Detection of Several Flicker Sources Using d-q Algorithm and Flicker Power

Transcription

1 2012, TextRoad Publiation ISSN Journal of Basi and Applied Sientifi Researh Detetion of Seeral Fliker Soures Using d-q Algorithm and Fliker Power Jalal Khodaparast, Ali Dastfan Department of Eletrial and Roboti Engineering, Shahrood Uniersity of Tehnology, Shahrood, , Iran ABSTRACT Detetion of fliker soures is the first step to mitigate the effet of fliker in power system. In this literature, existene of seeral fliker soures is studied and proposes a tehnique for deteting all existing tones in oltage and urrent enelope. Half wae retifier demodulation is improed by using d-q transformation, to alulate phase differene between oltage fliker tones and urrent fliker tone. These phase differenes are onsidered as index of fliker soures detetion. By using phase differene between oltage fliker tones and urrent fliker tones, signs of fliker power are obtained in seleted branhes. By omparing the sign of fliker power with diretion of fundamental atie power, the plae of fliker soure is obtained. For alidation, the 6-bus network is simulated and algorithm for fliker soures detetion is tested. The simulations results show that by using the proposed algorithm, all fliker soures in a power system an be deteted orretly. KEYWORDS: d-q Algorithm, Fliker power, Fliker soures, Fliker tones, Improed half wae retifier, Power quality. 1. INTRODUCTION In reent years, by Proliferation of non linear load in power network, power quality beame of great importane for both onsumers and utilities. One of the most important power quality eents is fliker. Fliker defined as pereption of the human eyes to the light flux of lamps whih depend on oltage RMS alue. So fliker is defined as root mean square (RMS oltage ariation with low frequeny. Due to ompetition in power market, it is neessary to eliminate or redue negatie effets of fliker. Detetion of fliker soure s plae is the first step to mitigate fliker in power system. After this stage by using the appropriate instrument or improing the network struture, fliker is mitigated. Sine now, many different methods for detetion of plae of fliker soures hae been presented. In one of them, with determining the slop of V-I harateristi, the plae of fliker soure is determined in obseration point [1]. Another method is proposed for determining the diretion of a fliker soure by alulating the sign of the fliker power [2], []. Detetion of fliker soure in multi side supplied network has been onsidered in [4], and intelligent identifiation of fliker soure is proposed by using S-transform and neural network in [5]. The first step of fliker soures detetion is separation of oltage and urrent enelope. There are many methods for detetion of enelope. Square method (Square demodulation is one of these tehniques [6], but disadantage of this method is its low auray as it generates additional low frequeny omponents beause square method is not linear []. The Fast Fourier Transformation (FFT is another method whih is used to detet the fliker tones inluded in oltage enelope [7]. Howeer, the auray of the FFT is affeted by the leakage effet [8] and it inreases when system frequeny deiates from nominal alue. Another method is Phase Shifting method [9]. Howeer in this method there would be a small phase shift in fliker tones when main signal is fed to phase shifter so it dereases auray. Another tehnique is Hilbert Transform [10]. Howeer this method has a high mathematial burden, beause impulse response of the filter orresponded to Hilbert Transformation is infinite and so it needs to know all the samples of a signal [9]. Waelet Transform an separate a signal into different bandwidth and then reonstruts speial bandwidth, so it is introdued as tool for enelope traking [11]. But drawbak of this tehnique is diffiulty of interpretation of its returns *Corresponding Author: Jalal Khodaparast is with the Department of Eletrial and Roboti Engineering, Shahrood Uniersity of Tehnology, Shahrood, , Iran. jalal_khodaparast@yahoo.om 2519

2 Khodaparast and Dastfan, 2012 results [12]. Kalman Filter is optimal reursie estimators whih estimate the magnitude and frequeny of oltage ariation and is used to trak fliker enelope [1]. Howeer, its omputational burden is high [8]. d-q Transformation has been applied for detetion and lassifiation of power quality disturbanes [12], [14]. By using this method, a three phase signal an be onerted into a two dimensional frame (d-q whih diret axis (d is at angle of 90 0 to quadrature axis (q and this frame is rotating with angular eloity. By ombining ab-dqo Transformation algorithm and 90 0 phase shift algorithm, a fast and aurate algorithm to reognize all oltage disturbanes and faults harateristis hae been presented in [12]. The utility input oltages are sensed and then onerted to DC quantities in the d-q referene frame. Thus any disturbanes at the utility input oltage will be refleted as disturbane in d-q alues. Using these disturbed alues, it is possible to detet the power quality eents like sag, swell, fliker and so on [12], [14]. In this paper, a method is proposed for alulating all fliker tones phases in oltage and urrent enelope using d-q transformation and half wae retifier. The plaes of fliker soures are determined by using sign of fliker powers in seleted branhes. And sign of fliker power is determined by using phase differene between oltage fliker tone and urrent fliker tone whih is obtained by d-q transformation and half wae retifier. 2. DQ TRANSFORM The d-q transformation is a transform that maps balaned three phase oltages to a synhronous rotating frame in order to extrat a set of three-phase oltages with speial frequeny and to represent them as a DC omponent. The d-q Transformation is defined as follows [15]: sin( sin( sin( a d t t t (2 / b q os os( t os( t (1 sin( t os( t a 2 2 d b (2 / sin os( t q sin( t os( t (2 And also inerse of d-q transformation is defined as (2, where ω is d-q Transformation frequeny, d and q are synhronous rotating frame parameters, a, b, and are ab parameters oordinate system. If d-q Transformation frequeny (ω is equal to frequeny of the phase oltage (ω, d-q synhronous rotating frame lok to ab synhronous rotating frame. Under this ondition, relation between balaned three phase oltages as gien in ( and d-q synhronous rotating frame parameters has been defined as (4. d b ω q θ a Fig. 1. ab-dq Transform 2520

3 a V os( t b V os( t V os( t ( d V sin( q V os( 2 2 V d q d artan( q where V, ω, θ are amplitude, frequeny, phase angle of the phase oltage. Conept of d-q transformation is shown in Fig. 1. (4. PROPOSED METHOD FOR FLICKER TONES DETECTION In a power system, due to the presene of different fliker soures, usually more than one tone is presented in oltage enelope. In this paper a new method has been proposed for frequeny identifiation. In general, onsider oltage of phase a with N fliker tones in its oltage enelope as gien in (5: k N a V 1 V os( t os( t (5 k 1 where V, ω, θ are amplitude, frequeny and phase angle of the kth oltage fliker tone respetiely. So there is more than one fliker tone in oltage waeform and the aim is to separate all fliker tones. First of all, it is neessary to identify frequenies of these tones existed in enelope. Aording to (5, at AM (amplitude modulation, eah fliker tone produes one sub-harmoni in (f f and one inter-harmoni in (f +f as follow: V os 2 ( f t k N 1 (6 a V V os 2 ( f f t 2 k 1 k N 1 V V os 2 ( f f t 2 k 1 So aording to this onept, frequenies of fliker tones are determined. By multiplying (5 and os(2πf t, the result is: os(2 f t mul1 Low-pass filter out1 D Component f Input Voltage D Component sin(2 f t out 2 mul 2 Low-pass filter f Fig. 2. Detetion of signal frequenies 2521

4 Khodaparast and Dastfan, 2012 f f t f f t os 2 ( os 2 ( k N k N 1 V V os 2 ( f f f t os 2 ( f f f t (7 1 2 k 1 k 1 mul 1( t os(2 ft a V 2 k N k N 1 V os 2 ( f f f t os 2 ( f f f t 2 k 1 k 1 where f is ariable frequeny in proposed algorithm. Then, this output is passed to a low-pass filter to gie out DC omponents at eah aailable frequeny. Sine fliker is a slow hange in the oltage magnitude with frequenies between 0.5 to 0 Hz [16], by hanging f from 0 to 2f (f = 50 or 60 Hz, there will be a DC omponent in (7. The output is gien as (8 when f hanges from 0 to 2f. V out 1( f os f f 2 VV (8 out 1( f os f f f 4 VV out 1( f os f f f 4 So the argument of os is zero at frequenies of (f f m and (f +f m, and thereby there will be DC omponent in the output. In general, (8 an identify fliker frequenies but there are some speial onditions, in whih the argument of os would be 90 o, and thereby output would be zero, so oltage of phase a is multiplied to sin (2πf t to sole this problem and this proposed algorithm would be generalized method. So after multiplying to sin (2πf t and then passed to a low-pass filter, output is obtained as (10 by hanging f from 0 to2f. mul 2 sin(2 f t a (9 V out 1( f sin f f 2 VV (10 out 2( f sin f f f 4 VV out 2( f sin f f f 4 Blok diagram of proposed algorithm for detetion of fliker tones frequenies is depited in Fig. 2. Now there are two frequeny spetrums whih should be analyzed. Aording to amplitude modulated, eah fliker tone produes one sub-harmoni in (f f m and one inter-harmoni in (f +f m. out 2 d 2 f ( i V 0 out 1 d1 f (2 f i V 0 out 1 d1 f ( i V 0 f (2 f i V 0 out 2 d 2 OR OR AND Fliker tone Fig.. Blok diagram of reognizer unit 2522

5 The next step is designing a reognizer unit to separate frequenies of fliker tones. Blok diagram of proposed algorithm is shown in Fig.. This blok diagram inludes four onditional statements. Two of them erify output of first and seond spetrum in f (i and another two of them erify outputs of spetrums in f (2f -i. If there are DC omponents in (f f m and (f +f m in both spetrums, reognizer unit would gie out this frequeny as fliker tone. So four onditions are examined at a moment and if output of AND logial gate is one, that frequeny will be fliker tone. Also this algorithm an identify system frequeny howeer power system frequeny is usually 50 or 60 Hz. If system frequeny and frequenies of fliker tones are onstant during the time, they should be deteted one in time but if they are not fixed, they should be deteted any time for using the d-q transformation orretly. Outputs of the reognizer unit are sent to next part whih will be explained in setion IV. 4. PROPOSED METHOD FOR DETECTION OF FLICKER SOURCES PLACES Fliker power method is a method in fliker soure detetion issue. If there are seeral fliker soure in power system, Fliker power is obtained as follow: 1 FP T T 0 T 0 fp dt 1 N N ( V 1 os( t k.( I k 1 os( t dt T N V I N V I os( os( k 1 k where I, ω, α are amplitude, frequeny and phase angle of the kth urrent fliker tone respetiely. φ is phase differene between oltage fliker tones and urrent fliker tone. fp(t is instantaneous fliker power and FP is fliker power. This quantity is alulated from the low frequeny amplitude ariations of the oltage and urrent. With omparing fundamental power flow diretion with diretion of power flikers in eah fliker tone, plaes of all fliker soures are determined. If fliker power was positie, it implies that fliker soure is upstream with respet to fundamental power flow diretion and if fliker power was negatie, it implies that fliker soure is downstream with respet to fundamental power flow diretion. Aording to (11, depending on if sign of os is negatie or positie, sign of fliker power is negatie or positie thereby sign of fliker power is determined by using phase differene between oltage fliker tones and urrent fliker tone. As you know, if argument of os is between π/2 and +π/2, sign of os is positie. (11 Fig. 4. Improement of half wae retifier method 252

6 Khodaparast and Dastfan, 2012 For alulating of phase differene between oltage fliker tones and urrent fliker tone, a method based on half wae retifier and d-q transformation is proposed in this paper, to be able to extrat phases of all fliker tones. Half wae retifier is a method whih has been used for traking fliker enelope []. In half wae retifier method, by feeding main oltage to a half wae retifier blok, fliker enelope is appeared independently. Then fliker enelope an be traked by using filter hain. In this paper, this method is improed by d-q transformation. Proposed method flowhart is shown in Fig. 4. Assume three phase oltages that hae N tone as (12. kn a( t V 1 V os( t os( t k 1 k N b( t V 1 V os( t os( t k 1 k N ( t V 1 V os( t os( t k 1 (12 Fliker tones are obtained aording to Fig. 5. Consider (1 as a oltage of phase a with N fliker tones. After passing half wae retifier, (14 is obtained as []: k N k N a V 1 V os( t os( t V 1 os( t k 1 k k N k N os(2 w t os(4 w t ' ( 1 os( 1..5 a t V t 1/2waeret V 1 os( w t k 1 k os(6 w t (1 It an be seen that many different frequenies are produed. Outputs of the d-q transformation (ω=ω m1 for traking of the first tone are obtained as (15: (14 VV m1 d1( t sin( m1 ACd1( retifier ( t VV m1 q1( t os( m1 ACq 1( retifier (15 where AC d1 (retifier and AC q1 (retifier refer to alternating omponents after passing d-q transformation in retifier method. Then, by using low-pass filter d1 and q1 are: V V m 1 d 1 sin( m 1 V V m 1 q 1 os( m 1 So, phase of the first oltage fliker tone is obtained as follow: m 1 d 1 artan( So, phase of the all oltage fliker tones are obtained. For example, kth fliker tone is obtained as: dk artan( Also the same proedure should be applied for urrent. Assume three phase urrents that hae N tone as (19. q1 qk (16 (17 (

7 k N i a( t I 1 I os( t os( t k1 kn i b( t I 1 I os( t os( t k 1 kn i ( t I 1 I os( t os( t k 1 (19 So, phase of the all urrent fliker tones are obtained. For example, kth fliker tone is obtained as: i dk artan( (20 i qk So phase differene between oltage fliker tone and urrent fliker tone in eah line for kth fliker tone is alulated as: (21 This algorithm an detet more than one fliker soure beause this proposed algorithm an alulate phase of all fliker tones separately. 5. SIMULATION RESULT In this setion simulations for two ases hae been arried out to erify the effetieness of the proposed algorithms in deteting seeral fliker soures in power network. Spot welders are one of the main loads whih an produe fliker in the power network. There are some methods to simulate the spot welder. The performane of a spot welder an be simulated by ariable resistor shown in Fig. 5. The resistor (R onneted with an ideal swith whih is ontrolled by pulse generator are treat as ariable resistor. Current of a single phase spot welder with enelop frequeny 10 Hz is shown in Fig. 6. A simulation based on 6-bus test system is used to demonstrate the proposed approah. Fig. 7 shows a 20 KV power network whih is supplied by three generators and its data is gien in APPEDIX [17]. In this study, two different ases are onsidered and analyzed; one fliker soure, two fliker soures. First of all, some lines should be seleted to analyze, so two lines ended to eah bus are seleted. Line between bus 1 and 2, line between bus 2 and, Line between bus and 6, Line between bus 5 and 6, Line between bus 4 and 5 and Line between bus 1 and 4 are six onsidered lines for power system shown in Fig Case one (one fliker soure In the first ase, the appliability of the proposed method for network with only one fliker soure has been examined. In this setion, only one spot welder in power network onneted to bus has been onsidered. Enelope frequeny is 5 Hz and resistane(r is 70 Ω. Phase differene between oltage fliker tone and urrent fliker tone in eah seleted line is alulated using proposed algorithm shown in Fig. 4. Simulation results are presented in table I. Based on the signs of the powers shown in Table I, diretion of fundamental and fliker power flows are shown in Fig. 8. In order to improe the isualization of simulation results, only six seleted lines are shown in Fig. 8. In Fig. 8, dashed arrows represent positie diretion of fundamental power flow and bold arrow represent fliker power. By following the diretion of fliker power (bold arrow with head number 5, the plae of fliker soure an be found. Thus, the load onneted at bus is a fliker soure. 2525

8 Khodaparast and Dastfan, 2012 with an ideal swith whih is ontrolled by pulse generator are treat as ariable resistor. Current of a single phase spot welder with enelop frequeny 10 Hz is shown in Fig. 7. A simulation based on 6-bus test system is used to demonstrate the proposed approah. Fig. 8 shows a 20 Fig. 5. Modeling of spot welder as ariable resistor Fig. 6. Single phase spot welder urrent Bus Bus2 Bus6 Bus1 Bus5 Bus Case two (two fliker soures Fig. 7. Power network In the seond ase, the appliability of the proposed method for network with two fliker soures has been examined. In this setion, first spot welder is onneted to bus with enelope frequeny of 5 Hz and resistane is 70 Ω and the seond spot welder is onneted to bus 4 with enelope frequeny of 10 Hz and resistane is 100 Ω. Simulation results are presented in table II and table III. Based on the signs of the powers, diretion of fundamental and fliker power flows are shown in Fig. 9. In Fig. 9, dashed arrows represent positie diretion of fundamental power flow and bold arrow with head number 5 represent fliker power of 5Hz and bold arrow with head number 10 represent fliker power of 10Hz and. By following the arrows with head number 5, the plae of fliker soures with enelope frequeny 5 Hz an be found and by following the arrows with head number 10, the plae of fliker soures with enelope frequeny 10 Hz an be found. Furthermore, this method is implied for more soures too but beause of spae limitation, they were not presented. 2526

9 Fig. 8. Graphial simulation result of the first ase Fig. 9. Graphial simulation result of the seond ase 6. CONCLUSION Identifiation of the fliker soures is the first step in the proess of improing power quality. In this paper, a method is proposed for alulating all fliker tones in oltage and urrent enelope using d-q transformation and half wae retifier. The proposed method is apable of getting more than one tone in enelope. Then phase differene between oltage fliker tone and urrent fliker tone is obtained and so sign of fliker power is determined. The plaes of fliker soures are determined by using signs of fliker powers in seleted branhes. Simulation has been arried out whih results show aurate deteting of the fliker soures plaes. APPENDIX Table IV presents Generators and loads data of power system and table V presents lines data. REFERENCE [1] A. B. Nassif, E. E. Nino, W. Xu, A V-I Slope-Based Method for Fliker Soure Detetion: IEEE Trans. Power Deliery 21, [2] P. G. V. Axelberg, M. H.J. Bollen, An Algorithm for Determining the Diretion to a Fliker Soure: IEEE Trans. Power Deliery, 21 (2: [] P. G. V. Axelberg, M. H. J. Bollen Trae of Fliker Soure by Using the Quantity of Fliker Power: IEEE Trans. Power Deliery, 2 (1: [4] A. Dastfan, M. R. Mirzayi, Identifiation of Dominate Fliker Soure in Multi side Supplied Power Systems: Int. Re. Eletr. Eng., (5: [5] N. Eghtedarpour, E. Farjah, A. Khayatian, Intelligent Identifiation of Fliker Soure in Distribution Systems: IET Gener. Transm. Distrib., 4 : [6] P. G. V. Axelberg, 200. Measurement methods for alulation of the diretion to a fliker soure, Li. Eng. dissertation, Dept. Elet. Power Eng., Chalmers Uni. Tehnol., Gothenburg, Sweden. [7] K. Sriniasan, Digital Measurement of the Voltage Fliker: IEEE Trans. Power Deliery, 6: [8] M. I. Marei, E. F. El-Saadany, M. A. Salama, Enelope Traking Tehniques for Fliker Mitigation and Voltage Regulation: IEEE Trans. Power Deliery, 19 (4 :

10 Khodaparast and Dastfan, 2012 [9] X. Jia, Q. Chen, A Method of Traking Voltage Fliker Enelope Real-Time: IEEE Power & Energy Soiety General Meeting, PES,: 1-6. [10] T. K. Abdel-Galil, E. F. EI-Saadany, and M. M. A. Salama, Online Traking of Voltage Fliker Utilizing Energy Operator and Hilbert Transform: IEEE Trans. Power Deliery 19: [11] T. Zhang, E.B. Makram, Waelet Representation of Voltage Fliker: Eletr. Power. Syst. Res. 48 : [12] E. Pouresmaeli, M. F. Akorede, M. Hojabri, A Hybrid Algorithm for Fast Detetion and Classifiation of Voltage Disturbane in Eletri Power System: Eur. Trans. Eletr. Power, 21: [1] A. A. Girgis, J.W. Stephens, E. B. Makran, Measurement and Predition of Voltage Fliker Magnitude and Frequeny: IEEE Trans. Power Deliery, 10: [14] O. C. Montero-Hernande, P. N. Enjeti, A Fast Detetion Algorithm Suitable for Mitigation of Numerous Power Quality Disturbanes: EEE Trans. Ind. Appl., 41 (6 : [15] P. C. Krause, O. Wasynzuk, S. D. Sudhoff, Analysis of Eletri Mahinery and Drie System, Wiley, New York. [16] M. Mazadi, S. H. Hosseinian, W. Rosehart, Instantaneous Voltage Estimation for Assessment and Monitoring of Fliker Indies in Power System: IEEE Trans. Power Deliery, 22 ( : [17] A.J.Wood, B.F.Wollenberg, Power Generation, Operation and Control, 2nd ed., Wiley, New York. [18] A.B. Nassif, J. Yong, W. Xu, Interharmonis: Signaling Proessing Issues and Appliations: IEEE Power and Energy Soiety General Meeting,: Table. I. Simulation result of first ase from to Fundamental Power(MW Phase differene Sign of fliker power Negatie Negatie Negatie Positie Negatie Negatie Table. II. Simulation result of seond ase (5Hz Table. III. Simulation result of seond ase (10Hz from to Fundamental Power(MW Phase differene Sign of fliker power from to Fundamental Power(MW Phase differene Sign of fliker power Negatie Negatie Negatie Positie Positie Positie Positie Positie Negatie Negatie Negatie Negatie 2528

11 Table. IV. Generators and loads data Table. V. lines data Bus number GEN (pu (S=100 oltage (pu P load (pu Q load (pu From To R(pu X(pu BCAP(pu MW