Article history: Received Oct 24, 2016 Revised Feb 5, 2017 Accepted Feb 19, 2017

Transcription

1 International Journal of Electrical and Coputer Engineering (IJECE) Vol. 7, No., April 017, pp. 610~618 ISSN: , DOI: /ijece.v7i.pp Optiizing Tri-Core Peranent-Magnet-Linear-Generator Direct-Drive Wave-Energy-Conversion Syste Design for Sea Wave Characteristics in South Coast Yogyakarta Fransisco Danang Wijaya, Sarjiya, Muhaad Rifa i Putra Sugita Departeent of Electrical Engineering and Inforation Technology, Faculty of Engineering Universitas Gadjah Mada, nd Grafika Street, Mlati, Slean 5581, Yogyakarta, Indonesia Article Info Article history: Received Oct 4, 016 Revised Feb 5, 017 Accepted Feb 19, 017 Keyword: Ocean waves Optiized design Overall weight Power losses Tri-core PMLG Wave-energy-conversion ABSTRACT According to statistical data, the south coast Yogyakarta has significant ocean wave height which can be used to generate electricity by using waveenergy-converter syste. One of the siplest way to convert wave energy to electricity is using direct-drive wave-energy-conversion (WEC) syste with peranent-agnet-linear-generator (PMLG). This ethod is siple because it does not need to convert linear otion to rotational otion. However, PMLG has large electric power losses, has great weight in both of the stator and rotor, and expensive to ake. In this paper, a tri-core PMLG was designed. The electric power losses in the winding, translator weight, and aterial cost were ideally iniized using ultiobjective optiization cobined with siulated annealing (SA) algorith. Then, the design was verified using finite eleent analysis. The optiized design of this PMLG was siulated using sinusoidal ocean waves which usually occur in the south coast of Yogyakarta to analyze the perforance of this linear generator. Siulation result has been shown that this generator can generate 911 watt peak output power at the rated condition and at the optiu load with 81.14% efficiency. This confirs that the optiized design of PMLG is suitable for direct-drive WEC with low power losses and aterial cost. Copyright 017 Institute of Advanced Engineering and Science. All rights reserved. Corresponding Author: Muhaad Rifa i Putra Sugita, Departeent of Electrical Engineering and Inforation Technology, Faculty of Engineering, Universitas Gadjah Mada, nd Grafika Street, Mlati, Slean, Yogyakarta, Indonesia Eail: uhaad.rifai.p@ail.ug.ac.id 1. INTRODUCTION One of the world s renewable energy resources which available in a great aount is ocean waves energy. It is estiated that the total ocean waves energy in the world can be used to fulfill % fro the world s total energy deand [1]. According to [], south coast of Yogyakarta has potential to generate electric power about 1.9 MW. This is also supported based on the apping result which conducts by Indonesian Agency for Meteorology Cliatology and Geophysics that south coast of Yogyakarta has average ocean waves between eters height [3]. It eans that south coast of Yogyakarta has vertical velocity of the ocean waves about 0.5 /s [4]. This velocity is able to ake otion which can drive a generator. There are several technology of wave-energy-conversion (WEC) syste.one of the siple way to convert the ocean waves to electricity is using direct-drive WEC syste technology [5]. Direct-drive WEC syste is siple because it doesnot require to convert linear otion of the ocean waves to rotational otion to generate electricity, so the efficiency of the WEC is higher [6]. It happens because direct-drive WEC is using floater (buoy) coupled with linear generator to convert ocean waves otion into electricity, so that it is Journal hoepage:

2 IJECE ISSN: only needs a few part to develop the WEC. Linear generator which is used to convert the linear otion of the ocean waves is classified based on the shape, that is: tubular, tri-core, square-core and tri-coil linear generator [7], [8]. Tri-core linear generator is cheaper, provide higher ef, and suitable with ocean waves between 0.5 until 1.5 eters height which is siilar to the characteristic of ocean waves in south coast of Yogyakarta rather than the others linear generator topologies. To obtain a good design of linear generator, an optiization procedure should be done to get global optial solutions [9], [10]. Based on the literature review, there are still only a few research which related to the optiization of linear generator design, although the optiization procedure has been succesfully used to optiize several electric achine, such as transforer, direct-current achine, synchronous achine, and the others electric achine [11-14]. Result was shown that optiization procedure could deliver better electric achine design rather than without optiization procedure. This research proposes direct-drive WEC applied with tri-core peranent-agnet-linear-generator (PMLG) for converting ocean waves otion to electricity.to get a good design of PMLG, the optiization was perfored in this linear generator. Three objectives function was used in this optiization, there are: electric power losses in the winding, translator weight, and aterial cost of the PMLG. That three objectives function then iniized using siulated annealing (SA) algorith to get the best design. Then, the overall design of the PMLG was verivied using finite eleent analysis (FEA) to get soe paraeters of the linear generator which could not be calculated analytically and to confired the agnetic paraeter of the PMLG. Finally, the optiized design of tri-core PMLG was siulated using sinusoidal ocean waves characteristic to analyze and investigate the perforance of the linear generator if the PMLG installed in south coast of Yogyakarta.. RESEARCH METHOD.1. Noenclature R w internal resistance of AWG 11 (oh/), P dt expected power of PMLG (watt), nuber of phase, μ agnetic pereability,m s nuber of arature, A c area of the coil ( ), K cu winding filling factor, D w wire diaeter (), C air gap flux density coefficient, K c Carter s coefficient, B g air gap flux density (T), B r perenent agnet residual flux (T), H c agnetic flux coercive... Site Selection for Installing Direct-Drive WEC As entioned above, the south coast of Java Island, especially in the south coast of Yogyakarta has large potential of ocean waves energy. According to the onthly data which has beencollected by IndonesianAgency for Meteorology Cliatology and Geophysicsfro 000 until 010, in this location, the significant wave height has average value of 1.44 eters. Fro that data, the WEC syste was designed to work in rated condition with 1 eters wave height, so that the WEC could work well throughout the year. H v ( ) s s t cos( t) (1) Using Equation (1), where angular velocity of the ocean wave (ω) was 1.76 rad/s, significant wave height (H s ) was 1, and ocean wave period was 3.55 s, so that the average vertical velocity of the ocean waves (v s ) was 0.6 /s. It was assued that the linear velocity of linear generator was the sae as vertical velocity of the ocean waves since the direct-drive WEC becae wave followers [15]..3. Matheatical Model of Tri-Core PMLG Peranent-agnet-linear-generator (PMLG) with tri-core topology is one of the linear generator type which has triangular shape core in both of the stator and translator as shown in Figure 1(a).The design paraeters of tri-core PMLG was shown in Figure 1(b). The ai of this shape was to siplify anufacturing process because of the siple shape and construction, so that it would iniizethe anufacturing cost. There are several atheatical odels of tri-core PMLG which derived fro the agnetical Equation of linear generator [16]. This atheatical odels then optiized, so a good design of tri-core PMLG was obtained. The atheatical odels are as follows: 1) Electrical current produced by PMLG (i ph ): fro the expected power and ef produced by PMLG (E ph ), the Equation to deterine the electrical current is given by P dt i ph () E ph Optiizing Tri-Core PMLG Direct-Drive WEC Syste Design for Sea Wave (Fransisco Danang Wijaya)

3 61 ISSN: ) Winding resistance (R ph ): per phases winding resistance is influenced by internal resistance and length of wire (L c ), the Equation is as follows below Rph R w L c N ph (3) (a) (b) Figure 1. (a) Tri-core PMLG structure [8]; (b) Design paraeters of tri-core PMLG [4] where N ph is nuber of turns per phase which is derived fro nuber of turns per slot N c. The Equation is shown by (4) and (5) respectively N ph N pq (4) c Ac K N cu c (5) ( Dw / ) 3) Geoetry of tri-core PMLG: the geoetry of tri-core PMLG will affect the phase resistance which also related to the electrical power losses in the winding, overall weight, and aterial cost needed to construct the linear generator. The atheatical geoetry odel of tri-core PMLG is entioned in (6)-(11) with soe of the noenclature is entioned in the next chapter. Ac h s b s (6) Lc h ( Y s s ) M s W s (7) t (5g b ) K s c t (5g b s ) b s (8) geq K c g (9) g eq B g B r h o H c ( B r Bg ) (10) IJECE Vol. 7, No., April 017 :

4 IJECE ISSN: C p (11).3. Optiization Procedure using Siulated Annealing The design optiization of tri-core PMLG was done using ultiobjectives function such as: electric power losses in the winding, translator weight, and aterial cost of the PMLG. Then, besause the use of ore than one single objectives function, so all of the objectives function ust be transfored to a single objectives function using weighted su approach. The ethod worked by using soe weighted value w 1, w, w 3,.... w n with w 1 + w + w w n = 1. Each of the objective function was ultiplied with the weighted value, the Equation for transforing the ultiobjectives function to a single objectives function was shown below F t n w F i i i 1 (1) where, w i is the weighted value of the objectives function, F i is each of single objectives function and F t was the total single objectives function. Each of the objectives function was entioned below: 1) Electrical power losses in the winding (P loss ): the electrical power losses in the winding was influenced by winding resistance, the Equation of the first objectives function was given by F P loss i ph R 1 ph (13) ) Translator weight ( tr ): translator of the tri-core PMLG consisted of peranent agnet and translator core, to iniize the translator weight, the weight of peranent agnet and translator core ust be iniized. The Equation was entioned below respectivelly F tr p tc (14) where p is peranent agnet weight and tc is the translator core weight. Each of the coponent could be calculated using Equation as entioned below p tc p Msh W slt (15) M Y L W s h g tc s r t eq (16) where ρ is the ass density of peranent agnet, ρ tc is the ass density of translator core, and L t is translator length which was designed to be 1 eters. 3) Material cost of the PMLG (c tot ): the aterial cost of tri-core PMLG consisted of: cost for both of translator and stator core, cost for peranent agnet, and cost for coper wire. Cost function could be calculated using weight of each part of the generator and the price per kilogras according to [17]. The Equation for calculating coper weight and stator core weight was entioned below cu sc ( r ) L N (17) cu wire c ph h Y L b h pq M W sc s s s s s s s (18) where ρ cu is the ass density of coper wire, ρ sc is the ass density of stator core. Then the aterial cost of tricore PMLG could be calculated using Equation (10). F c 3 tot c p p tc sc c c fe cu cu (19) Optiizing Tri-Core PMLG Direct-Drive WEC Syste Design for Sea Wave (Fransisco Danang Wijaya)

5 614 ISSN: where c p is peranent agnet cost per kilogra, c fe is steel cost per kilogra, and c cu is coper cost per kilogra. 3. RESULTS AND ANALYSIS 3.1. Optiizing Tri-Core PMLG Design It was decided to designa pico-scale tri-core PMLG with three-phase wye configuration and the expected output power is 1 kw. Several initial paraeters were used in the design process, such as: average translator speed to be 0.6 /s, the use of AWG 11 wire, and NdFeB peranent agnet to constructed the linear generator. AWG 11 wire was choosenbecause the expected power of the generator was in pico-scale, so the current that flow through the winding was not exceed the current capability of AWG 11 which was 1 A. Then, NdFeB 5 MGOe was used because this type of peranent agnet has large flux density rather than the others type, so it would generate higher electrootive force (ef). With the sae generated power, the higher ef woulddecrease the current generated fro the PMLG, so it would decrease the electrical power losses in the winding. The NdFeB peranent agnet wasounted in the translator core surface. Table 1. Specification of Tri-Core PMLG Values Variables Sybol Analytical Design Optiized Design Stator width () W s Nuber of slot per pole per phase q 1/3 1/3 Nuber of poles p 1 1 Average flux density in air gap (T) Bav Air gap () g 0.4 Flux density in the stator yoke (T) Bys 1.8. Flux density in the rotor yoke (T) Byr 1. Stator length () L s Pole-pitch () τ p Tooth-pitch () τ t Stator slot width () b s Stator tooth width () b t Real air gap () g eq.86 1 Peranent agnet thickness () h Peranent agnet length () τ Stator yoke thickness () Y s Rotor yoke thickness () Y r Slot height () h s Nuber of stator turns per slot N c Nuber of stator turns per phase N ph Average length of stator wire per turn () L c Weight of peranent agnet (kg) p Weight of core (kg) fe Weight of copper (kg) cu Cost for peranent agnet ($) C p Cost for core ($) C fe Cost for copper ($) C cu Winding resistance (oh) R ph Ef (volt rs) E ph Max coil current (A) i ph Electrical frequency (Hz) f Winding power losses (watt) P loss Efficiency (%) η Table 1 showed coparison between analytical and optiized design specification of tri-core PMLG. The optiized design of tri-core PMLG wasdelivered using siulated annealing algorith cobined with objective function which had been entioned in (13)-(19) and several atheatical Equation of PMLG IJECE Vol. 7, No., April 017 :

6 IJECE ISSN: design entioned in ()-(11). The weighting factor to optiize tri-core PMLG design had chosen to be 0. for weighting the electrical power losses objectives function, 0.7 for weighting the translator weight objectives function, and 0.1 for weighting the aterial cost objectives function.this weighting value was chosenbecause it could deliver the ost optiu design of tri-core PMLG rather than the other weighting value. As seen in Table 1, the optiized design delivered lower winding resistance copared to the analytical design, it wouldgive effect to the electrical power losses in the winding, so that the efficiency of the optiized design result of tri-core PMLG was higher copared to the analytical design result. Then, thetranslator weight delivered fro optiization process was kg, and it was less higher than the analytical result which gave 7.1 kg, it was due to the spreading of weighting value, so that all of the objectives function was in the optiu value. However, the overall weight of PMLG drawn fro optiization result gave a great different. The overall weight fro optiization result was 6.53 kg, but the analytical result gave kg. This overall weight affected the aterial cost of tri-core PMLG, the aterial cost needed by optiized design was$ 1.3, but the analytical result needed$ to build tri-core PMLG. It hass been shown that optiization procedure succesfully gives tri-core PMLG with optiu design specification. 3.. Optiized Tri-Core PMLG Design Verification using FEA The use of FEA was to analyze the two diensional agnetic phenoenon in the optiized design, including to show unknown paraeters which could not be deterined using atheatical Equation such as winding inductance. It happened because FEA can deliver ore accurate agnetical odel result due to the use of differential Equation in agnetic field odel. Fro the siulation result drawn fro FEMM software, the axiu flux in the stator yoke was 1.83 T and the axiu flux in the translator yoke was.1 T. The agnetical flux in the translator yoke was larger than the assuption, but it wasacceptable since the axiu flux only happened in a sall area, so it did not ake the translator yoke becae excessive saturation.another result delivered fro FEMM showedthat the phase inductance of tri-core PMLG was H, this value was used to siulate and analyze the perforance of tri-core PMLG in dynaic condition. Figure shows flux density distribution in optiized tri-core PMLG design Figure 3. Flux density distribution in optiized tri-core PMLG design 3.3. Electrical Characteristic of PMLG To analyze the perforance of tri-core PMLG, a Siulink/MATLAB had been build. The first was to analyze the effect of load resistance to the output power and efficiency of tri-core PMLG. Siulation was conducted by varying load resistance of tri-core PMLG fro 0 until 50 ohs with 1 oh step size, and the translator speed was set to 0.6 /s. Figure 4(a) showed the siulation result. As seen, the axiu output power was 589 watt, and it happened in 4 ohs load. The input power of the generator was 76 watt, so that since the echanical power losses and core losses was neglected, the axiu efficiency of tri-core PMLG was 81.14%. This value was closest with the efficiency calculated in optiized design of PMLG. When the optial load was connected to the tri-core PMLG, the output characteristic of PMLG showed in Figure 4(b). It showed that the current waslagging to the ef produced by tri-core PMLG, it happened because of the phase inductance which present in the PMLG winding. However, the terinal voltage was in phase with the terinal current because of the resistive load. In the optiu load condition, the phase terinal voltage typically had a great different with the phase ef due to the resistance value of the PMLG. When the load resistance was increasing, the terinal voltage would also be increasing near the ef, Optiizing Tri-Core PMLG Direct-Drive WEC Syste Design for Sea Wave (Fransisco Danang Wijaya)

7 616 ISSN: but when the load resistance was not in optial value, the output power generated by PMLG was decreasingas shown in Figure 4(a). In fact, the ocean waves never happened in a steady state condition. Usually the characteristic of the ocean waves was assue to be sinusoidal waves with a certain height and period. This condition ade the translator speed changed sinusoidally, and it gave effect to the phase ef, phase current, and phase terinal voltage. The siulation result using sinusoidal ocean waves with constant height and period was shown in Figure 5(a). As seen, the peak agnitude of the ef changed sinusoidally according to the ocean waves oveents. Since the vertical ocean wave speed changed sinusoidally, the input and output power of tri-core PMLG was represented in Figure 5(b). The peak input power of tri-core PMLG was 1,1 watt and the peak output power of tri-core PMLG was911 watt. This value was near the expected value of initial design paraeters of the generator. The input and output power of tri-core PMLG increased when the vertical ocean waves speed was increasing due to the increase of wave height. The effect of ocean wave height to the output power of tri-core PMLG was shown in Table. It has been shown that the generated output power of PMLG is interittent since the ocean waves height is changed periodically. (a) (b) Figure 4. (a) Effect of load resistance to the output power and efficiency of tri-core PMLG; (b) Output characteristic of tri-core PMLG in steady state condition (a) (b) Figure 5. (a) Output characteristic of tri-core PMLG in sinusoidal ocean waves; (b) Input and output power of tri-core PMLG IJECE Vol. 7, No., April 017 :

8 IJECE ISSN: Table. Effect of ocean wave height to the output of PMLG Wave height () Peak output power (watt) Peak phase ef (V) Peak phase terinal volatge (V) Peak phase current (A) Frequency (Hz) Peak power losses (W) Efficiency (%) CONCLUSION A sall scale of wave-energy-conversion syste has been designed and optiized using siulated annealing algorith. Fro the optiization process, a good design of tri-core PMLG which has low electrical power losses in the winding, low translator weight, and low aterial cost has been obtained. The optiized design of tri core PMLG can generate 911 watt peak output power at the rated condition and at the optiu load with 81.14% efficiency. It has been shown since the translator is light in weight, the buoy will easily follows the ocean wave otion, it also happens in translator because the buoy and the translator is coupled. Furtherore, this condition will ake the generator ore easily to produce electrical power, rather than the PMLG which has heavy translator. Then, since the south coast of Yogyakarta is wavy and support the foration of waves between 0.5 to 1.5 eters height, the optiized design of PMLG is suitable for direct-drive WEC with low power losses and cost to install. REFERENCES [1] J Faiz and M Ebrahii-Salari, "Design and Siulation of a 50 kw Linear Peranent Magnet Generator for Wave Energy to Electric Energy Conversion in Caspian Sea", Sustainable Power Generation and Supply, 009. SUPERGEN '09. International Conference on, pp. 1-6, April 009. [] Raha S, "The potential study of ocean wave power plant using oscillating water colun syste in thirty Indonesian aritie territory", Universitas Indonesia, Jakarta, Thesis Report 010. [3] Susilohadi, "Mapping of Ocean Energy in Indonesia", 013, pp [4] Muhaad R.P.S, F.D Wijaya, and Sarjiya, "Design and Analysis of Tri Core Peranent Magnet Linear Generator for Wave Energy Conversion in South Coas of Java Island", 6th International Annual Engineering Seinar, pp. 1-6, August 016. [5] Hosna Titah Benbouzid and Mohaed Benbouzid, "Ocean wave energy extraction: Up-to-date technologies review and evaluation", 014 International Power Electronics and Application Conference and Exposition (PEAC), Noveber 014. [6] Iraide Lópeza, Jon Andreua, Salvador Ceballosb, Iñigo Martínez de Alegríaa, and Iñigo Kortabarriaa, "Review of wave energy technologies and the necessary power-equipent", Renewable and Sustainable Energy Reviews, vol. 7, pp , 013. [7] M.A.F.M. Hai, T Ibrahi, and N.M. Nor, "Design of Portable Pico Linear Peranent Magnet Generator for Wave Energy Conversion", Inforation Technology and Electrical Engineering (ICITEE), pp. 1-4, 014. [8] A.H. Meon, T. Bin Ibrahi, and N. Perual, "Portable and pico-scale linear generator for wave energy conversion", Intelligent and Advanced Systes (ICIAS), 014 5th International Conference on, pp. 1-4, 014. [9] Yuxin Sun, Qinghua Wu, and Xuesong Yan, "An Iproved Constrained Engineering Optiization Design Algorith", TELKOMNIKA Indonesian Journal of Electrical Engineering, vol. 1, no. 11, pp , 014. [10] Qinghua Wu and et al, "Research of Function Optiization Algorith", TELKOMNIKA Indonesian Journal of Electrical Engineering, vol. 10, no. 4, pp , 01. [11] N Bianchi and S. Bolognani, "Design optiisation of electric otors by genetic algoriths", IEE Proceedings - Electric Power Applications, vol. 145, no. 5, pp , Septeber [1] Ait Kuar Yadav and et al, "Design Optiization of High-Frequency Power Transforer by Genetic Algorith and Siulated Annealing", International Journal of Electrical and Coputer Engineering (IJECE), vol. 1, no., pp , 011. [13] Reza Ilka and S. Asghar Gholaian, "Optiu Design of a Five-Phase Peranent Magnet Synchronous Motor for Underwater Vehicles by use of Particle Swar Optiization", TELKOMNIKA Indonesian Journal of Electrical Engineering, vol. 10, no. 5, pp , 01. [14] Hany M. Hasanien, "Particle Swar Design Optiization of Transverse Flux Linear Motor for Weight Reduction and Iproveent of Thrust Force", IEEE Transactions on Industrial Electronics, vol. 58, no. 9, pp , Septeber 011. [15] Ean A Aon, Ted K. A. Brekken, and Alphonse A. Schacher, "Maxiu Power Point Tracking for Ocean Wave Energy Conversion", IEEE Transactions on Industry Applications, vol. 48, no. 3, pp , March 01. [16] Rajkuar Parthasarathy, "Linear PM Generator For Wave Energy Conversion", Thesis in Louisiana State University and Agricultural and Mechanical College, 01. Optiizing Tri-Core PMLG Direct-Drive WEC Syste Design for Sea Wave (Fransisco Danang Wijaya)

9 618 ISSN: [17] Ozan Keysan and et al, "A direct drive peranent agnet generator design for a tidal current turbine (SeaGen)", 011 IEEE International Electric Machines & Drives Conference (IEMDC), pp. 4-9, May 011. [18] Kondapalli Siva R.R and Azrul Hisha B.O, "Design Optiization of A BLDC Motor by Genetic Algorith and Siulated Annealing", International Conference on Intelligent and Advanced Systes, pp , 007. BIOGRAPHIES OF AUTHORS Fransisco Danang Wijaya was born on February 1974 in Slean, Indonesia. He received his Bachelor and Master degree, both fro Electrical Engineering Major, Gadjah Mada University in 1997 and 001 respectively. He then got his Doctor of Engineering in Tokyo Institute of Technology in 009. He is currently Associate Professor at the Departent of Electrical Engineering and Inforation Technology, Gadjah Mada University. His research is specialized in power syste engineering, energy conversion, also transission and distribution syste. He is also expert in power syste control technique using Magnetic Energy Recovery Switch (MERS). Sarjiya received the bachelor and aster degrees fro Gadjah Mada University, Yogyakarta, Indonesia, in 1998 and 001, respectively, and the Ph.D. degree fro the Chulalongkorn University, Thailand, in 008, all in electrical engineering. Presently, he joint with the Electrical Engineering and Inforation Technology Departent at Gadjah Mada University. His research interests include reliability evaluation of electric power systes, econoic and secure operation of electric power systes, power systes and energy planning, and renewable energy integration. Muhaad Rifa i Putra Sugita was born in Bantul, special district of Yogyakarta in He received Bachelor Engineering degree in Departent of Electrical Engineering and Inforation Technology, Faculty of Engineering, Universitas Gadjah Mada in 016. He is currently an M.Eng. students in the sae departent. His research interest include renewable energy fro ocean waves, hybrid energy, optiization technique, dissolved gas analysis in power transforer, distributed generation, and sart grid application. IJECE Vol. 7, No., April 017 :