Synchronous Power Controller Merits for Dynamic Stability Improvement in Long Line by Renewables

Transcription

1 Synchronou Power ontroller Merit for Dynamic Stability Improvement in Long Line by Renewable Motafa Abdollahi, Joe Ignacio andela, Joan Rocabert, Raul Santiago Munoz Aguilar, and Juan Ramon Hermoo Technical Univerity of atalonia Barcelona, Spain Abtract Growing univeral demand for greater ue of renewable energy reource (RES) intalled in dipered location, ha caued to more urgent need for long tranmiion ytem to ue thee green ource. Small ignal tability i one of vital challenge for tranmiion ytem operator in preence of big RES. In thi paper, a Static Synchronou Generator baed in Synchronou Power ontroller (S-SP) i preented and it capabilitie for dynamic tability enhancement in a long A tranmiion ytem i invetigated. Dynamic modelling of a long line and then tudy in Uncompenated and ompenated cae by uing time domain imulation and modal analyi, and then experimental tet confirm that S-SP can be effective in active damping of power ocillation and reducing mall ignal tability challenge. eyword Static ynchronou generator; Synchronou power controller; Long A tranmiion ytem; Dynamic modelling; Small ignal tability I. INTRODUTION Renewable Energy Reource (RES) technologie, epecially wind power and olar have grown rapidly in the pat decade []. up to now the main reaon for uing major part of RES wa jut following nature and environmental benefit, cot aving and market iue and main advantage of thi reource wa their local controllability through the ue of power electronic device. But nowaday, ome challenge i appeared for power ytem in preence of high penetration of RES [2]. Impact on mall ignal tability i one of the main technical challenge. thi challenge i more critical epecially after growing the rated power of RES and able to have big participation ratio in generation area, a well a working in higher voltage level and connecting directly to the tranmiion level, becaue each of them need to comply with the pecial grid code [3].According to variety of primary ource and outpread poition of intalled RES in Europe (wind farm in North ea and Mediterranean olar plant in outh), a well a their different effect on the power ytem operation, the uper gird idea i one of the bet exiting olution for reducing thi challenge [4]. Moreover, to have better balance between generation and conumption by greater ue of RES, having a long tranmiion line with acceptable tability level i very important to achieve a high performance uper gird. When a major part of big traditional ynchronou generator () be replaced by new RES technologie which work a non-ynchronou generation unit, reduction of global rotational inertia and lack of overall damping are the firt drawback which will arie in the global power network and ha ignificant damaging effect on dynamic tability [5]. According to a few tudie [5], the bet current procedure for thi challenge i to mimic the characteritic of a big conventional ynchronou generator by RES converter. In thi cae, uch electronic generation unit can be called a Static Synchronou Generator (S). In S concept, ome virtual meaning uch a Virtual Inertia, Virtual Damping and Virtual Admittance (VA) will be implemented in controller of a voltage ource converter (VS). Thee concept are not phyical quantity o can be defined and tuned flexible during control proce. Synchronou Power ontroller (SP) technique [6] i one of the highly efficient idea among S concept. Synchronizing with the grid without uing Phaed Locked Loop (PLL) ytem and at the ame time, impoing flexible big virtual damping and virtual inertia in the grid, are the main advantage of thi technique which can be very ueful for mall ignal tability enhancement in power ytem. In thi work, after introducing a S baed in SP (S-SP), capabilitie of thi technology for improving dynamic tability in a long A tranmiion ytem i analyzed. In continue, firtly VA modelling and then grid ynchronization by SP are decribed. After that, dynamic modelling of a long ac tranmiion ytem i expreed in two cae: Uncompenated Line (UL), and ompenated Line (L) by S-SP. Then by uing time domain imulation in MATLAB/Simulink and modal analyi, and finally by a caled down laboratory platform, S-SP capabilitie for improving mall ignal tability i invetigated. II. STATI SYNHRONOUS GENERATOR A. Virtual Admittance Modelling Baed on idea uggeted in [6], control of a VS at current mode by uing the VA concept can be hown a Fig.. e v v Rv + Lv Virtual Admittance i * i vabc P Vgrid Zgrid Main Grid Fig.. Virtual admittance idea [6]. The main role of VA i that to provide a current reference for a current controlled VS. The idea i argued that, firt of all the voltage at point of common coupling (P) will be meaured. Then, thi voltage will compared with an internal electromotive force a an internal virtual voltage. The error of voltage

2 will pa through the VA model. Finally current reference can be calculated a: () = ()() + () () B. Grid Synchroniziation by SP Baed on SP idea [6], Electromechanical model (EM) of in combination with VA idea, can be ued for grid connection of VS a depicted in Fig. 2. EM part will control active power exchanging between VS and external grid, and in thi way grid phae tracking will be achieved without uing PLL. Reactive power control channel will provide reference magnitude, and finally reference value of internal virtual emf will generated for VA model. + _ I dc Vdc Limiter VS PWM * Vabc PR ontroller i * iabc VA Model L v + R v Filter v v abc αβ vabc e * P abc θref E * E θ oup. Tran Vconv Eref E Vgrid Intantaneou Power alculation Q ontroller Qmea 2H + D ωref EM Model Implementation Fig. 2. Grid ynchronization of VS baed on SP idea [6] If VA model ha enough big inductive part and a well a P voltage be connected to an external grid by a coupling tranformer, overall impedance between VS and external grid ha inductive characteritic 9. So, can be controlled by voltage angle, and by voltage magnitude, repectively a bellow: Pmea () (2) [ ()] (3) III. LONG LINES DYNAMI MODELING A long A tranmiion ytem for mall ignal tability analyi i conidered a i hown in Fig. 3. In thi ytem, conventional a weaker generation area i connected to trong grid through a long line, and local load "# i upplied jointly by the line and. For more implicity by uing dc power flow, exchanged active power through the line $% in per unit ytem, auming voltage are '(, can be calculated a below [7]: Pref Grid B3 X3 P3 Fig. 3. Long A tranmiion ytem (Uncompenated Line) $% $% )* $ * % + (4) Where, * $ and * % are phae angle at Sending-end bu and Receiving-end bu,, repectively. Alo, $% repreent reactance of tie line between thee bue. Dynamic modelling of thi tranmiion ytem when i operated a UL, and alo when i working a L, can be done a follow. A. Uncompenated Line Modelling By conidering a econd order wing equation (SOSE) to decribe dynamic of (ignoring PSS effect), dynamic model of UL, can be depicted a Fig. 4. Where, -# i tranferred power from B3 to B and can be calculated by (4) when $% = -# and indicate reactance of all length of line. Alo,. # i inertia contant and / # i damping coefficient of, repectively. 23 i mechanical input power of prime mover and 4 5 i nominal grid frequency in radian. L mech 2H + D θ θ 3 B PL X 3 Fig. 4. Dynamic model of uncompenated line connected to Baed on thi model, tranfer function of cloed loop ytem in Laplace domain i: -# (4 5 / -# ) = 23 "# 2. # 8 +/ # +(4 5 / -# ) omparing characteritic equation of ytem (5), with tandard form of econd-order ytem in normal operating point with damping ratio 9 le than one, give the root a bellow: #,8 = / # ±= 4 5 > 2. # / 8 # 4. # 2. # 4 5 -# 4 B. ompenated Line Modelling Baed on everal tudie [8], middle point intallation i the bet option for optimal compenation of a long line. So, here a bae cae, L i conidered a i hown in Fig. 5 where S- SP i connected to B2. 3 (5) (6)

3 Grid B3 B2 X32 X2 B P32 P2 PL Global contant inertia i combination of and S-SP effect and alo will be affect by location of S-SP (? 88 ), iii) Frequency of ocillation will change by variation in / 8 and alo location of S-SP (? 88,? #,? A ) and iv) Location of S-SP will affect the overall damping of ytem (? # ). S-SP Fig. 5. Long tranmiion line compenated by S-SP. In thi cae, 8# and -8 indicate active power in Receivingend and Sending-end bue, repectively. A well a, -8 = 8# = ( -# /2). onidering SOSE for both and S-SP, dynamic modelling of L can be diplayed by Fig. 6. Where,. 8 and / 8 are the virtual inertia contant and virtual damping coefficient of S-SP, repectively. A well a, * # to * - are phae angle in B to B3. After doing ome mathematical operation, characteritic equation of cloed loop ytem from "# to -# can be +? - - +(? 8# +? 88 ) 8 +? # # +? A = (7) = 2. # # -8 mech? - = (2. # / / # ) 8# -8? 8# = (/ # / 8 8# -8 )? 88 = ( # 8# +2. # -8 ) 4 5 L mech2? # = (/ 8-8 +/ # 8# +/ # -8 ) 4 5 2H + D 2H 2 + D2? A = Fig. 6. Dynamic model of compenated line with S For more implicity and a well a having better analyi about the effect of S-SP on overall dynamic behavior of L, all Power Sytem Stabilizer (PSS) i conidered to be deactivated. So, ince the internal damping of are very mall [7], / # can be conidered equal to zero. By thi aumption and alo convert (7) to tandard repreentation format of a 4th order equation we have new coefficient a =? - = / 8? 2. 8# = 8? 88 = D # ( 8# + -8 ) E 4 2. #. 8 8# 5-8? / # =? 2. # A = 8# 2. # # -8 According to to? A ome preliminary forecat about S-SP effect on dynamic model of L can be done: i) Overall damping of the ytem i caued only by / 8 (? -,? # ), ii) θ θ2 θ2 θ3 X 2 X IV. S-SP APABILTIES FOR SMALL SIGNAL STABILITY Both UL and L model are imulated by MATLB/ Simulink for evaluation S-SP capability in dynamic tability improvement. A tep change in local load "# =.5'( i applied then Sending-end and Receiving-end power are analyzed. A well a, eigenvalue are extracted in each tudy by uing (7) to confirm imulation reult. A. Variable Damping in S-SP Flexibility in / 8 i one of the main S-SP capability for control global damping of ytem. Thi election can be done purpoefully, meaning that firtly it can be elected to have a proper internal damping ratio in S (for example to have 9 HHI =.7) and then it can be modified to have better global damping on ytem. To clear thi feature, imulation reult of UL () and L (+S) for. # = 3.7, / # =,. 8 = 5, / 8 = /, -# =.75'(, and -8 = 8# = -# /2, are hown in Fig. 7 and eigenvalue for elected magnitude of / are reported in Table I. A can be een, in UL due to deactivation of PSS and alo / # =, repone i undamped ocillatory. A well a from (6), UL model ha two pure imaginary eigenvalue which are located on the,4 axe and frequency of ocillation i.97 Hz. But on the contrary, in L even when the S parameter are deigned independently to have low internal damping (/ 8 = 78.47, 9 HHI =.44), mode which are trongly linked to and UL model, have coniderable damping. Power ocillation will damped more by proper increaing in / and in wort cae during 2.7 ytem will arrive to new teady tate operating point. Deeper invetigation of 8# and -8 how that, although by increaing /, ocillation between B2 and B3 will be reduced fater and when / = 77 we have not big overhoot, but elect very big / can lead to new inter area ocillation between B2 and B. So, it will be very ueful to invetigate L mode when / i variable. Pole poition of L for / = L M and L M 2 with L M = 5, i hown in Fig. 8. A can be een, in beginning mode will take more damping along the improving damping in S mode. Although, the ocillation frequency of S mode alo will be affected but finally overall damping will be enhanced properly. Internal invetigation of S how that 9 HHI = will occur when L M = 77.25, which mean in thi value S ha an overdamped dynamic. It eem after thi critical damping value, S cannot have ignificant interaction with other part of ytem in L to be able participate properly in damping of ocillation. Moreover, when S i connected to network and provide initially damping for L, thi critical point will occur ooner (L M = 55). So after thi internal breaking point, S a rigid ytem ha critical pole from ytem point of view and by increaing more L M firtly ocillation with new frequency will be made and then overall damping of ytem will be reduced. Beginning of thi new ocillation can be een clearly in 8# when / = 77 (Fig. 7).

4 B. Several Location of S-SP In thi ection, effect of S-SP location on mall ignal tability of L i invetigated. Internal parameter of S-SP i conidered to be fixed (/ 8 = 24.7,. 8 = 5). By auming 8# = L -# and -8 = (L) -# when L, S located at everal poiton along line can be modeled. So, maller L mean location cloer to. Dynamic model repone and eigenvalue for L = 5, L =.5 and L =.75 are preented in Fig. 9 and Table II. A can be een, maller L lead to repone with very lower overhoot, no coniderable wing and very fat (.75) damped at -8. Reaon i that, when location i cloe to diturbance, according to fat reaction capability of S and maller electrical ditance, S i able to participate even in very mall wing and in thi way wing outpread will be prevented over big part of line. Early mall wing with higher frequency in 8# for when L = 5, and very oft tranient in -8 at the ame time confirm thi reaon. Sending-end Power (pu) S (78) +S (24) +S (77) Fig. 7. Sending and receiving active power for everal damping Im 2 d=2 - S Re Fig. 8. Impact of S-SP damping variation on ytem eigenvalue Table I. Eigenvalue of UL and L Several damping +S By increaing L, two event will occur. Firtly, changing in phae angle will be ened by S with more delay a reult of bigger electrical ditance from diturbance place; econdly S effectivene on diturbance will be reduced due to bigger reactance of tie line. That i why, for L =.75 after firt big peak ocillation tart to damped and moreover damping duration i longer (7). In addition, according to lower effect of S, frequency of ocillation i impoed by mode, a well a diturbance are pread in total length of line and thu we have imilar ocillation at receiving and ending end bue. Variation in eigenvalue for. L.9 when L =.5 i plotted in Fig.. A would be expected, mode move cloer to,4 axe by increaing L which mean having lower damping and converting to dominant pole for overall ytem. But on the other ide, S mode will get more damping by increaing L due to the location cloer to infinite bu and ha maller interaction with ocillation of weaker area. In thi plot it Receiving-end Power (pu) Pole Poition - Increaing S Damping d=55 D= HHI =.44 D= HHI =.7 D= HHI =. S Eigenvalue d=2 ±7.524 d d d= 2.39± ± ± ± ± ±6.5 +S (78) +S (24) +S (77) can be een that, overall damping will be cloe to zero for big L. Although, damping ratio of mode 9 HI and thu overall damping will be improved by decreaing L but enhancement in damping ratio ( 9 HI ) i not linear. In area cloe to B3 (.8 L.9), 9 HI i not very big, but in central area (.5 L.75) thi variation i ignificant. Finally, after croing from middle of line (L =.5), 9 HI will be mall again. The reaon can be expreed by attention to the variation in S mode. In (.8 L.9), variation in S damping ratio ( 9 HHI ) i mall due to being too cloe to external trong grid. After that, 9 HHI i better by moving away from B3 and will affect 9 HI ignificantly up to L =.45. At thi critical point the dominant pole of ytem will change from mode to S mode. So, from thi point onward overall behavior of ytem i impoed by S. Sending-end Power (pu) S (.5) +S (5) +S (.75) Receiving-end Power (pu) Fig. 9. Impact of S-SP location on power ocillation Im 4 2 = Re Fig.. Sytem mode affected by intallation location of S-SP Table II. Eigenvalue of UL and L for different location. S-SP Participation Level Effect of S-SP power ratio in providing active power, when S-SP intalled at fixed location, will be analyzed in thi ection. According to dynamic modelling and auming (2. 8 +/ 8 ) L O, S-SP contribution can be controlled by L O. Bigger L O mean S-SP i emulating behavior of a bigger and vice vera. In thi tudy, / 8 = 24.7 and. 8 = 5, are elected a bae value and S-SP i intalled at PQR middle of line ( -8 = 8# = -# /2). If HHI repreent bae Pole Poition - S in Variou Location =.5 =.5 S S =. =.5 =. =.9 +S (.5) +S (5) +S (.75) Eigenvalue ± ±9.26 =5 3.47± ±3.488 = ± ±4.52 =.75.78±7.42 After thi critical point, 9 HHI i not a big a before and in continue very big change will happen in frequency of S mode. A reult, mode will alo affected by thi frequency variation and finally new ocillation with bigger frequency will appear in ytem (uch a ocillation for L = 5 in Fig. 9). It can be concluded that, for thi pecified S, central area of long line (.5 L.7) are bet place to intall which will lead to have ufficient effect on dynamic tability without cauing any new inter-area ocillation. +S

5 value of power ratio, then new power ratio can be expreed a VW HHI = L O PQR HHI. The L and UL repone to "# are hown in Fig., when L O = ±.5. A can be een, by reducing 5% of HHI and thu lower effect of S-SP on total active power, in fact with very mall damping will get major contribution, o finally overall damping will be weaker. A reported in Table III, for L O =, S mode with 9 HHI = 48 and mode with 9 HI =.393 are very cloe together, o in thi cae all of them are dominant pole for L and will provide large enough damping jointly. When L O =.5, while S mode ha a bit larger damping ratio (9 HHI = 59), but now mode with 9 HI =.68 are new dominant pole and new 9 HI i 2.5 time maller than previou. So, having maller overall damping in L for L O =.5 i reaonable compared to when L O =. Sending-end Power (pu) S (.5) +S () +S (.5) Receiving-end Power (pu) S (.5) +S () +S (.5) V. EXPERIMENTAL TEST To experimental validation of theoretical ection, a cae tudy of tranmiion ytem with detail reported in Table IV i implemented in laboratory platform a i hown in Fig. 5. Long line ytem i modeled in Real-Time Digital Simulator environment (OPAL-RT56), and three phae voltage of B2 caled down and then ued for reference of PUISSANE PU 2VA amplifier. So thi amplifier emulate characteritic of B2 in long line model. A 5kVA, 4V, khz converter which i upplied by a 32kW REGATRON D power upply i connected to amplifier a a S-SP. SP baed controller which i implemented in dspae 3, will control the converter. By meauring three phae exchanged current between amplifier and real converter and then tranfer it to inide of OPAL-RT model by proper gain, the behavior of a real S- SP will be modelled in middle bu of mentioned long line ytem. A ample, L with everal damping value (/ 8 = 5 and / 8 = 24) are teted by thi platform and experimental reult are hown in Fig. 3 and Fig Fig.. Impact of S-SP power ratio on power ocillation 5 Pole Poition - S Power Ratio S p p= p=.5 p=. Im p=. p=.5 p S p=.5 p= Re Fig. 2. Sytem mode affected by S-SP power ratio a) +S Table III. Eigenvalue Several power ratio p=.5 p= p=.5 Eigenvalue \ ]^ \ ]]^ ± ± ± ± ± ± ± b) However, comparing reult of L for L O =.5 with UL, how that even by mall participation of S-SP, overall damping of ytem can be improved dramatically. For L with L O =.5, although S mode are more prevail, but mode ha alo major effect on ytem repone (due to maller difference between real part of thi two group mode compare to when L O =.5). So compare to bae cae (L O = ), although a mall reduction in 9 HHI ( 9 HHI =.79) ha afflicted the ocillation, but HHI = +5% give a coniderable damping ratio to mode (9 HI =.635). Finally thi modification in ytem mode will reult in ocillation with maller firt overhoot in ending-end bu and a well a active power ocillation with lightly maller damping in receiving-end bu. Variation in ytem mode poition for. L O.5 and L O =. i plotted in Fig. 2, and confirm thi analyi. c) Fig. 3. Repone of tranmiion ytem for / 8 = 5, a) load, S-SP and bue power, b) ocillation in active power of bue, c) boundary of power ocillation in lat 2 econd. A bae active load "# = 26YZ and imilar S-SP participation level HHI = 2YZ i conidered at primary teady tate condition for both cae. A well a, [ - = 264YZ and [ # = 645YZ are fixed, which mean S-SP ha 8.7% participation ratio in generation compare to generation level. For invetigate tranmiion ytem repone a

6 tep change in active load "# = 29YZ i applied for both cae. A can be een in Fig. 3, for / 8 = 5 a mall damping effect of S-SP, ocillation in all power remain for long time (more than 9) a well a will affect HHI. Ocillation of ending-end ([ - ) and receiving-end ([ # ) power in lat two econd are 22.4YZ and 2YZ, repectively. But in / 8 = 24 a i hown in Fig. 4, tranferring of ocillatory mode from one area to another i prevented ignificantly by having more damping in S-SP. Where, amplitude of ocillation in [ - i reduced remarkable from firt cycle ( [ -`ab = 327.3YZ and [ -`a#8@ = 224.2YZ), while magnitude of ocillation in [ # i fixed on ame cycle ( [ #`ab = [ #`a#8@ = 34.9YZ). Moreover, all ocillation in thi cae are damped.5 time fater than previou cae ( [ #`ab = 2YZ at c = 8.28 and [ -`ab = 22.4YZ at c = 8.22 comparing with [ #`a#8@ = 2YZ at c = 5.65 and [ -`a#8@ = 22.4YZ at c = 5., where c how time interval after "# connection). A tranmiion ytem. Dynamic modelling of compenated line how that, a) poibility of having big and variable virtual damping, a well a b) flexibility in power participation rate, are two main freedom degree at S-SP which by affecting dominant pole will caue to adding ignificant overall damping and control of inter-area ocillation. Moreover, optimal placement tudie approve that by chooing an appropriate mounting location for S-SP (a a bu with tiff dynamic), will reult to prevent preading of diturbance to other area. Finally it can be concluded that S-SP could be a ueful practical technology for dynamic tability enhancement only by relying on it controller capabilitie. ANOWLEDGMENT Thi work wa upported by the Spanih Minitry of Science under Project ENE R. Grid Line S- SP Table IV. ae Study Data Long Line A Tranmiion Sytem 5 MVA, 5 kv, 5 Hz j =.755 Ω/km m =.8737 mh/km =3.33 nf/km mcph =7km MVA, 3.8 V, 5 Hz, Qr =25Mvar M =.35 pu, w =.474 pu,. =3.7 Tran: yz = pu, yz =.2 pu 5{VA, Q2 =4V, M2 =6V, khz. 8 = 5, / 8 =24, =.pu, =.3pu a) b) c) Fig. 4. Repone of tranmiion ytem for / 8 = 24, a) load, S-SP and bue power, b) ocillation in active power of bue, c) boundary of power ocillation in mentioned 2 econd interval on part (b). VI. ONLUSION In thi work, aim to dynamic tability enhancement, capabilitie of a Static Synchronou Generator equipped with Synchronou Power ontroller (S-SP) i invetigated a a highly efficient technology for integration renewable into long OPAL-RT (Grid+Monitoring) PU 2VA (Amplifier) DSP + ontrol Panel D Power Supply Fig. 5. Platform for experimental validation REFERENES 5VA onverter [] The European offhore wind indutry key trend and tatic t half 25, the European Wind Energy Aociation (EWEA), July 25. [2] J. Quintero, V. Vittal, G. T. Heydt and H. Zhang, "The Impact of Increaed Penetration of onverter ontrol-baed Generator on Power Sytem Mode of Ocillation," in IEEE Tranaction on Power Sytem, vol. 29, no. 5, pp , Sept. 24. [3] IEEE Standard for Interconnecting Ditributed Reource With Electric Power Sytem, IEEE Std , 23. [4] J. Blau, Europe plan a north ea grid, IEEE Spectrum, vol. 47, no. 3, pp. 2 3, 2. [5] H. Bevrani, T.Ie, Y.Miura, Virtual ynchronou generator: A urvey and new perpective, Int. J. Electr. Power Energy Syt, Vol. 54, pp , 24. [6] P. Rodriguez, I. andela and A. Luna, "ontrol of PV generation ytem uing the ynchronou power controller," 23 IEEE Energy onverion ongre and Expoition, Denver, O, pp , 23. [7] P. undur, Power Sytem Stability and ontrol, N.Y: M-Hil, 994. [8] T. Moger, T. Dhadbanjan, A novel index for identification of weak node for reactive compenation to improve voltage tability, IET Generation, Tranmiion & Ditribution, vol. 9, no. 4, pp , 25.