CURENT Course Power System Coherency and Model Reduction

Transcription

1 CURENT Course Power System Coherency and Model Reduction Prof. Joe H. Chow Rensselaer Polytechnic Institute ECSE Department November 1, 2017

2 Slow Coherency A large power system usually consists of tightly connected control regions with few interarea ties for power exchange and reserve sharing The oscillations between these groups of strongly connected machines are the interarea modes These interarea modes are lower in frequency than local modes and intra-plant modes Singular perturbations method can be used to show this time-scale separation References: J H. Chow, G. Peponides, B. Avramovic, J. R. Winkelman, and P. V. Kokotovic, Time-Scale Modeling of Dynamic Networks with Applications to Power Systems, Springer-Verlag, J. H. Chow, ed., Power System Coherency and Model Reduction, Springer, c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

3 Klein-Rogers-Kundur 2-Area System Gen Gen 11 Local mode Local mode Gen Gen 12 Load 3 Interarea mode Load 13 c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

4 Coherency in 2-Area System Disturbance: 3-phase fault at Bus 3, cleared by removal of 1 line between Buses 3 and Gen 1 Gen 2 Gen 11 Gen Machine Speed (pu) Time (sec) c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

5 Power System Model An n-machine, N-bus power system with classical electromechanical model and constant impedance loads: m i δi = P mi P ei = P mi E iv j sin(δ i θ j ) x di = f i (δ,v) (1) where machine i modeled as a constant voltage E i behind a transient reactance x di m i = 2H i /Ω, H i = inertia of machine i Ω = 2πf o = nominal system frequency in rad/s damping D = 0 δ = n-vector of machine angles P mi = input mechanical power, P ei = output electrical power c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

6 Power System Model the bus voltage V j = ( ) Vjre 2 +V2 jim, θ j = tan 1 Vjim V jre (2) V jre and V jim are the real and imaginary parts of the bus voltage phasor at Bus j, the terminal bus of Machine i V = 2N-vector of real and imaginary parts of load bus voltages c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

7 Power Flow Equations For each load bus j, the active power flow balance ( ) N Vjre + jv jim P ej Real (V jre + jv jim V kre jv kim ) R k=1,k j Ljk + jx Ljk V j 2 G j = g 2j 1 = 0 (3) and the reactive power flow balance ( ) N Vjre + jv jim Q ej Imag (V jre + jv jim V kre jv kim ) R k=1,k j Ljk + jx Ljk V j 2 B j +Vj 2 B Ljk 2 = g 2j = 0 (4) R Ljk, X Ljk, and B Ljk are the resistance, reactance, and line charging, respectively, of the line j-k P ej and Q ej are generator active and reactive electrical output power, respectively, if bus j is a generator bus G j and B j are the load conductance and susceptance at bus j Note that j denotes the imaginary number if it is not used as an index. c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

8 Electromechanical Model M δ = f(δ,v), 0 = g(δ,v ) (5) M = diagonal machine inertia matrix f = vector of acceleration torques g = power flow equation c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

9 Linearized Model Linearize (5) about a nominal power flow equilibrium (δ o,v o ) to obtain M δ = f(δ,v) δ + f(δ,v) δo,v o V = K 1 δ+k 2 V (6) δo,v o 0 = g(δ,v ) δ + g(δ,v) δo,v o V = K 3 δ +K 4 V (7) δo,v o δ = n-vector of machine angle deviations from δ o V = 2N-vector of the real and imaginary parts of the load bus voltage deviations from V o K 1, K 2, and K 3 are partial derivatives of the power transfer between machines and terminal buses, K 1 is diagonal K 4 = network admittance matrix and nonsingular. the sensitivity matrices K i can be derived analytically or from numerical perturbations using the Power System Toolbox c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

10 Linearized Model Solve (6) for to obtain where V = K 1 4 K 3 δ (8) M δ = K 1 K 2 K 1 4 K 3 δ = K δ (9) K ij = E i E j (B ij cos(δ i δ j ) G ij sin(δ i δ j )) δo,v o, i j (10) and G ij +jb ij is the equivalent admittance between machines i and j. Furthermore, n K ii = K ij (11) j=1,j i Thus the row sum of K equals to zero. The entries K ij are known as the synchronizing torque coefficients. c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

11 Slow Coherent Areas Assume that a power system has r slow coherent areas of machines and the load buses that interconnect these machines Define δ α i = deviation of rotor angle of machine i in area α from its equilibrium value m α i = inertia of machine i in area α Order the machines such that δ α i from the same coherent areas appears consecutively in δ. c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

12 Weakly Coupled Areas We attribute the slow coherency phenomenon to be primarily due to the connections between the machines in the same coherent areas being stiffer than those between different areas, which can be due to two reasons: The admittances of the external connections Bij E much smaller than the admittances of the internal connections Bpq I ε 1 = BE ij B I pq (12) where E denotes external, I denotes internal, and i,j,p,q are bus indices. This situation also includes heavily loaded high-voltage, long transmission lines between two coherent areas. c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

13 The number of external connections is much less than the number of internal connections ε 2 = γe γ I (13) where γ E = max α {γe α }, γi = min{γ I α α }, α = 1,...,r (14) γα E = (the number of external connections of area α)/nα γα I = (the number of internal connections of area α)/n α where N α is the number of buses in area α. c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

14 Internal and External Connections For a large power system, the weak connections between coherent areas can be represented by the small parameter Separate the network admittance matrix into ε = ε 1 ε 2 (15) K 4 = K I 4 +ǫk E 4 (16) where K4 I = internal connections and KE 4 = external connections. The synchronizing torque or connection matrix K is where K = K 1 K 2 (K I 4 +ε(ki 4 )) 1 K 3 = K 1 K 2 (K I 4 (I +ε(ki 4 ) 1 K E 4 )) 1 K 3 = K I +ǫk E (17) K I = K 1 K 2 (K I 4) 1 K 3, K E = K 2 K E 4εK 3 (18) In the separation (17), the property that each row of K I sums to zero is preserved. c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

15 Slow Variables Define for each area an inertia weighted aggregate variable y α = n α i=1 m α i δα i /mα, α = 1,2,...,r (19) where m α i = inertia of machine i in area α, n α = number of machines in area α, and m α = n α i=1 is the aggregate inertia of area α. m α i, α = 1,2,...,r (20) c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

16 Denote by y = the r-vector whose αth entry is y α. The matrix form of (19) is where y = C δ = M 1 a U T M δ (21) U = blockdiag(u 1,u 2,...,u r ) (22) is the grouping matrix with n α 1 column vectors [ ] T u α = , α = 1,2,...,r (23) M a = diag(m 1,m 2,...,m r ) = U T MU (24) is the r r diagonal aggregate inertia matrix. c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

17 Fast Variables Select in each area a reference machine, say the first machine, and define the motions of the other machines in the same area relative to this reference machine by the local variables z α i 1 = δα i δ α 1, i = 2,3,...,n α, α = 1,2,...,r (25) Denote by z α the (n α 1)-vector of z α i and conside z α as the αth subvector of the (n r)-vector z. Eqn. (25) in matrix form is z = G δ = blockdiag(g 1,G 2,...,G r ) δ (26) where G α is the (n α 1) n α matrix G α = (27) c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

18 Slow and Fast Variable Transformation A transformation of the original state δ into the aggregate variable y and the local variable z [ ] y [ ] C z = G δ (28) The inverse of this transformation is ( )[ y δ = U G + z ] (29) where is block-diagonal. G + = G T (GG T ) 1 (30) c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

19 Slow Subsystem Apply the transformation (28) to the model (9), (17) where M a ÿ = ǫk a y +ǫk ad z M d z = ǫk da y +(K d +ǫk dd )z (31) M d = (GM 1 G T ) 1, K a = U T K E U K da = U T K E M 1 G T M d, K da = M d GM 1 K E U K d = M d GM 1 K I M 1 G T M d,k dd = M d GM 1 K E M 1 G T M d (32) K a, K ad, and K da are independent of K I System (31) is in the standard singularly perturbed form ε is both the weak connection parameter and the singular perturbation parameter Neglecting the fast dynamics, the slow subsystem is M a ÿ = εk a y (33) c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

20 Formulation including Power Network Apply (28) to the model (6)-(7) where M a ÿ = K 11 y +K 12 z +K 13 V M d z = K 21 y +K 22 z +K 23 V 0 = K 31 y +K 32 z +(K I 4 +εk E 4 ) V (34) K 11 = U T K 1 U, K 12 = U T K 1 G +, K 13 = U T K 2, K 21 = (G + ) T K 1 U K 22 = (G + ) T K 1 G +, K 23 = (G + ) T K 2, K 31 = K 3 U, K 32 = K 3 G + (35) Eliminating the fast variables, the slow subsystem is M a ÿ = K 11 y +K 13 V 0 = K 31 y +K 4 V (36) This is the inertial aggregate model which is equivalent to linking the internal nodes of the coherent machines by infinite admittances. c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

21 2-Area System Example Connection matrix K = Decompose K into internal and external connections K I = εk E = (37) (38) (39) c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

22 EM Modes in 2-Area System λ(m 1 K) = 0, , , (40) with the corresponding eigenvector vectors v 1 = 0.5, v = , v = , v = Interarea mode: = ±j3.779 rad/s Local modes: = ±j7.795 rad/s and = ±j7.890 rad/s (41) c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

23 Slow and Fast Subsystems of 2-Area System Slow dynamics: M a = 1 2π 60 [ ] [ , εk a = ] (42) The eigenvalues of Ma 1 K a are 0 and an interarea mode frequency of = ±j3.816 rad/s. Fast local dynamics: [ ] [ ] M d =, K 2π d = (43) The eigenvalues of M 1 d K d are and local modes of ±j and ±j rad/s. c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

24 Finding Coherent Groups of Machines 1 Compute the electromechanical modes of an N-machine power 2 Select the (interarea) modes with frequencies less than 1 Hz 3 Compute the eigenvectors (mode shapes) of these slower modes 4 Group the machines with similar mode shapes into slow coherent groups 5 These slow coherent groups have weak or sparse connections between them c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

25 Use V s1 to form a new coordinate system c J. H. Chow (RPI-ECSE) Coherency November 1, / 29 Grouping Algorithm Plot rows of the slow eigenvector V s of the interarea modes Use Gaussian elimination to select the r most separated rows as reference vectors and group them into V s1 and reorder V s as [ ] [ ] [ ] [ ] [ ] Vs1 Vs1 I I 0 V s = Vs1 1 = = + L O(ε) V s2 V s2 L g (44)

26 Coherent Groups in 2-Area System V s = Then Gen 1 Gen 2 Gen 11 Gen 12, V s1 = 1 0 V s V s = [ Gen 1 Gen 11 Gen 2 Gen 12 ] Gen 1 Gen 11 (45) (46) c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

27 Coherency for Load Buses V θ = Bus 1 Bus 2 Bus 3 Bus 10 Bus 11 Bus 12 Bus 13 Bus 20 Bus 101 Bus 110 Bus 120 V θ V 1 s1 = (47) Bus 101 c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

28 17-Area Partition of NPCC 48-Machine System c J. H. Chow (RPI-ECSE) Coherency November 1, / 29

29 EQUIV and AGGREG Functions for Power System Toolbox L group: grouping algorithm coh-map, ex group: tolerance-based grouping algorithm Podmore: R. Pormore s algorithm of aggregating generators at terminal buses i agg: inertial aggregation at generator internal buses slow coh: slow-coherency aggregation, with additional impedance corrections These functions use the same power system loadflow input data files as the Power System Toolbox. c J. H. Chow (RPI-ECSE) Coherency November 1, / 29