Austalasian Univesities Powe Engineeing onfeence (AUPE 24) 26-29 Septembe 24, Bisbane, Austalia SELF EXITED INDUTION GENERATORS FOR BRAKE VAN APPLIATIONS Abstact Dawit Seyoum* and Pete Wolfs* *ente fo Railway Engineeing Faculty of Engineeing and Physical Systems ental Queensland Univesity The suga cane industy uses bake vans, coupled to the end of cane tains, to poduce a given constant baking foce to impove the cane bin ide dynamics and to assist the baking of tains. Unlike othe ailway vehicles, cane bins often have neithe bakes o suspensions. uently bake vans opeate using compessed ai, supplied by an on boad compesso, which activates a bake callipe, which clamps a ventilated disc oto on each of the fou wheel sets. This system needs maintenance due to wea on the bake pads and otos. An electical baking implemented using self-excited induction geneato is poposed. With the application of electical baking enegy is dissipated is in well ventilated esistos. Maintenance is futhe educed by a bushless design. 1. INTRODUTION To assist vehicle ide, the couples on cane bins in the suga cane industy ae kept in a stetched condition using a bake van, coupled to the end of the tain to poduce a given constant baking foce. uently the bake van opeates using compessed ai which activates a bake callipe, which clamps a ventilated disc oto on each of the fou wheel sets. This mechanical baking mechanism needs maintenance due to wea on the bake pads and otos. Electical baking using self-excited induction geneato, (SEIG), [1-7] can povide a maintenance advantage. With the application of electical baking, instead of conveting kinetic enegy diectly into heat enegy by fiction, the kinetic enegy is conveted to electical enegy and then dissipated as heat enegy in well ventilated esistos. The mechanical and electical connections fo the application of a SEIG in bake van systems is shown in Figue 1. While it is possible to use a vaiable fequency dive to achieve the same esult a good deal of modification would be equied. The eal powe can be absobed by a dynamic bake esisto on the D bus of a VSD invete. In a commecial poduct the pesence of a thee phase system does help stabilise the D bus, cetainly in tems of potential bus collapse. Othe issues that would need to be solved include initial powe fo machine excitation and the povision of invete contol powe that may be mains deived. A bake van pesents a elatively difficult envionment in tems of vibation and tempeatue vaiation so a elatively ugged invete would be equied. 2. SIMULATION MODELS Any induction machine equies excitation cuent to magnetise the coe and poduce a otating magnetic field. An isolated induction geneato without any excitation will not geneate voltage and will not be able to supply electic powe iespective of the oto speed. In geneal an induction geneato equies eactive powe fo its opeation. Thee chaged capacitos connected to the stato teminals of the thee phase induction geneato can supply the eactive powe equied by the induction geneato. Van wheel Belt o Gea oupling Induction geneato Figue 1. Application of SEIG in bake van system Povided that the conditions fo self-excitation ae satisfied the chaged capacitos cause the teminal voltage to build up at the stato teminals of the induction geneato. When the chaged capacitos ae connected to the teminals a tansient exciting cuent will flow and
poduce a magnetic flux. This magnetic flux will geneate voltage and the geneated voltage will be able to build the chage in the capacitos. As the chage inceases, moe exciting cuent is supplied to the induction geneato. The magnetic flux continues to incease hence poducing a highe geneated voltage. In this way voltage is built up. Models will be pesented to investigate the self excitation and steady state opeations of the SEIG. The equation govening the capacitance voltage in Figue 2, without the load esistos, can be epesented as [7]: Vcq = 1 iqsdt+ Vcqo (1) Vcd = 1 idsdt+ Vcdo (2) Whee Vcqo = V cq and V t= cdo = V cd ae the initial t= voltage along the q-axis and d-axis capacitos, espectively. With L s = L ls +L m and L = L l +L m the oto flux linkage is given by: λ = Li + Li + λ (3) q m qs q qo λ = Li + Li + λ (4) d m ds d do Whee λqo λ q t= = and do = d ae the t λ λ = emnant o esidual oto flux linkages along the q-axis and d-axis, espectively. Then, with an electical oto speed of ω, the otational voltage in the oto cicuit along the q-axis is: ( Li Li ) ω λ = ω + + ω λ d m ds d do ( ) ωλ = ω L i + L i + K (5) d m ds d q and the otational voltage in the d axis of the oto cicuit is: ( L i L i ) ( ) ω λ = ω + + ω λ q m qs q qo ωλ = ω L i + L i + K (6) q m qs q d whee Kd ω λqo = and Kq = ω λdo ae constants, which epesent the initial induced voltages along the d- axis and q-axis, espectively. The constants K d and K q ae due to the emnant o esidual magnetic flux in the coe. And ω is the equivalent electical oto speed in adians pe second. V cq V cd R s R s L ls λ qs L ls λ ds L m (a) L m (b) L l λ q L l λ d R R ω λ d + - -ω λ q + - Figue 2. d-q model of a loaded SEIG in a stationay efeence fame (a) q-axis cicuit (b) d-axis cicuit. The matix equation fo the d-q model of a self-excited induction geneato, in the stationay stato efeence fame, using the SEIG model given in Figue 2 and fom Equations (1) to (6), is given as: Rs + pls + 1 p plm Rs + pls + 1 p = plm ω Lm R + pl ω Lm plm ω L i plm i ω L i R + pl s i qs ds q d V cqo + V cdo K Kd q (7) The above equations fom the basis fo a simulation model. Fo self excitation, it suffices to analyse only the expession fo i ds. Using Equation (7) the expession fo i ds can be given as: i ds = 2 2 2 2 2 2 [ R + pl + 1 p ( p L ( R + pl ) + pl ω L ) ] + ( pl R ω ) 2 s whee: s m U = ( R + pl ) + ω L = R + 2p + p L + ω L 2 2 2 2 2 2 2 2 In Equation (8) U epesents all the tems on the numeato and is dependent on the initial chage in the capacitos, the emnant flux in the coe, capacitance, oto speed and the machine paametes. U only has an effect m m (8)
on the coefficients of the patial faction expansion of i ds, which detemine the constants that will be multiplied with the exponential cuent expession in the time domain, and does not affect the behavio of the cuent. The detail of the expession of U is long and it is not necessay to conside when detemining whethe thee is self-excitation o not. The numeato of Equation (8) helps to detemine the multiplying constants fo the solution of the cuent in the time domain and these constants ae dependent on the machine paametes, capacitance value, oto speed and initial conditions. Setting the denominato of Equation (8) equal to zeo, gives the chaacteistic equation. If any of the oots of the chaacteistic equation has a positive eal pat then thee will be a gowing tansient indicating that thee will be selfexcitation. 3. PARAMETER ESTIMATION Machine modeling equies knowledge of the paametes of the machine. Whethe the thee-phase induction machine is modeled using the conventional equivalent cicuit o d-q method the paametes of the machine ae equied. To have an accuate model of the machine, which epesents all the chaacteistics of the physical machine, the paametes need to be detemined accuately. An in depth analysis and simulation of an induction machine can be caied out only with accuate paametes that epesent the actual machine. onsequently to accuately model a thee-phase induction machine, accuate paamete values which epesent the actual opeating conditions being modeled should be known. The paametes used in the SEIG can be obtained by conducting tests on the induction geneato when it is used as a moto. The taditional tests used to detemine the paametes ae the open cicuit (no load) test and the shot cicuit (locked oto) test. The induction machine used as the SEIG in this investigation is a thee-phase squiel cage 22kW WEG induction moto with specification on the name plate: 4 pole, 4V delta connected, 41A, 22kW, 5Hz, 147pm. The paametes ae obtained by conducting paamete detemination tests on the above mentioned induction machine. The paametes obtained fom tests at ated values of voltage and fequency ae L ls =L l =8.5mH, L m =14mH, R s =.5Ω R =.36Ω. The no-load tests wee conducted at ated speed with the machine unde test being diven at synchonous speed using a connected pime move and vaiable speed dive. Fo motoing application these paametes can be used diectly. Howeve, fo SEIG application the vaiation of L m with voltage o cuent should be taken into consideation. The vaiation of the magnetizing inductance with phase voltage, measued at ated fequency, fo the induction machine used in this investigation is given in Figue 3, whee the dots ae expeimental esults and the cuve is a fifth ode cuve fit. Figue 4 shows the vaiation of magnetizing inductance with magnetizing cuent. In Figue 4 the magnetizing inductance as a function of RMS magnetizing cuent is epesented by thee 5 th ode polynomial cuve fits. Lm (H).2.18.16.14.12.1.8.6.4 Lm vs phase voltage.2 1 2 3 4 5 6 phase Voltage Figue 3. Vaiation of magnetizing inductance with phase voltage Lm (H).2.18.16.14.12.1.8.6.4 Lm vs phase cuent.2 1 2 3 4 5 6 7 8 9 phase cuent Figue 4. Vaiation of magnetizing inductance with magnetizing cuent 4. SIMULATION RESULTS The minimum oto speed and minimum capacitance fo self-excitation calculated at no load is given in Figue 5. The self excited induction geneato was simulated fo loaded conditions while speed amped ove the ange 85RPM to RPM.
1 2 3 4 5 6 7 16 Roto speed vs capacitance fo self excitation (22KW induction machine) 16 14 Speed (pm) 12 1 8 6 4 Retadation powe dissipation (kw) 14 13 12 11 1 2 9 apacitance (µf) (micof) Figue 5. Values of minimum capacitance and oto speed fo self-excitation at no load 8 9 1 11 12 13 14 Angula speed (pm) Figue 9 Simulated Dynamic Powe 16 2Ω 3µF 25Ω 3µF 28Ω 22µF 34Ω 22µF 32Ω 16µF Retadation powe dissipation (kw) 14 13 12 11 1 9 3Ω 3µF 38Ω 16µF 36Ω 125µF 42Ω 125µF 8 857 9 11 1 12 13 14 Angula speed (pm) 42Ω 11µF 45Ω 11µF Figue 6. Geneated steady state powe cuves Retadation toque (Nm) 1 5 857 11 13 Angula speed (pm) Figue 1. Dynamic etading toque poduced by the SEIG 5 3 45 4 apacitance (mico Faad) 25 2 Line cuent (A) 35 3 25 2 1 5 1 5 9 1 11 12 13 14 Roto speed (pm) Figue 7. apacito Selection Steps 9 1 11 12 13 14 Roto speed (pm) Figue 11. Vaiation of line cuent with oto speed 6 45 5 Load Resistance (ohm) 4 35 3 25 Line Voltage (V) 4 3 2 2 1 8 9 1 11 12 13 14 Roto speed (pm) Figue 8. Resistance Selection Steps 9 1 11 12 13 14 Roto speed (pm) Figue 12. Vaiation of line voltage with oto speed
Figue 6 shows the steady state geneated powe fo diffeent combinations of the excitation capacitance and load esistance. Resistance and capacitance combinations will be selected by a PL accoding to machine otational speed as shown in Figues 7 and 8. Figue 9 shows the simulated etading powe dissipation fo those combinations while Figue 1 shows the etading toque. Some toque ipples exist which may ultimately be undesiable. A possible avenue fo futhe wok will be the use of phase o bust contolled thyistos to allow smoothe toque contol. An impotant consideation is the contol of the load cuents so that they emain within the machine atings as shown in Figue 11. At the machine ated speed of RPM a powe of appoximately kw is equied fom a nominal 22kW machine but the excitation equiements educe the powe facto consideably. Figue 12 shows the vaiation in machine voltage ove the load ange. Again it is impotant to ensue the voltages emain in a easonable ange. A 22kW switchable esistive load bank. A PL based contolle to allow load stepping with speed is cuently being constucted. Duing manual load stepping loss of excitation has been obseved and on occasion the use of eithe D cuent injection of the connection of a chaged capacito bank has been equied to ecove excitation. (a) Line cuent (A) 6 4 2-2 -4 A n=14pm and =9mico Faad (chaged capacito) B ms cuent Loss of excitation in a pactical system is an impotant issue that can be eadily obseved. If the machine speed falls and additional capacitance is not connected in a timely way the subsequent excitation collapse can thooughly demagnetize the machine. Povision is made fo capacito pe-chaging to ecove a demagnetized machine. 4 2-6 1 2 3 4 5 6 time (sec) Detail A Detail B 4 2 The no load dynamic voltage build up pocesses duing self-excitation of an initially unexcited machine at 14RPM following the paallel connection of a 9uF chaged capacito set is shown as Figue 13. haging is povided by a small D to D convete that pechages two lines of a delta connected bank to 6Vdc. The SEIG, exciting capacitos and the load esistos ae connected in delta. 5. EXPERIMENTAL WORK The simulation models the self excitation cuves and load steps have been confimed by numeous steady state tests using the expeimental system shown in Figue 14. The equipment includes fom left to ight: A vaiable speed dive Two back to back 22kW induction machines An equipment enclosue containing capacitos, switching contactos and powe and eactive measuement facilities -2-4 5 5.2 5.4 5.6 5.8 5.1 5.12 5.14 5.16 5.18 (b) -2-4 16.3 16.4 16.5 16.6 16.7 16.8 Figue 13. Simulated self excitation pocess fo a chaged capacito (a) dynamic line cuent (b) detail just when the capacitos ae connected (c) detail fo the cuent build pocess. At this stage the hadwae fo the PL based contolle is not finalized. Hence the only dynamic expeimental esult is the stato voltage build up fom emnant flux in the machine ion is given as shown in Figue. The machine was bought to 14RPM without capacitos connected. The data collection was tiggeed fom the holding coil voltage of the capacito contacto. A twenty second ecoding was captued and self excitation is visible fom 6 seconds and completes at appoximately 11 seconds. The initial line to line voltage due to emnant flux pio to the capacito connection was between 5 and (c)
6Vms as measued by a digital powe mete. The oscilloscope tace appeas lage than this but ecodings at a time base of 2 seconds a division esulted in elatively noisy data that ovestates the initial voltage. It does illustate the envelope of the voltage and the time esponse vey well. based vaiable speed dive to implement a egeneative solution. Invete dives will still find applications whee enegy ecovey and e-use is equied. Seveal inteesting applications of battey electic and hybid electic locomotives fo shunting o banking applications (assisting tains ove gades) exist. Some futhe wok is cuently being conducted on the laboatoy demonstation of a PL based system and the development of a capacito pe-chaging system to ecove excitation in the event of de-magnetisation. Futhe wok will continue on the development of continuously vaiable thyisto contolled esistive loads and point on wave switched capacitos to educe the etading toque vaiation with speed. 6. AKNOWLEDGEMENTS This wok has been suppoted by SR and the ente fo Railway Engineeing at QU. Geneated Voltage (V) 6 4 2-2 -4 Figue 14. Expeimental System 9micofaad and 14mp -6 2 4 6 8 1 12 14 16 18 2 time (sec) Figue. Expeimental build up of phase voltage with 9µF at 14pm 5. ONLUSIONS A method fo applying a self excited induction geneato fo bake van applications has been pesented. The solution can be implemented eadily with a PL and is consideably simple than the use of an invete 7. REFERENES [1] Basset, E.D. and Potte, F.M., apacitive excitation of induction geneatos, Tansactions of the Ameican Institute of Electical Enginees, 54, 1935, pp.54-545. [2] Gantham,., Sutanto, D. and Mismail, B., Steadystate and tansient analysis of self-excited induction geneatos, Poceedings IEE, Vol. 136, Pat B, No. 2, 1989, pp. 61-68. [3] Elde, J.M., Boys, J.T. and Woodwad, J.L., Selfexcited induction machine as a low-cost geneato, IEE Poceedings Pat, 131, (2), 1984, pp.33-41. [4] Salama, M. H. and Holmes, P.G., Tansient and steady-state load pefomance of stand-alone selfexcited induction geneato, IEE Poceedings on Electical Powe Applications, Vol. 143, No. 1, Januay 1996, pp. 5-58. [5] D. Seyoum,. Gantham and F. Rahman, Analysis of an Isolated Self-Excited Induction Geneato Diven by Vaiable Speed Pime Move, Poceedings AUPE 1, Peth, Austalia, 21, pp. 49-54. [6] Shidha, L., Singh B., Jha,. S., Singh, B. P. and Muthy, SM., Selection of capacitance fo the self egulated shot shunt self excited induction geneato, IEEE Tansactions on Enegy onvesion, Vol.1, No.1, Mach 1995, pp. 1-17. [7] Seyoum, D. The Dynamic Analysis and ontol of a Self-Excited Induction Geneato Diven by a Wind Tubine, Ph.D. Thesis, Univesity of New South Wales, Mach 23.