Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Jurnal Otomasi, Kontrol &<br />
Instrumentasi<br />
Journal of Automation, Control and<br />
Instrumentation<br />
Volume 1, No.2, Tahun 2009<br />
Diterbitkan oleh/Published by :<br />
Masyarakat Otomasi, Kontrol dan Instrumentasi<br />
Society of Automation, Control and Instrumentation<br />
2<br />
ISSN : 2085‐2517
Jurnal Otomasi, Kontrol &<br />
Instrumentasi<br />
Journal of Automation, Control and<br />
Instrumentation<br />
Volume 1, No.2, Tahun 2009<br />
Masyarakat Otomasi, Kontrol dan Instrumentasi<br />
Alamat : Litbang (ex.PAU) Lt.8 Jl. Ganesa 10 Bandung 40132, Indonesia<br />
Tel. +62-22-2514452 Tel / Fax. +62-22-2534285.<br />
Email : jurnal_oki@instrument.itb.ac.id
Tim Editor<br />
(Board Editor)<br />
[Ketua/Chairman]<br />
Deddy Kurniadi<br />
(Institut Teknologi Bandung, Indonesia)<br />
[Anggota/Member]<br />
Bambang Lelono<br />
(Institut Teknologi Sepuluh November Surabaya, Indonesia)<br />
Bolo Dwiartomo<br />
(Politeknik Manufaktur Bandung, Indonesia)<br />
Estiyanti Ekawati<br />
(Institut Teknologi Bandung, Indonesia)<br />
Mitra Djamal<br />
(Institut Teknologi Bandung, Indonesia)<br />
Parsaulian Siregar<br />
(Institut Teknologi Bandung, Indonesia)<br />
Riza Muhida<br />
(International Islamic University Malaysia, Malaysia)<br />
Satriyo Nugroho<br />
(PT. Petrokimia Gresik, Indonesia)<br />
Sudarto Ramli<br />
(PT.Yokogawa Indonesia)<br />
Suprijadi<br />
(Institut Teknologi Bandung, Indonesia)<br />
Togar MP Manurung<br />
(PT. Pertamina, Indonesia)<br />
Titon Dutono<br />
(Politeknik Elektronika Negeri Surabaya, Indonesia)<br />
Son Kuswadi<br />
(Institut Teknologi Surabaya)<br />
Waskita Indrasutanta<br />
(PT. Wifgasindo Dinamika Instrument, Indonesia)<br />
Yudi Samyudia<br />
(Curtin University of Serawak, Malaysia)<br />
ii
Kata Pengantar<br />
Dewasa ini penelitian-penelitian dalam bidang otomasi, kontrol dan instrumentasi<br />
telah mendorong perkembangan yang sangat pesat pada teknologi industri<br />
manufaktur, energi, makanan, kesehatan, transportasi, militer dan sebagainya.<br />
Teknologi tersebut mempunyai peranan yang sangat penting untuk meningkatkan<br />
kualitas baik produk maupun proses di industri serta untuk menjaga kelestarian<br />
lingkungan dan kesehatan, yang pada akhirnya sangat menentukan daya saing<br />
suatu industri maupun daya saing bangsa.<br />
Dengan maksud untuk lebih mengenalkan hasil-hasil penelitian dalam bidang<br />
otomasi, kontrol dan instrumentasi ke masyarakat serta memberikan sarana untuk<br />
bertukar informasi bagi para peneliti, praktisi dan pengguna teknologi tersebut,<br />
kami dari Masyarakat Otomasi, Kontrol dan Instrumentasi (Society of Automation,<br />
Control and Instrumentation) menerbitkan jurnal ilmiah yang dikhususkan pada<br />
bidang yang disebutkan di atas.<br />
Pada penerbitan jurnal kedua ini, makalah-makalah ilmiah yang dipublikasikan<br />
merupakan hasil kegiatan riset di lingkungan Institut Teknologi Bandung,<br />
perguruan tinggi baik dalam dan luar negeri. Kami sebagai pengurus jurnal telah<br />
mengundang para ahli/pakar baik sebagai peneliti dari berbagai universitas<br />
maupun praktisi dari industri sebagai anggota tim editor. Kami juga mengundang<br />
para peneliti, praktisi dan pengguna teknologi di bidang yang dimaksud untuk<br />
menerbitkan makalah hasil penelitian yang dilakukan di jurnal ini.<br />
Jurnal ini akan didistribusikan ke berbagai universitas, lembaga penelitian, industri<br />
serta individu terkait. Kami berharap dalam beberapa tahun ke depan, jurnal ini<br />
sudah dapat diajukan untuk diakreditasi. Dengan demikian, diharapkan misi jurnal<br />
ini sebagai bagian dalam pengembangan dan diseminasi ilmu dan teknologi<br />
otomasi, kontrol dan instrumentasi dapat tercapai.<br />
Terakhir, kami sampaikan terima kasih yang sebesarnya kepada para penulis yang<br />
telah mengajukan makalahnya untuk diterbitkan di jurnal ini dan kepada para<br />
pengurus Masyarakat Otomasi, Kontrol dan Instrumentasi yang telah menginisiasi<br />
jurnal ini. Tidak lupa kami sampaikan juga terima kasih kepada para ahli/pakar<br />
atas kesediaanya menjadi editor jurnal ini.<br />
iii
Preface<br />
Automation, control, and instrumentation are the key to the advancement of<br />
science, engineering, and technology. Progress in these fields have become<br />
thoroughly integrated into all aspects of technology in our modern lives - from<br />
agriculture, health to national security. These have a vital role to improve the quality<br />
of both industrial products and processes and to preserve the environment and<br />
health of which ultimately determines the competitiveness of an industry and a<br />
nation. Therefore, the Society of Automation, Control and Instrumentation has<br />
published this journal as a part of our responsibility to introduce the results of<br />
researches to the community and as a means to exchange information between<br />
researchers and users of this technology.<br />
In this second edition, published papers are the results of research activities within<br />
the Bandung Institute of Technology and other universities. We have invited<br />
researchers and practitioners from industries as a member of the editorial team.<br />
We also invite researchers and practitioners to publish their research in this journal.<br />
This journal will be distributed to various universities, research institutions, industry<br />
participants and contributing authors. We hope in the next few years, this journal<br />
has been filed for accredited. Thus, the mission of this journal as a part of the<br />
development and dissemination of science and technology of automation, control<br />
and instrumentation can be achieved.<br />
Finally, we thank to the authors who have submitted papers in this journal and to<br />
the Society of Automation, Control & Instrumentation who has initiated this journal.<br />
Our gratitude also goes to the researchers and practitioners on the willingness to be<br />
the editor of this journal.<br />
iv
Daftar Isi/List of Content<br />
Tim Editor/Editor Board ii<br />
Kata Pengantar/Preface iii<br />
Daftar isi/List of Content iv<br />
1 Development of Circularly Polarized Synthetic Aperture Radar<br />
Sensor Mounted on Unmanned Aerial Vehicle<br />
M. Baharuddin, P.R. Akbar, J.T.S. Sumantyo, H. Kuze<br />
2 Electric Traction Motor Drive Modelling for Electric Karting<br />
Application Using Matlab / Simulink Software<br />
D. Istardi<br />
3 Feasibility Study of Solar Power Massive Usage in Indonesia :<br />
Yield versus Cost Effective<br />
M.A. Setiawan<br />
4 Modelling and Designing The Model Predictive Control System of<br />
Turbine Angular Speed at Hydropowerplant UBP Saguling PT<br />
Indonesia Power<br />
R.K.A. Kusumah, E. Joelianto, E. Ekawati<br />
5 Performance Analysis of Finger Flexor and Finger Extensor<br />
Muscles on Wall Climbing Athletes trough Electromyography<br />
Measurement, Handgrip Strength, Handgrip Endurance and<br />
Lactate Acid<br />
H. Susanti, Suprijanto, F. Idealistina, T. Apriantono<br />
6 Study on Voltage Controller of Self-Excited Induction Generator<br />
Using Controlled Shunt Capacitor, SVC Magnetic Energy Recovery<br />
Switch<br />
F.D. Wijaya, T. Isobe, R. Shimada<br />
v<br />
1<br />
7<br />
21<br />
29<br />
41<br />
49
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Development of Circularly Polarized Synthetic Aperture Radar<br />
Sensor Mounted on Unmanned Aerial Vehicle<br />
M. Baharuddin, P.R. Akbar, J.T.S.Sumantyo, and H. Kuze<br />
Microwave Remote Sensing Laboratory, Center for Environmental Remote Sensing, Chiba<br />
University, 1-33, Yayoi, Inage, Chiba 263-8522 Japan, merna5@graduate.chiba-u.jp<br />
Abstract<br />
This paper describes the development of a circularly polarized microstrip antenna, as a part of the<br />
Circularly Polarized Synthetic Aperture Radar (CP-SAR) sensor which is currently under developed at<br />
the Microwave Remote Sensing Laboratory (MRSL) in Chiba University. CP-SAR is a new type of sensor<br />
developed for the purpose of remote sensing. With this sensor, lower-noise data/image will be<br />
obtained due to the absence of depolarization problems from propagation encounter in linearly<br />
polarized synthetic aperture radar. As well the data/images obtained will be investigated as the Axial<br />
Ratio Image (ARI), which is a new data that is expected to reveal unique various backscattering<br />
characteristics. The sensor will be mounted on an Unmanned Aerial Vehicle (UAV) which will be aimed<br />
for fundamental research and applications. The microstrip antenna works in the frequency of 1.27<br />
GHz (L-Band). The microstrip antenna utilized the proximity-coupled method of feeding. Initially, the<br />
optimization process of the single patch antenna design involving modifying the microstrip line feed to<br />
yield a high gain (above 5 dBi) and low return loss (below -10 dB). A minimum of 10 MHz bandwidth is<br />
targeted at below 3 dB of Axial Ratio for the circularly polarized antenna. A planar array from the<br />
single patch is formed next. Consideration for the array design is the beam radiation pattern in the<br />
azimuth and elevation plane which is specified based on the electrical and mechanical constraints of<br />
the UAV CP-SAR system. This research will contribute in the field of radar for remote sensing<br />
technology. The potential application is for landcover, disaster monitoring, snow cover, and<br />
oceanography mapping. Especially for Indonesia which is the largest archipelago country in the world,<br />
the need for surface mapping and monitoring is demanding.<br />
Keywords: synthetic aperture radar, circular polarization, microstrip antenna<br />
1 Introduction<br />
A circularly-polarized Synthetic Aperture Radar (CP-SAR) to be launched onboard a microsatellite<br />
is currently developed in the Microwave Remote Sensing Laboratory (MRSL) of the<br />
Center for Environmental Remote Sensing, Chiba University. SAR is a multipurpose sensor<br />
that can be operated in all-weather and day-night time. As part of the project, an airborne<br />
CP-SAR development is also undertaken in order to gain sufficient knowledge of CP-SAR<br />
sensor systems. An L-band CP-SAR system will be designed for operation onboard an<br />
unmanned aerial vehicle (UAV).<br />
Historically, synthetic aperture radars (SAR) have used linearly polarized (LP) antenna<br />
systems. However, there are limitations due to the propagation phenomenon namely the<br />
variation of geometric differences between earth and the radar, the occurrence of a phase<br />
shift as a result of radio wave strike the smooth reflective surface, etc. These phenomenon<br />
leads to a backscatter variation, random redistribution of returned signal-energy and in the<br />
end the formed image would encounter a spatially variant blurring and defocusing as well as<br />
ambiguous identification of different low-backscatter features in a scene.<br />
1
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
As compared with the conventional linear polarization SAR, most of the above-mentioned<br />
effects can be alleviated through the use of CP-SAR. Thus, a CP-SAR sensor would provide a<br />
greater amount of information about scenes and targets being imaged than a linear SAR<br />
sensor. The present work focuses on the design of an L-band CP-SAR antenna. We consider<br />
the SAR system requirements to achieve an excellent performance of the overall CP-SAR<br />
system, including optimization of the single element patch and array designing.<br />
2 Circularly Polarized SAR Antenna Requirements<br />
Table 1. Specification of CP-SAR onboard Unmanned Aerial Vehicle<br />
Parameter Specification<br />
Frequency f 1.27 GHz (L band)<br />
Chirp bandwidth 10 MHz<br />
Polarization Transmitter : RHCP<br />
Receiver : RHCP + LHCP<br />
Gain G > 20 dBic<br />
Axial Ratio AR < 3 dB (main beam)<br />
Antenna size 1.75 m (azimuth)<br />
0.5 m (range)<br />
Beam width 8º (azimuth)<br />
25º (range)<br />
Altitude range 3,000 - 10,000 m<br />
The capability of a SAR antenna can be described by its sensitivity, spatial resolution in<br />
range and azimuth directions, image quality, ambiguities, and swath coverage [1]. Table 1<br />
shows the specifications and targets desired for the present CP-SAR system, which in turn<br />
influence the specification of the L-Band CP-SAR antenna.<br />
The operation frequency of 1.27 GHz (L-band) has been chosen, since its relatively longer<br />
wavelength ensures better penetration through vegetation canopies. The drawback<br />
associated with this choice, however, is the relatively large dimension of microstrip<br />
elements. The requirements for the range resolution (15 m) determine the antenna<br />
bandwidth of 10 MHz, or less than 1% of the operation frequency of 1.27 GHz. This<br />
bandwidth requirement must be compatible with a low axial ratio (AR) (below 3 dB) for<br />
ensuring transmitting/receiving circularly-polarized waves. To satisfy the matching of input<br />
impedance, the return loss must be smaller than 10 dB in this bandwidth range.<br />
3 CP-SAR Antenna Concept Analysis<br />
The primary considerations in the design and fabrication process are low cost, light weight<br />
and ease of manufacturing. The CP-SAR antenna is conceived in the way that every single<br />
element microstrip patch is a circularly polarized antenna. Feed network will be<br />
implemented in different layer substrate as the feeding method is proximity coupled.<br />
2
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Figure 1. Configuration of a CP-SAR antenna array consisting of microstrip elements.<br />
The primary considerations in the design and subsequent fabrication processes are low<br />
cost, light weight and ease of manufacturing. The CP-SAR antenna consists of an array of<br />
single antenna elements, each being a microstrip antenna for circular polarization. The<br />
single element patches which have been optimized are then spatially arranged to form a<br />
planar array (see Figure 1 for illustration). The planar array configuration is widely employed<br />
in radar systems where a narrow pencil beam is needed [2]. The beam pattern for optimum<br />
ground mapping function is cosecant-squared beam in the elevation plane (E-plane) which<br />
can correct the range gain variation and pencil beam in the azimuth plane (H-plane) [3]. The<br />
antenna side lobe levels in the azimuth plane must be suppressed in order to avoid the<br />
azimuth ambiguity. To deal with reflection, the antenna side lobes and back lobes also must<br />
be suppressed. A better control of the beam shape and position in space can be achieved<br />
by correctly arranging the elements along a rectangular grid to form a planar array. The<br />
antenna gain is mostly determined by the aperture size and inter-element separation.<br />
To maximize the array performance, certain characteristics of feed networks have to be<br />
taken into account. These are the conductor and dielectric losses, surface wave loss, and<br />
spurious radiation due to discontinuities such as bends, junctions, and transitions [2]. The<br />
loss due to the coupling of the adjacent element have to be considered, therefore isolation<br />
between adjacent elements must be higher than 20 dB. The spacing between elements is<br />
measured as the distance between the midpoints of each element. A maximum directivity<br />
will occur for approximately spacing between elements in the range of 0.8 – 0.9 times the<br />
free space wavelength [4].<br />
4 Analysis and Design of Radiating Elements<br />
The configuration of the radiating element together with the microstrip line feed and ground<br />
plane is shown in Figure 2(a), where important parameters are labelled. Side view is<br />
depicted in Figure 2(b). The equilateral triangular radiator will generate a left-handed<br />
circular polarization (LHCP) by employing the dual feed method as shown in Figure 2(a). In<br />
order to generate a 90 o phase delay on one of the two modes, the line feed on the left side<br />
is approximately λ/4 longer than the other.<br />
Simulations with a finite ground plane model have been undertaken to optimize the size<br />
parameters using a full wave analysis tool (IE3D Zeland software) based on the method of<br />
moment (MoM) algorithm. The dimensions of the radiator, microstrip feed line and the<br />
ground plane for the equilateral triangular patch are a = 102.75, w = 6.8, ld = 21.5, le = 27,<br />
ld1 = 6.9, lc = 9.2, ls = 10.1, lm = 3.9, lst = 21.5, ws = 10.2, la = 146.1, and lr =163.1 in units<br />
3
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
of mm. The geometry model is implemented on two substrates, each with thickness t = 1.6<br />
mm, with the conductor thickness tc ≈ 0.035 mm, relative permittivity εr = 2.17 and loss tan<br />
δ (dissipation factor) 0.0005. The equilateral triangular microstrip antenna model has been<br />
fabricated to verify the simulation results. The reflection coefficient and input impedance<br />
were measured with a RF Vector Network Analyzer (Agilent, E5062A, ENA-L). The antenna<br />
gain, AR, and radiation patterns were measured inside the anechoic chamber of MRSL,<br />
having a dimension of 4×8.5×2.4 m.<br />
The experimental results are shown in Figures 3 – 5 in comparison with the simulation.<br />
Figure 3 shows the S-parameter which indicate an impedance bandwidth of more than 15<br />
MHz.<br />
In Figure 4, it can be seen that whereas the gain of the antenna is simulated to be 7.04<br />
dBic at 1.27 GHz, the experimental result shows a smaller value by about 0.6 dB. This<br />
difference may be ascribed to the fabrication imperfections (such as inaccuracy in the<br />
milling and etching processes and connector soldering) and the substrate loss. The 3-dB AR<br />
bandwidth of the simulation is 7.2 MHz and from observation it is 7.4 MHz which is still<br />
narrower than the target specification (10 MHz). To improve this situation, the next work will<br />
consider the technique to extend the 3-dB AR bandwidth.<br />
Figure 5 shows the radiation pattern in terms of gain an azimuth angle Az = 0 o (x-z plane) at<br />
the frequency of f = 1.27 GHz. A difference of around 0.7 dB is seen between simulated<br />
model and the measured antenna on the gain radiation pattern. There are some differences<br />
between the simulated and measured pattern of the antenna. This may be due to the<br />
imbalance in current distribution affected by the configuration of the antenna (such as holes<br />
and plastic screws in substrate) and the measurement system.<br />
Figure 2. Configuration of equilateral triangular patch antenna with proximity coupled feed;<br />
(a) top view and (b) side view.<br />
4
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
S 11 - Reflection Coefficient (dB)<br />
- Gain (dBic)<br />
G<br />
0<br />
-10<br />
-20<br />
-30<br />
Simulation<br />
Measurement<br />
-40<br />
1.25 1.26 1.27<br />
Frequency (GHz)<br />
1.28 1.29<br />
Figure 3. Reflection coefficient vs. frequency<br />
Gain (Simulation) AR (Simulation)<br />
8<br />
Gain (Measurement) AR (Measurement)<br />
8<br />
7<br />
7<br />
6<br />
6<br />
5<br />
5<br />
4<br />
4<br />
3<br />
3<br />
2<br />
2<br />
1<br />
1<br />
0<br />
1.25 1.26 1.27<br />
Frequency (GHz)<br />
1.28<br />
0<br />
1.29<br />
- Gain (dBic)<br />
G<br />
Figure 4. Gain and AR vs. frequency at θ (theta) angle = 0o<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
......<br />
Simulation<br />
Measurement<br />
← = 180<br />
θangle (degrees)<br />
o<br />
90 60 30 0 30 60 90<br />
= 0 o Az Az →<br />
Figure 5. Gain vs. theta angle (radiation pattern) in the theta plane (Az = 0o and 180o) (x – z plane) at<br />
f = 1.27 GHz.<br />
5<br />
AR - Axial Ratio (dB)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
5 Conclusion<br />
A circularly-polarized antenna has been developed for implementing antenna for circularlypolarized<br />
synthetic aperture radar (CP-SAR) sensor operated in L-band. The design and<br />
optimization process was carried out using a MoM analysis software. The model was<br />
actually fabricated and measured in MRSL. Although the AR bandwidth is slightly smaller<br />
than the requirement for an airborne CP-SAR system, the present work has indicated that<br />
the goals can be met through a precise adjustment in the design and fabrication process in<br />
the near future.<br />
6 References<br />
[1] Pokuls, R. , Uher, J. , and Pozar, D.M., September 1998. Dual-Frequency and Dual<br />
Polarization Microstrip Antennas for SAR Applications, IEEE Trans. Antenna<br />
Propagation, volume 46, no. 9, pp 1289-1296.<br />
[2] Garg, R. , Bhartia P. , Bahl I. , and Ittipiboon, A., 2001. Microstrip Antenna Design<br />
Handbook, Artech House, pp 720, 737.<br />
[3] Vetharatnam, G. Kuan, C.B, and Teik C.H., Microstrip Antenna for Airbone SAR<br />
Application<br />
http://www.remotesensing.gov.my/images/default/publication_3rdmicrowave/3rdmic<br />
rowave_paper5.pdf<br />
[4] Levine, E., Malamud, G., Shtrikman, S., and Treves, D., April 1989. A Study of<br />
Microstrip Array Antennas with the Feed Network, IEEE Trans. Antenna Propagation,<br />
volume 37, no. 4, pp 426-434.<br />
7 Acknowledgement<br />
The authors would like to thank Victor Wissan, Basari, Fauzan, Ilham A. and Zhang Jia-Yi for<br />
assisting in the antenna fabrication and measurement; the Japan Society for the Promotion<br />
of Science (JSPS) for Grant-in-Aid for Scientific Research - Young Scientist (A) (No.<br />
19686025); Venture Business Laboratory - Chiba University for Project 10th Research<br />
Grant; National Institute of Information and Communication Technology (NICT) for<br />
International Research Collaboration Research Grant 2008, and other research grants that<br />
have supported this research.<br />
6
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Abstract<br />
Electric Traction Motor Drive Modeling for Electric Karting<br />
Application Using Matlab®/Simulink® Software<br />
D. Istardi<br />
Batam Polytechnics<br />
Parkway st. Batam Centre, Batam Indonesia<br />
E-mail: istardi@polibatam.ac.id<br />
An electric traction motor drive for an electric karting application was modeled for efficiency studies<br />
and simulated using the MATLAB ®/Simulink ® software. The model includes models of a battery, power<br />
electronic converter, electric motor, and vehicles dynamic of go-karts to a typical 48 seconds track<br />
driving schedule. The losses of each component of the electric traction motor drive were modeled and<br />
simulated over the entire speed range. In the battery was also calculated the state of charge (SOC) of<br />
the battery over the driving cycle. The regenerative braking energy captured was also considered in the<br />
simulation. Finally, the overall electric traction motor drive system efficiency and energy consumed<br />
were estimated based on the individual model based efficiency and energy consumed analysis.<br />
Keywords : Index Terms—Battery, electric go-kart, efficiency maps, loss modeling, and regenerative<br />
braking.<br />
1 Introduction<br />
The first electric vehicle was made in the 1830s and was popular for almost a century<br />
[1],[2],[3]. However, since 1933 the numbers of electric vehicles have decreased due to the<br />
improvements of the internal combustion engine (ICE) that has become better and cheaper.<br />
Nowadays, environmental considerations, energy costs, and improvements in control and<br />
battery technology have inspired an increasing amount of research and development of<br />
electric vehicles [3].<br />
One of the developments in the electric vehicle is research on electric kart racing or karting.<br />
Research in this area is interesting due to their characteristics that were different compare<br />
to the normal electric go-kart such as energy consumption and acceleration. In electric<br />
karting, the energy is higher than in the electric go-kart. The drive cycle that was used in the<br />
electric karting have a high acceleration and deceleration. In this paper, the electric karting<br />
that would be used as an input for simulation is the ICE karting for children (8-12 years old).<br />
Therefore, the research is focus on energy consumption and losses of an electric small<br />
karting.<br />
Electric karting is a variant of an open-wheel motor sport, with small four-wheeled vehicles<br />
called karts. These karts are simple and usually raced on a scaled-down track. Since the<br />
electric kart engine is powered by an electric motor instead of an internal combustion<br />
engine and the motor is operated using the power stored in batteries [4], its engine has<br />
many advantages over the ICE. It is pollution-free, has higher energy conversion efficiency<br />
and less vibration, requires low maintenance, its speed is easy to control and it can use the<br />
energy from regenerative braking [1], [3]. An electric karting race was first started in 1989<br />
in Italy [5] and is currently getting popular in the United States and Europe due to<br />
7
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
improvements in control and battery technology.<br />
The components of electric karting are chassis made of a steel tube, a propulsion system<br />
that includes an electric motor that drives the wheels, a power electronic converter that<br />
regulates the energy flow to the motor and a transmission system, a battery that provides<br />
energy, and a control unit that ensures a proper operation of the power electronic converter<br />
[3],[6], [7].<br />
The paper is organized as follows: In the next section, a brief review of electric traction drive<br />
system components and simulation of each component are presented. In Section III, a<br />
description of the ICE karting drive cycle is given. Section IV and V presents results of<br />
simulation using MATLAB®/Simulink® software and discussion of the results. Finally, the<br />
conclusions are made in section VI.<br />
2 Electric Traction Drive System<br />
A simple electric traction drive system consists of a drive system (transmission, electric<br />
motor, and power electronics) and energy storage (battery) [6] as seen in Fig. 1. Examples of<br />
some papers that describe the modeling of an electric traction drive system using MATLAB®<br />
or other software [7]-[10].<br />
Within the system, energy is stored in the battery. A power electronic converter connects the<br />
battery to an electric motor. The voltage and current output of the battery are maintained to<br />
match the ratings of the electric motor. The electric motor converts electrical energy<br />
supplied by the battery into mechanical energy. A transmission transforms the mechanical<br />
energy into a linear motion. Speed and torque are adjusted using the gear box. The gear can<br />
transmit the rotational force at different speeds, torque, and directions. A controller and<br />
energy management control the speed and direction of the electric karting, and optimize the<br />
energy conversion from the battery to the transmission. The battery can be charged from the<br />
line power and also from regenerative braking energy.<br />
Figure 1. Block diagram of drive system for electric karting.<br />
In order to calculate the efficiency of the electric karting, it is essential to understand how<br />
the losses of each electric drive system component change with speed.<br />
8
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
2.1 Vehicle Dynamic Model<br />
Mechanical energy provided by the electric traction drive system is used to drive the wheels<br />
of the electric karting. The supplied energy must be large enough to overcome the “traction<br />
resistance” (Ft), i.e. the sum of rolling resistance (Frr), aerodynamic drag (Fad), climbing<br />
resistance and acceleration force (Faf) [8], [11]. Rolling resistance is a deformation process<br />
mechanism which occurs at the contact patch between the tires and road surface.<br />
Aerodynamic drag is the viscous resistance of air upon the vehicle. In this paper, the race<br />
track is assumed to be flat, thus the climbing resistance is neglected. Those forces can be<br />
calculated using:<br />
F<br />
t<br />
= F<br />
ad<br />
+ F<br />
rr<br />
+ F<br />
cr<br />
+ F<br />
af<br />
(1)<br />
F<br />
ad<br />
=<br />
1<br />
2<br />
δ C<br />
ad<br />
Av<br />
2<br />
a<br />
(2)<br />
F<br />
rr<br />
= C<br />
rr<br />
mg<br />
(3)<br />
F<br />
af<br />
= ma<br />
(4)<br />
Where δ : front surface area of vehicle [m2]<br />
Cad : coefficient aerodynamic drag<br />
va : relative vehicle speed with respect to air [m/s]<br />
Crr : coefficient rolling resistance<br />
m : total vehicle mass, include the driver [kg]<br />
g : gravitational acceleration [m/s2]<br />
a : acceleration [m/s2]<br />
Calculation of the torque generated by the electric motor is based on energy considerations<br />
in terms of inertias (J), load acceleration, coupling ratio (B), and the load torque (TL) or force<br />
as shown below:<br />
d ω<br />
T<br />
r<br />
r<br />
= J + B ω<br />
dt r<br />
+<br />
The losses in this model are neglected. The parameters of vehicle dynamics for a small<br />
electric karting can be seen in Table 1.<br />
2.2 Electric Motor Losses<br />
9<br />
T<br />
L<br />
Table 1 . Vehicle dynamic parameter for small go kart<br />
Total mass 110 kg<br />
Rolling resistance coeff. 0.03<br />
Drag coefficient 0.6<br />
Air density 1.202 kg/m3 Vehicle cross section 0.5 m2 Driving wheel radius 0.14 m<br />
The losses in the electric motor can be divided into 4 components [12]: copper losses, core<br />
losses, mechanical losses and stray losses. In this paper, only copper and iron losses will be<br />
used in simulation and analysis. Mechanical and stray losses are disregarded. The electric<br />
(5)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
motor that used in this paper is induction motor due to its simplicity, minimum maintenance<br />
requirement, and low costs [13]-[15]. Per-phase equivalent circuit of the induction motor at<br />
steady state is shown in Figure 2 [15]-[18].<br />
Figure 2: Equivalent circuit of induction motor<br />
From the equivalent circuit in Figure 2 and power flow in induction motor, the rotor current at<br />
rated condition can be calculated by<br />
I<br />
r<br />
Where Td : developed torque of electric motor [T]<br />
s : slip [%]<br />
p : pole pairs<br />
ωs : angular speed of stator [rad/sec]<br />
Using the current divider laws, the stator current is calculated by:<br />
Then the air gaps voltage can be expressed as<br />
I<br />
s<br />
2 T<br />
d<br />
ω<br />
s<br />
s<br />
= (6)<br />
3 pR<br />
r<br />
R R<br />
2<br />
2<br />
⎛ m r<br />
R<br />
ω<br />
2 ⎞ ⎡ ⎛ r ⎞⎤<br />
⎜ L L ⎟<br />
s r m ⎢ω<br />
⎜<br />
s<br />
R<br />
m<br />
L<br />
m<br />
L<br />
m<br />
R<br />
m<br />
L ⎟<br />
⎜<br />
−<br />
s<br />
⎟<br />
+<br />
⎜<br />
+ +<br />
s r ⎟⎥<br />
⎝<br />
⎠ ⎢⎣<br />
⎝<br />
⎠⎥⎦<br />
= I<br />
(7)<br />
ω L R<br />
r<br />
s m m<br />
s ω T ⎛<br />
⎞<br />
⎜ R<br />
2<br />
⎟<br />
V =<br />
s d r<br />
⎜ + L<br />
2<br />
g<br />
ω<br />
3 R<br />
⎟<br />
⎜ 2 s r<br />
r<br />
⎟<br />
⎝<br />
s<br />
⎠<br />
The total rated losses of the motor can be obtained as<br />
P<br />
loss<br />
=<br />
⎡<br />
3 ⎢ R I<br />
2<br />
⎢ s s<br />
⎢⎣<br />
+ R I<br />
2<br />
r r<br />
+<br />
V<br />
2<br />
g<br />
R<br />
m<br />
⎤<br />
⎥<br />
⎥<br />
⎥⎦<br />
(9)<br />
The induction motor can operate above the rated speed by using frequency or voltage<br />
control variations because of its rugged mechanical construction [10], [18]-[21]. The torque<br />
and power capabilities as a function of rotor speed can be seen in Figure 3. In the below<br />
rated speed area, the flux in the air gap is kept constant by controlling Vs/f and the value of<br />
10<br />
(8)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
the slip is small. Then the electric motor can produce torque until up to its rated torque and<br />
as a result, the rotor current may be assumed to be proportional to the torque.<br />
Figure 3. Induction motor characteristic and capabilities<br />
To increase the motor speed above the rated speed, the stator voltage is kept at the rated<br />
voltage and the stator frequency is increased to a value above the rated frequency. So, the<br />
Vs/f is reduced and the flux is also reduced by the ratio of the instantaneous operating<br />
speed to the rated speed. The rating of the motor can be seen in Table 2.<br />
Table 2. Induction motor rating<br />
Power 6000 W<br />
Voltage 3 x 27 V<br />
Frequency 100 Hz<br />
Speed 2850 rpm<br />
Current 168 A<br />
Weight 19.2 kg<br />
From this rating of the motor, efficiency map can be calculated and the result can be seen in<br />
Figure 4.<br />
11
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Figure 4. Efficiency maps for the used induction motor<br />
2.3 Power Electronic Converter<br />
The power electronic converter used in this simulation is a standard six-switch three-phase<br />
bridge inverter. The aim of this component is to provide appropriate mean values of<br />
parameters commonly used in electric motor. The power switching device used in this paper<br />
is MOSFETs (Metal Oxide Semiconductor Field Effect Transistors) and anti-parallel power<br />
diodes. The controller pulse used in this simulation is a three-phase pulse-width modulation<br />
(PWM). The main losses in the converter are the conduction losses (Pcond) and switching<br />
losses (Psw) for each switching component [10],[19]. Switching process of power electronic<br />
components depends on the load and switching strategies [19],[21]-[25]. The losses for the<br />
MOSFET are<br />
P<br />
Q<br />
where M : Modulation index (0
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
The parameters of this component can be found in Table 3.<br />
Table 3. Power electronic converter parameter<br />
Power MOSFET MTD3055VL<br />
Strain drain to source on-resistance 0.012 ohm<br />
Rise time 85e-9 seconds<br />
Fall time 43e-9 seconds<br />
Constant voltage drop 0 V<br />
Power Diode QuietIR series 20 ETF<br />
Forward voltage drop 1.2 V<br />
On-resistance 0 ohm<br />
rms reverse voltage 21 V<br />
Snappiness factor 0.6<br />
Rate of fall forward current 100e6 A/s<br />
Reverse recovery time 60e-9 seconds<br />
Controller<br />
Frequency switching 10 kHz<br />
Modulation index 0.5<br />
Power factor motor 0.8<br />
2.4 Battery<br />
A battery is a device that converts chemical energy into electrical energy and vice versa. In<br />
this paper, a generic battery model will be used. The model is a modification of the<br />
Sheppard discharge battery model introduced by [26]. The battery is modeled using a<br />
controlled voltage source in series with internal resistance, as shown in Figure 5.<br />
Figure 5. Generic battery model.<br />
This model can represent the behavior of different battery types. The parameters of this<br />
model can be extracted from the discharge curve data. This model is based on several<br />
assumptions: the model has the same characteristic of charge and discharge cycles, the<br />
model has constant internal resistance and there is no Peukert effect; the battery capacity<br />
13
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
does not change with the amplitude of the current.<br />
The simulation used eight batteries as energy storage with the capacity was 36 Ah and the<br />
internal resistance of the batteries was 0.045 Ω.<br />
2.5 Regenerative Braking<br />
Regenerative braking is a mechanism to reduce the vehicle speed by converting some of its<br />
kinetic energy to other useful form of energy [27]-[29]. This converted energy can be used to<br />
charge the energy storage in the system, such as a battery or a capacitor. The regenerative<br />
braking is different from an auxiliary drive braking, where the electrical energy is dissipated<br />
as heat by passing current through large bank of variable resistors.<br />
The total energy dissipation is limited by either the capacity of the supply system to absorb<br />
this energy or by the SOC of the battery. If SOC of the battery is full, the auxiliary drive<br />
braking will absorb the excess energy. In order to capture the regenerative braking energy,<br />
the total traction torque must be negative.<br />
3 Load Profile of the Electric Drive Systems<br />
The load used in this paper was drive cycle of ICE karting at race day for one lap (48<br />
seconds). This drive cycle had previously been measured by the Flap track software at<br />
Göteborg karting ring track. The speed profile of the ICE karting for one lap can be seen in<br />
Figure 6. The drive shall be optimized for 10 minutes heat in full race and 3 minutes for in<br />
and out laps.<br />
Figure 6. Speed profile of ICE karting at Göteborg karting ring track<br />
According to this speed characteristic, the traction torque at the wheel is varying between 47<br />
Nm and -78 Nm. The negative torque indicates that the ICE karting is in deceleration or in<br />
regenerative braking region. It is clear that the regenerative braking or deceleration torque<br />
is higher than the acceleration torque. Therefore, the regenerative power used in charging<br />
the battery must be limited due to the limitation of the electric motor, power electronic<br />
converter and battery capability [27].<br />
4 Modeling of the Electric drive system<br />
The model of electric traction drive system for electric karting is implemented using<br />
14
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Matlab ® /Simulink ® software. A steady-state model is used to get raw data that is helpful<br />
during the design stage and for long-term analysis over an extended drive cycle. The<br />
advantage of this modeling is fast computation. The steady-state model is suitable to model<br />
the efficiency and performance of the system.<br />
This simulation ran in 48 seconds and used variable-step ode45 (Dormant-Prince) solver.<br />
The relative tolerance was 1e-3. There were also three m files that each represents the<br />
system parameters, displays the post processing, and includes calculation on the<br />
performance of each component of the system.<br />
5 Result and Discussion<br />
According to the data sheet of the electric motor, the rated of the motor is 2850 rpm, 6000<br />
W and 20.1 Nm. The speed and torque at the wheel were between 886 - 1801 rpm and 0 -<br />
47 Nm respectively, as explained. This speed and torque must be geared to values that have<br />
a high efficiency of the electric motor. By using, the gear ratio used in this drive system has<br />
been selected to 45/21. With this gear ratio, the induction motor operates at speeds<br />
between 1889 – 3859 rpm and torques between 0 – 21.9 Nm. The slip of the induction<br />
motor was assumed to be constant because the electric motor operates almost in the<br />
torque constant region. The average efficiency of the electric motor was 86.2%. The average<br />
efficiencies of the power electronic converter and the battery were 92.7% and 83.5%,<br />
respectively as seen in Figure 7. Thus, the total average efficiency of this system was 66.7%.<br />
Figure 7. Efficiency of the drive system<br />
The lower efficiency at the battery occurred as a result of the large current in the equivalent<br />
internal resistance of the battery. At regenerative braking condition, the efficiency of the<br />
electric drive system was lower than that of normal operation due to the larger current in<br />
regenerative braking. Therefore, at regenerative braking condition, the losses increased in<br />
the power electronic converter, battery and electric motor.<br />
The total energy used by the drive system is the subtraction of the regenerative energy from<br />
the used driving energy and can be seen in Figure 8.<br />
15
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Figure 8. Energy used of the drive system.<br />
Figure 8 shows that the regenerative braking supplies a higher energy level in short time<br />
compared to normal operation. Consequently, this energy must be considered in designing<br />
an electric traction drive system. The comparison of energy used can be found at Table 4.<br />
Table 4. Comparison of energy usage<br />
Reg W/o reg [Wh] With reg [Wh]<br />
[Wh] a lap 13 min a lap 13 min<br />
Transmission 18.8 48.7 796.3 30.2 490.8<br />
Motor 15 53.5 869.4 38.5 626.2<br />
PEC 13.7 57.2 929.4 42 706.1<br />
Battery 11.4 67.9 1104.5 56.6 920<br />
Table 4 shows that the total regenerative energy at the battery is lower than the total<br />
regenerative braking energy at the electric motor due to losses at the components of the<br />
electric drive system. The energy needed to operate the electric karting for 13 minutes was<br />
920 Wh with regenerative braking, or 1104.5 Wh without regenerative braking. Therefore,<br />
the energy available in the battery (48Vdc) must be at minimum 19.2 Ah with regenerative<br />
braking and 23 Ah without regenerative braking.<br />
The performance of the battery in the drive system can be evaluated using the SOC of the<br />
battery during the drive cycle that can be seen in Figure 9. At the end of simulation, the SOC<br />
was 0.9639. If the drive cycle is assumed to be the same for other times, the storage energy<br />
can support the electric traction drive system for 23 minutes. It is clear that the regenerative<br />
braking energy can be used to charge the battery and reduce the energy usage in the<br />
system.<br />
A power and torque characteristics as function of the speed can be seen in Figure 10. The<br />
transmitted power in the electric traction drive cycle increases with the increasing speed<br />
due to increasing frequency and voltage until the rated speed. The maximum power occurs<br />
in the rated speed. At above the rated speed, the transmitted power decreases again as a<br />
16
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
result of the decreasing stator and rotor currents. The torque is almost constant at rated<br />
torque below the rated speed and will decrease at above the rated speed due to decreasing<br />
power transmitted in the electric karting. At regenerative braking condition, the torque is<br />
almost constant above the rated torque of the electric motor.<br />
6 Conclusion<br />
Figure 9. State of charge of battery at the drive system<br />
Figure 10. Speed – power and torque characteristics of the drive system<br />
The complete electric traction drive system was simulated and observed. The total average<br />
efficiency of the system was 66.7% and the total efficiency depended on the efficiency of<br />
the electric motor and the battery. In this simulation, there was no limitation for regenerative<br />
braking energy feed into the electric motor. The average power of the electric motor was<br />
found to be 5.4 kW. The total energy consumed for this electric traction drive system was<br />
56.61 Wh in one lap with the regenerative braking energy and 920 Wh in the whole race.<br />
17
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
The type of battery can be changed to other types of battery which have a lower internal<br />
resistance but the price will be higher.<br />
In this paper, the main focus has been placed on the induction motor modeling using the<br />
efficiency model on steady state. Thus, there exists much future research scope in<br />
improving the behavior of the electric motor, power electronic converter, and battery for<br />
dynamics simulation. Furthermore, the use of an advanced traction motor such as the<br />
permanent magnet DC motor, permanent magnet synchronous motor, series DC motor,<br />
brushless DC motor, and switched reluctance motor, which have even higher efficiencies,<br />
might lead to higher system efficiencies.<br />
7 References<br />
[1] J. Larminie, J. Lowry, Electric Vehicle Technology Explained, John Wiley and Sons,<br />
2003<br />
[2] C. C. Chen, “An overview of electric vehicle technology,” Proceeding of IEEE, vol. 81,<br />
no. 9, pp. 1202-1213, Sept 1993.<br />
[3] M. Ehsani, K. M. Rahman, H. A. Toliyat, “Propulasion system design of electric and<br />
hybrid vehicles,” IEEE Trans on Industrial electronics, VOL. 44, NO. 1, pp. 19-27,Feb<br />
1997<br />
[4] http://en.wikipedia.org/wiki/Go_karting last visited 25-2-09<br />
[5] http://www.kartelec.com/f/en_actu.htm last visited 25-2-09<br />
[6] C. Cardoso, J. Ferriera, V. Alves, R. E. Araujo, “ The design and implementation of an<br />
electric go-kart for education in motor control, “ IEEE International SPEEDAM 2006,<br />
pp. 1489 – 1494, May 2006.<br />
[7] F. J. Perez-Pinal, C. Nunez, R. Alvarez, M. Gallegos, “ Step by step design procedure of<br />
an independent-wheeled small EV applying EVLS,” IECON 2006-32nd Annual<br />
Conference on IEEE Industrial Electronics, pp. 1176-1181, Nov 2006.<br />
[8] M. Xianmin, “Propulsion system control and simulation of electric vehicle in MATLAB<br />
software environment, “Proceeding of the 4th World Congress on Intelligent Control<br />
and Automation 2002, pp. 815-818, June 2002.<br />
[9] J. M. Lee, B. H. Co, “ Modeling and simulation of electric Vehicle power system, “<br />
Proceeding of the 32nd Intersociety IECEC-97, vol. 3, pp. 2005-2010, August 1997<br />
[10] S.S. Williamson, A. Emadi, K. Rajashekara, “Comprehensive Efficiency Modeling of<br />
Electric Traction Motor Drives for Hybrid Electric Vehicles Propulsion Applications,”<br />
IEEE Transaction on Vehicular Technology, vol. 56, no. 4, pp.1561-1572, July 2007.<br />
[11] Robert Bosch GmbH, BOSCH-Automotive Handbook, Robert Bosch GmbH, German,<br />
2002<br />
[12] G. C. D. Sousa, B. K. Bose, “ Loss modeling of converter induction machine system for<br />
variable speed drive,” Proceeding of the 1992 International Conference on Power<br />
electronics and Motion Control, vol. 1, pp. 114-120, Nov. 1992<br />
[13] Mohamed A. El-Sharkawi, ”Fundamentals of Electric Drives,” Brooks/Cole Thomson<br />
Learning, 2000.<br />
[14] J.J. Cathey, ”Electric Machines Analysis and Design Applying Matlab®,” McGraw Hill,<br />
2001<br />
[15] C. Shumei, L. Cheng, S. Liwei,” Study on Efficiency Calculation Model of Induction<br />
Motors for Electric Vehicles,” IEEE Vehicle Power and Propulsion Conference, pp. 1-5,<br />
Sept 2008<br />
[16] J. Faiz, M. B. B. Sharifian, “Optimal design of an induction motor for an electric<br />
vehicle, “Euro. Trans. Electr. Power 2006, vol. 16, pp. 15-33, July 2005.<br />
[17] G. Pugsley, C. Chillet, A. Fonseca, A-L. Bui-Van, “New modeling methodology for<br />
18
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
induction machine efficiency mapping for hybrid vehicles,” IEEE International Electric<br />
Machines and Drives Conference 2003, vol.2, pp. 776-781, June 2003.<br />
[18] S.M. Lukic, A. Emado, “ Modeling of Electric Machines for Automotive Applications<br />
Using Efficiency Maps,” Electrical Insulation Conference and Electrical Manufacturing<br />
& Coil Winding Technology Conference 2003, pp. 543-550, Sept. 2003.<br />
[19] N. Mohan, T. M. undeland, and W. P. Robbins, ”Power Electronics: Converter,<br />
Applications, and Design,” Hoboken, NJ:Wiley, Oct 2002<br />
[20] B.K. Bose, “Power Electronics and Variable Frequency Drives,” IEEE Press, New York,<br />
1997.<br />
[21] B.K. Bose, “Modern Power Electronics and AC Drives,”Prentice Hall, New York, 2002.<br />
[22] I. Husain, M. S. Islam, “Design, Modeling and Simulation of an Electric Vehicle<br />
System, “SAE-Advanced in Electric Vehicle Technology, 1999-01-1149, March 1999.<br />
[23] Ali Emadi, “ Handbook of Automotive Power Electronics and Motor Drives,” CRC Press<br />
– Taylor and Francis Groups, Florida 2005<br />
[24] F. Casanellas ,”Losses in PWM inverter using IGBTs,” IEE Proc. Electr. Power Appl.,<br />
vol. 141, no. 5, pp. 235-239, September 1994.<br />
[25] P.A. Dahono, Y. Sato, T. Kataoka,” Analysis of conduction losses in inverter,” IEE Proc.<br />
Electr. Power Appl., vol. 142, no. 4, pp. 225-232, July 1995<br />
[26] O. Tremblay, L-A. Dessaint. A-I. Dekkiche, “A generic battery model for the dynamic<br />
simulation of hybrid electric vehicles, “IEEE Conference on Vehicles Power and<br />
Propulsion VPPC 2007, pp. 284-289, Sept. 2007.<br />
[27] J. Lee, D. J. Nelson, “Rotating inertia impact on propulsion and regenerative braking<br />
for electric motor driven vehicles,” IEEE conference on Vehicle Power and Propulsion<br />
2005, pp. 308-314, Sept 2005.<br />
[28] B. Cao, Z. Bai, W. Zhang, “Research on control for regenerative braking of electric<br />
vehicle,” IEEE International Conference on Vehicular Electronics and Safety 2005, pp.<br />
92-97, Oct 2005.<br />
[29] Z. Junzhi, L. Xin, C. Shanglou, Z. Pengjun, “Coordinated control for regenerative<br />
braking system,” IEEE Vehicle Power and Propulsion Conference (VPPC), pp 1-6, Oct.<br />
2008<br />
19
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Feasibility Study of Solar Power Massive Usage in Indonesia: Yield<br />
versus Cost-Effective<br />
Abstract<br />
M. A. Setiawan<br />
Politeknik Manufaktur Timah<br />
Jln. Timah Raya Air Kantung Sungailiat Bangka Indonesia<br />
Email : made_andik_s@plasa.com, made@polman-timah.ac.id<br />
The aim of this study is to analyze the cost production of solar power utilization comparing with its<br />
annual yield especially in Indonesia. Solar cell module employed is poly-crystalline silicone with Peak<br />
Power 20 Wp, Power Current (Imp) 1.17 A and Power Voltage (Vmp) 17.1 V. To obtain the maximum<br />
power of the sun, the module is static fixed in 1 0-2 0 N adjusting to the equator line. The<br />
measurement is conducted in Timah Manufacture Polytechnic which is situated in 1 020’ S and 106 0<br />
E. The output is observed by multimeter data logger for every hour average. The cable employed<br />
between solar module and the multimeter and the battery is NYA Eterna 2.5 mm 2 450/750 V with<br />
SNI number 04-2698 SPLN 42, the long is 40 meter and have resistance about 0.6-0.7 Ω. The<br />
measurement output indicates that the maximum solar power is at 11.00 to 14.00 WIB. In these<br />
times, the current output is more than Imp of the Module. By calculating the average of energy<br />
received in a day, the energy received is 65%-75% of Imp. Therefore by mathematically calculation,<br />
the annual yield of current is about 4,982.68 Ampere and in 25 years will be around 124,566.90<br />
Ampere. According to PLN statistic report in 2008, the average cost production in 2007 is<br />
Rp.706.62/KWH and the cost production of the solar energy is about Rp.440.819/KWH. This<br />
calculation is included the investment, overhead and 10% of inflation /year. By comparing to the<br />
regional minimum revenue (UMR) per month in 2008, the cost of investment for the solar power<br />
usage is about 6-7 times. Although break-even-point will be occurs in 10 years, the affordability of<br />
Indonesian for massive usage of solar power is still too hard and need funded by government and<br />
others funding groups.<br />
Keywords: Solar Cell, Cost-Effective, Poly-Crystalline-Silicone, Annual Yield<br />
1 Introduction<br />
Solar power is the one of potential renewable energy sources in the future as stated by<br />
Mark Clayton [1] "Solar power is the energy of the future - and always will be". The massive<br />
usage of the renewable energy is not always in term of technology but also in term of costeffective<br />
[2,3] : how the source could provide the affordable energy.<br />
Indonesia is situated in tropical area which receives a lot of solar energy every year.<br />
Unfortunately, the applying of solar energy as the one energy sources is still have many<br />
problems such low efficiency and more expensive than fossil fuel [4]. Another problem is<br />
that there is no actual data in experimental based which shows the actual solar energy<br />
received annually. Thus the massive usage of solar power in term of cost-effective is hard to<br />
be determined.<br />
This study is to determine and evaluate the annual yield of solar energy by experimentally<br />
measurement. Obtaining annual yield is useful to analyze the potential energy sources<br />
21
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
based on employing mass technology equipments. After that, the cost- effective of the solar<br />
power especially in Indonesia could be predicted.<br />
2 Literature Review<br />
The main advantages of solar power is “quite operation, good reliability”, and high scalable<br />
[4], “abundant, clean, renewable energy” [2], and also “almost pollution- free” [3]. Solar<br />
energy also will reduce greenhouse gas emission 1.7 million tons and 1.9 million tons car<br />
gas emission every year and estimated in the 2050, carbon dioxide emissions will reduce<br />
up-to 62% than emission in the 2005 [3].<br />
The most used solar energy is crystalline silicon, thin-film solar cell and solar concentrator<br />
[5]. Solar concentrator is technology which uses mirror to mix a light in “high-performance<br />
and sensitivity” area [6]. Another technique is presented by Marc Dalbo from MIT laboratory<br />
[7]. This technology is called dye molecule. Dye Molecule is to transmit a light from the sun<br />
in different wave length as well as be used in fiber optic communication. The energy of the<br />
light is more powerful than in original wave length.<br />
The investment of applying solar energy today is booming [6]. In 2008 the solar power<br />
usage is increased to 55% than in 2007. United States government has been allocated<br />
funding more than $400 billion for solar power investment up to 2050 [4].<br />
The scientists also conduct intensive research to increase the efficiency of solar cell<br />
technology. Scientists in National Renewable Energy Laboratory (NREL) claim that they have<br />
produced a solar cell which has efficiency up to 40.8% [8]. Marc Dalbo with his Dye<br />
Molecule claims that his technology efficiency can be 50% [7]. General efficiency<br />
development of solar cell is presented by Figure 1.<br />
Figure 1. Solar cell technologies and its efficiency [8]<br />
22
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
3 Materials And Methodology<br />
This study is conducted in Timah Manufacture Polytechnic, Sungailiat – Bangka - Indonesia.<br />
Bangka Island is situated in 1020’-307’ S and 1050-1070 E. The satellite map of Bangka<br />
Island from Google Earth in 2008 is illustrated in Figure 2.<br />
Figure 2. Satellite map of Bangka Island from Google Earth and solar cell fixed in campus.<br />
Solar cell employed is 2 modules of poly-crystalline silicon which has specification:<br />
a. Module Type : 1-051Z-00067<br />
b. Peak Power (pmp) : 20 W pmp<br />
c. Max Power Current (Imp) : 1.17 A<br />
d. Max Power Voltage (Vmp) : 17.1 V<br />
e. Short Circuit Current (Isc) : 1.29 A<br />
f. Open Circuit Volatge (Vop) : 21.5 V<br />
g. Nominal Temperature Cell (Tnoct) : 25 0 Celcius<br />
h. Test Data Condition : E=1000 W/m 2 , Tc=25 0 C<br />
This technology is employed due to the affordability of Indonesian, components reliability<br />
and space needed. The others technology need more expensive in investment, wider space<br />
in installation and harder in components reliability.<br />
To obtain the maximum sun light power, the module is fixed to the equator line. The module<br />
fixed in 1 0 – 2 0 N because the campus location is in 1 0 – 2 0 S. The module employed and its<br />
setup is illustrated in Figure 3.<br />
Configuration of components used in this study is solar cell module, multi-meter (current<br />
and voltage), battery, inverter and DC/AC loads. The equipments configuration of this study<br />
is presented in Figure 4.<br />
Figure 3. Setup and solar cell module<br />
23
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Figure 4. Equipments configuration<br />
The output of the solar cell module is in Ampere and in Volt, but the current received on the<br />
multi-meter will be depended and deducted by resistance of the cable. According to Ohm’s<br />
law, resistance of the cable depend on length, diameter and materials of the cable. To<br />
eliminate the resistance as much as possible, the diameter of the cable is chosen which has<br />
a bigger diameter. The affordable cable in Bangka’s market is maximum 2.5 mm 2 of<br />
diameter. In this study the cable used has type NYA Eterna 2,5 mm 2 450/750 V, SNI<br />
number 04-2698 SPLN 42 and 40 meter of length. This cable has 0,6-0,7 Ω of resistance.<br />
The battery employed is GS Astra with 70 AH.<br />
4 Results And Discussions<br />
The output of the solar module is sampled for every hour from 08.00 WIB to 16.00 WIB. The<br />
measurements will be conducted in a year to obtain actual annual yield. The results of these<br />
measurements are presented in Figure 5 and 6.<br />
Ampere<br />
2.50<br />
2.00<br />
1.50<br />
1.00<br />
0.50<br />
0.00<br />
Current Yield<br />
8 9 10 11 12 13 14 15 16<br />
Time (WIB)<br />
Figure 5. Current output of the solar cell module<br />
24<br />
July<br />
August<br />
Sept<br />
Imp
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Figure 6. Voltage output of the solar cell module<br />
The results indicate that:<br />
a. The maximum current is received between 11.00 WIB and 14.00 WIB.<br />
b. By mathematically calculation, the average energy received for a day is around 65%<br />
to 75% of Imp times by hours.<br />
c. The voltage output of the solar cell module is likely to be not much different each<br />
others.<br />
According to daily yield of the energy, it can be mathematically calculated that in toward 5<br />
year the production cost will be only for overhead and total ampere will be around<br />
6.411.000 A. The comparison of production cost and energy yield is presented in Table 1.<br />
Table 1. Cost production and energy yield estimation up to 25 years<br />
Year 1 5 10 15 20 25<br />
Ampere 4,982.68 24,913.38 49,826.76 74,740.14 99,653.52 124,566.90<br />
Cost (Rp) 4,940,000 6,411,000 8,172,100 9,964,110 12,080,521 12,130,521<br />
According to PLN statistic report in 2008, the average cost production of power in Indonesia<br />
in 2007 is Rp. 706.62/ KWH. By mathematically calculation, the cost production of the<br />
solar energy in this study is Rp. 440.819 /KWH. This cost is cheaper than the average<br />
production cost of the PLN but equal with the cost production of natural others resources.<br />
The cost of solar energy is determined by calculating the overhead, components<br />
replacements, and 10% of inflations.<br />
Based on those calculations, the comparison of the costs up to 25 years is illustrated in<br />
Figure 7. The Figure 7 indicates that the break-event-point of the solar energy will be<br />
occurred in 10-15 years. After that the cost production of the solar energy will be cheaper<br />
than fossil fuel energy used by PLN.<br />
Although the cost production of solar energy will be cheaper than fossil fuel in toward 10-15<br />
years, but the cost of investment in the beginning year is out of the affordable of General<br />
Indonesian. According to BPS Bangka province statistic report in 2008, the regional<br />
25
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
minimum revenue (UMR) in Bangka province is about 1,1 million rupiah. Thus solar energy<br />
cost investment is about 4-6 times than UMR and become harder for their affordability.<br />
Indonesian Rupiah<br />
18,000,000<br />
16,000,000<br />
14,000,000<br />
12,000,000<br />
10,000,000<br />
8,000,000<br />
6,000,000<br />
4,000,000<br />
2,000,000<br />
-<br />
The Cost Estimation<br />
1 5 10<br />
Year<br />
15 20<br />
Figure 7. The cost comparison of thePLN and the solar energy.<br />
26<br />
Solar Energy<br />
PLN (Fossil Fuel)<br />
5 Conclusions<br />
This study is to determine and evaluate the annual yield of solar energy by experimentally<br />
measurement. Obtaining annual yield is useful to analyze the potential energy sources<br />
based on employing mass technology equipments. After that, the cost- effective of the solar<br />
power especially in Indonesia could be predicted.<br />
The measurements are conducted by sampling average every hour and every day from<br />
08.00-16.00 WIB. The solar module employed is poly-crystalline silicone with Peak Power<br />
20 Wp and Power Current (Imp) 1.17 A. The average daily yield is about 65%-75% Imp times<br />
by sum of hours.<br />
The break-event-point of solar energy will be occurred in toward 10-15 years. The cost<br />
production of solar energy is Rp. 440.819/KWH cheaper than PLN cost production. But, the<br />
cost of investment of the solar energy in the beginning year is more than 4-6 times than<br />
regional minimum revenue (UMR), therefore the massive usage of solar energy for<br />
Indonesian will be harder to be realized. The government and funding group is needed to<br />
involve in this project.<br />
6 References<br />
[1] Mark Clayton, “Solar edges closer to 'grid parity'”, The Christian Science Monitor, p25,<br />
2008.<br />
[2] Nancy Spring, “Solar Performing Brilliantly”, Electric Light and Power, 5(86), p50, 2008.<br />
[3] Ken Zweibel, James Mason, Vasilis Fthenakis, “Solar Grand Plan”, Rachel's Democracy<br />
& Health News, (976), p1, 2008<br />
[4] Jeffrey Winters, “the sunshine solution”, Mechanical Engineering, 12(130), pp24-29,<br />
2008.
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
[5] Dean Takahashi, “Solar Boom”, Technology Review, 5(111), p30, 2008<br />
[6] Gerald Parkinson, “Cost-Effective Devices Open A New Window on Solar Energy”,<br />
Chemical Engineering Progress, 8(104), p14, 2008.<br />
[7] Kevin Bullis, “Intensifying the Sun”, Technology Review, 5(111), p104, 2008.<br />
[8] Suzanne Shelley, “Solar Cell Sets World Efficiency Record at Over 40%”, Chemical<br />
Engineering Progress, 9(104), p18, 2008.<br />
27
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Modelling and Designing The Model Predictive Control System of<br />
Turbine Angular Speed at Hydropowerplant UBP Saguling PT<br />
Indonesia Power<br />
R.K.A Kusumah, E. Joelianto, E. Ekawati<br />
Research Division of Instrumentation and Control – Faculty of Industrial Technology;<br />
Bandung Institute of Technology; Jl. Ganesha 10, Bandung, 40132, Indonesia<br />
Astract<br />
Saguling Generation Business Unit (GBU) is one of hydro powerplants under PT. Indonesia Power<br />
which has vital role to produce and distribute electricity in Indonesia. The demand for electricity in<br />
Indonesia, which is fluctuative, force the plant to operate in immediate and responsive pattern.<br />
Saguling need 2 minute to connect to the transmission system from its non operating state. Plant<br />
response is controlled by manipulating guide vane opening so the water entering the turbine chamber<br />
can be maintained. 5.6 % maximum overshoot still occurs in start up process due to manual<br />
mechanism. This paper provide a design of control system using Model Predictive Control (MPC) to<br />
optimize the plant performance which is indicated by faster response time and reduced overshoot.<br />
Neural Network with Back Propagation algorithm is used to model the turbine with guide vane<br />
opening as input variable and turbine angular speed as output variable. The model is then used in<br />
MPC algorithm to compute the optimum control signals.<br />
Keywords: Neural Network, Model Predictive Control, Guide Vane, Cost Function.<br />
1 Introduction<br />
Saguling GBU operates only when demand for electricity occurs. Fast response time (2<br />
minute) from non operating state to connected state has become an advantage for Saguling<br />
plant compare to similar plant with different generating source. The plant also contribute in<br />
maintaining transmission system frequency stability, besides its main role to produce<br />
electricity [1].<br />
Generally, in order to transmit voltage and at the same time maintain frequency stability,<br />
hydro powerplant must be connected to the transmission system. For that purpose, the<br />
plant frequency must be made equal to the transmission system frequency. This sequence<br />
is done manually. Once it is connected, the whole process, including electricity producing, is<br />
done automatically.<br />
Controller is modeled and designed with hope that it would increase plant’s performance<br />
rather than controlled manually. The criterion chosen to evaluate plant’s performance are<br />
settling time and the existence of maximum overshoot. The basic consideration for<br />
choosing these criterions is because system response is clearly seen based on these<br />
criterions. In addition, at present time, system response still has overshoot and long<br />
duration of settling time. System is modeled using neural network with backpropagation<br />
algorithm while Model Predictive Control is used to designed the controller.<br />
2 Hydro Powerplant Fundamentals<br />
29
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
2.1 Basic Principle<br />
Frequency is a significant aspect in hydro powerplant. Initially, frequency has to be made<br />
synchronized with the transmission system frequency, which is 50 Hz. Once it is connected,<br />
plant’s frequency will be influenced by transmission system frequency due to small capacity<br />
in power production that the plant’s generate, relative to overall power consumption [2]. If<br />
there is an increasing in demand for electrical power, there will be an increasing in amount<br />
of power transferred from the powerplant as well. Plant’s frequency will gradually decreased<br />
in result. This frequency fluctuation will affect the turbine angular speed.<br />
2.2. Control Principle<br />
Hydro powerplant has two sequence of control, there are control in start up condition and in<br />
synchronize condition. In start up condition, the plant is not connected to the transmission<br />
system. It must first maintain its frequency until it met the requirement to get to the next<br />
sequence. This should be done in order to avoid damage to the generator due to phase<br />
difference between generator frequency and transmission system frequency. In start up<br />
condition, the main purpose is to maintain turbine angular speed in its angular speed<br />
nominal, which is 333 rotation per minutes, equals to 50 Hz. Once this nominal is aqcuired,<br />
system is ready to connect to the transmission system. Fig.1 illustrate the control loop of<br />
hydro powerplant.<br />
3 Identification System<br />
Figure 1. Block diagram of Saguling GBU control system<br />
3.1 Identification Fundamentals<br />
Identification is a process to find model of a process or system based on experimental data<br />
provided. With such model, the characteristics of the system can be known and analyzed<br />
[3]. Fig.2 shows the identification steps.<br />
Identification process first begin by designing an appropriate experiment which later sets of<br />
data will be aqcuisited from the experiment. The data obtained from variables entering the<br />
process and variables leaving the process. The next step is to choose the model structure,<br />
30
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
including the order of the system and choosing the estimation parameter method.<br />
Afterwards, the model should be validated.<br />
Figure 2. Identification steps<br />
Turbine system was choosed because in startup condition, the process was focused in<br />
controlling the turbine angular speed. Fig.3 illustrate the relation between the input<br />
variable, the process and the output variable.<br />
Figure 3. Turbine system<br />
3.2 Neural Network<br />
Neural network is a system which modeled human nerves system as a continuous nonlinear<br />
dynamic system. This network has nodes analogues with neuron in brain.<br />
Mathematical processes used in this network are also an approach to the way how brain<br />
works and also has the ability to learn from experience. Fig.4 illustrate the structure of<br />
neural network,<br />
Figure 4. General neural network structure<br />
31
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
A simple neural network structure can be expressed in,<br />
y<br />
⎜<br />
⎛ n<br />
f W U + b⎟<br />
⎞<br />
⎝ i 1<br />
i i<br />
⎠<br />
= ∑ =<br />
W1, W2, ….Wn and Wb are neural network weighthing parameters. These values will be<br />
evaluated in each iteration process. U1, U2, ... Un are neural network inputs. Each one of<br />
them will be multiplied with weight W accordingly, which in turn will be summed and<br />
calculated with an activation function F. In result, the output Y will be known.<br />
3.3 Backpropagation Algorithm<br />
The main idea of this algorithm is to evaluate and modify the weights and bias in a way so<br />
the error value minimized. First step in this algorithm is choosing the cost function. Error is<br />
represented as the difference between desired output and output obtained from neural<br />
network learning. One of the cost functions used is the sum of quadratic error which defined<br />
in equation,<br />
( ) 2<br />
1<br />
E =<br />
2 ∑ O −<br />
d<br />
d y d<br />
(2)<br />
where Od and yd are the desired output and output obtained from neural network<br />
respectively. To minimize the cost function, the weights are evaluated with,<br />
∂E<br />
W(<br />
k + 1 ) = W(<br />
k)<br />
−η<br />
(3)<br />
∂W<br />
where W is weight, is learning rate and E is cost function in error quadratic form. On outer<br />
layer, the gradient of cost function to weight is,<br />
∂E<br />
∂W<br />
O<br />
dj<br />
= −<br />
32<br />
(1)<br />
d<br />
∑ ( Od<br />
− y d ) (4)<br />
O<br />
where is the weight connecting the d- neuron from outer layer to j-neuron from hidden<br />
layer.<br />
3.4 Identfication Using Neural Network<br />
System was identified with previous acquisited data in non real time condition. 2250<br />
number of data were used as turbine input and output, while guide vane opening was the<br />
input variable and turbine angular speed as the output variable. Time sampling used was<br />
0.1 seconds. Identification process was done using neural network. Assumed a first order<br />
system with time delay. Based on this apriori knowledge, a neural network structure can be<br />
formed to be identified.<br />
d<br />
∂y<br />
∂W<br />
Based on (1), Saguling turbine system model can be illustrated with Fig. 5,<br />
dj
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
And in mathematical equation,<br />
Figure 5. ADALINE structure<br />
Due to delay time in system response, (5) become,<br />
y ( k)<br />
= −y(<br />
t −1)<br />
a + u(<br />
t −1)<br />
b<br />
(5)<br />
33<br />
1<br />
y ( k + nk)<br />
= −y(<br />
k −1+<br />
nk)<br />
a + u(<br />
k −1)<br />
b<br />
(6)<br />
with nk is the delay time, which values nk = 4 seconds. Fig. 4 shows the model obtained<br />
from identification process while Fig. 6 shows validated model.<br />
Figure 6. Model from NN learning and actual response<br />
1<br />
1<br />
1
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Figure 7. Model validation graphic<br />
Root mean square error (RMSE) was used to analyze the closeness degree between<br />
identification result and the real value. RMSE defined by,<br />
RMSE =<br />
N<br />
∑ i=<br />
1<br />
34<br />
( y(<br />
ˆi<br />
) − yˆ<br />
( ˆi<br />
) )<br />
Table 1 shows the identification parameters result,<br />
Table 1 Identification process parameters<br />
PARAMETER VALUE<br />
Data used for learning 1100<br />
Data used for validation 2250<br />
Learning Rate 10.5 x 10 -7<br />
Epoch 1500<br />
RMSE from learning 1.72<br />
RMSE from validation 0.89<br />
RMSE parameter in Table I shows the error value for each tracing point obtained between<br />
the model and the actual data. Saguling GBU can tolerate error value to 2 % which equals to<br />
range of response the systems can handle, that is 326.4 RPM to 339.66 RPM. From<br />
validation result, the system response vary from 332.11 RPM to 333.89 RPM, which means<br />
the error range still tolerable. Therefore, the model can be used to represent the process<br />
dynamics of the system.<br />
The transfer function obtained from identification is,<br />
N<br />
2<br />
(7)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
2<br />
0.<br />
02847z<br />
− 0.<br />
1137z<br />
+ 0.<br />
881<br />
H ( z)<br />
=<br />
(8)<br />
2<br />
z −1.<br />
995z<br />
+ 0.<br />
9949<br />
Information about delay time was already included in above equation. To design the<br />
controller, the model above should be multiplied with the actuator transfer function which<br />
was already known. Below equation is the actuator transfer function[6],<br />
0.<br />
003221<br />
G ( z)<br />
= (9)<br />
z − 0.<br />
9968<br />
Thus, the whole process and the actuator transfer function become,<br />
9.<br />
16×<br />
10 z − 3.<br />
66×<br />
10 z + 2.<br />
83×<br />
10<br />
P ( z)<br />
= G(<br />
z)<br />
⋅H(<br />
z)<br />
=<br />
(10)<br />
3<br />
2<br />
z − 2.<br />
992z<br />
+ 2.<br />
983z<br />
− 0.<br />
9917<br />
35<br />
−5<br />
4 Cotroller Design<br />
4.1 Model Predictive Control (MPC)<br />
MPC is a control strategy which designed based on model of certain processes. The model<br />
is used to calculate a set of future prediction output based on set of control signals given to<br />
the model. By using an optimization algorithm to minimize MPC cost function, a set of<br />
control signal can be obtained. Thus, controler performance really depends on the<br />
availability of a good model[4].<br />
Model used in MPC in descrete state-space form can be represented as,<br />
x d<br />
y d<br />
z z<br />
2<br />
−4<br />
( k +1 ) = Ad<br />
x(<br />
k)<br />
+ B u(<br />
k)<br />
(11)<br />
( k)<br />
= C x(<br />
k)<br />
(12)<br />
( k)<br />
= C x(<br />
k)<br />
(13)<br />
is state of the system, Ad, Bd are output matrices, Cd and Cz is observable and<br />
controllable output matrice of the system respectively. Furthermore, prediciton output can<br />
be obtained by iterating a model defined by,<br />
ˆz<br />
( k + i|<br />
k)<br />
= C ˆx<br />
( k + i|<br />
k)<br />
= C<br />
z<br />
z<br />
A<br />
i<br />
d<br />
x +<br />
∑<br />
i<br />
j=<br />
1<br />
C<br />
z<br />
A<br />
j−1<br />
d<br />
−4<br />
B uˆ<br />
( k + i − j|<br />
k)<br />
Output prediction for the next k+j step, where state of the system on step k assumed to be<br />
known, can be recursively done and represented in matrice form,<br />
d<br />
(14)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
⎡ zˆ<br />
( k + 1|<br />
k)<br />
⎤<br />
C B<br />
⎡ C A<br />
x d<br />
x d ⎤<br />
⎡<br />
⎤<br />
⎢<br />
ˆ<br />
⎥<br />
z(<br />
k 2|<br />
k)<br />
⎢ 2<br />
C A<br />
⎥<br />
⎢<br />
C x Ad<br />
Bd<br />
+ C B<br />
⎥<br />
x d<br />
⎢<br />
+<br />
⎥ x d ⎢<br />
⎥<br />
= ⎢ ⎥ x(<br />
k)<br />
+<br />
u(<br />
k −1)<br />
⎢ M ⎥ ⎢ M ⎥<br />
⎢ M ⎥<br />
⎢ − 2 i<br />
⎢<br />
⎥ ⎢<br />
H<br />
H ⎥<br />
⎥<br />
v<br />
zˆ<br />
( k H | k)<br />
C A v<br />
⎢<br />
C A B<br />
⎣ + v ⎥⎦<br />
⎣ x d ⎦ ⎢∑<br />
i<br />
x d d ⎥<br />
= 0 ⎣<br />
⎦<br />
⎡ C xB<br />
d<br />
⎢<br />
C x Ad<br />
Bd<br />
+ C x B<br />
⎢<br />
d<br />
+ ⎢ M<br />
⎢ − 2 i<br />
Hv<br />
⎢∑<br />
C A B<br />
i 0<br />
x d d<br />
⎣ =<br />
L<br />
L<br />
O<br />
L<br />
0 ⎤ Δuˆ<br />
( k|<br />
k)<br />
0<br />
⎥<br />
⎥ Δuˆ<br />
( k + 1|<br />
k)<br />
M ⎥<br />
⎥<br />
M<br />
C xB<br />
d ⎥ Δuˆ<br />
( k + Hv<br />
−1|<br />
k)<br />
⎦<br />
With Δ u ˆ( k + i|<br />
k)<br />
is incremental input, which is Δ uˆ ( k + i|<br />
k)<br />
= uˆ<br />
( k + i|<br />
k)<br />
− uˆ<br />
( k + i −1|<br />
k)<br />
.<br />
(15) can be simplified into [5],<br />
Zˆ<br />
( k)<br />
= ψ x(<br />
k)<br />
+ γ u(<br />
k −1)<br />
+ ΘΔUˆ<br />
(16)<br />
144244<br />
4 3<br />
The objective cost function is,<br />
36<br />
past<br />
{ future<br />
Hp<br />
2 Hu<br />
−1<br />
JMPC ∑ zˆ<br />
( k + i)<br />
| k)<br />
− w(<br />
k + i)<br />
+<br />
i=<br />
H<br />
Q(<br />
i)<br />
∑<br />
u<br />
i=<br />
0<br />
2<br />
R(<br />
i)<br />
(15)<br />
= Δuˆ<br />
( k + i|<br />
k)<br />
(17)<br />
with Q and R are weighting matrices. The optimum of Δ û can be obtained by finding the<br />
gradient equals to zero of JMPC.<br />
4.2 System Constraints<br />
In practices, all processes have their constraints, such as input constraints, output<br />
constraints, incremental input constraints and state-space constraints. These constraints<br />
can exist in form physical constraints, like actuator limit.<br />
The constraints can be expressed in form,<br />
u<br />
y<br />
x<br />
min<br />
Δu<br />
min<br />
min<br />
min<br />
≤ uˆ<br />
( k + i|<br />
k)<br />
≤ u<br />
≤ yˆ<br />
( k + i|<br />
k)<br />
≤ y<br />
≤ xˆ<br />
( k + i|<br />
k)<br />
≤ x<br />
max<br />
≤ Δuˆ<br />
( k + i|<br />
k)<br />
≤ Δu<br />
with umin, umax, ymin, ymax, xmin and xmax are minimum and maximum input, output and state<br />
constraints respectively.<br />
4. 3 Simulation<br />
Due to the necessity of MPC for a discrete model, (10) was discretized so it changed into,<br />
max<br />
max<br />
max<br />
(18)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
−5<br />
9.<br />
16×<br />
10 z − 0.<br />
0003663z<br />
+ 0.<br />
0002837<br />
P ( z)<br />
=<br />
(19)<br />
3<br />
2<br />
z − 2.<br />
992z<br />
+ 2.<br />
983z<br />
− 0.<br />
9917<br />
37<br />
2<br />
Then (19) was transformed into state-space form so the value of each parameters in (11),<br />
(12) and (13) can be found, there are:<br />
⎡2 . 992 −1.<br />
492 0.<br />
4959⎤<br />
⎢<br />
⎥<br />
• A =<br />
⎢<br />
2 0 0<br />
⎥<br />
⎢<br />
⎣ 0 1 0 ⎥<br />
⎦<br />
•<br />
B =<br />
0.<br />
⎡<br />
⎢<br />
⎢<br />
⎢<br />
⎣<br />
01563⎤<br />
⎥<br />
0<br />
⎥<br />
0 ⎥<br />
⎦<br />
• C = [ 0 . 005868 − 0.<br />
01172 0.<br />
00908]<br />
• D = [] 0<br />
The next step was to apply the parameters into MPC algorithm. MPC was then tuned to<br />
obtain the desired transient response. The tuning was limited only on weight R because<br />
manipulating Q gave no significant effect. The tuning parameters used in this simulation<br />
were,<br />
• Maximum Horizon = 10<br />
• Control Horizon = 3<br />
• Minimum Input Constraint = -10 V<br />
• Maximum Input Constraint = 10 V<br />
• Weight Q = 40 X 40 Identity matrice<br />
Weight R was the parameter tuned to find the desired response. It is a 3 X 3 diagonal<br />
matrice which valued R=0.01, R=0.1 and R=1.0 respectively. Fig.8 and Fig. 9 shows the<br />
response and control signal comparation between manual control and system with MPC<br />
included as controller.<br />
(a)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
38<br />
(b)<br />
(c)<br />
Figure 8. Comparation between manual vs system with MPC response with (a) R=0.01, (b) R=0.1 and<br />
(c) R=1.0<br />
(a)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
39<br />
(b)<br />
(c)<br />
Figure 9. MPC control signal with (a) R=0.01, (b) R=0.1 (c) R=1.0<br />
Table 2 shows the simulation result,<br />
Table 2. System performance comparation with different weighting<br />
Performance Criterion<br />
R Settling Time<br />
(Sec)<br />
Maximum<br />
Overshoot<br />
0.01 55.6 None<br />
0.1 58.1 None<br />
1 70.1 None
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
As can be seen in Fig. 8, system response with MPC was faster compare to manual system.<br />
And it can be seen from Table II that settling time for MPC (55.6 sec, 58.1 sec and 70.1<br />
sec) was faster than manual control (120 sec). As R smaller, the response became faster<br />
yet the signal control became more fluctuative, which can be seen in Fig. 9. For mechanical<br />
system like turbine, an extreme fluctuation of control signals is undesirable due to actuator<br />
limitation. Therefore an appropriate value for R should be chosen carefully. In this<br />
experiment, the appropriate R would be 0.01. The control signal was qualitatively smooth<br />
enough and there were no overshoot.<br />
5 Summary<br />
From the simulation, a SISO turbine-generator system in Saguling GBU can be modeled,<br />
although there were many simplifications during identification process. The RMSE in<br />
identification process found to be 0.89, which is good enough to decide that the model<br />
accurately represent the real process. The model obtained was good enough to be used in<br />
simulation. Settling time criterion for system with MPC controller found to be 58.1 seconds,<br />
faster than using manual controller which is 120 seconds. In addition, system with MPC<br />
controller can eliminate the overshoot.<br />
6 References<br />
[1] Anonymous, Deskripsi Indonesia Power UBP Saguling, Februari 2009,<br />
www.indonesiapower.co.id<br />
[2] Zuhal. Dasar Tenaga Listrik, Penerbit ITB, Bandung, 1991.<br />
[3] Soderstrom, Torsten and Stoica, Peter. System Identification. Marylands Avenue :<br />
Prentice Hall International (UK) Ltd, 1989.<br />
[4] Maciejowski, J. M., Predictive Control with Constraints, England: Prentice Hall, 2000.<br />
[5] Ling, K. V., "Introduction to Model Predictive Control," course notes, Bandung Institute of<br />
Technology, May 2008.<br />
[6] Manual Handbook Sistem Governor UBP Mrica, HPC 610 Water Turbine Governor, Asea<br />
Generation, 1985.<br />
40
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Performance Analysis of Finger Flexor and Finger Extensor<br />
Muscles on Wall Climbing Athletes through Electromyography<br />
Measurement, Handgrip Strength, Handgrip Endurance and<br />
Lactate Acid<br />
H. Susanti * , Suprijanto * , F. Idealistina * , and T. Apriantono #<br />
∗ Medical Instrumentation Laboratory and Instrumentation and Control Research Division<br />
Bandung Institute of Technology - Ganesha 10, Bandung 40132, Indonesia<br />
E-mail: supri@tf.itb.ac.id, hesty133@students.itb.ac.id<br />
# Sport Science Research Division, School of Pharmacy<br />
Bandung Institute of Technology<br />
Ganesha 10, Bandung 40132, Indonesia<br />
Abstract<br />
One of activities involved gripping activity is wall climbing sport. Physiologically, gripping activity<br />
involves finger flexor and finger extensor muscles at the lower arm. Naturally, muscles performance<br />
will decrease if muscles are given static or dynamic weight during certain time. This performance<br />
decrease is related with fatigue condition, which is also related with lactate acid accumulation,<br />
insufficient of metabolic reserve, and the decrease of neural activity in stimulating contraction[2].<br />
Muscle fatigue characteristic experienced by participants were observed from the changing trend of<br />
Power Spectral Density (PSD) and median frequency of EMG signals, lactate acid level, and changing<br />
of handgrip strength and handgrip endurance values from four groups of measurement data.<br />
The experiment was involving two professional wall climbing athletes as participants. The<br />
measurements were performed four times, which were before and soon after climbing, after first 15<br />
minutes and after the second 15 minutes active recovery process, except for lactate acid level,<br />
measurements were performed twice, which were before and soon after climb activity. The athlete<br />
would have been ordered to climb the boulder for 15 minutes with a certain difficulty level.<br />
Muscle performance measurement involved in this gripping activity was very important to know how<br />
far the influence of difference of subject and period of active recovery to recover muscle condition. By<br />
this knowledge, a couch or an athlete can arrange an effective training strategy to reach a maximum<br />
achievement.<br />
Keywords: Electromyography, handgrip endurance, handgrip strength, Power Spectral Density,<br />
median frequency, finger flexor and finger extensor, lactate acid, fatigue<br />
1 Introduction<br />
Naturally, muscles performance will decrease if muscles are given static or dynamic weight<br />
during certain time. This performance decrease is related with fatigue condition, which is<br />
also related with lactate acid accumulation, insufficient of metabolic reserve, and the<br />
decrease of neural activity in stimulating contraction[2].<br />
41
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Muscle fatigue can be recovered by resting during certain time, which enable oxygen reavailable<br />
and lactate acid are transported by blood to the heart to retransform become<br />
pyruvic acid, and glucose are released for muscle glycogen supply[2].<br />
When performed gripping activity, such as wall climbing sport, fatigue condition were<br />
observed from the descent of handgrip endurance, while handgrip endurance was a longest<br />
time to endure 70% from handgrip strength value. Handgrip strength was measured from<br />
the maximum value of three times right hand grip forces.<br />
Another parameter could describe fatigue condition was the changing of electrical activity<br />
from recording of EMG signals. EMG measurements were performed by using two sets<br />
bipolar surface electrodes, in finger flexor and finger extensor muscles, when the subjects<br />
performed gripping activity using handgrip dynamometer. Then, they will be completed by<br />
doing spectral analysis (in this case were Power Spectral Density and median frequency) to<br />
quantify muscle fatigue condition.<br />
The last parameter was the blood lactate acid accumulation. The measurements were<br />
performed in the use of lactate analyzer by taking blood sample from auricle. The lactate<br />
accumulation in blood when the subject experienced fatigue was different to another, based<br />
on many factors, such as exercise factor.<br />
The objective of this paper was to correlate handgrip dynamometer measurement and<br />
blood lactate accumulation with EMG signals parameters to describe fatigue condition.<br />
Besides, it would be observed the influencing signification of subject difference factor and<br />
the collecting data session with the two ways ANOVA statistical analysis.<br />
2 Basic Theory<br />
2.1 Muscular and Nervous System<br />
Muscular system is including skeletal muscles that arranged in functional cluster, adapting<br />
to do particular moves. Skeletal muscles are moved consciously under central nervous<br />
system control. This control system is regulated by a group of electric activities in nervous<br />
system[2]. The information from whole nervous system is sent by electric signals that being<br />
produced by electrochemical reaction. Electrochemical reaction is a reaction that can<br />
produce electric current.<br />
Nervous cell (neuron) is covered by semi-permeable membrane that works selectively on<br />
passing ion. Important ions in nervous system are sodium (Na + ), calcium (Ca 2+ ), chlor (Cl - )<br />
and protein molecules with negative component. Those ions can move through semipermeable<br />
membrane thus effect electrostatic potential of nervous cell.<br />
When the nervous cell is having a rest, the potential intern cell will be more negative<br />
relatively to the outside (-70 mV). To produce a potential action, stimulus that should be<br />
given has to have intensity 15-20 mV, so it can exceed threshold level about -55 mV.<br />
2.2 Electromyography<br />
EMG signal comes from nervous electric activity when muscles contract or relax. Amplitude<br />
range of EMG signal is between 0-10 mV (peak to peak) or 0-1,5 mV (rms). While the rang<br />
42
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
frequency recorded by surface electrode is around 0-500 Hz, with dominant frequency<br />
between 50-150 Hz. EMG signal is affected by noise, they are electromagnetic radiance (50-<br />
60 Hz), motion artifact and instability because the randomness of unit motor activation<br />
phase.[3] The electric activity shape that measured in EMG (raw signals) is pictured in figure<br />
1.<br />
Figure 1. Recorded raw signal EMG<br />
2.3 Electrode<br />
Electrode used in this experiment was surface electrode Ag/AgCl. The configuration on<br />
putting bipolar electrodes is, positive and negative electrodes are put on the thick part of<br />
the muscle with gap space 1 cm, and reference electrodes put on the neutral which a bump<br />
bone relatively far from those other electrodes. Voltage that measured is the gap between<br />
positive electrode voltage to reference and negative electrode voltage to reference.<br />
2.4 EMG Signal Parameter<br />
EMG signal parameters that would be analyzed are mean Power Spectral Density (PSD) and<br />
median frequency (MF). Mean PSD is determined through classical spectral estimation<br />
technique with Fourier transform, which is periodogram Welch method. Estimation result is<br />
got from equation (1). Then, it is calculated its estimated PSD value in every experimented<br />
frequency and for every data. The estimated PSD values for every data will be calculated the<br />
average for every experimented frequency, it is what we called mean PSD. This mean PSD<br />
will relatively increase with the ascending of muscle fatigue[1].<br />
P<br />
ˆ<br />
~ (<br />
Pw<br />
( fi<br />
) P<br />
P ∑ −1<br />
1<br />
=<br />
43<br />
p=<br />
0<br />
MF is frequency that divides PSD into two parts which have equal power. MF can be<br />
abbreviated to equation (2). MF values will relatively decreasing with the ascending of<br />
muscle fatigue[4].<br />
p)<br />
xx<br />
( f)<br />
(1)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
fmed<br />
∫<br />
0<br />
S<br />
m<br />
( f)<br />
df = ∫ S<br />
44<br />
∞<br />
fmed<br />
m<br />
( f)<br />
df<br />
3 Experiment Method<br />
The experiment was involving two professional wall climbing athletes as the participants. At<br />
the beginning, subjects were asked to perform three times VC (Voluntary Contraction) with<br />
handgrip dynamometer by right hand. The maximum value from three measurements was<br />
taken as handgrip strength value. Then, handgrip endurance value would be determined<br />
from the period of time to endure gripping force, as 70 percents of handgrip strength value.<br />
When performed gripping activity, EMG signals was recorded from finger flexor and finger<br />
extensor muscles. The locations of these two muscles were shown in figure 2. Then, power<br />
spectral of these EMG signals would be analyzed by using algorithm on Matlab 7.0.<br />
Measurements of handgrip strength, handgrip endurance, and EMG were performed four<br />
times, which were which were before and soon after climbing, after first 15 minutes and<br />
after the second 15 minutes active recovery process. For lactate acid level, measurements<br />
were performed twice, which were before and soon after climb activity. The measurement<br />
was performed in the use of lactate analyzer by taking blood sample from auricle. The<br />
athlete will be ordered to climb the boulder for 15 minutes with a certain difficulty level.<br />
4 Results<br />
Figure 2. Measurements of Handgrip parameters and recording of EMG Signals<br />
The results of EMG signals spectral analysis would be compared with handgrip endurance<br />
parameter and lactate acid accumulation. In this case, handgrip endurance parameter and<br />
lactate acid accumulation are considered as reference to determine muscle fatigue level.<br />
(2)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
On subject 1, handgrip endurance value decreased after climbing and the first active<br />
recovery period, it then increased after the second active recovery period. The lactate acid<br />
accumulation increased after climbing. On finger extensor muscle, PSD value tended to<br />
increase and MF value tended to decrease when handgrip endurance measurement was<br />
performed. However, it happened vice versa on finger flexor muscle.<br />
Figure 3. Handgrip endurance and lactate acid accumulation on subject 1<br />
Figure 4. Handgrip endurance and lactate acid accumulation on subject 2<br />
Figure 5. PSD on subject 1 and its linearization<br />
45
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Figure 6. Median frequency on subject 1 and its linearization<br />
On subject 2, handgrip endurance value decreased after climbing, it then increased after<br />
the first active recovery period with the nearly equal value after the second active recovery<br />
period. The lactate acid accumulation increased after climbing, with a higher increase than<br />
subject 1. On finger flexor muscle, PSD value tended to increase and MF value tended to<br />
decrease when handgrip endurance measurement was performed. While, on finger<br />
extensor muscle, PSD value and MF value tended to decrease.<br />
Figure 7. PSD on subject 2 and its linearization<br />
Then, it was performed two ways ANOVA statistical analysis, with the source of variations,<br />
consist of data collecting session and subject. Recapitulations of the results of ANOVA<br />
analysis were shown on table 1 and 2. A variation source would be influenced significantly,<br />
if Fcrit value was less than F value. From these results, subject parameter influenced PSD<br />
significantly only on finger extensor muscle when handgrip endurance measurement was<br />
performed.<br />
46
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Figure 8. Median frequency on subject 1 and its linearization<br />
Table 1. ANOVA table (source of variation : session)<br />
Table 2. ANOVA table (source of variation : subject)<br />
47
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
5 Conclusions<br />
1. The increase of muscle fatigue was shown by tendency of PSD increase and MF<br />
decrease, correlate with handgrip endurance decrease and lactate acid increase.<br />
2. The muscles that experienced higher fatigue level when gripping were different one to<br />
another subject. Subject 1 (finger extensor), subject 2 (finger flexor).<br />
3. Subject parameter influenced PSD significantly only on finger extensor muscle when<br />
handgrip endurance measurement was performed.<br />
4. Generally, active recovery period gives positive influence to recover muscle condition,<br />
from handgrip endurance parameter.<br />
6 Reference<br />
[1] Tarata, Mihai T., “Mechanomyography versus Electromyography, in Monitoring The<br />
Muscular Fatigue”, Biomedical Engineering Online, February 2003.<br />
[2] “Sistem Muskular,” class notes, Sekolah Teknik Elektro dan Informatika, Institut<br />
Teknologi Bandung, Bandung, Indonesia, 2007.<br />
[3] De Luca, C.J., “Surface Electromyography: Detection and Recording”, DelSys<br />
Incorporated, 2002.<br />
[4] De Luca, C.J., “The Use of Surface Electromyography in Biomechanics”, Journal of<br />
Applied Biomechanics, pp. 13(2): 135-163, 1997.<br />
[5] De Luca, C.J.,”Electromyography in Encyclopedia of Medical Devices and<br />
Instrumentation (John G. Webster, Ed.)”, USA: John Wiley Publisher, 2006, pp. 98-109.<br />
[6] De Luca, Gianluca, “Fundamental Concepts in EMG Signal Acquisition”, DelSys<br />
Incorporated, 2001.<br />
[7] Susanti, Hesty, “Analisis Sinyal Respon Electromyography terhadap Stimulasi Terapi<br />
Akupuntur” Final Project, Institut Teknologi Bandung, 2008.<br />
[8] Muttaqien, Sjaikhunnas El, “Pengembangan Sistem Untuk Mengevaluasi Performansi<br />
Otot Pada Genggaman Tangan” Final Project, Institut Teknologi Bandung, Bandung,<br />
Indonesia, 2009.<br />
[9] Tjokronegoro, Harijono A.,” Analisis Spektral Digital”, Indonesia: Penerbit ITB, 2004.<br />
[10] Tjokronegoro, Harijono A.,”Pengolahan Sinyal”, Indonesia: Penerbit ITB, 2005.<br />
[11] DR. Sugiono, “Statistika untuk Penelitian”, Indonesia: Alfabeta Bandung, 2002.<br />
48
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Study on Voltage Controller of Self-Excited Induction Generator<br />
Using Controlled Shunt Capacitor, SVC Magnetic Energy Recovery<br />
Switch<br />
Abstract<br />
F.D. Wijaya, T. Isobe, R. Shimada<br />
Tokyo Institute of Technology,<br />
2-12-1 Ookayama Tokyo 152-8550 Japan<br />
Reactive compensation is required to maintain terminal voltage of induction generator under varying<br />
load and speed operation. A new variable shunt capacitor, which is called SVC magnetic energy<br />
recovery switch (SVC MERS), is proposed. The operation principle, characteristics of injected current,<br />
operating range of reactive compensation of SVC MERS in star and delta configuration were<br />
investigated. Application for induction generator voltage controller, which is required leading reactive<br />
compensator, is suitable for SVC MERS. Small scale experiments were conducted to verify the<br />
proposed system performance to control induction generator voltage in variable load and speed<br />
conditions. The advantage of this device is simple control with low switching frequency. Moreover in<br />
delta configuration, the SVC MERS current is low means downsizing of heatsink can be achieved.<br />
Keywords : Voltage controller, induction generator, reactive compensation, SVC MERS<br />
1 Introduction<br />
The global warming issue as well as the fossil fuel limitation has made human kind to do<br />
more research in the area of renewable energy sources. One of the promising research area<br />
is application of self-excited induction generator (SEIG) in micro-hydro power, wind power<br />
and diesel engine with bio-fuel. For example, in Indonesia as energy supply became a<br />
problem, the government projected 500 MW of micro-hydro to be installed, especially in<br />
rural areas to develop a green source power system [1]. Such a system would normally be<br />
operated as an isolated system supplying electricity to local un-electrified areas because it<br />
can save transmission and distribution capital investment cost. Some advantages can be<br />
achieved such as CO2 reduction and environmental awareness.<br />
SEIG consists of an ordinary three phase induction machine excited by a bank of capacitors<br />
and driven by a prime mover, such as hydro turbine, wind turbine, flywheel system or diesel<br />
engine [2]. Low cost, robustness and low maintenance need are some of the reasons to use<br />
this machine.<br />
However, there is a problem in the operation of SEIG, with poor voltage regulation in varying<br />
load conditions. Various approaches have been proposed for overcoming these problems.<br />
Availability of low cost controllable power devices, such as IGBTs, have made the application<br />
of power electronic based VAR compensation possible. Various controllable reactive power<br />
supplies exist such as TSC (Thyristor Switched Capacitor), TSC-TCR (Thyristor Controlled<br />
Reactor), STATCOM, and other variable shunt compensators [3-6]. TSC can only give a<br />
variation of capacitance in discrete steps. In transients conditions, charging and discharging<br />
of the capacitor will stress the thyristor. To avoid these problems, a combination of TCR and<br />
TSC is developed. It has a large reactor in the TCR, in order to have a large continuous<br />
control range. The latest technology is STATCOM, which uses PWM inverter as voltage source<br />
49
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
with high frequency switching to reduce harmonic and more complex control is required<br />
[4,5].<br />
In this paper, SVC MERS is proposed to control the voltage of induction generator with low<br />
switching and simple control using voltage feedback with PI controller.<br />
2 Induction Generator Characteristics<br />
To generate rated voltage of induction generator, VAR compensation is required. If the<br />
induction generator is connected to the grid, VAR can be supplied from the grid by other<br />
reactive power sources, such as synchronous generator. In isolated or stand-alone<br />
condition, capacitor is usually used. This system is called SEIG.<br />
Self excitation of the generator begins by the action of either a residual magnetism of the<br />
iron core and charge in the excitation capacitors. When the induction machine is driven by a<br />
prime mover, the residual magnetism of the iron core associated with an external capacitor<br />
that generates current by rotor movement inside of this magnetic field will produce induces<br />
voltages in the stator windings at a frequency proportional to the rotor speed. However, if<br />
there is no residual magnetism, induction generator voltage cannot be generated.<br />
A variable capacitor is required in order to realize voltage regulation of SEIG in varying load<br />
conditions or for variable speeds. From theoretical calculations [7], a range of fixed shunt<br />
capacitor sizes can be calculated to maintain rated voltage under load varying conditions.<br />
Fig. 1 shows load characteristic curve for a range of fixed capacitor sizes at synchronous<br />
speed and unity power factor load for a 200V 1.5 kW induction machine with magnetizing<br />
curve given in Eqn. (1) and induction generator parameter shown in Table 1.<br />
Lm= 0.6778Im 3 – 7.9931Im 2 + 16.231Im + 115.04 (1)<br />
The curve in Fig. 1 shows that variable compensation is needed to maintain rated voltage.<br />
Higher capacitance is required if load has low power factor as shown in Fig. 2.<br />
Voltage (Volt)<br />
200<br />
150<br />
100<br />
160uF<br />
150uF<br />
140uF<br />
130uF<br />
120uF<br />
110uF<br />
0 500 1000 1500<br />
Output Power (Watt)<br />
Figure 1. Effect of Excitation<br />
Capacitorof 1.5 kW 200 V Induction<br />
Generator.<br />
50<br />
Excitation capacitor ( μF)<br />
300<br />
240<br />
180<br />
120<br />
0 500 1000 1500<br />
Output power (W)<br />
pf=0.8<br />
pf=0.9<br />
pf=1.0<br />
Figure 2. Required excitation capacitor of<br />
SEIG for Inductive Load at Constant<br />
Voltage and Speed.
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Table 1. Parameters of induction generator<br />
Induction generator 1.5 kW, 200V, 6.8 A,<br />
50 Hz<br />
- Stator resistance, Rs 1.337 Ω<br />
- Rotor resistance, Rr 0.713 Ω<br />
- Stator & rotor<br />
inductance, Ls, Lr<br />
3.85 mH<br />
3 SVC Magnetic Energy Recovery Switch<br />
3.1 Star Configuration<br />
The configuration of this device is based on 4 IGBTs and a dc capacitor per phase. It is<br />
called magnetic energy recovery switch (MERS), and typically inserted in series between AC<br />
source and load, as series reactive compensation applied for power factor correction and<br />
power flow control [7,8,9]. In this paper, this device is used as shunt reactive compensation<br />
as shown in Fig.3. MERS is connected with an inductor in series as a filter to reduce the<br />
harmonic current flowing in the system and then it is called SVC MERS.<br />
3.1.1 Operational Principle<br />
Figure 3. Configuration of SVC MERS in Star Connection<br />
Operational states to control the injected current are shown in Fig.4. Two IGBTs are turned<br />
on and off in pairs one time each cycle of the ac power source (50 Hz) and controlled<br />
synchronously. In a half cycle, two switches (S1 and S3) are turned on, the current flowing<br />
is charging and discharging the dc capacitor with the same polarity. When the dc capacitor<br />
voltage is equal to zero, the current is flowing in parallel. The other half cycle, the other pair<br />
(S2 and S4) is turned on, with similar conditions, but with the opposite current flow direction<br />
The waveforms of phase voltage, shunt current, dc capacitor voltage, gate signal and IGBT<br />
current are shown in Fig. 5. It can be seen that the IGBT always turn on at zero current and<br />
turn off at zero voltage, therefore the low switching losses can be achieved. By controlling<br />
the switches as describe above, three different control can be achieved, which are balance<br />
mode when Xc = Xmers, dc-offset mode when Xmers > Xc and discontinuous mode when Xmers<br />
< Xc, where Xc is MERS capacitance and Xmers is variable capacitance. In the dc-offset<br />
mode, the IGBT will turn off at non zero voltage. This is because a small voltage still remain<br />
in the dc capacitor. For three phase systems there will be one SVC MERS per phase.<br />
51
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Source voltage (pu)<br />
MERS current (pu)<br />
MERS voltage (pu)<br />
Figure 4. The operational state condition of SVC MERS<br />
Gate 2, 4<br />
1<br />
0<br />
−1<br />
20<br />
0<br />
−20<br />
4<br />
0<br />
−4<br />
1<br />
dc−offset<br />
0<br />
0 0.01<br />
Time (s)<br />
0.02<br />
Figure 5. Typical waveforms of SVC MERS balance mode, discontinuous mode and dc-offset<br />
mode<br />
52<br />
balance<br />
discontinuous<br />
δ 0<br />
δ 1<br />
δ 2
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Variable reactive compensation can be achieved by controlling the current flowing to the dc<br />
capacitor by applying appropriate gate signals. The control is based on performing a phase<br />
shift of the gate signals. The control variable called δ phase, which is the phase difference<br />
between phase voltage (Vln) and the time of switching. The value of δ phase depends on how<br />
much reactive power must be supplied to the induction generator and the load.<br />
From other point of view, as illustrated in Fig. 6, SVC MERS is a capacitor controlled by<br />
semiconductor devices. The reactive/shunt current Isvc mers and the reactive power Qsvc mers,<br />
can be represented as follows:<br />
⎛ ⎞<br />
⎜ Vin<br />
I ⎟<br />
svc mers =<br />
(2)<br />
⎜ ⎟<br />
⎝<br />
X svc mers ⎠<br />
⎛ 2 ⎞<br />
⎜ Vin<br />
Q = ⎟<br />
svc mers ⎜ ⎟<br />
⎝<br />
X svc mers ⎠<br />
(3)<br />
X = X (δ ) − X<br />
(4)<br />
svc mers<br />
53<br />
( )<br />
mers<br />
Figure 6. Equivalent circuit of SVC MERS<br />
3.1.2 Control System<br />
In order to control the terminal voltage, voltage feedback with PI control is proposed as<br />
shown in Fig. 7. The control part starts with sensing the line to line voltage. Only two voltage<br />
sensors are used, which are fed to the control board. Phase lock loop (PLL) technique is<br />
applied to synchronize the gate switching time to the phase of the line voltage. However,<br />
zero detection can also be used to make this synchronization. However zero crossing<br />
detection of voltage can also be used for synchronization.<br />
For feedback control, the rms value of the line voltage is compared to the reference voltage.<br />
The error is given to the PI controller to determine the δ phase, and then it is fed to the gate<br />
controller to generate the gate signals. A δ phase limiter is to keep δ phase in the operating<br />
area.<br />
L
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Figure 7. Control system of SVC MERS<br />
3.1.3 Characteristics of Injected Current<br />
The characteristic of the injected current to the system is determined by the size of the<br />
capacitor and inductor. The selection of the operating range can be based on Fig.1. The<br />
minimum injected current should be equal to the magnetizing current of the induction<br />
generator to generate rated voltage at no load condition.<br />
Fig.8 shows the relationship of the injected current to δ phase and the relative reactance.<br />
The relative reactance is the ratio of the equivalent reactance Xsvc mers to actual reactance<br />
XC-XL .<br />
Relative reactance<br />
1.1<br />
1<br />
0.9<br />
0.8<br />
dc−offset<br />
X eq<br />
discontinuous<br />
balance<br />
−10 0 10 20<br />
δ (deg)<br />
I injected<br />
Figure 8. Characteristics of the injected<br />
current with 110uF and 10mH<br />
1.1<br />
1<br />
0.9<br />
0.8<br />
Current (pu)<br />
54<br />
Relative reactance<br />
Noload Fulload<br />
4<br />
3<br />
2<br />
1<br />
0<br />
balance<br />
X eq<br />
Over load for resistive load<br />
Resistive load Inductive load at 1 pu load (pf low)<br />
Low speed operation<br />
0 30 60 90<br />
δ (deg)<br />
I injected<br />
4<br />
3<br />
2<br />
1<br />
0<br />
Current (pu)<br />
Figure 9. Operating range area of SVC MERS for<br />
SEIG at rated voltage<br />
The operating range availability of SVC MERS to compensate reactive power of induction<br />
generator is shown in Fig.9. In the induction generation operation point, more reactive<br />
power is required to supply in inductive load or low speed operation
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
The injected current contains some harmonics; therefore inductor must be inserted as a<br />
filter. The relationship between the inductance and the harmonic of the injected current<br />
for various operating points is shown in Fig.10. Three combinations of capacitor and<br />
inductor were simulated. It can be seen that higher inductance will reduce the harmonic<br />
of the injected current; on the other hand the capacitance can be reduced.<br />
THD current (%)<br />
8<br />
6<br />
4<br />
2<br />
not−continuous dc−offset<br />
C=110uF; L=10mH<br />
C=105uF; L=15mH<br />
C=100uF; L=20mH<br />
0<br />
0.8 0.9 1 1.1<br />
Relative reactance<br />
Figure 10. Harmonic of the injected current<br />
3.1.4 Steady state and transient characteristics<br />
The experimental data results in presented in pu which base voltage is 200 V, base current<br />
6.8 A and base speed is 1500 rpm. Fig 11 and 12 show steady state voltage and current<br />
characteristics of the system.At no load condition, SVC MERS is supplied reactive current at<br />
about 0.56 pu in order to generate rated voltage. The reactive current represented by shunt<br />
current increased as load increased and terminal voltage always maintaned constant in<br />
load varying conditions.<br />
Output power (pu)<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
Output power<br />
Terminal voltage<br />
0<br />
−10 0<br />
Phase shift angle δ (deg )<br />
10<br />
55<br />
Balance point<br />
Figure 11. Steady state voltage characteristics at load varying conditions<br />
1<br />
0.5<br />
0<br />
Voltage (pu)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Current (pu)<br />
1<br />
0.5<br />
Stator<br />
Load<br />
0<br />
0 0.2 0.4 0.6<br />
Output power (pu)<br />
56<br />
Shunt<br />
Figure 12. Steady state current characteristics<br />
Fig. 13 shows the experimental transient response of the voltage, load current, shunt<br />
current, and dc capacitor voltage with a step change from no load to full resistive load<br />
condition. The voltage is recovered within two cycles. Operation changed from dc-offset<br />
mode to discontinuous mode in order to supply required reactive power.<br />
Terminal voltage (V)<br />
Current (A)<br />
DC Capacitor voltage (V)<br />
200<br />
0<br />
−200<br />
10<br />
0<br />
−10<br />
200<br />
100<br />
0<br />
Shunt<br />
0 time (s) 0.1<br />
Load<br />
Figure 13. Transient characteristics of the system
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
Variable speed condition was also experimented. The results is shown in Fig. 14. In this<br />
experiment, the rotor speed is changed from 0.8 pu to 1.3 pu, while the induction generator<br />
output power was setting at half of its rated power (750 W). It can be found that SVC MERS<br />
can control the voltage to its rated voltage keeping the induction generator to generate<br />
output power.<br />
For higher speed, the value of phase shift angle δ is small, meaning low reactive power<br />
is generated by SVC MERS. While for lower speed, phase shift angle δ is larger, meaning<br />
higher reactive power is required to maintain its rated voltage.<br />
Speed (pu)<br />
1.2<br />
1<br />
0.8<br />
Terminal voltage<br />
Speed<br />
0 16 32<br />
0<br />
48<br />
Phase shift angle δ (deg)<br />
57<br />
1<br />
0.5<br />
Figure 14. Variable speed characteristic using SVC MERS<br />
3.2 Delta Configuration<br />
In order to reduce the current rating of the switch, delta connection is configured for this<br />
application, however capacity rating will similar. By doing this, the capacitance can be<br />
reduced to 37 µF (1/3 of star). The control of the switch for MERS is the same as star<br />
configuration. Fig. 15 shows the waveforms of terminal voltage, shunt current, MERS<br />
current, stator current, IGBT and gate signal of this configuration at 0.7 pu load. In half<br />
cycle capacitor will charge (a), dis-charge (b) and have parallel path (c). It can be found<br />
that MERS current is smaller than shunt current. As a result IGBT losses will be lower,<br />
therefore downsizing of heat sink can be achieved.<br />
4 Conclusion<br />
The operation principle, characteristics of injected current, operating range of reactive<br />
compensation of SVC MERS in star and delta configuration was investigated. Application for<br />
induction generator voltage control, which is required leading reactive compensator, is<br />
suitable for the proposed system. Experimental result confirmed that the proposed system<br />
can controlled induction generator in load and speed varying conditions. This proposed<br />
system has the following advantages: i) simple control, where only two voltage sensors are<br />
required and voltage feedback control with PI controller gives a good response, ii) low<br />
Voltage (pu)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
switching frequency, which zero current turn on and zero voltage turn off and resulting in<br />
low switching losses.<br />
5 References<br />
Terminal voltage(V)<br />
Current (A)<br />
Capacitor voltage (V)<br />
Current (A)<br />
200<br />
0<br />
−200<br />
10<br />
0<br />
−10<br />
300<br />
200<br />
100<br />
0<br />
4<br />
0<br />
−4<br />
(a) (b)<br />
MERS<br />
0 0.01 0.02<br />
Time (s)<br />
0.03<br />
58<br />
(c)<br />
shunt<br />
Gate S2−S4<br />
stator<br />
Figure 15. Waveforms of SVC MERS in delta configuration<br />
[1] UNDP Report:<br />
"Connecting micro-hydro power Indonesia to the national grid", UNDP Report (2003)<br />
[2] E.D. Basset, F.M. Potter, “Capacitive excitation for induction generator”,AIEE Trans.<br />
pp.540–73., May 1935.<br />
[3] T. Ahmed, O. Noro, E. Hiraki, M. Nakaoka, “Terminal voltage regulation characteristics<br />
by static var compensator for a three-phase self-excited induction generator”, IEEE<br />
Trans. on IEEE Industry Applications”, Vol. 40, No. 4, July 2004.<br />
[4] Mustafa, A. Al-Saffar, Eui-Cheol N., T. A. Lipo, “Controlled shunt capacitor self-excited<br />
induction generator,” Proc.33-rd IAS IEEE annual meetings USA, 1989.<br />
[5] R. Leidhold, G. Garcia, M. I. Valla, “Induction generator controller based on the<br />
instantaneous reactive power theory”, IEEE Trans. on Energy Conversion, vol. 17, no. 3.<br />
pp. 368-373 September 2002.<br />
4<br />
0<br />
−4<br />
Gate voltage (V)
J.Oto.Ktrl.Inst (J.Auto.Ctrl.Inst) Vol 1 (2), 2009 ISSN : 2085-2517<br />
[6] M. Naidu and J.Walters, “A 4-kW 42-V induction-machine-based automotive power<br />
generation system with a diode bridge rectifier and a PWM inverter,” IEEE Trans. Ind.<br />
Applicat., vol. 39, pp. 1287–1293, September 2003.<br />
[7] T.F. Chan, “Analysis of self-excited induction generators using an iterative method”,<br />
IEEE Trans. Energy Conversion. vol. 10, pp. 502-507, September 1995.<br />
[8] T. Isobe, J.A. Wiik, F. Danang Wijaya, K. Inoue, K. Usuki, T. Kitahara, R. Shimada,”<br />
Improved performance of induction motor using magnetic energy recovery switch”,<br />
presented at the 4th PCC Nagoya, May 2007.<br />
[9] J.A. Wiik, F. Danang Wijaya, R. Shimada,” An innovative series connected power flow<br />
controller, Magnetic Energy Recovery Switch (MERS)”, presented at IEEE PES General<br />
Meeting USA July 2007.<br />
[10] J.A. Wiik, T. Isobe, T. Takaku, T. Kitahara, R. Shimada, “Reactive power compensation by<br />
using series connected current phase control switches”, presented at PCIM China,<br />
2006.<br />
59
Jurnal Otomasi, Kontrol & Instrumentasi<br />
Informasi bagi Penulis Naskah<br />
Penerimaan Naskah<br />
Naskah yang diterima berasal dari civitas academica baik dari Institut Teknologi Bandung maupun<br />
dari luar ITB, dan dari kalangan industri yang berhubungan dalam bidang otomasi, kontrol, dan<br />
instrumentsi. Naskah yang dikirim tidak pernah dipublikasikan pada jurnal atau majalah yang lain<br />
sebelumnya. Bahasa yang dipakai dalam penulisan adalah bahasa Indonesia atau bahasa Inggris.<br />
Naskah akan diuji kelayakannya melalui sistem refereeing oleh para penilai berdasarkan keaslian<br />
(orisinalitas), relevansi, dan validitas ilmiah.<br />
Penulisan Naskah<br />
Format penulisan dan jenis font mengikuti petunjuk berikut ini :<br />
Mirror margin, margin dalam 2.5 cm, margin luar 2 cm, margin atas dan bawah 2.54 cm.<br />
Ukuran judul 14 pt (bold), nama penulis 10 pt, alamat institusi 10 pt.<br />
Abstrak dan kata kunci 9 pt, judul bab / sub bab 12 / 10 pt, body text 10 pt.<br />
Judul gambar/tabel 9 pt (bold), ukuran kertas : B5 (JIS), jenis huruf : Franklin Gothic Book.<br />
Contoh penulisan judul dan penulis :<br />
A Measurement-Based Form of the Out-of-Place Quantum Carry<br />
Lookhead Adder<br />
A. Trisetyarso1) , R. Van Meter1) , K. M. Itoh2) 1) Department of Applied Physics and Physico-Informatics, KeioUniversity, Japan<br />
2) Yagami Campus, 3-14-1Hiyoshi, Kohoku-ku,Yokohama-shi, Kanagawa-ken 223-8522,<br />
Japan<br />
Panjang naskah termasuk tabel dan gambar maksimum 20 halaman. Naskah diterima oleh redaksi<br />
dalam bentuk softcopy dengan format dokumen. Pencantuman gambar diusahakan sejelas mungkin<br />
dan dalam warna hitam putih.<br />
Abstrak<br />
Penulisan abstrak untuk naskah ditulis dalam bahasa Indonesia atau bahasa Inggris disesuaikan<br />
dengan isi naskah. Panjang abstrak tidak lebih dari 200 kata dan dalam satu paragraf. Kata kunci<br />
dicantumkan setelah abstrak dan ditulis dengan huruf miring (italic). Jumlahnya tidak lebih dari 10<br />
kata yang ditulis berurutan, dengan tanda koma sebagai pemisah antar kata.<br />
Rujukan<br />
Penulisan rujukan dalam naskah didasarkan pada urutan pemunculannya dalam isi naskah, seperti<br />
contoh di bawah ini:<br />
[1] J.F. MacGregor, D.T. Fogal, Judul jurnal, Nama jurnal, 5 (3) (1995) hal 163-171.<br />
[2] J.F. MacGregor, D.T. Fogal, Judul buku, ed.2, Penerbit (1997), hal 5-10.<br />
[3] Nama penulis (jika ada), Judul makalah (jika ada), Organisasi, alamat situs web, (12 Juni 1998).<br />
Kepada setiap penulis naskah yang dimuat dalam jurnal akan mendapat satu reprint jurnal secara<br />
cuma-cuma. Untuk pengirim yang berdomisili di luar Bandung akan dikirim dalam bentuk softcopy.
Jurnal Otomasi, Kontrol &<br />
Instrumentasi<br />
Journal of Automation, Control and<br />
Instrumentation<br />
Volume 1, No.2, Tahun 2009<br />
1. Development of Circularly Polarized Synthetic Aperture Radar Sensor Mounted<br />
on Unmanned Aerial Vehicle<br />
M. Baharuddin, P.R. Akbar, J.T.S. Sumantyo, H. Kuze<br />
2. Electric Traction Motor Drive Modelling for Electric Karting Application Using<br />
Matlab / Simulink Software<br />
D. Istardi<br />
3. Feasibility Study of Solar Power Massive Usage in Indonesia : Yield versus Cost<br />
Effective<br />
M.A. Setiawan<br />
4. Modelling and Designing The Model Predictive Control System of Turbine<br />
Angular Speed at Hydropowerplant UBP Saguling PT Indonesia Power<br />
R.K.A. Kusumah, E. Joelianto, E. Ekawati<br />
5. Performance Analysis of Finger Flexor and Finger Extensor Muscles on Wall<br />
Climbing Athletes trough Electromyography Measurement, Handgrip Strength,<br />
Handgrip Endurance and Lactate Acid<br />
H. Susanti, Suprijanto, F. Idealistina, T. Apriantono<br />
6. Study on Voltage Controller of Self-Excited Induction Generator Using Controlled<br />
Shunt Capacitor, SVC Magnetic Energy Recovery Switch<br />
F.D. Wijaya, T. Isobe, R. Shimada