4 Power consumption of three type of open-
source 3D printer.
3.2 Warping deformation result
Figure 5 shows warping deformation percentage of
the different adhesive type when applied to 3D printing
bed. Epoxy resin based adhesive gives better adhesion
with about 0.5-1 times lower warping deformation
compares to cassava and soy based adhesive.
Figure 5 Warping deformation tests result by three
different adhesives.
This is expected because the epoxy resin adhesive
has better adsorption with plastic part and glass compare
to natural adhesives which do not have better adsorption
with the plastic part. In addition, humidity also plays
roles to the quality of natural based adhesive.
3.3 Tensile test result
Table 1 shows the result of the tensile test that had
been performed by using these adhesives. In
comparison, the epoxy resin based adhesive has better
tensile bond strength, σ = 1.2233MPa which is four
times better than the two natural adhesives. This
explains why the warping deformation is lower when
epoxy resin based adhesive was used in the experiment.
Table 1 Tensile test result at maximum load applied
Adhesive types Max. Stress,
σ (MPa)
Max. Strain,
ε
Epoxy 1.2233 1.0108
Cassava 0.3236 0.0221
Soy 0.1917 0.0046
4. CONCLUSION
As a conclusion, the investigation shows that the
power consumption has decreased to about half when
heat is not applied to the printing bed. Epoxy resin
based adhesive on printing bed without heat gives better
warping deformation compare to cassava and soy
adhesive because of its higher bonding strength.
However, both of the plant natural based adhesive
shows potential to solve the warping problem where the
obtained warping deformation is less than 10%.
ACKNOWLEDGEMENT
This research is funded under the research grants
FRGS/1/2015/TK03/FKM/02/F00270.
REFERENCES
[1] Y.H. Choi, C.M. Kim, H.S. Jeong, J.H. Yuon,
“Influence of bed temperature on heat shrinkage
shape error in FDM additive manufacturing of the
ABS-engineering plastic,” World Journal of
Engineering and Technology, vol. 4, no. 3, pp. 186-
192, 2016.
[2] F.R. Ramli, M.I. Jailani, H. Unjar, M.R. Alkahari
and M.A. Abdullah, "Integrated recycle system
Nazan et al., 2017
157
concept for low cost 3D-printer sustainability," in
Proceedings of Mechanical Engineering Research
Day, pp. 77-78, 2015.
[3] M.A. Nazan, F.R. Ramli, M.R. Alkahari and M.A.
Abdullah, "Optimization of warping deformation in
open source 3D printer using response surface
method," in Proceeding of Mechanical Engineering
Research Day, pp. 71-72, 2016.
[4] A.J. Gunorubon, “Production of Cassava Starch-
Based Adhesive,” Research Journal in Engineering
and Applied Sciences, vol. 1, no. 4, pp. 219-224,
2012.
[5] W.H. Wang, X.P. Li, X.Q. Zhang, “A soy-based
adhesive from basic modification,” Pigment &
Resin Technology, vol. 37, no. 2, pp. 92-97, 2008.
Proceedings of Mechanical Engineering Research Day 2017, pp. 158-159, May 2017
__________
© Centre for Advanced Research on Energy
Manufacturability of overhang structure using open source 3D printer M.R. Alkahari1,2,*, S.N.H. Mazlan1, O.I. Sun3, F.R. Ramli1,2, N.A. Maidin4 , M.N.Sudin1,2
1) Faculty of Mechanical Engineering, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 2) Centre of Advanced Research on Energy, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 3) Daikin Research &Development Malaysia Sdn. Bhd,
Taman Perindustrian Bukit Rahman Putra, 47000 Sungai Buloh, Selangor, Malaysia 4) Faculty of Engineering Technology, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
*Corresponding e-mail: [email protected]
Keywords: Fused deposition modeling (FDM); overhang structure; additive manufacturing
ABSTRACT - Fused Deposition Modeling (FDM) is
the most common 3D process where parts are created
layer by layer. Recently, researchers started to study on
the different structure that FDM cannot fabricate such as
overhang structure. The objective of this paper is to
investigate the manufacturability of the overhang
structure at various lengths. The overhang structure is
designed and built using the open source 3D printer.
Then, the capability of the structure to be produced
according to its design is evaluated. Based on this, some
recommendation is presented to provide a guideline to
the 3D printer user.
1. INTRODUCTION
Fused Deposition Modeling (FDM) creates a part
layer by layer and offers a lot of potential benefits to the
use of plastic material because it can fabricate
mechanical complex structures in a shorten time and
low cost of manufacturing process. However, the
drawbacks of this system, it cannot fabricate the
overhang structure. Adam et al studied on the element
transition and aggregated structure in order to provide
the design rules for 3D printer user. The study also
compared capability of different 3D printing techniques
to fabricate the overhang structure [1].
Meanwhile, Mahesh et al studied on the
benchmarking design for evaluation in order to compare
the RP systems and process. The purpose benchmarks
consist of features called flat beam (FB) to evaluate the
overhang, flatness and straightness [2]. Paolo et al,
studied on the benchmarking of FDM machine through
part quality using IT grades to measure the dimensional
accuracy of benchmarking design and evaluate the
standard ISO basic size and Geometric Dimension and
Tolerance (GD&T) value [3]. Bakar et al analyzed FDM
performance by fabricating the benchmarking design
using high end AM 3D printer and measure the surface
roughness and dimensional accuracy of the printed
part [4].
2. METHODOLOGY
In this study, an overhang structure was fabricated
using open source 3D printer originated from the Prusa
3D printer. The material is a thermoplastic material,
polylactic acid (PLA). The structure has twenty-two
different overhang values. After the part was fabricated,
it was measured using the profile projector to compare
the dimension of CAD (Computer Aided Design) data
and the printed parts. In order to identify the suitable
length at the element transition (overhang), the qualities
of the manufactured test specimens were examined by
visual inspection and evaluated. The dimension of the
overhang length was taken according to the guide of “x”
as tabulated in Table 1.
Table 1 Details descriptions of the overhang feature.
Overhang Description
Feature
Guide Length, x
Dimension range x, mm 0.2 to 24
3. RESULTS AND DISCUSSION
The visual inspection of an overhang structure was
tabulated in Table 2 after the part was fabricated. The
successfully built overhang structures were compared to
the nominal overhang length. The data were presented
in the terms of dimensional accuracy by comparing the
result from 3D CAD dimension and printed parts.
Figure 1 shows the 3D CAD data and the fabricated
parts of the overhang structure with different length, x
varied from 0.2 to 2 mm. Generally, the overhang can be
fabricated using a 3D printer but has limited maximal
length.
Based on the Figure 1, the overhang length, x
dimensional accuracy of overhang structure does not
deviate much as compared to the CAD data. The plotted
graph only contains the successfully fabricated
overhang by the range from 0.2 to 2.0 mm, meanwhile
for the rest, the measurements cannot be properly taken
due to the disturbance effect from the fabricated parts
because the parts has sagged and agglutinated filament
was formed underneath the layers. For overhang, the
Alkahari et al., 2017
159
hanging part should be short enough to ensure
manufacturability given by part layer filaments do not
“fall out” from its nominal position because built layer
by layer and each layer must be supported by something
underneath it or a build platform. The graphical trend
shows that, printed part from the 3D printer having the
value nearer to CAD data, especially when x≤2.0 mm.
Thus, some recommendations on the suitable length of
the overhang structure were proposed. Table 3 provides
the general guideline for 3D printer user for
manufacturing of overhang structure using the open
source 3D printer.
Table 2 Visual inspection of overhang quality.
Overhang length, mm Quality
0.2 Successfully fabricated
0.4 Successfully fabricated
0.6 Successfully fabricated
0.8 Successfully fabricated
1.0 Successfully fabricated
1.2 Successfully fabricated
1.4 Successfully fabricated
1.6 Successfully fabricated
1.8 Successfully fabricated
2.0 Successfully fabricated
2.2 Sagging
4.0 Sagging
6.0 Sagging
8.0 Sagging
10.0 Sagging
12.0 Sagging
14.0 Serious fall out
16.0 Serious fall out
18.0 Serious fall out
20.0 Serious fall out
22.0 Serious fall out
24.0 Serious fall out
Figure 1 Measured length of successfully fabricated
overhang structure.
Table 3 Length of overhang structure.
Design Not
Recommended Acceptable Recommended
Feature
Range,
mm x≥14 2.2≤x≥13 x≤2
Table 4 Visual images of overhang structure.
Design Not
recommended Acceptable Recommended
Parts
Quality Serious
fall out Sagging
Successfully
fabricated
4. CONCLUSIONS
FDM creates a parts layer by layer, thus it cannot
print hanging parts unless by generating the support
structure. However, overhang structure with minimal
overhang length of x≤2 is recommended to be
manufactured. When the length x>2, it may cause
sagging of filament and fall out which can cause the
structure to be built not be similar to the CAD design.
ACKNOWLEDGEMENT
The research was supported by research grant
RAGS/1/2015/2016/TK0/FTK/03/B00113.
REFERENCES
[1] G. Adam and D. Zimmer,"Design for Additive
Manufacturing—element transitions and
aggregated structures," CIRP Journal of
Manufacturing Science and Technology, vol. 7, no.
1, pp. 20-28, 2014.
[2] M. Mahesh, Y. Wong, J. Fuh and H. Loh,
"Benchmarking for comparative evaluation of RP
systems and processes," Rapid Prototyping
Journal, vol. 10, no. 2, pp. 123-135, 2004.
[3] P. Minetola, L. Iuliano, and G. Marchiandi,
“Benchmarking of FDM Machines through Part
Quality Using IT Grades”, Procedia CIRP, vol. 41,
pp. 1027–1032, 2016.
[4] N.S.A. Bakar, M.R. Alkahari, and H. Boejang,
“Analysis on Fused Deposition Modeling
performance,” J. Zhejiang Univ. Sci. A, vol. 11, no.
12, pp. 972–977, 2010.
Proceedings of Mechanical Engineering Research Day 2017, pp. 160-161, May 2017
__________
© Centre for Advanced Research on Energy
Design self-balancing bicycle N. Tamaldin1,2,*, H.I.M. Yusof1,2, M.F.B. Abdollah1,2, G. Omar1,2, M.I.F. Rosley2
1) Faculty of Engineering Technology, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia 2) Center of Advance Research on Energy, Faculty of Mechanical Engineering,
Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
*Corresponding e-mail: [email protected]
Keywords: Self-balancing; gyroscope; control system
ABSTRACT – Recently, many investigations have been
done regarding to the problems of controlling two-
wheeled self-balancing robot. This paper reviewed based
on five previous journal in order to find out which
method is suitable to design a self-balancing bicycle and
it will focus on the control system of the structure. There
are several ways in order to design an efficient self-
balancing bicycle which are by using control moment
gyroscope (CMG), mass balancing, steering control and
reaction wheel. Based on previous research, the usage of
CMG is the suitable choice since it can produce large
amount of torque, it has no ground reaction forces, and
the system can be stable even when the bicycle is
stationary.
1. BACKGROUND STUDY
Bicycle is a common form of transportation,
recreation, and also can be a medium in exercising which
have been used for years of many people. Bicycle also
can serve to provide physical therapy, as they are a low
impact form of exercise that can train balance, strength,
stamina and coordination. Though one may consider
riding a bicycle to be a fairly simple task, this is not the
case for many people especially for young children, and
adults who have never learned to ride a bicycle, injured
people, or people suffering from developmental or
cognitive disabilities. A system that could provide
balancing assistance to a bicycle rider without otherwise
affecting the experience of riding a bicycle could provide
great benefit to these groups of individuals. Such a
system could be used both as a teaching tool, and as a
physically therapeutic device.
Recently, many investigations have been done
regarding to the problems of controlling two-wheeled
self-balancing robot, which are widely taken into
applications in the field of autonomous robotics and
intelligent vehicles. First and the foremost problem is
bicycle keep falling when it is not in controlled especially
at low forward velocity [1-2]. Besides, the lack of
controlling flywheels in the horizontal position also
become a problem in this cases. Without controlling the
flywheels in the horizontal position, the bicycle will lose
its balance after a particular limited flywheel’s angle [3].
So that, many inventors have been made research and
development to encounter this problem such as
introducing the self-balancing bicycle. Self-balancing
bicycle use sensors to detect the roll angle of the bicycle
and actuators to bring it into balance as needed, similar
to an inverted pendulum where it is an unstable nonlinear
system and can be implemented in several ways. This
paper reviewed based on five previous journal in order to
find out which method is suitable to design a self-
balancing bicycle and it will focus on the control system
of the structure.
2. METHODOLOGY
The methods to achieve a self-balanced bicycle are
mainly classified into four types. The first type is using a
control moment gyroscope (CMG) [1-4]. This method
can provide a large torque, but energy consumption of
CMG is very high because the flywheel is spinning all
the time. The CMG consists of a spinning rotor with a
large, constant angular momentum, whose angular
momentum vector direction can be changed for a bicycle
by rotating the spinning rotor. The spinning rotor, which
is on a gimbal, applies a torque to the gimbal to produce
a precession, gyroscopic reaction torque orthogonal to
both the rotor spin and gimbal axes. A CMG amplifies
torque because a small gimbal torque input produces a
large control torque to the bicycle
The second type is mass balancing where
mechanical structure of mass balancing is simple, but the
torque this method could provide is small.
The third type is steering control where a controller
controls the amount of torque applied to the steering
handlebar to balance the bicycle. Advantages of this
system are low mass and low energy consumption, while
its disadvantages are it requires ground reaction forces
and it cannot withstand large tilt angle disturbance. The
energy consumption of steering control is low, but it
cannot balance the bicycle at low forward velocity.
The forth type is using a reaction wheel where speed
of a reaction wheel is increased or decreased to generate
a reactionary torque about the spin axis which is parallel
to the bicycle’s frame [5]. As the bicycle begins to fall to
one side, a motor mounted to the reaction wheel applies
a torque on the reaction wheel, generating a reactionary
torque on the bicycle, which brings back the bicycle’s
balance. Advantages of this system are it is low cost,
simple and no ground reaction, while disadvantages are
it consumes more energy and it cannot produce large
amount of torque.
A very well-known self-balancing bicycle robot
using a reaction wheel is the Murata Boy which was
developed by Murata Manufacturing Co., Ltd in 2005.
Tamaldin et al., 2017
161
3. RESULTS AND ANALYSIS
Among these methods, the CMG, a gyroscopic
stabilizer is a good choice because its response time is
short and the system is stable when the bicycle is
stationary. Gyroscopic stabilization, where one or more
motorized gimbals tilt the angular momentum of a
spinning rotor. As the rotor tilts, the changing angular
momentum causes a gyroscopic precession torque that
balances the bicycle.
Figure 1 An example of main components of self-
balancing control system.
Advantages of this system are it can produce large
amount of torque, it has no ground reaction forces, and
the system can be stable even when the bicycle is
stationary. Disadvantages are it consumes more energy
and it is physically complex. Research studies using this
concept include Narong et al. [1,5], Jiarui et al. [2], and
Sandeep et al. [4].
Figure 2 Control moment gyroscope (CMG).
4. CONCLUSIONS
In the nutshell, there are several ways in order to
design an efficient self-balancing bicycle which are by
using control moment gyroscope (CMG), mass
balancing, steering control and reaction wheel. Based on
previous research, the usage of CMG is the suitable
choice since it can produce large amount of torque, it has
no ground reaction forces, and the system can be stable
even when the bicycle is stationary. Unfortunately, CMG
consumes more energy and it is physically complex but
this disadvantage will be contained with further research
and development in the future.
REFERENCES
[1] J. He and M. Zhao, “Control system design of self-
balanced bicycles by control moment gyroscope,”
in Proceedings of the 2015 Chinese Intelligent
Automation Conference, pp. 205-214, 2015.
[2] S.K. Gupta and V. Gulhane, “Pose estimation
algorithm implication for bicycle using gyroscope
& accelerometer,” Design approach, vol. 4, 2014.
[3] N. Aphiratsakun and K. Techakittiroj, “Single loop
and double loop balancing control of AU Self-
balancing Bicycle (AUSB),” 2012 IEEE
International Conference on Robotics and
Biomimetics, 2012.
[4] N. Aphiratsakun and K. Techakittiroj,
“Autonomous AU bicycle: Self-balancing and
tracking control (AUSB2),” 2013 IEEE
International Conference on Robotics and
Biomimetics, 2013.
[5] K. Kanjanawanishkul, “LQR and MPC controller
design and comparison for a stationary self-
balancing bicycle robot with a reaction wheel,”
Kybernetika, pp. 173-191, 2015.
Proceedings of Mechanical Engineering Research Day 2017, pp. 162-163, May 2017
__________
© Centre for Advanced Research on Energy
Optimizing the design of n-channel trench power MOSFET device using Taguchi method
N.A. Ahmed Nazri, F. Salehuddin*, N.R. Mohamad, M.N.I.A. Aziz, M. Hadinan, A.S.M. Zain,
A.H. Afifah Maheran, A.R. Hanim, H. Hazura, S.K. Idris
Centre for Telecommunication Research and Innovation, Faculty of Electronics and Computer Engineering,
Universiti Teknikal Malaysia Melaka, Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
*Corresponding e-mail: [email protected]
Keywords: Trench power; Taguchi method
ABSTRACT – In this study, the impact of the process
parameters such as trench width, trench depth, epitaxial
thickness and epitaxial thickness on the response of
breakdown voltage and on-resistance for the n-channel
trench power MOSFET devices were investigated. The
virtual fabrication of the device was performed using
ANTHENA module while the device electrical
characteristics was simulated using ATLAS module.
These two modules were combined with L8 orthogonal
array in Taguchi method to aid design and optimize the
process parameters. In this work, the most significant
factor that affects the threshold voltage, breakdown
voltage and on-resistance was the epitaxial resistivity.
The result shows that the threshold voltage (VTH),
breakdown voltage (BV) and on-resistance (RON) for n-
channel trench power MOSFET devices after an
optimization approach was approximately 1.471V,
34.57V and 0.161mΩ/cm2 respectively.
1. INTRODUCTION
The planar and the trench designs are the two
common designs of power MOSFETs. Out of various
possible options, the trench gate design yields the most
efficient performance as a discrete power device for
relatively low voltage applications compared to any
other structure with similar specifications due to their
lower attainable on-resistance [1]. The trench gate
power MOSFETs has become the mainstream of low
voltage power switch since it has been developed about
20 years ago. In terms of resistance and gate charge, the
trench gate technology has an advantage in contrast with
the doubled diffused MOSFET transistor for the
products with a drain voltage capability less than 100V.
By using this technology, the gate electrodes are buried
inside the trench of the device and the channels are
created along the sidewall of the trench. For the power
device, there is a trade-off between the specific on-
resistance and breakdown voltage. Achieving the lowest
value of on-resistance without breakdown voltage
deterioration is crucial for the power devices [2]. This is
due to the on-resistance of the power MOSFET rise
immediately to the rise in breakdown voltage, causing
high conduction losses, even when using semiconductor
materials that are costly and ultimately lowers the
overall efficiency of the system [3]. The application
nowadays keeps demanding further reduction in on-
resistance with adequately high breakdown voltage even
though the on-resistance proposed by the trench power
MOSFET has been the lowest among diverse possible
structures [1]. To meet this ever continuous demand, the
optimization of the input process parameters that have
an impact on the on-resistance and breakdown voltage
of trench power MOSFET is needed. The optimization
was done to choose a set of variables (parameters)
values subject to the diverse limitations that will
produce the desired optimum response. In this study, the
impact of trench width, trench depth, epitaxial thickness
and epitaxial resistivity on the breakdown voltage and
on-resistance of the power MOSFET device was
investigated.
2. MATERIALS AND METHODS
The completed structure of n-channel trench power
MOSFET device is shown in Figure 1. After the
structure of n-channel trench power MOSFET device
was modeled in ANTHENA simulator, the electrical
characteristics of the structure such as threshold voltage
(VTH), breakdown voltage (BV) and on-resistance (RON)
is simulated using ATLAS simulator.
Figure 1 Cross section of n-channel trench power
MOSFET structure
For a power MOSFET with low (<50V)
breakdown voltages, the VTH within the range of 1 – 2V
is required [3]. According to Korec et.al. [4], a power
MOSFET device with an epitaxial thickness of 4μm,
epitaxial doping concentration of 3.5x1016 atom/cm3 and
gate oxide thickness of 600Å is suitable for obtaining a
breakdown voltage of 30V. Therefore, it is important to
achieve the minimum RON for a power MOSFET
structure that is capable of supporting a breakdown
voltage more than 30V [3]. The results of VTH, BV and
Ron were analyzed and processed to get the optimal
Ahmed Nazri et al., 2017
163
design by using L8(27) of orthogonal array in Taguchi
method. The optimized results obtained was simulated
in order to verify the predicted optimal design. The
values of the process parameter and noise factor at
different levels are listed in Tables 1 and 2 respectively.
Table 1 Process Parameter and their level. Symbol Process parameter Unit Level 1 Level 2
A Trench width µm 1.00 1.05
B Trench depth µm 1.00 1.10
C Epitaxial thickness µm 4.00 4.10 D Epitaxial resistivity cm 0.13 0.18
Table 2 Noise Factors and their level. Symbol Process parameter Unit Level 1 Level 2
Y Epitaxial Implant Temp
°C 1200 1205
Z Gate Oxide Temp °C 880 885
3. RESULTS AND DISCUSSION
VTH of the device belongs to nominal-the-best
quality characteristics, while BV belongs to larger-the-
best quality characteristics and RON belong to smaller-
the-best quality characteristics. The S/N ratio (SNR) for
VTH is selected to get closer or given target value
(1.5V), which is also known as nominal value. The SNR
(dB) for each level of the process parameters and factor
effects on SNR (%) for VTH, BV and RON is summarized
in Table 3.
Table 3 SNR for VTH, BV and RON in n-channel
trench power MOSFET. Output
Responses
Process
Parameter
SNR (dB) Factor Effect
on SNR (%) Level
1
Level
2
VTH
(Nominal-
the-best)
A 28.20 30.80 0
B 32.83 26.17 15
C 33.24 25.75 19 D 23.35 33.65 23
BV
(Larger-
the-best)
A 30.75 31.43 5
B 30.69 31.48 6
C 30.77 31.41 4 D 30.04 32.14 46
RON
(Smaller-
the-best)
A 72.73 72.54 0
B 71.61 73.67 23 C 72.64 72.63 0
D 74.16 71.12 51
The individual optimum condition for VTH, BV
and RON were compared to choose the best optimization
value for each of the process parameters of n-channel
trench power MOSFET device. The optimal value is
chosen due to the high percentage contribution of S/N
ratio for each process parameter. The full
recommendation for optimization is A1,B2,C1,D1 i.e.
trench width is 1.00μm (level 1), trench depth is 1.10μm
(level 2), epitaxial thickness is 4.00μm (level 1) and
epitaxial resistivity is 0.13Ωcm (level 1). Once the best
setting for a combination of the output response has
been identified, the confirmation test will be performed
to verify the accuracy of the prediction. Table 4 shows
the percentage of performance improvement in VTH, BV
and RON of the device before and after the optimization
is done by using Taguchi method. According to this
table, it was found that the VTH is increased by 3.29%
from 1.424V to 1.471V after optimization. This new
VTH value is within the range of 1 to 2V and is closer to
the nominal value (1.5V). The BV is increased by
47.11% from 23.50V to 34.57 after optimization. This
new value exceeds the targeted breakdown voltage for
the 30V class of trench power MOSFET device.
Meanwhile, the RON is reduced by 39.01% from
0.224mΩ/cm2 to 0.161mΩ/cm2 after optimization. This
new value exceeded the target and is lower than
0.199mΩ/cm2 reported by Su et al. [5].
Table 4 Device performance before and after
optimization. Output
Response
Targeted
Value
Before
Optimize
After
Optimize
%
Different
VTH (V) 1~2 1.424 1.471 3.29%
BV (V) >30 23.50 34.57 47.11%
RON (m/cm2) <0.199 0.224 0.161 39.01%
4. CONCLUSIONS
As a conclusion, to obtain a high BV with the
lowest RON is desirable. To achieve this, the
optimization of the input process parameters that have
an impact on the device performance was necessary.
Through this study, the factors that most affect the
output response of n-channel trench power MOSFET
device has been identified and the optimum factor levels
were also determined. Epitaxial resistivity was found as
the most significant factor that affects the VTH, BV and
RON of n-channel trench power MOSFET device. The
second most significant factor was the trench depth. At a
gate bias of 10V, the VTH, BV and RON for n-channel
trench power MOSFET device after optimization
approaches are about 1.471V, 34.57V and 0.161mΩ/cm2
respectively. These values have reached the target.
ACKNOWLEDGEMENT
The authors would like to thank the Ministry of
Higher Education (MOHE) and UTeM for sponsoring
this study under the research grant
(RAGS/1/2014/TK03/FKEKK/B00064).
REFERENCES
[1] S.S. Raghavendra and M.K. Jagadesh, “Trench
gate power MOSFET: Recent advances and
innovations,” Advance in Microelectronics and
Photonics, pp. 1-23, 2012.
[2] Y. Lei and B. Zhang, “A lateral power MOSFET
with the extended trench gate in substrate,” in
International Conf. Solid-State and Integrated
Circuit Technology, pp. 908-910, 2010.
[3] B.J. Baliga, “Advanced power MOSFET
concepts,” Springer Science, LLC, 2010.
[4] J. Korec, N.D. Mohamed and D.C. Pitzer, “Method
of fabricating trench-gated power MOSFET,” U.S.
Patent US6534366B2, March 2003.
[5] X. Su and Q. Feng, “Investigation of performance
optimized power Trench MOSFETs with double
epilayer,” in International Conf. on Advanced
Power System Automation and Protection, pp.
2166-2169, 2011.