Solution to Application Example 7.11

Calculation of the inverse of J¯0si370_e, that is. J¯01si371_e.

J¯01=AdjJ¯0detJ¯0,

si372_e

where Adj J¯0si373_e is the adjoint of matrix J¯0si374_e and detJ¯0si375_e = |J¯0si376_e| is the determinant of J¯0si377_e.

The adjoint of J¯0si378_e is obtained by first taking the transpose of J¯0si379_e, that is, J¯0tsi380_e, and then replacing each element in the transpose with its cofactor:

J¯0tij=J¯0ji.

si381_e

The transpose of J¯0si382_e is

J¯0t=J11J21J31J41J51J61J12J22J32J42J52J62J13J23J33J43J53J63J14J24J34J44J54J64J15J25J35J45J55J65J16J26J36J46J56J66

si383_e

J¯0t=61.6852.9411.7642.9440052.93261.9452.94411.7640011.7642.94451.77222.92640.007619.9822.94411.76422.91751.77319.98240.00760040.007619.98280.01539.9630019.9840.007639.96180

si384_e

AdjJ¯0=+AdjJ¯011AdjJ¯012+AdjJ¯013AdjJ¯014+AdjJ¯015AdjJ¯016AdjJ¯021+AdjJ¯022AdjJ¯023+AdjJ¯024AdjJ¯025+AdjJ¯026+AdjJ¯031AdjJ¯032+AdjJ¯033AdjJ¯034+AdjJ¯035AdjJ¯036AdjJ¯041+AdjJ¯042AdjJ¯043+AdjJ¯044AdjJ¯045+AdjJ¯046+AdjJ¯051AdjJ¯052+AdjJ¯053AdjJ¯054+AdjJ¯055AdjJ¯056AdjJ¯061+AdjJ¯062AdjJ¯063+AdjJ¯064AdjJ¯065+AdjJ¯066

si385_e

where

AdjJ¯011=J22J32J42J52J62J23J33J43J53J63J24J34J44J54J64J25J35J45J55J65J26J36J46J56J66,AdjJ¯012=J12J32J42J52J62J13J33J43J53J63J14J34J44J54J64J15J35J45J55J65J16J36J46J56J66,

si386_e

AdjJ¯013=J12J22J42J52J62J13J23J43J53J63J14J24J44J54J64J15J25J45J55J65J16J26J46J56J66,AdjJ¯014=J12J22J32J52J62J13J23J33J53J63J14J24J34J54J64J15J25J35J55J65J16J26J36J56J66,

si387_e

AdjJ¯015=J12J22J32J42J62J13J23J33J43J63J14J24J34J44J64J15J25J35J45J65J16J26J36J46J66,AdjJ¯016=J12J22J32J42J52J13J23J33J43J53J14J24J34J44J54J15J25J35J45J55J16J26J36J46J56,

si388_e

AdjJ¯021=J21J31J41J51J61J23J33J43J53J63J24J34J44J54J64J25J35J45J55J65J26J36J46J56J66,AdjJ¯022=J11J31J41J51J61J13J33J43J53J63J14J34J44J54J64J15J35J45J55J65J16J36J46J56J66,

si389_e

AdjJ¯023=J11J21J41J51J61J13J23J43J53J63J14J24J44J54J64J15J25J45J55J65J16J26J46J56J66,AdjJ¯024=J11J21J31J51J61J13J23J33J53J63J14J24J34J54J64J15J25J35J55J65J16J26J36J56J66,

si390_e

AdjJ¯025=J11J21J31J41J61J13J23J33J43J63J14J24J34J44J64J15J25J35J45J65J16J26J36J46J66,AdjJ¯026=J11J21J31J41J51J13J23J33J43J53J14J24J34J44J54J15J25J35J45J55J16J26J36J46J56,

si391_e

AdjJ¯031=J21J31J41J51J61J22J32J42J52J62J24J34J44J54J64J25J35J45J55J65J26J36J46J56J66,AdjJ¯041=J21J31J41J51J61J22J32J42J52J62J23J33J43J53J63J25J35J45J55J65J26J36J46J56J66,

si392_e

AdjJ¯051=J21J31J41J51J61J22J32J42J52J62J23J33J43J53J63J24J34J44J54J64J26J36J46J56J66,AdjJ¯061=J21J31J41J51J61J22J32J42J52J62J23J33J43J53J63J24J34J44J54J64J25J35J45J56J65.

si393_e

As an example, the calculation of the first entry of the adjonint AdjJ¯011si394_e will be outlined:

AdjJ¯011=J22J32J42J52J62J23J33J43J53J63J24J34J44J54J64J25J35J45J55J65J26J36J46J56J66=61.9452.94411.764002.94451.77222.92640.007619.98211.76422.91751.77319.98240.0076040.007619.98280.01539.963019.9840.007639.96180.0

si395_e

AdjJ¯011=61.94551.77222.92640.007619.98222.91751.77219.98240.007640.007619.98280.01539.96319.9840.007639.96180+2.9442.94422.91740.007619.9811.76451.77319.98240.0076019.98280.01539.963040.007639.9618011.7642.94451.77240.007619.9811.76422.91719.98240.0076040.007680.01539.963019.98239.96180.

si396_e

The determinant is

detJ¯=J11J22J23J24J25J26J32J33J34J35J36J42J43J44J45J46J52J53J54J55J56J62J63J64J65J66J12J21J23J24J25J26J31J33J34J35J36J41J43J44J45J46J51J53J54J55J56J61J63J64J65J66+J13J21J22J24J25J26J31J32J34J35J36J41J42J44J45J46J51J52J54J55J56J61J62J64J65J66J14J21J22J23J25J26J31J32J33J35J36J41J42J43J45J46J51J52J53J55J56J61J62J63J65J66+J15J21J22J23J24J26J31J32J33J34J36J41J42J43J44J46J51J52J53J54J56J61J62J63J64J66J16J21J22J23J24J25J31J32J33J34J35J41J42J43J44J45J51J52J53J54J55J61J62J63J64J65.

si397_e

The J1101si398_e entry is nowJ1101=AdjJ¯0detJ¯0si399_e.

7.3.9 Application Example 7.12: Inversion of a 3 × 3 Matrix

In order to demonstrate the entire procedure for the computation of the inverse of a matrix the following 3 × 3 matrix will be used

A¯=131242311.

si400_e

Solution to Application Example 7.12

Transpose of Ā is

A¯t=123341121.

si401_e

The cofactors of Āt are

A11t=11+14121=42=6,

si402_e

A12t=11+23111=3+1=4,

si403_e

A13t=11+33412=64=2,

si404_e

A21t=12+12321=2+6=4,

si405_e

A22t=12+21311=1+3=4,

si406_e

A23t=12+31212=2+2=4,

si407_e

A31t=13+12341=2+24=14,

si408_e

A32t=13+21331=19=8,

si409_e

A33t=13+31234=46=10.

si410_e

The determinant detA¯=A¯si411_e is

|A¯|=142113223112431=426+18212=26+18=8.

si412_e

The inverse of Ā is now

A¯1=AdjA¯A¯=1864244414810.

si413_e

7.3.10 Application Example 7.13: Computation of the Correction Voltage Vector

For the system of Fig. E7.8.1, compute the correction voltage vectorΔx¯0si414_e, updating the mismatch power vector, and check the convergence of the load flow algorithm (using a convergence tolerance of 0.0001).

Solution to Application Example 7.13

Based on the results of Application Examples 7.9 to 7.11, the correction voltage vector Δx¯0si415_e can be computed from

Δx¯0=J¯01ΔW¯x¯0=(2.68E4,2.3E3,2.79E3,3.57E3,3.39E3,4.03E3)t.

si416_e

The updated mismatch power vector is

ΔW¯1=(1.41E4,3.01E4,9.63E6,7.62E5,1.20E3,5.96E4)t.

si417_e

Since the components of ΔW¯1si418_e are larger than the specified convergence tolerance (e.g., 0.0001), the Newton–Raphson solution has not converged. However, it will converge in two iterations:

Fundamental power flow iteration summary (Fig. E7.8.1):

Absolute real power mismatchAbsolute reactive power mismatch
IterationAverage (%)Worst (%)Worst busAverage (%)Worst (%)Worst bus
011.672546.6710.004
10.050.1240.030.064
20.000.0040.000.004

t0035

Fundamental power flow output solution (Fig. E7.8.1), where δ angles are rounded (approximate) values:

Bus voltageBus generation (G) and load (L) powersLine powera
From bus|V|
(%)
δ
(°)
PG
(%)
QG
(%)
PL
(%)
QL
(%)
To
bus
Pline
(%)
Qline (%)
1100.00.0035.0920.150.000.00213.3510.69
421.739.47
299.76–0.020.000.0010.0010.001–13.32–10.66
33.320.66
399.64–0.160.000.000.000.002–3.32–0.65
43.320.65
499.59–0.200.000.0025.0010.003–3.32–0.65
1–21.68–9.35

t0040

a Active and reactive powers in the line between “from bus” and “to bus.”

7.4 Newton-based harmonic power flow

Classifications of solution approaches for the harmonic power (load) flow problem are summarized at the end of this chapter. In this section, the Newton-based harmonic power flow [5,6] is selected, analyzed, and thoroughly illustrated. There are two main reasons for this selection:

 The Newton-based harmonic power flow is relatively easy to understand and to implement since it is an extension of the conventional fundamental Newton–Raphson formulation; it is based on similar terminologies and equations. It is capable of including any type of nonlinear load assuming its v–i characteristic is available in the frequency (or time) domain.

 It was the first proposed approach for power system modeling with nonlinear loads and harmonic couplings [5,6].

Among the limitations of the Newton-based harmonic power flow in its present form is the assumption of balanced network and load conditions.

7.4.1 Harmonic Bus Admittance Matrix and Power Definitions

For nonsinusoidal operating conditions power system matrices will be defined at harmonic frequencies. These matrices include the harmonic bus admittance matrix Y¯bush,si419_e the harmonic bus impedance matrix Z¯bushsi420_e = 1/Y¯bushsi421_e, and the harmonic Jacobian J¯hsi422_e matrix.

The harmonic bus admittance matrix Y¯bushsi423_e is identical to the fundamental bus admittance matrix Y¯bush=Y¯bussi424_e with the difference that line admittances are evaluated at the hth harmonic frequency fh = h · f1:

Y¯bush=y11hy12h...y1nhy21hy22h...y2nh......yn1hyn2h...ynnh,

si425_e  (7-83)

where ωh = 2πhf1 = h · 377 rad/s.

Power Definitions at Harmonic Frequencies

Consider a voltage v(t) and a current i(t) expressed in terms of their rms harmonic components

vt=V0+h=12Vhsinhω1t+θhv,

si426_e  (7-84)

it=I0+h=12Ihsinhω1t+θhi.

si427_e  (7-85)

Then

 the active (real, average) power is given by

P=V0I0+h=1VhIhcosθhvθhi,

si428_e  (7-86)

 the reactive power is given by

Q=h=1VhIhsinθhvθhi,

si429_e  (7-87)

 and the apparent (voltampere) power is

S=h=0Ih2h=0Vh2.

si430_e  (7-88)

In case of sinusoidal v(t) and i(t) (h = 1 only), we have

S2=P2+Q2.

si431_e  (7-89)

For nonsinusoidal cases that relation does not hold and must be replaced by

D=S2P2Q2,

si432_e  (7-90)

where D is called the distortion power. Other definitions have also been proposed for harmonic powers, which are not discussed here.

Classification of Time Harmonics

Time harmonics can be classified into three types (Table 7.2):

Table 7.2

Classification of Time Harmonics

Positive (+) sequenceNegative (–) sequenceZero (0) sequence
123
456
789
101112
131415
161718
192021
si6_esi7_esi8_e

 positively rotating harmonics (+),

 negatively rotating harmonics (–), and

 zero-sequence harmonics (0).

Triplen (multiples of three) harmonics (3h, for h = 1, 2, 3, …) in a balanced three-phase system are predominantly of the zero-sequence type. With a few exceptions (discussed in preceding chapters), zero-sequence currents cannot flow from the network into a delta or an ungrounded wye-connected device (e.g., transformer), and therefore triplen harmonics are excluded from this analysis.

7.4.2 Modeling of Nonlinear and Linear Loads at Harmonic Frequencies

Before proceeding with the reformulation of the Newton–Raphson load flow to include harmonic frequencies, it is necessary to properly model linear and nonlinear loads at fundamental and harmonic frequencies.

Harmonic Modeling of Nonlinear Loads

The (vi) relationship of nonlinear loads (such as line-commutated converters) will be modeled as coupled harmonic current sources. The injected harmonic currents of a nonlinear load (Iinjected(h)) at bus m will be a function of its fundamental and harmonic voltages:

RealIinjectedh=gr,mhV~m1,V~m5,V~m7,..,V~mL,αm,βmImagIinjectedh=gi,mhV~m1,V~m5,V~m7,..,V~mL,αm,βm

si433_e  (7-91)

where αm and βm are the nonlinear load control parameters and L is the maximum harmonic order considered L = hmax. The currents of Eq. 7-91 are referred to the nonlinear bus m to which the nonlinear load is connected. The nonlinearity of g~mh=gr,mh+jgi,mhsi434_e is due to strong couplings between harmonic currents and voltages dictated by the nonlinear load. In some cases (e.g., decoupled harmonic power flow formulation), it is convenient to ignore harmonic couplings g~mhV~mhαmβmsi435_e and simplify the problem.

As an example, for a full-wave bridge rectifier connected to bus m (shown in Fig. 7.13), αm is the firing angle of semiconductor-controlled rectifiers (SCRs), and βm is either the commutating impedance (inductance) or the DC voltage (E). Note that the equivalent circuit models a three-phase full-wave rectifier/inverter with a general load, where Lcom is the commutating inductance, that is, the leakage inductance of the transformer, and

f07-13-9780128007822
Figure 7.13 Line-commutated converter with general load.

 for a passive load, only resistance R and filter inductance F are considered,

 for a DC motor drive, only R and E are considered (E > 0) when motoring, and

 for inverter operation with a DC generator, only R and E are considered (E < 0).

Modeling Linear Loads at Harmonic Frequencies

For the formulation of harmonic power flow, equivalent circuits of linear loads are required. At the fundamental frequency, linear loads are modeled as conventional PQ and PV buses. However, shunt admittances are used to model them at harmonic frequencies [7,8]. The admittance of a linear load connected to bus k at the hth harmonic is

y~kh=Pk1jQk1/h|V~k1|2.

si436_e  (7-92)

where j represents the operator 1si437_e.

Capacitor banks are modeled as a fixed shunt reactance. Transformers are approximated by linear leakage inductances, and their nonlinearities and losses due to eddy currents, hysteresis, and saturation are neglected.

7.4.3 The Harmonic Power Flow Algorithm (Assembly of Equations)

Power system components, loads, and generators that produce time harmonics in an otherwise harmonic-free network are termed nonlinear. Other buses such as the standard PQ and PV buses with no converter or other nonlinear devices connected are termed linear. Consider

 Bus 1 to be the conventional swing (slack) bus with specified values for voltage magnitude |V~1si438_e| and phase angle δ1.

 Buses 2 through (m – 1) to be the conventional linear PQ (or PV) buses. As with the conventional power flow problem, active Pi(1) and reactive Qi(1) powers are assumed to be known at these linear buses.

 Buses m through n to be nonlinear buses. At these buses, the fundamental real power Pi(1) and the total apparent power Si,total (or the total reactive power Qi,total) are specified.

Network voltages and currents are represented by Fourier series components, and the bus voltage vector x¯si439_e of the fundamental power flow (Eq. 7-42) is redefined as

U¯=[δ21,|V~21|,,δn1,|V~n1|,δ15,|V~15|,,δn5,|V~n5|,,δ1L,|V~1L|,,δnL,|V~nL|,αm,βm,,αn,βn]t=V¯1V¯5V¯LΦ¯t

si440_e  (7-93)

where (superscripts and subscripts denote harmonic orders and bus numbers, respectively)

FundamentalvoltagesareV¯1=δ21,|V~21|,,δn1,|V~n1|,

si441_e  (7-94a)

FifthharmonicvoltagesareV¯5=δ15,|V~15|,,δn5,|V~n5|,

si442_e  (7-94b)

LthharmonicvoltagesareV¯L=δ1L,|V~1L|,,δnL,|V~nL|,

si443_e  (7-94c)

NonlineardevicevariablesvectorisΦ¯=αm,βm,,αn,βn.

si444_e  (7-95)

Harmonic bus voltage magnitudes, phase angles, and the nonlinear device parameters (αm, βm) are unknowns; therefore the conventional formulation (forcing the fundamental mismatch powers to zero) is not sufficient to solve the harmonic power flow problem. Three additional relations are used:

 Kirchhoff’s current law for the fundamental frequency (fundamental current balance) at nonlinear buses.

 Kirchhoff’s current law for each harmonic (harmonic current balance) at all buses.

 Conservation of apparent voltamperes (apparent power or voltampere balance) at nonlinear buses.

Note that Kirchhoff’s current law for the fundamental frequency is not applied to linear buses since active and reactive power (mismatch power) balance has already been applied at these buses.

Current Balance for Fundamental Frequency

The current balance is written for fundamental frequency (h = 1) at the nonlinear buses as indicated in Fig. 7.14, where the nonlinear currents gr,m(1)gi,m(1) are referred to bus m. Therefore one obtains for the nonlinear buses (from bus m to bus n) for the fundamental current balance:

Ir,m1Ii,m1...Ir,n1Ii,n1=gr,m1V~m1V~m5V~mLαmβmgi,m1V~m1V~m5V~mLαmβm...gr,n1V~n1V~n5V~nLαnβngi,n1V~n1V~n5V~nLαnβn,

si445_e  (7-96)

where Ir,m(1) and Ii,m(1) are the fundamental real and reactive line currents at the mth nonlinear bus, respectively, and gr,m(1) and gi,m(1) are the fundamental real and reactive injected load currents at the same bus. Note that the line and injected load currents are positive if they leave the bus.

f07-14-9780128007822
Figure 7.14 Application of current balance Ir,m1+jIi,m1+gr,m1+jgi,m1=0si3_e at a nonlinear bus.
Current Balance at Harmonic Frequencies

For all (linear and nonlinear) buses one can write

Ir,1hIi,1h...Ir,m1hIi,m1hIr,mhIi,mh...Ir,nhIi,nh=0..swingbusandlinearbuses..0gr,nhV~m1V~m5V~mLαmβmgi,mhV~m1V~m5V~mLαmβm.nonlinearbuses.gr,nhV~n1V~n5V~nLαnβngi,nhV~n1V~n5V~nLαnβn

si446_e  (7-97)

The line currents Ir,m(h) and Ii,m(h) will be computed in the next section.

Apparent Power or Voltampere Balance

The third type of relation is the apparent power (voltampere) balance at each nonlinear bus j:

Sj2=Pjt2+Qjt2+Dj2,

si447_e  (7-98)

where j = m, …, n and

Pjt2=hPjh2=Vj1Ij1cosθj1+Vj5Ij5cosθj5+Vj7Ij7cosθj7+)2

si448_e  (7-99a)

Qjt2=hQjh2=Vj1Ij1cosθj1+Vj5Ij5sinθj5+Vj7Ij7sinθj7+)2

si449_e  (7-99b)

are the total (fundamental plus harmonics) real and reactive load powers at nonlinear buses. Dm to Dn can be computed from gr,m, gi,m, …, gr,n, gi,n; therefore, distortion powers are not independent variables. Equation 7-98 is used to update (between iterations) the nonlinear-device injected powers P and Q at the nonlinear buses.

The number of unknowns for the n buses are as follows (K = number of harmonics considered not including fundamental, e.g., for h = 1, 5, 7, 11 we get K = 3):

 Fundamental bus voltage magnitudes and phases for all buses but the swing bus

2(n – 1)

 Fundamental real and reactive powers at the swing bus (losses P1, Q1)

2

 Harmonic (not including fundamental) voltage magnitudes and phases at all (total) n buses

2nK

 Total reactive powers Qjt at each nonlinear bus, where n is the number of all buses and (m – 1) the number of linear buses

n – (m – 1) = (n – m + 1)

 Two variables (αj, βj) associated with each nonlinear bus

2(nm + 1)

 Total number of unknowns

2n(K + 1) + 3(n – m + 1)

t0045

The number of available equations are as follows:

 Fundamental real and reactive power balance at all (m – 1) linear buses but the swing bus

2(m – 2)

 Fundamental voltage magnitude and phase at the swing bus (|V~1si450_e| = 1 pu, δ1 = 0 radians)

2

 Fundamental real and imaginary current balance at
n – (m – 1) nonlinear buses (at (m – 1) linear buses the Pi(1) and Qi(1)are specified)

2(n – (m – 1)) = 2(nm + 1)

 Harmonic real and reactive current (except the fundamental) balance at all n buses

2nK

 Apparent voltampere balance at n – (m – 1) nonlinear buses to compute reactive power (if reactive power is specified at nonlinear buses, however, this apparent voltampere balance is not required at nonlinear buses)

(n – m + 1)

 Total number of equations

2n(K + 1) + (n – m + 1)

t0050

There remain 2(nm + 1) equations to be defined. These can be the total real and reactive mismatch powers at the n – (m – 1) nonlinear buses:

ΔPjnonlinear=Pjt+hFr,jh,

si451_e  (7-100a)

ΔQjnonlinear=Qjt+hFi,jh,

si452_e  (7-100b)

where j = m, …, n and Pjt and Qjt are the total (fundamental and harmonics) nonlinear load real and reactive powers at bus j, respectively. hFr,jhsi453_e and hFi,jhsi454_e (for h = 1, 5, 7, …, L) are the total real and reactive line powers, respectively. Note that the formulas for Fr,j(h) and Fi,j(h) are identical to those for Fr,j and Fi,j of the fundamental power flow problem with the difference that Y¯bushsi455_e and the hth harmonic voltages must be used. The application of the newly defined mismatch powers at the nonlinear buses results in 2(n – m + 1) additional equations, and the number of equations is identical to the number of unknowns.

7.4.4 Formulation of Newton–Raphson Approach for Harmonic Power Flow

The mismatch vector (consisting of mismatch power and mismatch currents) for harmonic power flow is defined as

ΔM¯=ΔW¯,ΔI¯5,,ΔI¯L,ΔI¯1t,

si456_e  (7-101)

where ΔW¯si457_e is the mismatch power vector (Eq. 7-56) and (ΔĪ(5), …, ΔĪ(L)ΔĪ(1)) is the mismatch current vector for the harmonics including the fundamental.

In Eq. 7-101,

ΔW¯=[P21+Fr,21,Q21+Fi,21,,Pm11+Fr,m11,Qm11+Fi,m11,ΔPmnonlinear,ΔQmnonlinear,,ΔPnnonlinear,ΔQnnonlinear]t

si458_e  (7-102)

where P21+Fr,21,Q21+Fi,21,,Pm11+Fr,m11,Qm11+Fi,m11si459_e applies to linear buses, and ΔPmnonlinear, ΔQmnonlinear, …, ΔPnnonlinear, ΔQnnonlinear applies to nonlinear buses. Pj(1) and Qj(1) are the real and reactive fundamental load powers for the linear bus j (as defined by the fundamental load flow analysis), respectively; Fr,j(1) and Fi,j(1) are the (line) fundamental real and reactive powers at the linear bus j (see fundamental load flow analysis), respectively.

The fundamental current mismatch is defined for nonlinear buses where all currents (e.g., line currents and nonlinear load currents) are referred to the swing bus:

ΔI¯1=[Ir,m1+Gr,m1,Ii,m1+Gi,m1,,Ir,n1+Gr,n1,Ii,n1+Gi,n1]t,

si460_e  (7-103)

and the harmonic current mismatch is defined for linear and nonlinear buses including swing bus:

ΔI¯h=[Ir,1h,Ii,1h,,Ir,m1h,Ii,m1h,Ir,mh+Gr,mh,Ii,mh+Gi,mh,,Ir,nh+Gr,nh,Ii,nh+Gi,nh]t,

si461_e  (7-104)

where the nonlinear load current components Gr,m(h) and Gi,m(h) are given (referred to swing bus) and the line current components Ir,m(h) and Ii,m(h) will be generated in the next section. The Newton–Raphson method is implemented by forcing the appropriate mismatches, ΔM¯si462_e, to zero using the Jacobian matrix J¯si463_e and obtaining appropriate correction terms ΔU¯ξ=U¯ξU¯ξ+1si464_e, where ξ represents the iteration number:

ΔM¯=J¯ΔU¯ξ.

si465_e  (7-105)

The Jacobian J¯si466_e is a 2(nK + n – 1) + 2(nm + 1) by 2(nK + n – 1) + 2(nm + 1) matrix, where n is the total number of buses, (m – 1) is the number of linear buses including the swing bus, and K is the number of harmonics considered, excluding the fundamental (h = 1):

ΔM¯=ΔW¯ΔI¯5ΔI¯7...ΔI¯LΔI¯1=J¯1J¯5..J¯L0YG¯(5,1)YG¯(5,5)..YG¯5LH¯5YG¯(7,1)YG¯(7,5)..YG¯7LH¯7..YG¯L1YG¯(7,5)..YG¯LLH¯LYG¯(1,1)YG¯(1,5)..YG¯1LH¯1ΔV¯1ΔV¯5ΔV¯7...ΔV¯LΔΦ¯

si467_e  (7-106)

Solving the equation for the correction vector yields

ΔU¯=ΔV¯1ΔV¯5ΔV¯7...ΔV¯LΔΦ¯=J¯1J¯5..J¯L0YG¯(5,1)YG¯(5,5)..YG¯5LH¯5YG¯(7,1)YG¯(7,5)..YG¯7LH¯7...........YG¯L1YG¯(7,5)..YG¯LLH¯LYG¯(1,1)YG¯(1,5)..YG¯1LH¯1ΔW¯ΔI¯5ΔI¯7...ΔI¯LΔI¯1,

si468_e  (7-107)

where all elements in this matrix equation are subvectors and submatrices partitioned from ΔM¯,J¯si469_e and ΔŪ(ζ).

The subvectors are

 ΔV¯1,ΔV¯5,,ΔΦ¯tsi470_e= correction vector,

 ΔW¯si471_e = mismatch power vector at all buses except the swing bus,

 ΔĪ(1) = fundamental mismatch current vector at nonlinear buses, and

 ΔĪ(h) = hth harmonic mismatch current vector at all buses.
The submatrices are

 J¯1si472_e = Jacobian 2(n – 1) square matrix corresponding to Jacobian of fundamental power flow written for all buses except the swing bus, and

 J¯hsi473_e = Jacobian of harmonic h ≠ 1 with dimensions 2(n – 1) × 2n, whereby there are zero entries for (m – 1) linear buses:

J¯h=0¯2m2×2nzeroentriesforlinearbusespartialderivativesofthehthharmonicmismatchtotalPandQatnonlinearbuseswithrespecttoVhandδh.Thesearesimilarlyformedasthoseforthefundamentalpowerflow.

si474_e  (7-108)

where 0¯2m2×2nsi475_e denotes a 2(m – 2) by 2n array of zeros and J¯hsi476_e is a 2(n – 1) by 2n matrix.

The partial derivatives are found using the harmonic admittance matrix Y¯bushsi477_e and the harmonic voltages:

YG¯hj=Y¯hh+G¯hhforh=j0¯+G¯hjforhj

si478_e  (7-109)

or

YG¯hj=Y¯hhG¯hhY¯hj0G¯hj.

si479_e  (7-110)

 Y¯hhsi480_e is an array of partial derivatives of the hth harmonic (linear) line currents with respect to the hth harmonic bus voltages,

 G¯hhsi481_e is an array of partial derivatives of the hth (nonlinear) harmonic load currents with respect to the hth harmonic bus voltages, and

 G¯hjsi482_e is an array of partial derivatives of the hth harmonic nonlinear load currents with respect to the jth harmonic bus voltages:

G¯hj=0¯2m1×2m1||0¯2m1×2nm+1−−−−−−−−||−−−−−−−−−−−−−−−−−−−−−−−−−||Gr,mhδmjGr,mhVmj..00||Gi,mhδmjGi,mhVmj..000¯2nm+1×2m1||......||00..Gr,nhδnjGr,nhVnj||00..Gi,nhδnjGi,nhVnj,

si483_e  (7-111)

where the partial derivatives of Gr,m(h) and Gi,m(h) are referred to the swing bus. Note that YG¯hjsi484_e is a 2n square matrix; however, for h = 1 only the last 2(n – m + 1) rows exist because the fundamental current balance is not applied to linear buses. YG¯1jsi485_e is a 2(nm + 1) × 2n matrix and for j = 1 the first two columns do not exist because |V~11|si486_e and δ1(1) are known, that is, YG¯h1si487_e is a 2n × 2(n – 1) matrix.

H¯hsi488_e is an array of partial derivatives of real and imaginary nonlinear load currents with respect to the firing angle α and the commutation impedance β:

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset