The 14th International Conference on Intelligent System Applications to Power Systems, ISAP 2007 November 4 - 8, 2007, Kaohsiung, Taiwan MFFN based Static Synchronous Series Compensator (SSSC) for Transient Stability improvement V.K.Chandrakar and A.G.Kothari Voltage sourced converter (VSC) based series connected FACTS controller can inject a voltage with controllable magnitude and phase angle at the line frequency and found to be more capable of handling the Power system problems. The benefits of using an SSSC are listed in [2]-[4].In ref.[3] &[4 ], the modeling and control aspects are given in details. In ref.[5], paper describes the independent control of series FACTS devices based on local measurable components with conventional PI controller. However, PI controller is less effective for nonlinear system reported in literature [6]. The Artificial neural networks (ANN) offers an alternative solution to the conventional PI controller[6].Ref.[6], presented the Radial basis function network model in coordination with extended kalman filter(EKF) based controller for nonlinear UPFC device. However, effect of damping schemes like PSS & POD is not included in analysis. Ref.[9], demonstrated the RBFN based FACTS devices in coordination of POD&PSS damping schemes. However, MFFN controller is not included for analysis. This paper presented the coordination of two control variables by conventional PI controller and Multilayer feed forward network (MFFN) controllers are tested on power handling capacity of the line and transient stability improvements of SMIB system. The damping of oscillations are conventionally handle by the power system stabilizer (PSS). This paper presented the power oscillations damping (POD) control [8] and PSS in coordination with SSSC for damping of oscillations improvement. Abstract-- The main aim of the paper is to analyze the performance of Multilayer feed forward network( MFFN) based SSSC on power handling capacity of the line, improvement in transient stability and damping of oscillations of the Single machine infinite bus system(SMIB). This paper presents the comparative performance studies of two different controllers namely: [i] Conventional PI controller , and [ii]Multilayer feed forward network (MFFN) .Controllers are designed to coordinate two control inputs : in-phase voltage and the qudrature voltage of SSSC. The coordinated action of proposed SSSC controllers with power oscillations damping (POD) control & power system stabilizer (PSS) are tested for dynamic performance of the system under various system conditions. The simulation results shows that the proposed controllers increases the power handling capacity of the line, significant improvement in transient stability of the system, and damping of oscillations. The results indicates that the coordinated POD & PSS action further improves the dynamic performance of the system. The proposed SSSC controllers are tested in multi-machine system. The MFFN based SSSC controller provides superior dynamic performance than PI controller. Index Terms-- Damping of oscillations, Power oscillations damping control, Power system stabilizer, Multilayer feed forward network, SSSC, Transient stability T I. INTRODUCTION he need for flexible and fast power flow control in the transmission system is anticipated to increase in the future in view of utility deregulation and power wheeling requirement . The utilities need to operate their power transmission system much more effectively, increasing their utilization degree. Reducing the effective reactance of lines by series compensation is a direct approach to increase transmission capability. However, power transfer capability of long transmission lines is limited by stability considerations [2]. The advent of fast acting FACTS devices[1] allows for fast and vernier control of series compensation using Thyristor controlled series capacitor(TCSC) and static synchronous series compensator (SSSC)[3],[4].TCSC is variable impedance device based on thyristor ,while SSSC is based on voltage source converter (VSC) . In recent years, II. SYSTEM MODEL The system depicted in Fig.1 and Fig.2 is used to validate the implementation of the proposed PI and MFFN model for SSSC controller. The detail system data is given in Appendix. The synchronous generator is represented by a 6th order machine model and the generator excitation system has a simple automatic voltage regulator (AVR) as shown in Fig. 3.For the transient stability analysis mechanical power input is assumed to be constant. The SSSC is located at the sending end of the line in the SMIB system chosen as typical case [5],[7],[9]. In case of the multi-machine system, SSSC is tested at three different locations as shown in Fig.3. V.K.Chandrakar is with Department of Electrical Engineering, G.H.Raisoni College of Engineering. Nagpur, India, 440016, (e-mail: vc_vkc@yahoo.co.in , vkchandrakar@satyam.net.in ). III. SSSC MODEL SSSC can operate in four different ways [2].To achieve real and reactive power flow control[5],[9], we need to inject series voltage of the appropriate magnitude and angle. The A.G.Kothari is with the Department of Electrical Engineering, Visvesvaraya National Institute of technology, Nagpur, India, 440011 376 The 14th International Conference on Intelligent System Applications to Power Systems, ISAP 2007 injected voltage can be split into two components which are in-phase and in-quadrature with the line current. The real power is controlled using the reactive voltage and the reactive power is controlled using the real voltage. The real and reactive power reference is obtained from the steady state load flow requirements. The real power reference can also be modulated to improve damping and transient stability. The voltage at bus C can be controlled readily by directly calculating the required real voltage to be injected . The design criteria for both the proposed controller for SSSC is based on local measurable component at SSSC location. The measured real power and measured bus C voltage. Measured value is compared with steady state reference value. The error signals are used to get injected voltage real and imaginary component of V pq . The rating of SSSC is calculated by the Where, November 4 - 8, 2007, Kaohsiung, Taiwan φ 2 = tan −1 (iD / iQ ) A.1.2 Bus C voltage control The voltage at bus C of test system as shown in Fig. 4 is algebraically related to that at bus B and the reactive voltage Bus Bus Bus A B SSSC C G Line 1 Bus 1 Bus 2 T1 G1 Damping Kf. s - (2) V(ref.)+ ∑ + (3) Ka 1+ sTa Main regulator Vt 1 + s Tf + ∑ Ke Kp 1+ sTe Exciter - Vfd Fig. 3 AVR model D b damping signal - subscripts ‘D’ and ‘ Q’ denote the variable in D-Q frame.[ e pq Bus B Bus C Bus D b , e pq ]: are the components of injected voltage V pq . [ vD , vQ ] V1 d , [ vD , vQ ] and [ vD , vQ ]are the components of voltage at eP = eDpq ∗ sin(φ 2 ) + eQpq ∗ cos(φ 2 ) (7) Line Impedance Vr injected for power flow control . The voltage relation is given by: Injected reactive and real voltage are written in terms of injected voltage in D-Q frame: (6) V2 Fig. 4 Simplified SMIB system (5) eR = e ∗ cos(φ ) − e ∗ sin(φ ) e pq Vs bus B, bus C and bus D respectively. For the simplification, it is assumed that the sending end voltage Vs is constant and power at receiving end or at bus D is approximately equal to that at bus C of the test system with SSSC therefore control of power at bus C is applied. The feedback signal is readily available. Real power at bus C in D-Q frame of reference is: 2 Load 3 Fig. 2. Multi-machine 7 bus system with SSSC (4) pq Q Load 4 Load 2 SSSC Location 'c' wb is the base frequency and w0 is system frequency . The 2 Infinite Bus Line 3 (1) vQc = vQb + eQpq pq D Location'a' c differential equation for the current at bus C in the D-Q frame of reference are given by[9] : P c = vDc ∗ iD + vQc ∗ iQ SSSC T2 G2 Bus 7 Bus 6 Load 1 SSSC Location 'b' Bus 3 Bus 4 Line 2 The real power control can be achieved by control of reactive voltage component of injected voltage V pq . The d Bus 5 b A.1.1 Power flow control c Line Fig.1. Single machine infinite bus system (SMIB) The mathematical relations for power flow control and voltage of bus C control are developed with the help of simplified system shown in Fig. 4. c Infinite Bus T Vdc A.1 Conventional PI controller Q Bus D VSC product of magnitude of max line current and magnitude of max. injected voltage V pq . diD = (− R * wb / x )iD − w0iQ + ( w0 / x)(vDc − vDd ) dt diQ = (− R * wb / x)iQ − w0iD + (w0 / x )(vQc − vQd ) dt c b pq Where, vD = vD + eD (8) 377 v c = {(vDc )2 + (vQc )2 } (9) v c = {(vRb + eRpq )2 + (vPb + ePpq ) 2 } (10) vRb = VDb ∗ cos(φ 1 ) − VQb ∗ sin(φ 1 ) (11) The 14th International Conference on Intelligent System Applications to Power Systems, ISAP 2007 vPb = vDb ∗ sin(φ 1 ) + vQb ∗ cos(φ 1 ) Where W ji are the weights connecting the input to node j in (12) Since all quantities are locally available, we can easily the hidden layer, b j is the bias to the node and Woi are the pq P calculate real voltage e to be injected to obtain the desired voltage at bus C. The SSSC does not exchange any real power with the system. However some power is drawn to compensate for the losses. The DC side capacitor voltage Vdc is described by the dynamical equation: dVdc = (− gc ∗ wb / bc ) ∗ Vdc − (wb / bc ) ∗ idc dt November 4 - 8, 2007, Kaohsiung, Taiwan weights from the hidden to output layer. Four neurons in the hidden layer are chosen for training. After having training for a period of time, the training error should have converged to a value so small that, if training was to stopped , and the weights frozen, then, the neural network would continue to identify the plant ,while the operating condition remain fixed. The training of NNs is said to have reached a global minimum after changing operating conditions, as well as freezing the weights, the response of the network is still reasonably acceptable. (13) where gc and bc are the conductance and susceptance of the DC capacitor respectively. By using “(1)”-“(13)” the conventional power flow controller block diagram is IV. SYSTEM CONFIGURATION FOR PLANT developed as shown in Fig.5. The signal τ 1 & τ 2 are applied IDENTIFICATION to transformed receiving end bus reference to the bus B reference. The gain Kse & Kpe are optimized by using The ANN model as shown in Fig.7 is used to represent the nonlinear control design block set. By adjusting the gain of feedback system damping ratio can be improved. The input output mapping of SMIB system. The input vector, maximum and minimum voltage limits are chosen for the U (k ) consists of the deviation in local measurable components safe operation of SSSC under abnormal system conditions. namely: change in bus C voltage ( ∆V2 (k )) & change in real V2(ref.) PI + power at bus C ( ∆ P2 ( k )) for the SSSC control. The U(k) = + % - - τ2 Vs Kse Series PI P2(ref.) + - Y (k ) = ∆eDpq , ∆eQpq . The neural ∧ ∧ Y ne t ( k ) = f ( X ( k )) , where (X(k)) is the network output, input vector to the identifier. Consisting of three time lag of system input and output respectively, that is T X (k) = [Y(k −1),U(k −1),Y(k − 2),U(k − 2),Y(k − 3),U(k − 3)] . The error - Vs τ1 Kpe P2 vector E ( k ) used for updating weights during training is Fig. 5. Injected series voltage controller ∧ given by E (k ) = Y ( k ) - Y net ( k ) . III. MULTILAYER FEED FORWARD NETWORK (MFFN) It is well know that a MFFN with back propagation (BP) algorithm is most widely used NN model for non linear control of a power system[6]. In this paper, the MFFN consists of three layers of neurons interconnected by weight as shown in Fig. 6. The MFFN transforms n inputs to m outputs ∧ n Input layer m Output layer X2 y Xn 1 k Fig. 6: The three layers neural network (14) V. POWER SYSTEM OSCILLATIONS DAMPING CONTROLLER A damping controller is provided to improve the damping of power system oscillations [8],[9].The damping controller and the output of the network is given by k y = ∑ (Woi .hi ) Hidden layer 1 X1 through nonlinear function, f : R → R . The weights of the MFFN are trained by the error back propagation algorithm in the batch by the error back propagation algorithm in the batch mode and the hidden layer neurons in the network uses a sigmoid activation function. The out put of the node j in the hidden layer is given by: n h j = g ∑ W ji . X i + b j j =1 ( k ), P2 ( k ) ] . The output & phase angle, that is e Qpq + % 2 vector, Y ( k ) consists of change in injected voltage magnitude Voltage Controls V2 [ ∆ V2 ( k ), ∆ P2 ( k ) ] and Re f .(k)= [V e Dpq (15) i =1 378 The 14th International Conference on Intelligent System Applications to Power Systems, ISAP 2007 be considered as comprising two cascade connected blocks. The speed deviation signal is derived from the difference of measured power at SSSC location and the set mechanical input power and the error signal is integrated and multiplied by 1 M .Fig. 8 shows the block diagram of POD control. The November 4 - 8, 2007, Kaohsiung, Taiwan 10 deg. The Fig.10 shows the power –transmission angle curve for SMIB system with SSSC controlled by various controllers, keeping the injected voltage constant at 15% of the operating voltage as case study. The result indicates that the PI based SSSC increases the power handling capacity of the line by nearly 5% and the MFFN based SSSC by 7% with respect to without SSSC in the system. Vmax Ref(k) Speed deviation Y(k) Plant Kpss sTw 1+sTw 1+sT3 1+sT4 1+sT5 1+sT6 U(k) -1 U(k-1) Vfd Vmin Fig. 9. Transfer function block diagram of the PSS z -1 MFFN -1 U(k-3) Y(k-3) z _ (NN Identifier) -1 z 5 + 4.5 Ynet(k) E(k) Real power , pu U(k-2) z Y(k-2) -1 z Y(k-1) -1 z 4 3.5 3 a b 2.5 2 c 1.5 1 0.5 0 Fig. 7 Block diagram for ANN identifier 0 optimized parameters of POD controller is used, so as to achieve the desire damping ratio of the electromechanical mode and compensate for the phase shift between the control signal and the resulting electrical power deviation. The output of the damping controller modulates the reference setting of power flow controller. In Fig.5, The P2 deviation signal is replaced by P2 + u in order to include the POD for multimachine system , P2 signal is replaced by P2+ K pod (∆w1 − ∆w2) ,where, (∆w1 − ∆w2) is the difference Transmission angle , deg. Fig. 10. Power-delta curve (a)Without SSSC (b) PI based SSSC (c) MFFN based SSSC VII. SIMULATION RESULTS A. Simulation in single machine infinite bus system of speed deviations of generator1 and generator2. Fig.9 shows the block diagram of power system stabilizer (PSS). The output of PSS is applied to the AVR as an additional signal. Digital simulation are carried out by the MATLAB software. For the simulation, different loading conditions with different fault locations in the SMIB system is included. The proposed SSSC control schemes performance in terms of transient stability issue and damping of oscillations are tested on SMIB. VI. POWER –ANGLE CURVE AND STABILITY The power –delta characteristics of the transmission line in the SMIB system is developed under steady state condition of the system. The transmission line voltage profile is kept constant and the sending end source is assumed as the stiff source so that the transmission angle can be vary whereas the infinite dδ dt Kdd sTw 1+sT1 1+sTw 1+sT2 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 A1. Results under light load condition with three phase fault at bus D The digital simulation results are shown in Fig. 11 to Fig. 12 under light load condition with three phase fault of 50 ms duration at receiving end of the line. The response of the PI based SSSC is depicted in Fig.11,result indicates that PI based SSSC with POD transient stability improves and oscillations are damped within 4 sec. Whereas with PSS oscillations are reduced and with the coordinated action of PSS & POD further improve the dynamic performance of the system. The post fault rotor angle excursion is arrested by MFFN based SSSC and the oscillations are damped within 3.5 sec. The transient stability is improved and the damping control scheme improved the dynamic performance as shown in Fig. 12. u Fig. 8. Transfer function block diagram of the POD bus voltage is kept constant. For each set of transmission angle, sending end power is measured, similarly for the transmission angle varies from 0 deg. to190 deg. in the step of 379 The 14th International Conference on Intelligent System Applications to Power Systems, ISAP 2007 of generator-1 and generator-2 ( ∆w1 − ∆w2) for SSSC location ‘a’ is comparatively smaller then for location ‘b’ & ‘c’. Therefore SSSC location at ‘a’ is considered to be more favorable location in multi-machine test system. c 50 b 150 a b 0 0 1 2 Time , sec. 3 Rotor angle , deg. Rotor angle , deg. 100 November 4 - 8, 2007, Kaohsiung, Taiwan 4 100 c 50 a 0 0 Fig.11. System with PI based SSSC (a)With POD (b) With PSS (c ) With POD &PSS A2. POD & PSS under heavily loaded condition with three phase fault at bus C The simulation result of SMIB system during heavily loaded condition with three phase fault of 50 ms at the sending end of the line is shown in Fig.13 & Fig. 14. The PI based SSSC helps to maintain transient stability of the system as depicted in Fig. 13.The PSS &POD independent action reduces the oscillations. The PSS is comparable better than POD under abnormal system condition. The results shown in Fig.14 depicts the response of the system with MFFN based SSSC under heavily loaded condition ,generator output Pg = 1.01 p.u. with three phase fault of 50ms duration. Result demonstrates that the MFFN based SSSC significantly improved the transient stability. The independent action of POD & PSS provides effective damping whereas its simultaneous action proved to be the superior damping performance. The MFFN based SSSC is superior than conventional PI controller. 0 .5 1 1.5 T im e, sec . 2 2.5 3 Fig. 13. System with PI based SSSC (a) With POD (b) With PSS (c) With POD & PSS 150 Rotor angle , deg. a 100 c b 50 0 0 1 2 3 Time, sec. 4 5 Fig.14. System with MFFN based SSSC (a) With POD (b) With PSS (c) With POD & PSS -3 120 4 a Rotor angle, deg. Speed deviation (w1-w2) rad. / sec. c 100 80 b 60 40 20 0 0 1 2 3 4 x 10 2 0 b a - 2-4-6 - 8- - 0 5 1 2 3 4 5 6 7 Time (sec.) 8 9 10 Time, sec. Fig. 12. System with MFFN based SSSC (a)With POD (b) With PSS (c ) With POD &PSS Fig.15.Multi-machine system with SSSC with POD (a)PI, and (b) MFFN controller B. Simulation in multi-machine system VIII. CONCLUSIONS The proposed controllers for SSSC performance is tested in multi-machine system environment. The simulation result presents the inter area oscillations during three phase fault of 50 ms duration at receiving end of the line 3 is shown in Fig. 15. The results indicates that the difference of speed deviation is very small with MFFN based SSSC. The MFFN based SSSC is tested at three different locations ‘a’, ‘b’ & ‘c’ in the multi-machine system. The simulation result is shown in Fig.16. The result indicates that variation in speed difference This paper presents a design of PI and MFFN controllers for the SSSC control in single machine infinite bus system. The controllers comparative performance in terms of power handling capacity of the line, transient stability improvement and damping of oscillations is demonstrated under various system conditions. The proposed MFFN model performance is comparatively better than PI controller. The results indicates that the applied power system stabilizer (PSS) performance is 380 The 14th International Conference on Intelligent System Applications to Power Systems, ISAP 2007 03 H/km, C1 = 12.74e-09 F/km ; Length of line = 450 km ; Length of line 1 = 350 km ;Length of line 2 = 150 km; Length of line 3 = 450 km ; CT: 260MVA, 50e6/1e6 ,R = 0.002 ,L = 0.04,Rm = 100, Xm= 200; PT: 6MVA, 208/345e3,R= 0.002,L = 0.04 ,Rm = 100; SSSC: 5MVA, Vop= 345KV, Vpq= ± 0.3Vop, Ks= 0.9, Kd= 1.0,Kse= 0.9, Kpe = 2.3; POD: Kd=144,Tw=15,T1=0.0518e-6s,T2 = 0.0221e-6s; PSS: Kpss =20,Tw=15,T3=T5= 0.02e-6s, T4=T6 = 0.035e-6s. better than independent action of power oscillations damping (POD) control in the system. The coordinated action of POD & PSS with MFFN model provide further improvement in dynamic performance of the system. The MFFN controller demonstrate the robust dynamic performance and easy to coordinate with damping schemes. The SSSC location at position ‘a’ is favorable location for the multi-machine system. Speed deviation (w1-w2) rad. /sec. 10 10 x 8 6 4 2 0 -2 -4 -6 0 Multi-machine test system data(in p.u.) -3 Base voltage:220kv,MVA(Base):100 MVA,f = 60c/s, G1: Similar to SMIB system , Generator 2: 300MVA, 22KV, 60c/s, Rs = 0.00045 , Ls = 0.14 ,Lmd = 1.51,Lmq = 1.45,Rf = 0.000096, Lfd =0.61168, H = 2.87882s; Exc. System –1& Exc. System-2:Ka=10 ,Ta = 0.001 s , Ke = 1.0 , Te = 0.001 s, Kp1 = 1,Kp2 = 2.0 ; CT: 500MVA, 60e6/1e6, R=0.002, L=0.04,Rm=100, Xm= 200 ;PT:10MVA,208/345e3 ,R = 0.002, L = 0.04, Rm = 100; SSSC: 5MVA, Vop=220KV , Vpq= ± 0.3Vop, Ks= 0.95, Kd=1.0, Kse=0.9, Kpe = 2.5, POD:Kpod=0.6, Tw=10,T1=0.051e-6s,T2= 0.022e-6 s ; PSS: Kpss=2,Tw=10, T3=T5=0.2e- 7s, T4=T6= 0.035e-6s; Load 1: 0.15 p.u , Load 2: 0.15p.u , Load 3: 0.40+j 0.10 p.u. Load 4 : 1.0+j 0.05 p.u. c a b 1 November 4 - 8, 2007, Kaohsiung, Taiwan 2 3 4 5 6 7 Time , (sec.) 8 9 10 Fig.16. Simulation results of multi-machine system with MFFN based SSSC located at : ‘a’, ‘b’, ‘c’ IX. REFERENCES [1]A.A Edris, R Aapa, M H Baker, L Bohman, K Clark, “ Proposed terms and definitions for flexible ac transmission system (FACTS),” IEEE Trans. on Power Delivery ,Vol. 12, No.4, 1997. [2] N.G.Hingorani and L.Gyugyi, Understanding FACTS , Piscataway, NJ: IEEE Press,2001. [3]K.Sen.“SSSC- Static Synchronous Series Compensator : Theory modeling and application,” IEEE Trans. on Power Delivery, Vol.13,No.1,pp. 241-246, 1998. [4]L. Gyugyi, C. Schauder, and K.Sen, “ Static Synchronous Series Compensator : A solid state approach to the series compensation of transmission lines,” IEEE Trans. on Power Delivery, Vol.2, No.1, pp.406417,1997. [5]K.R.Padiyar, A.M.Kulkarni,“ Development and Evaluation of controls for unified Power flow controller ,”IEEE Trans. on Power Delivery, Vol.13, No.4, pp. 1348-1354 , 1998. [6]P.K.Dash, S.Mishra& G. Panda,“A radial basis function neural network controller for UPFC,” IEEE Trans. Power system, Vol.15,No.4, pp.12931299, 2000. [7]V.K.Chandrakar,A.G.Kothari, “Fuzzy logic based static synchronous series compensator(SSSC) for transient stability improvement,”2nd IEEE DRPT Hong Kong , DRPT-245,2004. [8]N.Tambey & M.L.Kothari,“Damping of power system oscillations with unified power flow controller (UPFC),” IEE Proc. Genera. Trans. Distri. Vol.150, No. 2, pp.129- 140,2003. [9]V.K.Chandrakar,A.G.Kothari, Comparison of RBFN based STATCOM,SSSC and UPFC Controllers for Transient Stability improvement, IEEE PSCE 2006, Power Systems Conference & Exposition , P.No.1369, USA,2006 [10] The Math Works Inc. Simulink Users Guide, The Math Works Press, Natick, Mass, 1992. X. APPENDIX-1 SMIB system data ( in p.u.) G: 250MVA,13.8KV, 60c/s, Rs = 0.00045 , Ls = 0.14 , Lmd = 1.51 , Lmq = 1.45 , Rf = 0.000096 , Lfd =0.61168 , H = 0.87882s ; Exc. System: Ka=2, Ta = 0.001 s, Ke = 1.0 ,Te = 0.001 s, Kp = 1; T.line: R1=0.01273 omh/km, L1 = 0.9337e381 XI. BIOGRAPHIES A.G.Kothari received the B.E. and M.Tech. degrees from Nagpur University. He obtained Ph.D. degree from Indian Institute of Technology , Kanpur. At present, he is a Professor and Head of the Electrical Engineering Department. He was Dean(Academic) from 1999 to 2002 at Visvesvaraya National Institute of Technology , Nagpur. He has number of research publications in International / National referred Journals. His research areas include Power systems and Power electronics. He is Member of the Institution of Engineers (India). V.K.Chandrakar obtained B.E degree from R.U. Raipur and M.Tech and PhD from VNIT, Nagpur . Presently working as Professor and Head of Electrical Engineering, G.H.R.C.E, Nagpur. He has number of research publications in Journals & Conferences etc. He is Member of the Institute of Engineers (India). His research interests include FACTS devices and AI.