ELEC 4100

THREE PHASE CIRCUITS

ELEC 4100 ELECTRICAL ENERGY SYSTEMS REVIEW MATERIAL THREE PHASE CIRCUITS 1 Introduction.
For most electrical energy systems, electrical power is generated and transmitted in the form of three phase Alternating Current (AC), and in Australia the power system frequency is 50Hz. Throughout the electrical grid different voltage levels are used, with each level being appropriate for the various functions that must be performed for each section of the electrical network. So for example generation is typically performed at 20kV. To reduce losses during transmission the voltage is stepped up to 220kV, 330kV or 500kV. Sub-transmission is typically performed at 66kV, while for distribution the voltage is stepped down to 33kV, 22kV, 11kV, 6.6kV and 415V. There are a number of reasons that electrical energy is generated, transmitted and distributed as AC. The primary reason is that the varying voltage levels at different points within the electrical network means that it is necessary to have a device that can connect the different sections together. AC transformers can perform this function with a high degree of reliability and efficiency. Other reasons for using AC include the fact that it is considerably easier to interrupt fault currents in an AC network compared to DC, because the interruption devices can make use of the zero crossing of the fault current. Three phase power flow is also advantageous when compared to single phase power flow. This is because the real power flow in a balanced three phase circuit is constant. For a single phase circuit, however, the power flow pulsates at twice the system frequency. This means that for single phase systems the peak currents must be higher for the same power flow, and also for electrical machines supplied from a single phase network the torque produced by the machine also pulsates. This is why three phase power networks are the preferred choice for most applications. Note however that there are some applications such as rail systems where a single phase network is used, but this is because of the difficulty of feeding a train reliably with three separate energy circuits from an over-head line.

2 Generation of Three Phase Power.

The most common form of three phase generator is called the Synchronous Machine. The rotor of a synchronous machine spins at constant speed, and the speed is such that the frequency of the voltage wave produced by the generator is synchronous to the electrical grid. The following analysis is a simplified derivation of how synchronous machines generate a balanced set of three phase voltages. Figure 1 illustrates a simple single phase Synchronous Machine. A rotor winding is supplied with a DC field current, to create a magnetic field along the axis of the rotor. The rotor is driven at a constant speed of rad/sec. Angular Velocity = rad/sec =
60 rpm 2

(1)

The field created by the rotor is constant in magnitude, but the motion of the rotor leads to a rotating magnetic field in space. The stator of the machine is comprised of a single phase coil ( aa' ), and the rotating magnetic field of the rotor interacts with this coil such that the coil sees a field of the form:
= pk sin ( )

(2)

The voltage induced on the stator coil (i.e. across the coil aa' ) is then given by Faradays law as:

ELEC 4100

THREE PHASE CIRCUITS

a'
Figure 1 : Single Phase AC Generator.
e= d d =N dt dt
d dt

(3) (4)

e = N pk cos( )

In the above equations, N is the number of turns in the coil aa' . Since the rotational speed of the machine is constant, we can write:
=
d dt

(5) (6) (7)

e = N pk cos(t )

e = Em cos(t )

Note that the voltage magnitude, Em, is proportional to the rotor flux, and that the frequency of the voltage is determined by the angular velocity of the rotor. Hence to achieve a voltage of 240V at a system frequency of 50Hz, this two pole machine must be driven at 3000rpm with the field current set to produce a flux of 10.8mWb. Now consider the same synchronous machine, but with three sets of stator coils : aa ' , bb' , and 0 cc' . Each winding is spatially located 120 apart. This machine is illustrated in figure 2. The voltages generated in the two additional windings have the same magnitude as the aa' winding, but the spatial rotation leads to a phase shift of 1200 between the three voltages. The three stator voltages are given by:
ea = Em cos(t )

(8) (9) (10)

2 eb = Em cos t 3
4 ec = Em cos t 3

Figure 3 illustrates the three voltages for the simple three phase generator of figure 2, and also shows the phasor representation of the three voltages. The three phasors are given by: 2

ELEC 4100

THREE PHASE CIRCUITS

a b' c'

c b a'
Figure 2 : Three Phase AC Generator.
Em

ea

eb

ec Ea

1200

1200

Ec

1200

Eb

Figure 3 : The three voltages for the simple synchronous machine generator, and the phasor representation of these voltages.
Ea = E00 , Eb = E 1200 , Ec = E 2400

(11)

The RMS magnitude of the phasors is given by:

E= N pk

(12)

Another important point to note is that the direction of rotation of the machine determines the phase sequence of the three AC voltages. With the direction of rotation shown in figure 2 the phase sequence is (a,b,c), which means that the order in which each voltage reaches the peak value is (a,b,c). If the direction of rotation were to be reversed (i.e. a clockwise direction) the phase sequence would be (a,c,b). With a power network it is important to ensure that the phase sequence of a generator is correct before it is connected to the network. An incorrect phase sequence would result in a fault condition. Typically the colours RED, YELLOW (or WHITE) and BLUE are used to denote a phase sequence. The three phase voltage waveforms are often denoted as three independent voltage sources for circuit analysis as shown in figure 4. When these sources are connected in the configuration shown in figure 5(a) the system is said to be star or wye connected. When the system is connected in the configuration shown in figure 5(b) the system is said to be delta connected. When the phase shift between each voltage is 2400, and the three voltages have the same magnitude the source is said to be balanced. The common point in the star configuration of figure 3

ELEC 4100

THREE PHASE CIRCUITS

Ea

Ec Eb
Figure 4 : The circuit representation of the three phase source.
VR

VR

VY

VY
VB VN

VB

Figure 5 : (a) Star or Wye Connected source (b) Delta Connected source 5(a) is called the star point, and when a return wire is connected to this point it is called the neutral. If the currents of a three phase system are balanced, then the neutral current is zero. Clearly for a delta connected source there is no star point and it is not possible to connect a neutral wire.

3 Star Connected Generator.

Figure 6 illustrates a star connected source. This source supplies three currents, IR, IY and IB. In the case where the source is balanced the star point voltage is zero when measured with respect to earth, and with a balanced load the neutral current IN is zero. It is possible to construct a phasor diagram showing the three phase voltages, the line to line voltages, and the phase currents. The three phase voltages are defined by equation 11, and this allows the derivation of the three line to line voltages as:
VRY = VR VY = V0 V 120 = 3V30 = V (1 cos( 120)) jV sin ( 120)

(13)

Similarly it is possible to define the remaining two line to line voltages as:
VYB = VY VB = V 120 V 240 = 3V( 120 + 30) = 3V 90

(14)

ELEC 4100

THREE PHASE CIRCUITS

VRY
IR VR
300

VR IR

IB VB
IY IB IN VY VB VN

VYB IY VY

VBR

Figure 6 : A Star Connected Generator and Corresponding Phasor Representation.

VBR = VB VR = V 240 V0 = 3V 240 + 30

(15)

It can hence be seen that the line to line voltages are a factor of voltages, but also lead the phase voltages by 30.
I R = I , IY = I 120

larger than the phase

The phase currents lag the phase voltages by the angle and are defined by:

),

I B = I 240

(16)

Note that in this case the phase currents are the same as the line currents. The real power delivered by the source are given by:
Ptot = 3VPhs I L cos( ) = 3VLL I L cos( )

(17)

4 Delta Connected Generator.

For a delta connected source the phase and line to line voltages are equal. The line currents are however quite different to the phase to phase currents, as will now be demonstrated. Figure 7 shows the circuit diagram of a delta connected generator and its corresponding phasor representation. The line currents are then calculated as:
I R = I RY I BR = I LL (30 ) I LL ( 210 ) = I LL

= I LL (cos(30 ) cos( 210 )) + jI LL (sin (30 ) sin ( 210 )) = 3 I LL ( )

( 3 cos( )) jI ( 3 sin( ))
LL

(18)

The remaining two line currents can be calculated as:

ELEC 4100
VRY
300

THREE PHASE CIRCUITS

VRN IR IRY 300

IR IRY

VRN

IB

300

300 IYB

VYB

IBR IYB

IY VYN VBN IB

IBR IY VBR

Figure 7 : A Delta Connected Generator and Corresponding Phasor Representation.

IY = IYB I RY = I LL = I LL ( 90 ) I LL (30 ) = 3 I LL ( 120 )
I B = I BR IYB = I LL
LL

( 3 cos( 120 )) jI ( 3 sin( 120 ))

( 3 cos( 240 )) jI ( 3 sin( 240 ))
LL

(19)

= I LL ( 210 ) I LL ( 90 ) = 3 I LL ( 240 )

(20)

So the line currents can be seen to lag the phase to phase currents in the delta by 30, but also have a magnitude that is a factor of 3 times the phase to phase currents. The real power delivered by the source are given by:
Ptot = 3V phs I phs cos( ) = 3VLL I L cos( )

(21)

5 Power in Three Phase Circuits.

One of the advantages of a three phase AC supply is that the real power delivered by the source is constant. This is in contrast to a single phase source in which the real power delivery pulsates at twice the grid frequency. This effect will now be illustrated.

5.1 Single Phase Source.

Consider first a single phase source with a voltage v(t) supplying a load current of i(t). The supply voltage and the load current are given by:
v(t ) = V pk sin (t )

(22) (23)

i(t ) = I pk sin (t )

The instantaneous power delivered by this single phase source is then:

ELEC 4100
p(t ) = v(t )i(t )

THREE PHASE CIRCUITS

= V pk I pk sin (t )sin (t )

= V pk I pk [sin (t ){sin (t ) cos( ) cos(t )sin ( )}] = V pk I pk sin 2 (t ) cos( ) sin (t ) cos(t )sin ( ) = V pk I pk 2

(24)

[{1 cos(2t )}cos( ) sin (2t )sin ( )]

The peak values can be replaced by their RMS values:

V= V pk 2 I= I pk 2

(25)

This gives:
p(t ) = VI [1 cos(2t )]cos( ) VI sin (2t )sin ( )

(26)

Then by defining the real and reactive powers as P and Q, we can write:
P = VI cos( ) Q = VI sin ( )

(27)

p(t ) = P[1 cos(2t )] Q sin (2t )

The real power term P is the average power delivered to the load, and note that this term is always greater than or equal to zero. The reactive power term Q, on the other hand, always has an average value of zero. Note that both the real and reactive components of the power flow pulsate at twice the system frequency.

5.2 Three Phase Source.

Now consider a three phase source. The instantaneous power delivered by the source is given by:
p(t ) = vR (t )iR (t ) + vY (t )iY (t ) + vB (t )iB (t ) p(t ) = V pk sin (t )I pk sin (t ) + V pk sin (t 120)I pk sin (t 120) + V pk sin (t 240)I pk sin (t 240) p(t ) = V pk I pk [sin (t )sin (t ) + sin (t 120)sin (t 120) + sin (t 240)sin (t 240)] sin (t ){sin (t ) cos( ) cos(t )sin ( )} ( ) { ( ) ( ) ( ) ( ) } = V pk I pk + t t t sin 120 sin 120 cos cos 120 sin ( ) { ( ) ( ) ( ) ( ) } + t t t sin 240 sin 240 cos cos 240 sin V pk I pk 2

(28) (29)

(30)

[cos( ) cos(2t ) + cos( ) cos(2t 240) + cos( ) cos(2t 120)]

Now since:
cos(2t ) + cos(2t 240) + cos(2t 120) = 0

(31)

Then:
p(t ) = 3V pk I pk 2 cos( ) = 3V phs I phs cos( )

(32)

This result shows that the real power delivered by a balanced three phase source is constant, and equal to three times the real power in each phase. There is no double frequency pulsation.

ELEC 4100

THREE PHASE CIRCUITS

VRY
VR IR
300

VR IR

IB VB
VY VB VN IY IB IN Z Z

VYB IY VY

VBR

Figure 8 : A Star Connected Load and Corresponding Phasor Representation.

5.3 Apparent Power and Total Power.

Apparent power is a term that is often used to describe the rating of electrical equipment. The apparent power of a system is given by:
VA = P 2 + Q 2 = V I

(33)

Total or complex power is a phasor and is described by:

S = V I e j = P + jQ = VI *

(34)

6 Three Phase Loads.

While there are many possible circuit configurations that can be used for three phase loads, the two standard structures are again the star and the delta arrangements.

6.1 Star Connected Load.

The star connected load is illustrated in figure 8. The load in figure 8 is shown to be balanced, in that the three phase to neutral impedances are equal, and have an impedance value denoted by Z. The voltage across each impedance is simply the phase to neutral voltage, and the corresponding line currents are then given by:
IR =
IY = IB =

VR V phs 0 V phs = = Z Z Z
VY V phs 120 V phs = = ( 120 ) Z Z Z VB V phs 240 V phs = = ( 240 ) Z Z Z

(35) (36) (36)

The power consumed in each phase is :

PR = VR I R cos( ) = V phs I phs cos( )
PY = VY I Y cos( ) = V phs I phs cos( )

(37) (38) 8

ELEC 4100
VRY
300

THREE PHASE CIRCUITS

VRN IR IRY 300

VR

IR IBR

IB
Z VB VY IB IY IYB Z Z IRY

300

300 IYB

VYB

IBR IY VBR

Figure 9 : A Delta Connected Load and Corresponding Phasor Representation.

PB = VB I B cos( ) = V phs I phs cos( )

(39)

Hence the total power consumed is:

Ptot = 3V phs I phs cos( )

(40)

The total reactive power is:

Qtot = 3V phs I phs sin ( )

(41)

Since the voltage across each phase is equal to 1 3 times the line to line voltage the total power may also be written as:
Ptot = 3VLL I phs cos( )

(42)

Note that the term cos( ) is called the power factor, and it is determined by the impedance of the three phase load.

6.2 Delta Connected Load.

The delta connected load is illustrated in figure 9. The load in figure 9 is shown to be balanced, in that the three phase to phase impedances are equal, and have an impedance value denoted by Z. The voltage across each impedance is simply the line to line voltage, and the corresponding phase to phase currents are then given by:
I RY =
IYB = I BR =

VRY VLL 30 VLL = = ( 30 ) Z Z Z

VYB VLL 90 VLL = = ( 90 ) Z Z Z VBR VLL 210 VLL = = ( 210 ) Z Z Z

(43) (44) (45)

The power consumed in each phase is :

PRY = VRY I RY cos( ) = V phs I phs cos( )

(46) 9

ELEC 4100
PYB = VYB IYB cos( ) = V phs I phs cos( )
PBR = VBR I BR cos( ) = V phs I phs cos( )

THREE PHASE CIRCUITS (47) (48)

Hence the total power consumed is:

Ptot = 3V phs I phs cos( )

(49)

The total reactive power is:

Qtot = 3V phs I phs sin ( )

(50)

Since the phase voltage and the line to line voltage are the same, but the line current is equal to 3 times the phase to phase current, the total power may also be written as:
Ptot = 3VLL I L cos( )

(51)

6.3 Unbalanced Loads.

The load analysis presented so far considers solely the case where the three impedances are equal. However under the condition where these impedances are not equal it is necessary to use more complex expressions. Considering a star connected load, but without a neutral current path, the star point voltage is non-zero, and so the following equations apply:
IR = IY = IB = VRN VSN ZR VYN VSN ZY VBN VSN ZB

(52) (53) (54) (55)

I R + IY + I B = 0

To use the above expressions it is first necessary to calculate the star point voltage. This then allows the remaining currents and voltages to be determined. If there is a neutral current path, the star point voltage is taken to be zero, and the neutral current is given by:
I R + IY + I B = I N

(55)

6.4 Star Delta Transformation.

It is possible mathematically to transform a star connected load into an equivalent delta connected load, and vice versa. Figure 10 illustrates the concept in which the line to line impedances of a delta circuit have an equivalent phase impedance in a star circuit. The expressions for the delta to star transformation are given by:
ZR = ZY = Z RY Z BR Z RY + ZYB + Z BR ZYB Z RY + ZYB + Z BR

(56) (57)

Z RY

10

ELEC 4100

THREE PHASE CIRCUITS

ZR

ZBR

ZRY

ZY ZB ZYB

Figure 10 : The Star-Delta Transformation of a three phase load.

ZB = Z BR ZYB + ZYB + Z BR

Z RY

(58)

The expressions for the star to delta transformation are given by:
Z RY = ZYB = Z BR = Z R ZY + ZY Z B + Z B Z R ZB Z R ZY + ZY Z B + Z B Z R ZR Z R ZY + ZY Z B + Z B Z R ZY

(59) (60) (61)

Now consider the case where the load is balanced. Consider first the star to delta transformation with a phase impedance Z. From equations (59) to (61) this gives:
Z RY = ZYB = Z BR = 3Z

(62)

Now consider the delta to star transformation when the line to line impedance is 3Z. This gives from equations (56) to (58):
Z R = ZY = Z B = Z

(63)

So the equivalent star configuration has an impedance that is a factor of 3 smaller than the corresponding delta connected circuit. That is to say:
Z = 3ZY

(64)

11

ELEC 4100

PER UNIT ANALYSIS

ELEC 4100 ELECTRICAL ENERGY SYSTEMS REVIEW MATERIAL PER UNIT ANALYSIS 1 The Per-Unit System.
The Per-Unit (PU) system is commonly used in power systems network analysis. The PU system is used to express quantities such as impedance, voltage, current and power as a percentage of the specified base values or ratings. Key advantages of this system is that it gives an immediate indication of the magnitude of a quantity in relative terms, and it also significantly simplifies the analysis of systems involving transformers. PU values also make it trivial to spot bad data. A Per Unit quantity is calculated as:
p.u. quantity = actual quantity base value of quantity

(1)

To reiterate the key advantages of the PU system include: All quantities are normalised It is easy to check whether the various statutory requirements for a system are met (i.e. allowable voltage ranges are normally specified as a percentage). Ideal transformers are eliminated. Per Unit impedances of similar electrical devices normally lie in a reasonably narrow range, and this makes it possible to check analysis data very quickly. Manufacturers normally give impedances of machines and transformers in Per Unit or Percentage of name plate ratings, and so it is therefore essential to know and use the correct base quantity. For an electrical circuit there are two degrees of freedom available, which means it is possible to select two arbitrary and independent base quantities. For three phase applications it is customary to set a base line to line voltage (VbaseLL) and a base three phase complex power (Sbase3). Then to ensure the conservation of electrical laws in the Per Unit system, the following relationships must hold:
Sbase3 = Pbase3 = Qbase3
Sbase1 = VbaseLN = Sbase3 3 VbaseLL 3

(2) (3) (4)

I base =

Sbase1 VbaseLN

Sbase3 3VbaseLL

(5) (6) (7)

Z base =

2 VbaseLN VbaseLN V2 = = baseLL I base Sbase1 Sbase3

Rbase = X base = Z base =

1 Ybase

In the above equations the subscript LL denotes a line to line quantity, LN denotes a phase or line to neutral quantity, 1 denotes a single phase quantity and 3 denotes a three phase quantity. 1

ELEC 4100 The following conventions are applied for base quantities: The value of Sbase3 is maintained to be the same for the entire system.

PER UNIT ANALYSIS

For a transformer the ratio of the primary side voltage base and the secondary side voltage base is set to be the same as the transformer turns ratio. These two conventions ensure that the Per Unit impedance of a transformer is the same on both the primary and the secondary side of the transformer. This has significant advantages for performing analysis of large electrical systems that involve multiple voltage levels (e.g. three levels of distribution voltage).

1.1 Example : Per Unit Impedance of a Single Phase Transformer.

Consider a single phase transformer with the following ratings: Rated transformer power throughput 20kVA. Primary Volts = 480V. Secondary Volts = 120V. System Frequency = 50Hz Secondary Referred Leakage Inductance = Zeq2 = 0.052578.12 .

Using the above transformer ratings as base values, the Per Unit leakage impedance on both primary and secondary side of the transformer can be determined as follows: The base power and voltages by convention are: Sbase = 20kVA, Vbase1 = 480V, Vbase2 = 120V. Therefore the base impedance on the secondary side of the transformer is:
Z base 2 =
2 Vbase 120 2 = = 0.72 Sbase 20,000

The Per Unit leakage referred to the secondary side of the transformer is then:
Z eq 2, pu = Z eq 2 Z base 2 = 0.052578.12 = 0.072978.12 p.u. 0.72

If the leakage is referred to the primary side of the transformer, then in Ohms:
N1 480 Z eq1 = N Z eq 2 = 120 0.052578.12 = 0.8478.12 2
2

The base impedance on the primary side of the transformer is :

Z base1 =
2 Vbase 480 2 1 = = 11.52 Sbase 20,000

The per unit leakage impedance referred to the primary side of the transformer is then:
Z eq1, pu = Z eq1 Z base1 = 0.8478.12 = 0.072978.12 p.u. = Z eq 2, pu 11.52

This result shows clearly that in the Per Unit system the transformer impedance is the same on both sides of the transformer. This aspect of the Per Unit system is achieved by specifying that:

ELEC 4100

PER UNIT ANALYSIS

jXeq.p.u.
I1p.u. E1p.u. I2p.u. E2p.u.

Req.p.u. I2p.u. E2p.u.

I1p.u. E1p.u.

(a)

(b)

I1p.u.

jXeq1.p.u. Req1.p.u.

Req2.p.u. jXeq2.p.u. I 2p.u. Icp.u.

E1p.u.

jXmp.u.

Rcp.u.

E2p.u.

(c) Figure 1 : Per Unit System Single Phase Transformer Models. (a) Ideal Case, (b) Leakage only (c) Complete Model.
Vbase1 Vrated1 480 = = Vbase 2 Vrated 2 120

For an ideal transformer, if the ratio of the selected voltage bases equals the ratio of the voltage ratings of the windings then:
V1 pu = V2 pu
I1 pu = I 2 pu

Therefore, ideal transformers can be eliminated from electrical networks using a Per Unit system. Figure 1 shows the possible transformer models that can be used in a Per Unit system when considering an ideal transformer, a transformer with a series leakage impedance only, and a complete representation accounting for magnetising effects as well as primary and secondary referred leakage effects.

1.2 Example : Three Zone Single Phase Network.

Consider a simple single phase network shown in figure 2. This network consists of three different voltage zones, connected together with two transformers. To analyse this simple system the first step is to find the Per Unit system representation. First find the base values in each zone. Arbitrarily select the power throughput base as 30kVA, corresponding to the maximum rating of the transformer T1. So Sbase = 30kVA. Similarly the voltage base for zone 1 is selected to be Vbase1 = 240V. The voltage base for zone 2 can then be calculated as:
480 Vbase 2 = Vbase1 = 480V 240

ELEC 4100
Zone 1 Vs = 220V Zone 2 Zone 3 Rload = 0.9 Xline = j2 T1 30kVA 240/480 V Xeq = 0.1 p.u. T2 20kVA 460/115 V Xeq = 0.1 p.u.

PER UNIT ANALYSIS

Xload = j0.2

Figure 2 : Single Phase Three Zone Network. Similarly the voltage base for zone 3 is given by:
115 Vbase3 = Vbase 2 = 120V 460

The base impedances for zones 2 and 3 are then given by:
Z base 2 = Z base3 =
2 Vbase 4802 2 = = 7.68 Sbase 30,000 2 Vbase 120 2 3 = = 0.48 Sbase 30,000

The base current in zone 3 is also given by:

I base3 = Vbase3 120 = = 250 A Z base3 0.48

It is now possible to calculate the Per Unit impedances of the single phase three zone network. The Per Unit impedance of the transmission line is given by:
X line p.u. = X line j2 = = j 0.2604 p.u. Z base 2 7.68

The Per Unit impedance of the load is given by:

Z load p.u. = Z load 0.9 + j 0.2 = = 1.875 + j 0.4167 p.u. 0.48 Z base3

Now it is necessary to consider the transformer impedances. The Per Unit impedances for the transformers quoted in figure 2 assume that the bases used are the transformer ratings. This is certainly valid for T1, but it is not the case for T2. Therefore the impedance of the transformer T2 must be re-scaled using the new system bases. Now recall that Per Unit impedance is defined as:
Z p.u. =
2 2 Z p.u.Vbase Z p.u.rateVbase Z actual Z actual I base Z actual Sbase _ rate = = , = = or Z actual 2 Sbase Sbase _ rate Z base Vbase Vbase

So to rescale this impedance for the new system base:

X T 2 p.u. = Z actual Sbase
2 Vbase

= Z p.u.rate

2 Vbase _ rate 2 Vbase

Sbase Sbase _ rate

460 2 30,000 = 0.1 = 0.1378 p.u. 480 2 20,000

This makes it possible to redraw the three zone network, in terms of the Per Unit system as shown in figure 3. 4

ELEC 4100
Zone 1 Vs = 0.9167 p.u. Zone 2 Zone 3 Rload p.u. = 1.875 p.u. XT1eq p.u. = j0.1 p.u. Xline p.u. = j0.2604 XT2eq = j0.1378 p.u.

PER UNIT ANALYSIS

Xload p.u. = j0.4167 p.u

Figure 3 : Per Unit System Representation of the Single Phase Three Zone Network. From figure 3 it is now possible to solve for the Per Unit load current, which is given by:
I load p.u. = = j X T 1eq p.u. + X line p.u. + X T 2eq p.u. + X load p.u. + Rload p.u. Vs p.u.

0.9617 j (0.1 + 0.2604 + 0.1378 + 0.4167 ) + 1.875 0.9167 = 1.875 + j 0.9149 = 0.4395 26.01 p.u.

The actual load current is then:

I load = I load p.u. I base3 = (0.4395 26.01 p.u.)250 = 109.9 26.01 A

The use of the Per Unit system simplified the analysis involving multiple transformers on a line, and will be used extensively for analysing power systems involving many elements.

1.3 Res-scaling of base values.

In the above example it was shown how the Per Unit impedance of a transformer needed to be re-scaled because the system bases did not match the ratings of the transformer. This situation occurs quite commonly because manufacturers often quote the impedance of transformers, machines, etc. in Per Unit quantities with the bases drawn from the component rating. To ensure that the same power base is used for an entire system this means it is necessary to rescale the impedances for a number of elements of a power system. The conversion formula can be formally stated as:
Z p.u., new = Z p.u., old Z base, old Z actual = Z base, new Z base, new I base, new Vbase, new
2 Sbase, new Vbase Sbase, new , old =Z p.u ., old 2 2 Vbase, new Sbase, old Vbase, new

Vbase, old = Z p.u., old I base, old

2 Vbase , old = Z p.u., old Sbase, old

1.4 Transformers with Off-Nominal Turns Ratios.

In the above it has been shown that when a transformer is used in applications where the ratio of the selected voltage bases is equal to the turns ratio of the transformer then the transformer can be eliminated from a P.U. circuit, with only the leakage and magnetising branch impedances remaining. The question arises as to what happens in systems when this condition can not be guaranteed? As an example consider two transformers connected in parallel. The first transformer 5

ELEC 4100

PER UNIT ANALYSIS

b:1

c:1

Figure 4 : Two Transformer Representation of an Off-Nominal Turns Ratio Connected Transformer. T1 is rated at 13.8kV/345kV, while the second transformer T2 is rated at 13.2kV/345kV. It is simply not possible to eliminate both transformers from a Per Unit model since at least one transformer will have base values that do not correspond to its turns ratio. This type of situation is very common with tap-changing transformers. It is necessary to develop a per unit model of the transformer operating with an off-nominal turns ratio, so as to be able to accommodate this scenario in which the voltage ratings are not in proportion to the selected voltage bases. So consider a transformer with primary and secondary rated voltages denoted by Vrated1 and Vrated2. The nominal transformer turns ratio is given by:
at = Vrated 1 Vrated 2

Note that this turns ratio need not be real. It can be complex to account for phase shifts across a transformer. Now assume that the selected voltage bases satisfy:
Vbase1 = bVbase 2

Then define c such that c = at b , so :

a Vrated 1 = b t b Vrated 2 = bcVrated 2

This expression can be represented by two transformers in series. The first transformer has a turns ratio of b, while the second has a turns ratio of c, as illustrated in figure 4. Note, however that the ratio of the voltage bases is b, so in a Per Unit Circuit the transformer with the turns ratio of b can be eliminated. This leaves the second transformer with a turns ratio of c in circuit. 1.4.1 Example A Tap Changing Transformer. Consider a tap changing transformer with ratings: 1000MVA 13.8kV/345kV Zeq = j0.1 p.u. The transformer HV winding has 10% taps. Sbase3 = 500MVA VbaseLV,LL = 13.8kV VbaseHV,LL = 345kV

This transformer is operated in a system with the following base quantities:

Determine the p.u. equivalent circuit for the following tap settings (a) Rated tap, (b) 10% tap (this provides a 10% voltage reduction for the HV winding). Solution: 6

ELEC 4100

PER UNIT ANALYSIS

ILV VLV

j0.05

IHV VHV

Figure 5 : Tap Changing Transformer Model Case (a).

ILV VLV

j0.05

IHV VHV

1.1111 : 1
Figure 6 : Tap Changing Transformer Model Case (b). The first step is to change the transformer impedance so it corresponds to the new base.
13.8kV 500 MVA Z p.u., new = j 0.1 = j 0.05 p.u. 13.8kV 1000 MVA
2

For case (a) with the tap set to the rated position, this gives:
at =
13.8kV , 345kV

b=

VbaseLV , LL VbaseHV , LL

13.8kV , 345kV

c = 1

This gives the simple model shown in figure 5. For case (b) with the tap set to 10%, this gives:
at =
13.8kV = 0.0444, 0.9 345kV

b=

VbaseLV , LL VbaseHV , LL

13.8kV = 0.4, 345kV

c =

at = 1.11111 b

This model is shown in figure 6.

ELEC 4100

THREE PHASE TRANSFORMERS

ELEC 4100 ELECTRICAL ENERGY SYSTEMS THREE PHASE TRANSFORMERS

Previously it was stated that the different sections of a power network utilise different voltage levels because of the different requirements of the system at that part of the network. For example it is advantageous to use a high voltage for long distance transmission so as to reduce the I2R losses. Note that the resistance of a transmission line increases linearly with length. The typical transmission voltages in Australia are 220kV, 330kV and 500kV. For a generator, on the other hand, it is difficult to physically construct the windings to withstand voltages in excess of 25kV. Hence there must be an interface between the sections of the network that have different voltage levels, and this function is performed by the transformer.

1 Single Phase Transformers Review.

A simple transformer consists of at least two galvanically isolated electrical circuits, and a magnetic circuit that is used to couple the electrical circuits. This is illustrated in figure 1. The operation of a transformer is such that the driving electrical circuit induces a m.m.f in the magnetic circuit. This m.m.f produces a flux in the low reluctance circuit, which induces a voltage in the winding of the second electrical circuit. The relationship between the voltages and currents in the transformer windings can be found by applying Amperes and Faradays laws.

1.1 Amperes Law.

Figure 1 shows a simple model of a single phase transformer. Two windings are shown wrapped around the core of the transformer, having N1 and N2 turns respectively. The core of the transformer has a cross sectional are of Ac, and the flux developed in the core travels a mean path of lc. The windings have voltages of V1 and V2, and carry currents I1 and I2. Faradays law states that the line integral of the tangential component of the magnetic field intensity vector H to the integral path, and the sum of the current enclosed by this path are related according to:

tan

dl =

I
loop

(1)

Assuming that the magnetic field intensity is constant along this path, and given by Hc, then the field intensity is given by:
H clc = N1I1 N 2 I 2

(2)

The magnetic flux density vector B and the magnetic field intensity vector H are related by:
AC
C
Core Cross Sectional Area

V2 V1 N2 turns N1 turns

dl
Magnetic Circuit Mean flux path

lC

Figure 1 : Structure of a Single Phase Transformer. 1

ELEC 4100
Bc = c H c

THREE PHASE TRANSFORMERS (3)

The core flux and the magnetic flux density are related according to:
c = Ac Bc

(4)

Combining this expression with equation 2 gives:

lc c = N1I1 N 2 I 2 c Ac

(5)

So defining the core reluctance as:

Rc = lc c Ac

(6)

Then:
Rc c = N1I1 N 2 I 2

(7)

This is the so called Ohms law for magnetic circuits. It relates the m.m.f created by a flux linkage (i.e. the number of turns in winding multiplied by the current carried by that winding) and the reluctance in the magnetic circuit. The reluctance impedes the creation of the magnetic flux and is analogous to resistance in electric circuits. An important point to note with equation 7 is that the flux produced by the current in the secondary winding subtracts from the flux produced by the current in the primary winding. This is dictated the winding direction (i.e. clockwise verses anti-clockwise) and then the direction of current flow in the winding (i.e. the Right Hand Rule). This is usually indicated with the DOT convention. In circuit schematics the windings of a transformer have a DOT indicator. The DOT convention indicates that when current enters the DOT terminals of each winding, the flux produced by the windings adds constructively. It is therefore simple to see that in figure 1 the current in the secondary winding is exiting the DOT terminal, but the current in the primary winding is entering the DOT terminal. Hence the flux produced by the secondary will subtract from that produced by the transformer primary. Ideally the transformer core reluctance is zero, and so the primary and secondary winding currents are related according to:
N1I1 = N 2 I 2

(8)

1.2 Faradays Law.

The voltage relationship between each winding can be determined using Faradays law. Recall that Faradays law states that the relationship between a time varying magnetic field and the voltage induced in an electrical circuit placed in this field is given by:
e(t ) = N d (t ) dt

(9)

Then if we assume sinusoidal operation with a constant angular frequency and using the phasor representations for the voltage and flux, Faradays law can be re-written as:
E = N j

(10)

So for an ideal transformer that has two windings with N1 and N2 turns on each, and linked by the same magnetic flux, the voltages on each winding can be written as:
V1 = N1j

(11) 2

ELEC 4100
V2 = N 2j

THREE PHASE TRANSFORMERS (12)

The voltage transfer ratio is then given by:

V1 V2 = N1 N 2

or

V1 N1 = = at V2 N 2

(13)

This is the classic turns ratio relationship for an ideal transformer. Note that the key assumption in this case is that the same magnetic flux links both windings. If there is flux leakage from the magnetic core then there is an apparent reduction in turns ratio. Another important result for sinusoidal transformers is the Volts/Turn characteristic. This derives from equation 10, and relates the Voltage induced on a winding (in RMS) to the applied frequency, maximum magnetic flux density and cross sectional area of the core:
VRMS N
VRMS N

= RMS
= 2 2 Ac B pk f = 4.443 Ac B pk f

(14) (15)

This is a particularly important result for transformer design. This is because for power system applications the grid frequency is fixed (to 50Hz in Australia), and the peak magnetic flux density is also fixed for the core material used. For Grain Oriented Steel this value is typically in the range of 1.3T to 1.7T. This way the transformer core operates in saturation so as to reduce the cost of material used, but the transformer does not operate so heavily into saturation that the core excitation losses lead to excessively high operational costs over the lifetime of the transformer. So for a transformer design this means that the required Volts/Turn for a given system level voltage and available core window size, the only parameter that can be varied is the core cross sectional area which is defined by equation 15. Example: A 22kV / 66kV 50Hz transformer can be constructed using a core operating at 1.5T peak flux, and with a cross sectional area of 0.75m2. The required Volts/Turn is therefore:
VRMS N = 4.443(0.75)(1.5)(50) = 249.9V

So to achieve the LV voltage the number of turns required is:

N LV = 22kV = 88 turns 249.9

For the HV winding :

N HV = 66 kV = 264 turns 249.9

The above relationships then define the classic ideal transformer characteristics. From equation (13) we have the voltage turns ratio as:
V1 N1 = = at V2 N 2

(16)

From (8) the primary and secondary currents are related by:
I1 N 2 1 = = I 2 N1 at

(17)

ELEC 4100

THREE PHASE TRANSFORMERS

Then if the transformer secondary is loaded with an impedance Z2, then the equivalent primary impedance is given by:
Z2 = V2 V 1 = 1 = Z1 I 2 at 2 I1 at 2

(18)

So there secondary side load impedance can be modelled as a primary impedance, but reduced by the inverse square of the turns ratio.

1.3 The Practical Transformer Model.

The ideal transformer characteristics have been defined assuming the following: (i) (ii) The transformer winding resistances are negligible. The core permeability is infinite, and hence the core reluctance is zero.

(iii) The magnetic flux is confined with the transformer core. (iv) There are negligible real and reactive power losses in the transformer core. For a practical transformer these assumptions must be discarded. It will now be shown how the effects of a finite core reluctance, flux leakage, winding losses and core losses can be accounted for to create a practical transformer model. 1.3.1 Excitation Branch. Since the transformer core has a non-zero reluctance, the transformer primary current is related to the secondary current according to equation 7 as:
I1 = Rc c N 2 + I2 N1 N1

(19)

By Faradays law the core flux is related to the primary excitation voltage according to:
E1 = jN1 c

(20)

Hence the primary current is given by:

I1 = Rc E1 jN12 + N2 I2 N1

(21)

If we then define the magnetising current as Im, the we can write:

I1 = I m + N2 I2 N1

where

Im =

Rc E1 jN12

(22)

From this equation it can be seen that the effect of the non-zero core reluctance can be modelled as a shunt conduction path on the primary side of the transformer. Furthermore the current in the conduction path lags the applied voltage by 900, and hence can be regarded as a shunt inductance with a value given by:
X m = jLm = jN12 Rc

or

Lm =

N12 Rc

(23)

In practice there is an additional element in the shunt conduction path, and this relates to the core loss. There are two major elements in the core loss, the first being magnetic hysteresis, while the second mechanism are eddy currents induced in the transformer core. Hysteresis losses relate to the fact that the cyclical variation of the flux direction in the core requires the magnetic dipoles of the core to change direction at the grid frequency. This requires an energy exchange which is dissipated in the core as heat. The magnitude of the hysteresis losses are determined by the level of saturation 4

ELEC 4100
I1 Ie V1 jXm Rc E1 E2 V2 I2' I2

THREE PHASE TRANSFORMERS

Figure 2 : Transformer Model showing Excitation Phenomena. that the core is driven to. Eddy current losses occur because the transformer core is conductive, and so the eddy currents, perpendicular to the field direction, are induced in the core. These currents can be significantly reduced by manufacturing the transformer with electrically isolated steel laminations. Both loss mechanisms can be modelled as an equivalent shunt resistance path. The combined currents drawn by the core magnetisation reactance, and the core loss resistance is called the excitation current of the transformer, and the equivalent transformer model with the excitation branch is shown in figure 2. 1.3.2 Leakage and Winding Losses. The effect of winding losses is easily modelled. These losses are a direct result of the resistance of the conductors used to fabricate the windings, and so a primary and secondary resistance is simply added to the primary and secondary paths. The effects of leakage flux can be accounted for by applying Faradays law. Consider the transformer primary winding which has an applied sinusoidal voltage V1 and so the flux induced in the core is given by:
1 = V1 jN1

(24)

Similarly the flux that links the secondary winding induces a voltage on the secondary winding given by:
2 = V2 jN 2

(25)

The flux that leaks from the core and does not link to the secondary winding is denoted leak, and is given by
leak = 1 2 =

1 j

V1 V2 N1 N 2

(26)

This expression can be rearranged as:

j N1 leak = V1 N1 V2 = V1 N2

(27)

Now the leakage flux only links the primary winding, and so Amperes law can be used to relate the linkage flux to the primary winding current according to:
Rleak leak = N1I1

(28)

Then the primary current responsible for the leakage flux is related to the difference between the primary voltage and the primary referred secondary voltage as:
j N12 I1 = V1 Rleak

(29) 5

ELEC 4100
I1 jXeq1 Req1 Ie V1 jXm Rc E1 E2 I2' Req2

THREE PHASE TRANSFORMERS

jXeq2 I2

V2

Figure 3 : Transformer Model with Leakage, Winding Resistance and Excitation effects. In the above, Rleak denotes the effective reluctance for the average path followed by the leakage flux. Hence the leakage flux can be modelled by an inductance with the reactance and inductance values given by:
X leak = j N12 Rleak

and

Lleak =

X leak N2 = 1 j Rleak

(30)

Note that in this case the leakage effect has been completely referred to the primary winding, however it is also possible to treat the leakage effect for both windings separately. This is illustrated by the complete transformer model shown in figure 3.

1.4 Transformer Characterisation.

For a practical transformer it is often necessary to determine the circuit parameters for modelling purposes. These parameters can be determined via a short circuit and an open circuit test on the transformer. The equivalent transformer models in both cases are shown in figure 4 below. In an open circuit test there is no secondary current, and so the primary current that flows is the excitation current alone. The secondary voltage under these conditions is then the secondary referred excitation branch voltage. These measurements then allow for the determination of the excitation branch parameters as:
Z excitation = N1 E2 N 2 I1

(31)

In a short circuit test there is no secondary voltage, and so the short circuit appears across the excitation branch elements, ensuring that there is no excitation current. The primary voltage appears solely across the leakage and winding resistance elements, and so the measurement of the primary voltage and current in this case determines the series impedance elements as:
Z leakage = V1 I1

(32)

I1 Ie V1 jXm Rc E1 E2

I1

jXeq1

Req1

I2

V1

Figure 4 : Transformer Models during open and short circuit tests. 6

ELEC 4100

THREE PHASE TRANSFORMERS

1.5 Physical Construction.

Two typical structures for single phase transformers are shown in figure 5. The first structure referred to as a Core Type transformer, uses a two limb core, and splits the windings to occupy both limbs. Note that the primary and secondary windings are wound concentrically, and this is to ensure that the coupling between the windings is as effective as possible and minimises flux leakage. The inner section of the winding is usually the low voltage winding, and this is because it makes the task of insulating the high voltage outer winding easier. The second transformer shown in figure 5(b) is referred to a Shell Type transformer. In this case the core consists of three limbs, but the inner-most limb is physically larger than the outer limbs to accommodate the double flux which passes through this limb. The winding in this case is not split but is placed around the central core limb. While it is possible to construct a three phase transformer using three single phase transformers using one of the structures outlined in figure 5, it is possible to use a single core in either a Shell or Core type structure. Figure 6 illustrates a Core type transformer. Here the transformer core consists of three limbs, and the windings for each phase are placed over each transformer limb. This structure is possible for a transformer supplied by a set of balanced three phase voltages since the sum of the three fluxes (A, B, C) sum to zero. For power systems transformers paper is used to insulate the conductors of the windings, and the core and winding arrangement are placed in an oil filled tank. This is done for a number of reasons, including:

2C
(a) (b)

Figure 5 : Single Phase Transformer Construction (a) Core Type (Split Windings) and (b) Shell Type (Split Core).
A B C

Figure 6 : Three Phase Core Type Transformer.

ELEC 4100 Improved Insulation properties compared to air. Improved heat transfer from the windings and core. Noise reduction.

THREE PHASE TRANSFORMERS

2 Three Phase Transformer Connections.

To adequately model a power system it is necessary to consider how to model three phase transformers. One way to construct a three phase transformer is to use three separate single phase transformers, all with separate and distinct magnetic cores, and simply connect the windings in either a star or a delta circuit. Clearly this leads to four possible ways to connect the transformer windings as: Y-Y, Y-, -Y or -. Figure 7 illustrates the case in which the primary and secondary windings are connected in Y-Y. Note the DOT conventions on the windings. For three

H1

X1

IA2 Van

VAN

H2

X2

IB2 Vbn

VBN

IC1 VCN

H3

X3

IC2 Vcn Vn

VN

Figure 7 : A Star Star Connected Transformer using three separate cores.

H3 IC1 X3 IC2

H1 N VAN IA1 n Van

X1 IA2

IB1

H2

IB2

X2

(a)

(b)

Figure 8 : Circuit Schematic for a Star Star Connected Three Phase Transformer , and an equivalent Single Line Diagram. 8

ELEC 4100

THREE PHASE TRANSFORMERS

VAN VN

H1 VA

X1

Va Vb Va

VBN VN VB VCN VC VN VN

H2

X2

Vb Vc Vb

H3

X3

Vc Vc Va

Figure 9 : A Star Delta Connected Transformer using three separate cores. phase transformers a different system is often used to the DOT convention, in which the high voltage DOT terminals are denoted by H1, H2 and H3 for the A, B, and C phases respectively, while the low voltage DOT terminals are denoted by X1, X2 and X3. Figure 8(a) illustrates a schematic representation of the Y-Y transformer of figure 7. In this representation windings on the same physical core are drawn in parallel to one another, and the angle between the windings on either side of the transformer indicates the phasor relationship between the applied voltages. For single line diagrams the Y-Y transformer is often represented by the symbol shown in figure 8(b). It is straightforward to show that the voltages on either side of the transformer are in phase (i.e. Van and VAN) and that their ratio is given by the transformer turns ratio. It can also be shown that for a - transformer there is no phase shift between the voltages across the transformer. For Y- and -Y transformers there is always a phase shift between the voltage waveforms across the transformer. Figure 9 shows a Y- transformer, and the phase shift between the transformer windings can be determined as follows. On the primary windings of the transformer in figure 9 the three voltage phasors representing VAN, VBN and VCN are shown. Similarly on the secondary winding the three voltage phasors representing Vab, Vbc and Vca are shown. Now since the secondary voltage phasors described above are on the same core as the primary voltage phasors, then by definition these phasors must be in phase with one another, and this is also illustrated in figure 9. Now for the secondary side phasors, the head of the Vbc phasor must connect to the tail of the Vab phasor because of the electrical connection, and this leads to the phasor representation of the transformer shown in figure 10. A careful examination of this diagram clearly shows that the Van phasor on the secondary lags the VAN phasor by 300. Hence the use of the Y- connection introduces a 300 phase shift between the primary and secondary side voltage waveforms. 9

ELEC 4100

THREE PHASE TRANSFORMERS

VC Vc

VN

VA Vb

Vn
300

Va

VB
Figure 10 : The Phasor Representation of the Y- Transformer. The convention that is used in power systems analysis states that in either a Y- or a -Y transformer, the positive sequence quantities on the high voltage side shall lead the corresponding quantities on the low voltage side by 300. Delta windings have many advantages for power system transformers, since they provide a current path for third harmonic currents. These currents are produced by in the excitation current of transformers which are driven into saturation. Third harmonic currents are referred to as zero sequence currents, and can flow within the delta arrangement, but do not appear in the line currents leaving the transformer. Without the current path for these harmonics the voltage waveforms of the transformers distort and contain the third harmonic terms. For this reason Y-Y transformers are seldom used. The star winding has advantages for high voltage connections since the star point can be conveniently be connected to earth. This reduces the voltage isolation requirements for a single winding, and as such -Y transformers are often used in generation applications.

2.1 Common Transformers and Winding Arrangements.

Thus far the Y-Y, Y-, -Y and - three phase transformer arrangements have been described, but there are many other possible configurations that can be used, especially when a tertiary winding is added. Below is a list of common transformer winding structures.
C1 c1

N A2 B2 C2 A1 a1 B1 b1 C1 c1

n a2 b2 c2
N

A1

a1

B1

b1

Figure 11 : Star - Star Transformer.

C1 c1

A2 B2 C2

A1 a1 B1 b1 C1 c1

a2 b2 c2
B1 A1 b1 a1

Figure 12 : Delta - Delta Transformer. 10

ELEC 4100
C1

THREE PHASE TRANSFORMERS

c1

A2 B2 C2

A1 B1 C1

n a1 b1 c1 a2 b2 c2

a1

B1

A1

b1

Figure 13 : Delta - Star Transformer.

C1 A1 c1

A2 B2 C2

A1 B1 C1

n a1 b1 c1 a2 b2 c2

a1

B1

b1

Figure 14 : Delta - Star Transformer.

C1 b1

N A2 B2 C2 A1 a1 B1 b1 C1 c1 a2 b2 c2
B1 c1 N A1 a1

Figure 15 : Star - Delta Transformer.

C1 c1 A1 a1

N A2 B2 C2 A1 a1 B1 b1 C1 c1 a2 b2 c2
B1 b1 N

Figure 16 : Star - Delta Transformer.

n N A2 B2 C2
C1

A1 a1 B1 b1 C1 c1
c1

a2 a3 b2 b3 c2 c3

a4 b4 c4

A1 b1

n a1

B1

Figure 17 : Star Zig-Zag Transformer. 11

ELEC 4100

THREE PHASE TRANSFORMERS

I1p.u.

jXeq1.p.u. Req1.p.u.

Req2.p.u. jXeq2.p.u. I 2p.u. Icp.u.

E1p.u.

jXmp.u.

Rcp.u.

E2p.u.

Figure 18 :Per Unit Model of Y-Y and - connected transformers.

I1p.u.

jXeq1.p.u. Req1.p.u.

Req2.p.u. jXeq2.p.u. Icp.u.

I2p.u.

E1p.u.

jXmp.u.

Rcp.u.

E2p.u.

ej30 : 1
Figure 19 :Per Unit Model of Y- connected transformers.

2.2 Per Unit Equivalent Models.

Recall that when the Per Unit system is used, and when the ratio of voltage bases on either side of the transformer matches the transformer turns ratio, then it is sufficient to model the transformer by its equivalent Per Unit impedance model. For three phase applications it is now necessary to consider how the transformer should be modelled, accounting for phase shifts from primary side to secondary side. For Y-Y and - transformers, since there is no phase shift between the primary and secondary side voltages it is sufficient to apply the per unit model presented for single phase transformers and shown in figure 18. For transformers that introduce a phase shift, such as the Y- transformer, it is a simple matter to include an ideal transformer in the model, that has a unity magnitude turns ratio, but with a 300 phase adjustment. This is illustrated in figure 18. Note that the convention applied to phase shifting transformers is that the positive sequence quantities on the high voltage side lead the equivalent positive sequence quantities on the low voltage side by 300. In this way it is not necessary to memorise a series of models for the many configurations possible. It is simply a matter of determining whether it is the positive sequence that is of interest, and then the Per Unit ideal transformer turns ratio is selected to ensure that the high voltage quantities lead.

12

ELEC 4100

LOAD FLOW

ELEC 4100 ELECTRICAL ENERGY SYSTEMS POWER FLOW ANALYSIS LOAD FLOW. 1 Introduction.
Load flow or power flow analysis is used to investigate the steady state operation of power systems for both system design and operational planning purposes. The main constraints on the operation of a power network under steady state conditions are: The generators must supply the loads plus all network losses. Bus voltage magnitudes must remain close to the rated values ( < 6% from the nominal). The generators must operate within the specified real and reactive power limits. Transmission lines and transformers must not be overloaded.

Load flow analysis is a standard power systems analysis task where the voltage magnitudes and angles at every bus in the system are computed for given loading and supply conditions. Load flow analysis also involves the calculation of the real and reactive power flows on all equipment which connect buses within the network as well as all equipment losses. In load flow analysis it is normal to assume that the system is balanced and that the network is composed of constant linear lumped elements (i.e. no distributed transmission line models). The most basic form of load flow analysis also assumes that the transformer tap settings are fixed, although for a real utility this assumption is relaxed. Load flow also assumes that the supply consists of the positive sequence only. Nodal analysis is generally used to describe the network, however closed form solutions to the nodal equations is difficult. This is because the loads at each bus are described by real and reactive power values, and not impedances. Similarly generators are also described by real and reactive power values as opposed to voltage or current sources. This leads to a set of non-linear equations, and this will be illustrated for a simple two bus network. Figure 1 shows a simple two bus power system with a generator connected to bus 1, supplying P1+jQ1, while the load is connected to bus 2 and draws P2+jQ2. The load is supplied through a lossy transmission line with an equivalent impedance R+jX. The load flow problem is to calculate the voltage (magnitude and angle) at the load bus, given that the bus 1 voltage is fixed. The power at the load end is given by:
P2 + jQ2 = V2 I

(1)

This can be rearranged as:

I = P2 jQ2 V2

(2)

V1 00 R+jX P1+jQ1
Bus 1

V2 0

I 0

Bus 2

P2+jQ2

Figure 1 : Simple Two Bus Power System. 1

ELEC 4100 At the generator end:

V1 = V2 + (R + jX )I

LOAD FLOW

(3)

or :
V1 V2 cos( ) = V2 + (RP2 + XQ2 )
2

(4a) (4b)

V1 V2 sin ( ) = ( XP2 RQ2 )

The two equations given in (4) defines the load end voltage, but the equation is clearly nonlinear. Conventional circuit analysis is not suitable to solve the power flow problem. Numerical techniques are required to solve this set of equations.

2 Constraints at Nodes.
Each node within the power system has four variables associated with it. These are: The bus voltage magnitude (V), The bus voltage angle (), The real power flow through the bus (P), The reactive power flow through the bus (Q).

Also, as we shall show later, each node introduces two equations, namely the real and reactive power balance equations. To obtain (isolated) solutions for a set of simultaneous equations, it is necessary to have the same number of equations as unknowns. Therefore two of the variables associated with each bus must be specified, i.e. given fixed values. The remaining variables are then free to vary during the solution process. The traditional way of specifying busbar quantities allows buses to be identified as follows: PQ Busbar : These buses are sometimes referred to as load buses. The net active and reactive powers are specified. The net power entering a busbar is the power supplied to the system from a generating source minus the power consumed by the load at that busbar. PV Busbar : These are often called voltage controlled buses. Here the net active power is specified, and the voltage magnitude is also known. The net reactive power is not known, and must be determined as part of the power flow solution process. This type of busbar typically represents a node in the system at which a synchronous source (generator or compensator) is connected, and where the sources reactive power output is varied to control the voltage magnitude to a specified value. Note that the reactive power is often constrained to lie with specific VAR limits. Slack or Swing Bus : On the Slack bus the voltage magnitude and angle are specified. Generally the voltage angle is set to zero. Unlike the other two bus types, which represent physical system conditions (e.g. load on a bus or source on a bus), this bus is used to satisfy a mathematical requirement. It is simply needed as a reference to which all other angles in the network can be set to. Furthermore this bus absorbs any real power mismatch that occurs across the system. (Note that it is not possible to specify the net active power at all buses in the system because transmission losses are unknown until the power flow solution is complete). Normally there can only be only one slack bus bar in the system. It is generally chosen from among the voltage controlled bus bars.

Note that even though these bus types are the most common, others are also possible, eg. Slack buses with constrained reactive power. In fact they are required at times. Also note that in general 2

ELEC 4100

LOAD FLOW

transformer taps are not fixed, but they vary to regulate the bus voltage. This introduces other bus types. However, so that the important aspects of power flows are clearly highlighted, only PQ, PV and slack buses will be considered here.

2.1 Minimum Data Requirements.

The minimum data required to fully specify system conditions is: Impedances (usually in Per Unit) for all series and shunt branches of the transmission network. Network elements are represented as lumped complex impedances at rated frequency (e.g. transmission lines, in-phase transformers, series and shunt reactors and capacitors). Transmission lines with non-negligible charging capacitance are represented by their simple equivalent networks. Active power (PQ and PV) and reactive power (PQ) busbar generations and loads. Voltage magnitudes at PV and slack bus bars.

2.2 Solution Outputs.

A load flow solution typically provides the following : Voltage magnitude and angle at each busbar. Real and Reactive power generation and load at each busbar. Power flows and MVA loadings at both ends of each transmission line and transformer. Power generation or consumption of each static shunt compensating device. Total system losses.

Note that load flow is a steady state solution tool. It does not provide any dynamic performance indication for the electrical network.

3 Component Models.
Load flow studies provide the planning engineer with information concerning the system when it is operating under normal balanced steady state conditions (dynamics are not considered). The mathematical model for the different components that are needed, therefore, are per phase models that apply under normal steady state conditions. The components of a load flow problem are : Generators Loads Transmission Lines Synchronous Condensers. Static Capacitors. Reactors. Voltage Regulators. Phase Shifters, etc.

3.1 Generators.
Since load flow is concerned with steady state characteristics, it is not necessary to model generators to any significant degree of complexity. Generators can be regarded as simple sources 3

ELEC 4100

LOAD FLOW

that can supply a demanded real power and reactive power within certain maximum and minimum limits. Similarly the generator can produce the required terminal voltage, again within certain limits.

3.2 Loads.
In power flow studies loads are assumed to be constant in magnitude at a given voltage bus. However it is instructive to note that the loads in a power system can be broadly classified according to the general character of the end use as: 3.2.1 Residential or Domestic Loads: Energy used in the home. This type of load is dispersed over large geographical areas in the power network. This type of load shows daily and seasonal fluctuations. The daily cycle typically shows two peak demands, in the morning and again in the evening. 3.2.2 Industrial Loads: These loads use large amounts of energy for manufacturing and large scale processes. This type of load is usually localised at relatively few points in the system, and does not fluctuate significantly in general. 3.2.3 Commercial Loads: In between residential and industrial loads. This type of load may have the variety of residential devices, but also requires moderately large amounts of energy for lighting, heating, and cooling large areas. These loads are more dispersed than industrial loads, but not to the same extent as residential. Key examples include hospitals, airports, shopping centres etc.

3.3 Transmission Lines.

Transmission lines are the main energy corridor in power systems. They transfer energy from the generating buses to the different load buses where the actual loads connect. The magnetic field, electric field and ohmic resistance are the three main phenomena that need to considered, in order of importance, when modelling a transmission line for load flow purposes. The magnetic field and the ohmic resistance lead to a series volt-drop across the transmission line, while the electric field effect leads to leakage currents to earth, and this can be modelled as a shunt admittance to earth. A transmission line consists of distributed circuit elements, as shown in figure 2. The distributed circuit elements represent the series resistance per unit length r, the series inductance per unit length L, the shunt conductance per unit length G, and the shunt capacitance per unit length C. The analytical determination of these parameters will be addressed later in this course. In figure 2, x represents the distance measured from the sending end of the transmission line. Therefore :
v(0, t ) = Vs (t ) = sending end voltage v(d , t ) = Vr (t ) = receiving end voltage

(5a) (5b)

i(0,t)

i(x,t) Ldx Rdx

i(x+dx,t)

i(d,t)

v(0,t)

v(x,t) Cdx

Gdx

v(x+dx,t)

v(d,t)

Figure 2 : A Distributed Parameter Transmission Line. 4

ELEC 4100
i(0, t ) = I s (t ) = sending end current i(d , t ) = I r (t ) = receiving end current

LOAD FLOW (5c) (5d)

The following equations can be derived from figure 2.

v(x, t ) i(x, t ) = ri(x, t ) L x t

(6a) (6b)

i(x, t ) v(x, t ) = Gv(x, t ) C x t

In the above equations the signals v(x,t) and i(x,t) are the voltage and current waveforms at the time t, and distance x along the transmission line. For load flow we are interested in a steady state relationship, and not a dynamic relationship, so it is possible to simplify these expressions. Consider a steady state waveform as follows:
v(x, t ) = V (x )e jt

(7a) (7a)

i(x, t ) = I (x )e jt

Substituting these expressions into the differential equations in 6, gives:

e jt dV (x ) = [ rI (x ) jLI (x )]e jt dx

(8a) (8b)

e jt

dI (x ) = [ GV (x ) jCV (x )]e jt dx

By removing the time dependence from the above equations, we get two ordinary coupled differential equations as:
dV (x ) = [ r jL ] I (x ) = zI (x ) dx
dI (x ) = [ G jC ] V (x ) = yV (x ) dx

(9a) (9b)

So here z = r +jL represents the equivalent series impedance per unit length, while y = G +jC represents the equivalent shunt admittance per unit length. Differentiating both expressions with respect to x and combining yields:
d 2V (x ) dx
2

= z = y

dI (x ) = zyV (x ) dx dV (x ) = zyI (x ) dx

(10a) (10b)

d 2 I (x ) dx
2

The solutions to the above equations are :

V (x ) = Vs cosh ( x ) Z C I s sinh ( x ) Vs I (x ) = I s cosh ( x ) Z C sinh ( x )

(11a) (11b)

Where the Zc and the are known as the characteristic impedance and the propagation constant respectively, and are related to z and y by:
Zc = z y

(12a)

ELEC 4100

LOAD FLOW

Is

Y1

Ir

Vs

Y2

Y3

Vr

Figure 3 : Equivalent Network Representation.

= zy = + j

(12b)

The real component (i.e. ) of the propagation constant represents the attenuation along the line, while the imaginary component (i.e. ) of the propagation constant represents the phase shift that occurs along the line. Since our only interest is what happens at the sending and receiving end of the transmission line, substitute the x = d into equations 12. This gives:
Vr = Vs cosh ( d ) Z C I s sinh ( d ) Vs I r = I s cosh ( d ) Z C sinh ( d )

(13a) (13b)

These expressions show that the relationship between the receiving end voltage and current, and the sending end voltage and current is a simple phase shift and attenuation. It may therefore be possible to replace the distributed model expressions with equivalent lumped element models. Figure 3 shows a - network, consisting of a series impedance and two shunt impedances. From figure 3 it can be shown that:
Vr = Vs

(I s VsY2 )
Y1

(14)

Rearranging:
Y I Vr = Vs 1 + 2 s Y1 Y1

(15)

Equating this expression with equation 13(a) gives:

Y1 =
1 Z C sinh ( d )

(16)

Using this value it is possible to get the value of Y2 as:

Y2 =

[cosh( d ) 1] = 1 tanh d Z C sinh ( d ) Z C 2

(17)

Similarly :
(I Y V ) I r = I s Y2Vs Y3Vr = I s Y2Vs Y3 Vs s 2 s Y1
Y2 Y3 Ir = Is Vs Y2 + Y3 1 + Y 1 + Y 1 1

(18) (19)

ELEC 4100 Equating this expression with equation () gives:

Y3 = Y1[cosh ( d ) 1] =

LOAD FLOW

[cosh( d ) 1] = 1 tanh d = Y2 Z C sinh ( d ) Z C 2

(20)

So the shunt branches are equal to one another, and we now have expressions for both the shunt branches and the series branch. This gives the model for the section representation of a transmission line. For short transmission lines of up to about 200km, generally d << 1 . Therefore the section model has the admittance branches :
Y1 = 1 Z C sinh ( d ) 1 z zy d y = 1 zd

(21)

Y2 = Y3 =

1 d 1 d tanh = ZC 2 ZC 2

y d yd zy = z 2 2

(22)

This is the simplified model suitable only for short transmission lines.

3.4 Reactors and Capacitor Banks.

A standard planning practice is to design transmission lines to be capable of carrying future loads that are typically two to five times the normal load. A nominal load the voltages at the sending end of a long transmission line, especially for EHV and UHV, increase to very high values due to large shunt branch charging currents. It can become impractical to close the circuit breakers to energise the substation transformers at this high voltage level. The solution to this problem is the introduction of three-phase reactors as loads at the receiving end of the line. These reactors compensate for the shunt capacitive effect of the transmission line which causes the high charging currents. These reactor banks are usually step adjustable, and are used with EHV and UHV lines. For load flow the reactor banks can either be represented by a lumped inductive element, or by a constant purely reactive load with the value:
QL = V2 V2 = XL L

(23)

In transmission lines where the series reactance is quite large, (e.g. old transmission lines designed for low voltage applications which are upgraded for higher voltage use) the reactive voltage drop across the line can be quite large. Similarly industrial plants that draw large reactive currents degrade the power factor, which means that the transmission line feeder must carry a large overall current. The combination can lead to very large voltage drops across the transmission line. One approach to overcome these problems is to use a three phase shunt capacitor bank. Capacitor banks are routinely used to improve the power factor at most industrial sites, and thereby reduce the overall current drawn. Capacitor banks can be modelled by either the lumped element, or by a constant purely reactive load with the value:
QL = V2 = CV 2 XC

(24)

Capacitor banks are routinely step adjustable, and part of the planning operation requires the determination of required setting to account for the daily and seasonal variation of the load.

ELEC 4100

LOAD FLOW

3.5 Synchronous Condensers and Static VAR Compensators.

Synchronous condensers and Static VAR Compensators (SVCs) are used to compensate for poor power factor loads at various points in the power network. Synchronous condensers are over-excited synchronous motors, operated with no real power output. In the over-excited state, the armature current of the synchronous motor leads the terminal voltage by 900, and the motor will absorb capacitive reactive power from the system. In other words it will appear that the motor supplies reactive power to the network. The motor will behave in exactly the opposite sense in the under-excited state, which means the motor will appear to sink reactive power from the network (i.e. an inductive load). A synchronous condenser can therefore be modelled as a synchronous generator that supplies reactive power, but with no capacity to supply real power to the network. SVCs behave in exactly the same fashion as a synchronous condenser, in that they can supply reactive power to the network, but do not have the capacitor to supply real power. However, SVCs are a power electronic solution, and use a converter to modulate the reactive energy. For load flow SVCs can be modelled in exactly the same fashion as synchronous condensers.

4 Nodal Analysis of the Transmission Network.

Conventional power flows assume balanced three phase networks. Therefore the network may ne represented in single phase (positive sequence) form. Circuit elements are usually represented in Per Unit by linear lumped complex impedances at rated system frequency, e.g. in-phase transformers, transmission lines, series and shunt capacitors and reactors. The transmission line is represented by its equivalent network. The equations describing the transmission network are therefore linear. In theory there are an infinite number of ways of describing analytically an electrical network, the common factor being that Kirchoffs current and voltage laws, and the volt-ampere relations of the network branches must be satisfied. In practice, two methods mesh (or loop) and nodal analysis have emerged as the most useful. Of these the latter has been found to be more suitable for power systems analysis.

4.1 The Basic Nodal Analysis Method.

As far as power networks are concerned, the major advantage of the nodal analysis technique may be listed as follows: Data preparation is straight forward. The number of variables and equations in the nodal representation is usually less than for the equivalent mesh representation. Parallel branches do not increase the number of variables or equations. Node voltages are available directly from the solution and branch currents are easily calculated. Off-nominal transformer taps can easily be represented.

In the application of the nodal method to power system networks, the variables are the complex node (busbar) voltages and currents, for which some reference must be designated. In fact, two different references are normally chosen : for voltage magnitudes the reference is Earth, and for voltage angles the reference is usually chosen as one of the busbar voltage angles which is fixed at the angle zero (generally). This busbar is also called the Swing or Slack bus, as discussed previously. A nodal current is the net current entering (injected into) the network at a given node, from a source and/or load external to the network. From this definition a current entering the network (from a source) is positive in sign, while a current leaving the network (to a load) is 8

ELEC 4100
I1 S1=E1I1* I2 S2=E2I2* I3 S3=E3I3*

LOAD FLOW

y12 E1 y10 E2

y23 y30 E3

Figure 4 : Simple Network showing nodal quantities. negative. The net nodal injected current is the algebraic sum if these. Nodal injected powers S = P+jQ are defined similarly. Figure 4 gives a simple network example showing the nodal currents, voltages and powers. In the nodal analysis method, it is convenient to use branch admittances rather than impedances. Denoting the voltages of nodes i and k as Ei and Ek respectively, and the admittance of the branch between them as yik, then the current flowing in this branch from node i to node k is given by:
I ik = yik (Ei Ek )

(25)

Then let the nodes in the network be numbered from 0,1, 2, " , n , where 0 denotes the reference node (earth). By Kirchoffs current law, the injected current Ii must be equal to the sum of the currents leaving node i, hence:
Ii =

k =0

I ik =

y
k =0

ik

(Ei Ek )

(26)

Since E0 is zero, and if the system is linear, then:

Ii =

k =0 k i

yik Ei

y
k =1 k i

ik Ek

(27)

Similar expressions can be written for each node in the network excluding the reference, allowing the matrix expression to be developed:
I1 Y11 Y12 " Y1n E1 I 2 = Y21 Y22 " Y2 n E2 # # # # # I n Yn1 Yn 2 " Ynn En

(28)

Where:
Yii =

y
k =0 k i

ik

self admittance of node i.

(29a) (29b)

Yik = yik =

mutual admittance between nodes i and k.

In matrix notation equation 28 can be conveniently written as:

I bus = Y bus Ebus

(30) 9

ELEC 4100 Or in summation notation:

Ii =

LOAD FLOW

Y
k =1

ik Ek

for

i = 1, " , n

(31)

The nodal admittance matrix in 28 and 30, has a well defined structure which makes it very easy to construct. This matrix has the following properties: It is square of order n n . It is symmetrical since Yik = Yki. The entries may be complex. The diagonal elements Yii, are equal to the sum of all admittances of the branches which terminate at node i, including any branches to ground. Each off diagonal element Yik, is the negative of the branch admittance between the nodes i and k. Very few non-zero mutual admittances exist in practical networks. Therefore the matrix Ybus is highly sparse.

4.1.1 Example: The admittance matrices for the example networks shown in figure 5 can be developed as by inspection, simply by applying the rules outlined above. So for the network in figure 5(a) the nodal expressions are:
Y1 V1 I1 Y1 + Y2 I = Y Y1 + Y3 1 2 V2

Similarly for the network shown in figure 5(b) the nodal expressions are:
Y5 Y4 I1 Y1 + Y4 + Y5 V1 = Y6 Y5 Y2 + Y5 + Y6 I2 V2 + + I Y Y Y Y Y 4 6 3 4 6 3 V3

4.2 Conditioning the Admittance Matrix.

The set of nodal equations I bus = Y bus Ebus may or may not have a solution. If not, there is
V1 Y1 Y5 Y4 Y6

Y1 V1 Y2 Y3 V2

V2 Y2

V3 Y3

0
(a) Figure 5 : Example Networks. 10 (b)

ELEC 4100
I1 I2

LOAD FLOW

y12 E1 y13 y23 E2

E3 0

I3

Figure 6 : Example of a Singular Network. generally a simple physical explanation relating to the formulation of the network description. Any numerical attempt to solve such equations is bound to break down at some stage of the process. (In practice this usually results in a finite number being divided by zero.) The example network in figure 6 illustrates this. The nodal equations are constructed in the usual way with the result:
I1 y12 + y13 I = YE = I 2 = y12 y13 I3 y12 y12 + y23 y23 y13 E1 y23 E2 y13 + y23 E3

(32)

Suppose that the injected currents are known and the nodal voltages are unknown. In this case, no unique solution for the latter is possible. The Y matrix is singular (i.e. has no inverse). This condition is easily detected in this example, since the sum of the elements in each row and column is zero. This is a sufficient condition for singularity. Hence, if it is not possible to express the voltages as E = Y-1I, then it is clearly not possible to solve (32) by any method, whether involving inversion of the admittance matrix or otherwise. The reason is that we are attempting to solve a network whose reference is disconnected, so there is no effective reference node in the network. This is why there exists an infinite number of possible solutions. However when a shunt admittance from at least one busbar is connected to the reference then the problem of insolubility immediately vanishes (in theory, but not necessarily in practice). Practical computation can not be performed with absolute accuracy, and during a sequence of arithmetic operations, rounding errors due to working with a finite number of decimal places accumulate. If the problem is well-conditioned and the numerical solution technique is suitable, these errors remain small and do not mask the eventual results. If the problem is ill-conditioned, and this usually depends upon the properties of the system being analysed, any computational errors introduced are likely to become large with respect to the true solution. It can be seen intuitively that if a network having zero shunt admittances cannot be solved, when working with absolute computational accuracy, then a network having very small shunt admittances may well present difficulties when working with limited computational accuracy. This reasoning provides a key to the practical problems of network (Y matrix) conditioning. A network with shunt admittances which are small with respect to the other branch admittances is likely to be illconditioned. The conditioning tends to improve with the size of the shunt admittances, i.e. with the electrical connection between the network busbars and the reference node.

11

ELEC 4100

LOAD FLOW

4.3 The Case Where One Voltage is Known.

In load flow, it is normal for at least one voltage in the network to be specified, with the current at that busbar to be determined. This immediately alleviates the problem of needing at least one good connection to ground, since a fixed busbar voltage can be interpreted as an infinitely strong ground tie. If it is represented as a voltage source with a series impedance of zero, then when converted to the Norton equivalent, the fictitious shunt admittance is infinite, as is the injected current. This approach is not computationally feasible, however. The usual way to deal with a voltage which is fixed is to effectively eliminate it as a variable from the nodal equations. For the sake of analytical convenience, let this busbar be numbered 1 in an n-busbar network. The nodal Equations are then:
I1 = Y11E1 + Y12 E2 + "Y1n En I 2 = Y21E1 + Y22 E2 + "Y2 n En # I n = Yn1E1 + Yn 2 E2 + "Ynn En

(33)

However note that the terms involving E1 on the right hand side of equation 33 are known, and so can be transferred to the left hand side of equation 33 as:
I1 Y11 E1 = Y12 E2 + "Y1n En I 2 Y21 E1 = Y22 E2 + "Y2 n En # I n Yn1E1 = Yn 2 E2 + "Ynn En

(34)

The first row of this equation set may now be neglected, leaving (n-1) equations with (n-1) unknowns : E2 to En. In matrix form this becomes:
I 2 Y21E1 Y22 " Y2 n E2 = # # # # I n Yn1E1 Yn 2 " Ynn En

(35)

Or :
=Y E I

(36)

is obtained from the full admittance matrix Y, merely by removing the row The new matrix Y and column corresponding to the fixed voltage busbar. In the present case this refers to the first row and column, though this need not be the case in general.

In summation notation, the new equations are:

I i Yi1E1 =

Y
k =2

ik Ek

for i = 2 , " , n

(37)

This is a set of (n-1) equations with (n-1) unknowns. The equations are then solved for the unknown voltages by any suitable solution technique. Note that the problem of singularity when there are no ground ties disappears if one row and column are removed from the Y matrix. Eliminating the unknown current I1 and the equation in which it appears is the simplest way of dealing with the singularity problem, and reduces the order of the equations by one. I1 is evaluated after the solution using the first equation in (28).

12

ELEC 4100

LOAD FLOW

5 Analytical Definition of the Load Flow Problem.

The power flow into any bus of the power network can be written as follows:
Si = Pi + jQi

(38)

and:
Si = Vi I i

(39)

However since we have a linear relationship between the nodal voltages and currents, equation (39) can be re-written as:
S i = Pi jQi = Vi

y V
k =1

ik k

(40)

With:
Vi = Vi i

(41) (42)

yik = yik ik

Then :
Si = Pi jQi =

y
k =1

ik

Vi Vk ( k i + ik )

(43)

Or:
Pi =

y
k =1

ik

Vi Vk cos( k i + ik ) =

y
k =1 n k =1

ik

Vi Vk cos( i k ik ) Vi Vk sin ( i k ik )

(44a)

Qi =

k =1

yik Vi Vk sin ( k i + ik ) =

ik

(44a)

These expressions relate the real and reactive power at the nodes in the power network to the voltage magnitudes and angles at each node in the network, as well as the elements of the global network admittance matrix. Figure 7 shows a bus within the power network, and it is clear that the power injected into the bus through generation, the power leaving the bus as a load, and the power entering the network are related according to:
Si = PGi PLi + j (QGi QLi )

(45)

Combining equations 44 and 45:

PGi+jQGi Ei i Pi+jQi

PLi+jQLi

Figure 7 : Bus i of a Power Network. 13

ELEC 4100

LOAD FLOW

Pi = PGi PLi =

y
k =1 n k =1

ik

Vi Vk cos( i k ik ) Vi Vk sin ( i k ik )

(46a)

Qi = QGi QLi =

ik

(46a)

These are the general load flow equations and for an n bus system there are 2n such equations. It is necessary to solve these expressions according to the bus type. Swing bus Solve for Pi and Qi. Load bus Solve for Vi and i Voltage controlled bus Solve for Qi and i .

Note that this formulation does not require a solution of equation (28) to get the load current first. However equation (46) is non-linear, and so the solution of this equation is non-trivial.

6 Gauss-Seidel Method of Solution.

The Gauss and Gauss-Seidel methods is an iterative technique for solving a set of equations. The technique relies on formulating the problem as:
x = h( x )

(47)

Where h is a vector of functions h1 , " , hn in the variables x1 , " , xn . This is called a fixed point formulation. In the Gauss method (47) is solved iteratively with the (p+1)th iteration is given by:
x p +1 = h x p

( ) (

(48)

That is:

x1 p +1 = h1 x1 p , x2 p , " , xn p xn
p +1

) ) )
(49)

# = hn x1 p , x2 p , " , xn p

The Gauss-Seidel method is a modification to this technique in which the updated values for the xn variables are incorporated into the solution process as soon as they become available. That is:
x1 p +1 = x2 p +1 = xi p +1 xn p +1

( h (x
2

h1 x1 p , x2 p , " , xn p
p +1 p 1

, x2 , " , xn p

) )
(50)

# p +1 p = hi x1 p +1 , x2 p +1 , " xip 1 , xi , " , xn # p p +1 = hn x1 p +1 , x2 p +1 , " xn 1 , xn

( (

The advantage of the Gauss-Seidel method is that by using the most recent value for the state variables there is less data to store in memory, and this has advantages in coding the algorithm, and also in terms of efficiency of implementation. Note also that the Gauss-Seidel method generally converges faster than Gauss (though this is not a general result), but it should be noted that convergence is not guaranteed. An iterative solution process can be regarded as a discrete time dynamical system, and as such there is a region of convergence around the solution point. In non-linear problems, such as the power flow problem, there may be more than one solution point. If the iterative solution procedure starts outside the convergence region of a particular solution point, then the process will diverge, or converge to a different solution point. 14

ELEC 4100 6.1.1 Example : Application to Linear Systems. Consider the linear system of equations:
A x=b

LOAD FLOW

Where A is an N N matrix, while x and b are N vectors. With an initial approximation x1 to the unknown vector x , then a new approximation can be computed by applying the Gauss method according to:
xi p +1 = bi aik p xk aii k =1 aii
k i

i = 1, 2, " , n ,

p = 1, 2, "

Similarly the Gauss-Seidel approximation can be developed as:

xi p +1 = bi aik p aik p +1 xk xk aii k = i +1 aii a k =1 ii

i 1

p = 1, 2, "

7 Application of Gauss-Seidel to the Power Flow Problem.

To apply the Gauss-Seidel method to the load flow problem, consider the power flow problem as it relates to each bus type in the network. 7.1.1 PQ Bus. From equation (40) the total power at a PQ bus (i.e. a load type bus) is given by:
Si = Pi jQi = Vi

y V
k =1

ik k

(51)

This expression can be rewritten as :

Pi jQi Vi

y V
k =1 k i

ik k

+ yiiVi

(52)

This gives:
n 1 Pi jQi Vi = yikVk yii Vi k =1 k i

(53)

So the Gauss-Seidel implementation of a voltage calculation is:

Vi p +1 = 1 yii
n P p jQ p i 1 p +1 p i i y V y V ik k ik k Vp k =1 k = i +1 i

( )

(54)

So for a PQ bus, equation (54) calculates the voltage magnitude and angle for the ith bus in the network. 7.1.2 PV Busbar. For a PV bus (i.e. a voltage controlled bus) the voltage angle and the reactive power must be calculated for a defined real power and voltage magnitude. The reactive power is calculated from equation 46(a) according to:
Qi =

y
k =1

ik

Vi Vk sin ( i k ik )

(55)

15

ELEC 4100

LOAD FLOW

The calculated value for the reactive power is then substituted into equation (54) to recalculate the voltage at the PV bus. However note that since the voltage magnitude at the ith bus is specified, this voltage magnitude calculated from equation (54) is reset, but the voltage angle is updated based on this calculation. 7.1.3 Slack Busbar. Since the voltage magnitude and angle is known for the slack bus, no calculation is needed. However the slack bus is used to make up the difference between the power generated by the machines in the network and the sum of the network loads and losses. Hence when convergence has been achieved it is necessary to calculate the swing bus total power. 7.1.4 Iteration Commencement and Termination. Usually when the Gauss-Seidel method is applied to the power flow problem, the initial guess for the bus voltages is set to the so-called flat start, in which:
Vi = 1.0 p.u. 00

for all load buses, and the slack bus. for all voltage controlled buses.

Vi = Vspec p.u. 00

The convergence is tested after each iteration, and terminated when the specified level of convergence has been attained according to:
Vi p +1 Vi p Vi p +1 < tolerance

(56)

7.1.5 Aids to Convergence. It is important to be able to control the rate at which states vary each iteration, i.e. the magnitude of the updated x p = x p +1 x p . At times x p can be small, causing a slow rate of convergence, yet at other times x p can be so large that divergence results. This control can be achieved by scaling the calculated update so that:
p p x act = x calc

(57)

p p Where x calc is the update given by equation (48), and x act is the update that is actually applied to the state variables. The multiplier is called the acceleration factor. Notice from (48) that p x calc = x p +1 x p

= h xp xp

( )
=

(58)

So from (57) :
p +1 x act

p +1 p x act x act = h xp xp p x act

(( ) ) + (h(x ) x )
p p

(59)

For convenience this illustration of the acceleration factor concept has been based on the Gauss method. It is easily extended to the Gauss-Seidel approach. 7.1.6 Convergence Behaviour. The Gauss-Seidel method is in general quite slow. This is because at each iteration the updates are calculated using only local information. From equation (54) it can be seen that the update of each voltage is based only on the old value of that voltage, together with those voltages at adjacent buses. Therefore, at each iteration the effects of a change at a bus can only be passed on to a limited number of other buses. (Depending on the ordering of the buses, this may only be the adjacent buses). So the effects can only slowly propagate throughout the network. In large radially connected 16

ELEC 4100

LOAD FLOW

networks, rounding error effects can actually overtake the solution process so that convergence is never achieved.

8 Newton-Raphson Method of Solution.

The Newton-Raphson method of solving a set of non-linear equations is frequently used in load flow analysis, and has a better convergence rate than either of the Gauss and Gauss-Seidel methods. Consider a set of N-functions which can be expressed as:
f1 (x1 , x2 , " , x N ) f (x , x , " , x ) N f (x ) = 2 1 2 # f N (x1 , x2 , " , x N )

(60)

Where the f1 , f 2 , " , f N are each non-linear functions of x1 , x2 , " , x N . The system of equations to be solved can be written as:
f (x ) = y

(61)

Where y is an N-vector. Let x0 be an approximation to x and then by expanding (61) as a Taylor series we can write:
y = f ( x ) = f (x 0 + x ) = f (x 0 ) + df dx
x0

x + "

(62)

By neglecting the higher order terms it is possible to estimate the x as:

d f x = dx x0
1

[y f (x )]
0

(63)

Then :
x = x0 + x

(64)
1

Therefore:
d f x = x0 + dx x0

[y f (x )]
0

(65)

This expression can be rewritten in the iterative form:

x p + 1 = x p + J 1 x p y f x p

( )[

( )]
"

(66)

Here the matrix J(x) is called the square Jacobian matrix of f(x) and is defined as:
f1 x 1 d f (x ) f 2 = x1 J (x ) = dx # f n x1 f1 x2 f 2 x2 # f n x2 f1 xn f 2 " xn # f n " xn

(67)

By this definition, it is clear that at each iteration of the Newton-Raphson method, the non-linear problem is approximated by a linear set of equations. 17

ELEC 4100

LOAD FLOW

f(x)

tangent to f(x)

xp

f(x)= y

xp+1 solution

xp

Figure 8 : Graphical Representation of the Newton-Raphson Method. The linear approximation employed in the Newton-Raphson method can best be visualised considering the single variable problem f(x) = y, as shown in figure 8. Figure 8 shows a graphical interpretation of the problem, in which one iteration step has been shown. Here the gradient of the function at the iteration step p is used to calculate the x-value at the next iteration step according to x p = [ f (x p ) y ] f (x p ) . The Newton-Raphson method converges rapidly (quadratically) if the initial estimate is good, and if the function f(x) is well behaved.

9 Application of Newton-Raphson Method to Load Flow.

From equation 46 we have:
PGi PLi =

y
k =1 n k =1

ik

Vi Vk cos( i k ik ) Vi Vk sin ( i k ik )

(68a)

QGi QLi =

ik

(68a)

At each bus the power mismatch is the difference between the generated power input and the sum of the load power and the power flowing to the other buses:
f i = Pi = PGi PLi

k =1 n

yik Vi Vk cos( i k ik ) = Pi

y
k =1 n k =1

ik

Vi Vk cos( i k ik ) Vi Vk sin ( i k ik )

(69a)

g i = Qi = QGi QLi

y
k =1

ik

Vi Vk sin ( i k ik ) = Qi

ik

(69a)

Equation (69) represents the functions that can be used in the Newton-Raphson method. Note that the power mismatches Pi and Qi will be zero for the values of the bus voltages and bus power flows corresponding to the required solution. Therefore the aim of the Newton-Raphson method is to determine the bus voltages such that the power mismatches in equation (69) are zero. So applying the Newton-Raphson expressions of equation (66), it is possible to write equation (69) as:

18

ELEC 4100
f p +1 p p +1 = p g V V f p p p V P = p + p p g Q V V V
1

LOAD FLOW

(70)

Alternatively this can be expressed as:

f P p p = g Q f p p V J 1 = p p g V J 3 V J 2 p p p J4p V

(71)

Where P p are the real power mismatches at all PQ and PV buses, Q p are the reactive power mismatches at all PQ buses, p are the voltage angle corrections for all PQ and PV buses, and V p are the voltage magnitude corrections for all PQ buses. Note that since a PQ bus has the real and reactive power specified, it is necessary to calculate both the voltage angle and magnitude, but for PV buses the voltage magnitude is specified so it is only necessary to calculate the voltage angle. Similarly since both the voltage angle and magnitude is specified for the slack bus no voltage calculation is required. This means that the Jacobian matrix in equation (71) is somewhat simplified, and is given by:
f 2 2 f 3 = 2 # f n 2
g m +1 2 g m + 2 = 2 # g n 2

J1p

f 2 3 f 3 3 # f n 3

f 2 n f " 3 n # f " n n "

g m +1 n g " m+ 2 n # g " n n "

J2p

f 2 V m +1 f 3 = Vm +1 # f n V m +1 g m +1 V m +1 g m + 2 = Vm +1 # g n V m +1

f 2 Vm + 2 f 3 Vm + 2 # f n Vm + 2 g m +1 Vm + 2 g m+2 Vm + 2 # g n Vm + 2

f 2 Vn f " 3 Vn # f n " Vn " "

(72a)

J 3p

g m +1 3 g m + 2 3 # g n 3

J4p

g m +1 Vn g m + 2 " Vn # g " n Vn

(72b)

Note that the Swing bus is denoted bus number 1, and as such is not included in the Jacobian matrix. Buses 2 to m are the PV or voltage controlled buses, while buses m+1 to n are the PQ or load buses. Also note that in equation (72) the partial derivatives in the Jacobian matrix have been evaluated at the pth iteration values. The partial derivatives in the Jacobian matrix are defined according to equation (69) with eight generic types. These are:
f i = yik Vi Vk sin ( i k ik ), k ik

(73) (74)

f i = yik Vi Vk sin ( i k ik ) i k =1

k i

f i = yik Vi cos( i k ik ), Vk

ik

(75) 19

ELEC 4100
f i = Vi

LOAD FLOW

y
k =1

ik

Vk cos( i k ik ) + yii Vi cos( ik )

k i

(76) (77) (78)

g i = yik Vi Vk cos( i k ik ), k g i = i

y
k =1 k i

ik

Vi Vk cos( i k ik )

g i = yik Vi sin ( i k ik ), Vk g i = Vi

k i

(79) (80)

y
k =1

ik

Vk sin ( i k ik ) yii Vi sin ( ii )

Note that all of the partial derivative elements of the Jacobian matrix in equations (71) through to (80) involve the branch admittance terms. Hence if there is no admittance branch between two buses i and k, then the Jacobian elements are zero. For real power networks where there is limited inter-connectivity between the nodes of the network, this means that the Jacobian matrix is highly sparse. So in a practical sense there may be advantages to exploiting sparsity solution methods, and this will be detailed later. The solution process using the Newton-Raphson methodology is as follows: 1. Initialise the bus voltages to either 10 or Vspec 0 . 2. Form the Jacobian matrix as outlined in equations (71) to (80), evaluated for the initial bus voltages and bus voltage angles. 3. Using equation (69) calculate the power mismatches Q and P . 4. Invert the Jacobian matrix and solve for:
p p p = J V

( )

1 P

p p Q

5. Determine the new estimates for the bus voltages and bus voltage angles:
p +1 = p + p ,
V p +1 = V p + V p

6. Calculate the bus real and reactive powers, and check the Q limits. Adjust if necessary. 7. Jump back to step 2 and repeat until the Q and P are within the tolerance limits. 8. Calculate the Swing bus real and reactive power.

9.1 Practical Solution Considerations.

The efficient solution of the Newton-Raphson expressions at each iteration is crucial to the success of the load flow analysis technique. If conventional matrix manipulation methods were to be exploited, the storage (an2) and computing time (an3) would be prohibitive for large systems. In practice, the sparsity of the Jacobian matrix can be exploited, using sparsity-programmed ordered elimination. In this scheme, the way in which the rows and columns of (29) are written is different to that shown, and only the non-zero elements of the Jacobian matrix are stored and operated upon. Alternative formulations of the Newton-Raphson load flow approach have also been used. One approach is to express Pi and Qi in terms of the busbar voltages in rectangular form rather than 20

ELEC 4100
f(x) unlimited xp tangent to f(x)

LOAD FLOW
f(x) unlimited xp tangent to f(x)

limited xp xp solution x

limited xp xp solution x

Figure 9 : Limitation of the Correction Update, and the aid to Convergence. the polar form. The Cartesian coordinates of the voltage then become the solution variables (rectangular power mismatch formulation). The expressions for the Pi and Qi are then partially differentiated with respect to each of the voltage Cartesian coordinates. Extra equations are needed for the PV busbar constraints to ensure the correct magnitude is maintained, and this gives rise to a larger and substantially different Jacobian matrix to the polar form derived above. Other variations may be formulated by using current mismatches rather than power mismatches. The complex current mismatch equation at a PQ busbar denoted i is obtained as:
I i = Pi Qi Ei

Y
k =1

ik Ek

(81)

The two coordinates of the current (obtained in either polar or Cartesian form) are then used to obtain the two required sets of equations. The convergence behaviour is different in each formulation.

9.2 Aids to Convergence.

The Newton-Raphson method can diverge very rapidly or converge to the wrong solution of the equations are not well behaved or if the starting voltages are poorly chosen. Such problems can often be overcome by a variety of techniques. The simplest device is to impose a limit on the side of each i and Vi update correction that is allowed at each iteration. Figure 9 illustrates two one dimensional cases that would diverge without the assistance of this limitation device. Another method is to obtain good starting values for the i and Vi , possibly from a previously obtained solution. This also reduces the number of iterations required.

10 Fast Decoupled Power Flow.

Numerical methods tend to be most successful when they take advantage of the physical properties of the system being solved. This idea can be applied to the Newton-Raphson (NR) power flow method (which is an entirely general mathematical approach), to produce new algorithms with various computational advantages. The decoupled methods approximate the NR power flow algorithm by using knowledge of the physical characteristics of electrical systems. The decoupling principle recognises that in the steady state, active powers are strongly related to voltage angles, and reactive powers to the voltage magnitudes. However, there is quite weak coupling between active powers and voltage magnitudes, and between reactive powers and voltage angles. Referring back to the Jacobian matrix equation (72) of the NR method, it is seen that the Jacobian sub-matrices J2 and J3 represent the weak P-V and Q- couplings. It can be shown that in general the elements of J2 and J3 are small compared to J1 and J4. 21

ELEC 4100

LOAD FLOW

If we therefore neglect the sub-matrices J2 and J3 we shall then have approximated the Jacobian matrix, and in particular will have decoupled the P-V and Q- problems. Equation (71) can be broken down into two separate matrix problems with:

[ P ] = [J ][ ] [Q ] = [J ][V ]
p p p

(82) (83)

This not only reduces the computation and storage for the construction and solution of these equations, but it enables successive-displacement iteration to be performed between (82) and (83). In this scheme (82) is first solved, and then is updated. With the new voltage angle data, (83) is then calculated to solve for the new voltage magnitudes. The initial rate of convergence of this block-successive type of scheme is very fast in the load flow application. It takes only a small number of iterations to reach normal convergence accuracies. However because the quadratic convergence property has been sacrificed by the decoupling approximation, a high accuracy solution may require many more iterations. (Note: The Jacobian matrix defines the directions in which the algorithm progresses. It does not make the solution more approximate simply because the Jacobian matrix has been approximated.) Some additional assumptions can now be made to further simplify the structure of the decoupled Jacobian sub-matrices. These assumptions include: The admittance angles are approximately 900. This assumes that the inductive elements of a network dominate over resistive elements. The angle between adjacent voltage buses is approximately the same. This means that the difference terms i - k = 0. The sparsity of the Jacobian matrix adds validity to this assumption since branch admittances connect only very few voltage buses.
f i = yik Vi Vk , k f i = yik Vi Vk i k =1

Hence the derivative elements within the relevant sub-Jacobian matrices become:
ik

(84) (85)

k i

g i = yik Vi , Vk g i = Vi

k i

(86) (87)

y
k =1

ik

Vk + yii Vi

With:
f 2 2 f 3 = 2 # f n 2 f 2 3 f 3 3 # f n 3 f 2 n f " 3 n # f " n n

"

J1p

J4p

g m +1 V m +1 g m + 2 = Vm +1 # g n V m +1

g m +1 Vm + 2 g m+2 Vm + 2 # g n Vm + 2

g m +1 Vn g " m+2 Vn # g n " Vn

"

(88)

The fast decoupled solution process then follows the same development as outlined in section 9.

22

ELEC 4100

LOAD FLOW

10.1 Comparison with the Newton-Raphson Method.

The main computational effort of the fast decoupled method is the calculation at each iteration step of the mismatch vectors Q and P . This is much less computation than is required by the NR method where the full Jacobian is built and factorised each iteration. Typically a NR iteration takes around 5 times as long as a fast decoupled iteration. However the fast decoupled method requires more iterations than the NR method, taking in the order of two times as many iterations for normal power systems with normal loading conditions. Consequently the fast decoupled method is much faster for normal systems and for moderate accuracy. Under these circumstances it is also very reliable. However if the system is stressed (i.e. is operating close to its limits), or if it contains a significant proportion of lines with high R/X ratios, then convergence of the fast decoupled method can become slow and unreliable. This occurs because the assumptions upon which the fast decoupled method is based are no longer valid. Since the NR technique does not rely on any such assumptions, it is more robust and will often converge reliably in situations where the fast decoupled method would not converge. If high accuracy is required for the solution, then the NR method is more suitable than the fast decoupled method. The NR method recalculates the Jacobian at each iteration, so near the solution point the voltage angle and magnitude updates always drive the process towards the solution point. In contrast the fast decoupled method uses an approximate relationship between the power mismatches and the voltage angle and magnitudes. It therefore cannot be guaranteed that at each iteration the voltage magnitude and angle updates will drive the solution process closer to the solution point. The values for the bus voltages obtained at each iteration step may in fact oscillate around the true solution point. Convergence is slowed, and in fact is not guaranteed. The fast decoupled method has an advantage over the NR method if storage requirements are critical. Since the fast decoupled method does not store the J2 and J3 sub-matrices, its storage requirements are typically only 70% of those required for the NR method.

23

ELEC 4100

SYMMETRICAL COMPONENTS

ELEC 4100 ELECTRICAL ENERGY SYSTEMS SYMMETRICAL COMPONENTS AND FAULT CALCULATIONS. 1 Introduction.
Under normal conditions, a power system network is usually considered to be operating in a balanced three phase sinusoidal steady state mode. Analysis of the network under these conditions commonly involves reduction to a single phase, line to neutral network representation, before any computation takes place. Most system load flow studies, for example, are computed on this basis. When a power system network is operating under unbalanced conditions (i.e. under fault conditions for example), it is no longer possible to analyse the network by a single line equivalent circuit representation, and more powerful analysis techniques must be used. One of the most widely accepted calculation methods for solving unbalanced systems involves the use of symmetrical components. This method is particularly useful when a network has limited impedance unbalance at one or two points, and the source voltages are balanced. An unbalanced fault on an other-wise balanced network falls into this category. The use of the symmetrical component method as applied to unbalanced faults, or in the application of unbalanced voltages to rotating machinery, allows the problem to be reduced to a consideration of combinations of balanced three phase networks, which can in turn be represented in equivalent single-phase form. Analysis by symmetrical components can also be extended to deal with networks having a general impedance and source voltage unbalance. The method then requires the resolution of the network impedances into their symmetrical components. However, under these circumstances the method of symmetrical components offers no advantage over direct solution of the original network.

2 The Concept of Symmetrical Components.

2.1 General Concepts.
The general theorem of symmetrical components was first developed in 1918 by Dr C. L. Fortescue. Briefly, it states that any unbalanced set of three phase phasor quantities (usually voltage or current) can be resolved into three balanced systems of phasor quantities, namely: 1. A Positive-Sequence system consisting of a balanced set of three phase components, usually of the same phase sequence as the original unbalanced set (phase sequence A-BC-A-B-C ). Positive sequence phasors are equal in magnitude with 1200 phase angle displacements, and are denoted with the sub-script 1. 2. A Negative-Sequence system consisting of a balanced set of three phase components of opposite phase sequence to the positive-sequence set (phase sequence A-C-B-A-C-B ). Negative-sequence phasors are equal in magnitude with 1200 phase angle displacements, and are denoted by the sub-script 2. 3. A Zero-Sequence system consisting of three single-phase components, equal in magnitude and all having the same phase angle. Zero sequence phasors are denoted in these notes by the sub-script 0. It should be noted that the angle between the reference phasor of each of the sequence component systems and a common reference axis is in general undefined, and is determined by the way in which the sequence components combine to form any particular unbalanced set of phasor quantities.

ELEC 4100

SYMMETRICAL COMPONENTS

Ic1 Ib2 Ic Ia1 Ic2 Ia2 Ic0 Ib0 Ia0 Ib Ia

Ib1

Figure 1 : Symmetrical Components combined to create a set of unbalanced three phase phasor quantities.
For any defined set of sequence systems, the original unbalanced set of phasor quantities can be expressed as the phasor sum of the sequence components, viz: I a = I a 0 + I a1 + I a 2 I b = I b 0 + I b1 + I b 2 I c = I c 0 + I c1 + I c 2 These phasor combinations are illustrated in figure1. (1) (2) (3)

2.2 The operator a.

For convenience, a complex operator a (also denoted , h or ) is now defined as a unit phasor with a phase angle of 1200 and a magnitude of 1, i.e. a = 1120 = e
j 2 3

= 0.5 + j 0.866

(4)

Thus when a phasor I0 is multiplied by the operator a, the phasor is rotated through 1200 with no change in magnitude. Note that the operator -a does not represent a phase rotation of 1200, but rather is the product of the operators -1 and a, and hence represents a phase rotation of 1800 plus 1200, to give a total angle rotation of 3000 (or 600). The following functions of a are particularly useful when applied to symmetrical components:

a 2 = 1240 = 1 120 a 3 = 10

(5) (6) (7)

1 + a + a2 = 0

3 Relationship between phase and sequence components.

The components of the positive, negative and zero sequence systems defined in section 2.1 above have a rigid phase and magnitude relationship between the phasors of each system. Using the operator a and the nine a, b and c phase sequence components of the sequence systems described in (1), (2) and (3) can be expressed in terms of the three a phase sequence components as follows:
I b1 = a 2 I a1 I c1 = aI a1 I b 2 = aI a 2 Ib0 = I a0

(8) (9)

Ic2 = a2Ia2

Ic0 = I a0

Equations (1), (2) and (3) can be now be combined with equations (8) and (9), according to:
2

ELEC 4100

SYMMETRICAL COMPONENTS (10) (11) (12)

I a = I a 0 + I a1 + I a 2 I b = I a 0 + a 2 I a1 + aI a 2 I c = I a 0 + aI a1 + a 2 I a 2
In matrix form, these equations can be expressed as:
I a 1 1 I = 1 a 2 b Ic 1 a 1 I a0 a I a1 = a2 I a2 I a0 A I a1 I a2

(13)

Or:

I abc = A I 012

(14)

Inverting the matrix A, the inverse transformation from the phase components to sequence components can be obtained as :
1 1 I a0 I = 1 1 a a1 3 2 1 a Ia2 1 Ia a2 Ib a Ic

(15)

Or :

I 012 = A1 I abc
The above relationships also hold for three phase voltages, i.e.

(16)

Vabc = A V012 V012 = A1 Vabc

Where A is the symmetrical component transformation matrix, defined as:
1 1 2 A= 1 a 1 a 1 a a2 1 a2 a

(17) (18)

(19)

The matrix A-1 is the inverse symmetrical component transformation matrix, defined as:
1 1 1 A1 = 1 a 3 2 1 a

(20)

In a three-wire three phase system, the phase voltages and currents sum to zero at all times. Hence zero sequence voltages (Va0) and currents (Ia0) are all zero. If a fourth wire (neutral) or a connection to earth is provided, zero sequence currents can then flow, and from (13):
I a0 = 1 [I a + I b + I c ] = I N 3 3 (21)

Where IN is the neutral current, which by definition must be the sum of the three phase currents. Correspondingly, IN = 3Ia0 = 3Ib0 =3Ic0, for a four wire connection.

ELEC 4100

SYMMETRICAL COMPONENTS

4 Power Relationships.
At any defined point in a three phase power network, the total complex power is the sum of the complex powers of the individual phases. For phase voltages of Va, Vb, Vc and phase currents Ia, Ib, Ic, the complex power is given by:
S = Va I a + Vb I b + Vc I c

(22)

Where Ij* is complex conjugate of the current Ij. In matrix form this power expression can be written as:
S = [Va Vb Ia T Vc ] I b = [Vabc ] [I abc ] Ic

(23)

Substituting symmetrical components for Vabc and Iabc gives:

S = [ A.V012 ] T [ A. I 012 ] S = V012 A T A I 012
T

(24) (25)

By inspection :
AT= A

(26)

Also:
a = a 2 A = 3 A1

and

(a )

=a

(27) (28)

Hence:
S = V012 A.3. A1I 012
T

(29) (30)

S = 3V012 I 012
T

S = 3Va 0 I a 0 + 3Va1I a1 + 3Va 2 I a 2

(31)

This shows that the total power in an unbalanced three phase system is three times the sum of the total powers in each of the sequence systems. Note that the transformation from phase to symmetrical components is NOT power invariant. To obtain a power invariant transformation, the A matrix must be changed to:
1 1 1 A= 1 a2 3 1 a 1 1 1 1 a = A 3 2 1 a
1

1 a a2 1 a2 a

(32)

The inverse symmetrical component transformation matrix then becomes: (33)

The sequence components of voltage and current evaluated with: 4

ELEC 4100

SYMMETRICAL COMPONENTS

Ia

Vaa' Za Vbb' Zb Vcc' Zc

Ib

Xab

Ic

Xbc

Xac

Figure 2 : Unbalanced Mutually Coupled Impedances within a Power System Network.

1 V =A V 012 abc 1 I =A I 012 abc

(34)

are all

3 times greater than the Fortescue components found from:

I 012 = A1 I abc

V012 = A1 Vabc

(35)

The power invariant forms of the symmetrical component transformation matrix are useful in electric machine analysis, while the Fortescue forms entrenched in the analysis of power systems. The major application of symmetrical components is in the analysis of faulted power systems, and power calculations are seldom required.

5 Impedance Transformations.
One of the major advantages of the symmetrical component transformation is the way in which it allows the unbalanced operation of a three phase network to be considered as a combination of three single-phase networks. This advantage comes about because of the effect of the symmetrical transformation on the three-phase (unbalanced) impedance matrix of the original network.

5.1 Network Impedance Transformations.

Consider three unequal series impedances with mutual coupling which is in part of a three phase network as shown in figure 2. Each impedance has a current flowing through it, and a voltage drop across it as indicated. These voltages and currents are related by an impedance matrix of the form:
Vaa ' Z a V = X bb ' ab Vcc ' X ac X ab Zb X bc X ac I a X bc Ib Zc Ic

(36)

Or:
Vaa 'bb 'cc ' = Z abc I abc

(37)

Where the Za, Zb, Zc are the self impedances of the three series elements, and the Xab, Xbc, Xac are the mutual reactances between the elements. By transforming the voltages and currents of (37) to symmetrical component form, the impedance matrix Zabc can be reduced to a decoupled formulation for balanced phase impedances, viz.: 5

ELEC 4100
A.Vaa '012 = Z abc . A. I 012 Vaa '012 = A1. Z abc . A. I 012 Vaa '012 = Z 012 . I 012

SYMMETRICAL COMPONENTS (38)

(39) (40)

The matrix Z012 is a symmetrical component representation of the original impedance matrix, and can be expressed as:
Z 012 Z s 0 + 2Z m 0 = Z s1 Z m1 Z s2 Zm2 Z s1 Z m1 Z s0 Z m0 Z s1 + 2 Z m1 Z s1 Z m1 Z s1 + 2 Z m1 Z s 0 Z m0

(41)

Where:
Z s0 =

1 (Z a + Z b + Z c ) 3

Z m0 =

1 ( X ab + X bc + X ac ) 3

(42)

Z s1 = Zs2 =

1 Z a + aZ b + a 2 Z c 3 1 Z a + a 2 Z b + aZ c 3

) )

Z m1 = Z m2 =

1 X ab + aX bc + a 2 X ac 3 1 X ab + a 2 X bc + aX ac 3

) )

(43) (44)

If the series impedances are balanced, with equal mutual coupling between the phases: Z s1 = Z s 2 = 0 This reduces the Z012 matrix to :
Z 012 Z s 0 + 2Z m0 = 0 0 0 Z s 0 Z m0 0 Z0 =0 0 Z s 0 Z m0 0 0 0 Z1 0 0 0 Z2

Z m1 = Z m 2 = 0

(45)

(46)

The Z0, Z1, Z2 are commonly referred to as the zero sequence, positive sequence, and negative sequence impedances of the original impedance matrix, respectively. Note that when the self and mutual impedances are balanced in this way there is complete decoupling between the sequence voltages across, and currents through, the three phase impedances, i.e. from (40): Vaa '0 = Z 0 I a 0 Vaa '1 = Z1I a1 Vaa '2 = Z 2 I a 2 (47) (48) (49)

This is the condition that exists for most of a power system under fault conditions (viz: unbalanced currents flowing through balanced impedances), and hence the advantage of transforming a faulted network into symmetrical component form can be clearly seen.

5.2 Machine Impedance Matrix Transformation.

For a synchronously rotating machine, the impedance matrix coupling the phase voltages and currents is usually of the form:

ELEC 4100 Z = Xb X f Xf Z Xb Xb Xf Z

SYMMETRICAL COMPONENTS

Z abc

(50)

Where Z represents the self impedance in each phase, Xf and Xb represent the forward and reverse mutual couplings between the phases, respectively. Transforming the impedance matrix to symmetrical component form, gives: Z 012 = A1 Z abc A Z + X f + X b 0 = 0 0 Z + a X f + aX b 0
2

Z + aX f + a 2 X b 0 0

(51)

Again, the sequence systems of the original three phase network have been decoupled under unbalanced operation, although the sequence impedances developed are somewhat different.

5.3 Sequence Impedance Networks.

Power system networks have many items of plant interconnected together, each of which can be represented by a sequence impedance matrix. Since these plant items are balanced (usually), their sequence impedance matrices will be of the form expressed by (46) or (51). Hence the sequence voltage drops across each plant item due to unbalanced currents will be determined solely by the respective sequence impedances for each sequence current component as (47) to (49). For the complete network, with all plant items having a sequence impedance matrix of the form defined by (46) or (51), the total sequence voltage drops over all plant items will be the sum of the sequence voltage drops for each of the plant items. The network can therefore be described as being the interconnection of the positive, negative and zero sequence impedances of the plant items in the system. The operation of the network under unbalanced fault conditions is then determined by the interconnection of the sequence impedance networks of the system.

6 Analysis of Asymmetric faults on a Power System Network.

When a fault occurs at a particular point on a power system network, the network behind the fault can be represented as a set of three phase balanced voltage sources behind three balanced series impedances (with mutual coupling between the impedances), i.e. a three phase Thevenin equivalent voltage source. Unbalanced fault operation comes about because of the asymmetrical connection of the network at the point of the fault, which causes unbalanced currents to flow back through the Thevenin equivalent of the remaining network. This is illustrated in figure 3. For any fault condition, the network phase voltages at the point of fault may be calculated from the Thevenin equivalent source voltages for the network, less the unbalanced voltage drops across the network impedances caused by the unbalanced fault current. Hence the network phase voltages at the point of fault may be expressed as : Vabc = Eabc Z abc I abc Where : Vabc = The unbalanced phase voltages at the point of fault. Eabc = The network Thevenin equivalent source voltages. (52)

ELEC 4100

SYMMETRICAL COMPONENTS

Ia Ea Ec Eb Ib Ic
Thevenin Equivalent Network

Z Zfa Zfb Z Zfc

Zfn
Unbalanced Fault Impedances

Figure 3 : Unbalanced fault connected to the Thevenin Equivalent of a Power Network. Z abc = The network Thevenin equivalent impedance matrix. I abc = The unbalanced fault currents. The transformation to symmetrical components gives: A.V012 = A. E012 Z abc A. I 012 V012 = E012 A1 . Z abc A. I 012 V012 = E012 Z 012 I 012 (53)

(54) (54)

Where:
V012 = The unbalanced phase voltages at the point of fault in symmetrical component form. E012 = The network Thevenin equivalent source voltages in symmetrical component form. I 012 = The unbalanced fault currents in symmetrical component form. Z 012 = The transformed network Thevenin equivalent impedance matrix for the network. E012, the network Thevenin equivalent source voltage, is usually a set of balanced three phase voltages, and may be represented as a positive sequence voltage only. Z012 is the representation of the system impedance as three sequence impedances. The system impedance is the sum of the individual impedance elements in the network, each of which reduces to a sequence form similar to (46). Hence the overall lumped system sequence impedance matrix will have a similar form.

Substituting these expansions into (54) gives a relationship between the symmetrical components of the fault currents and the voltages at the point of fault as follows:
V012 Va 0 0 Z 0 I a 0 = Va1 = Ea Z1 I a1 Va 2 0 Z2 Ia2

(55)

ELEC 4100

SYMMETRICAL COMPONENTS

Ia Ea Ec Eb Ib Ic

Z Zf

Figure 4 : Single Line to Ground Fault Connected to a three phase system.

Where Ea is the positive sequence source voltage, and equals the original network phase voltage. Equation (55) can be evaluated for any particular fault condition by considering the constraints on the components of the fault currents and fault voltages, as discussed in the next sections.

6.1 Single Line to Ground Fault.

Figure 4 shows a three phase system subject to a single line to ground fault through an impedance. For this fault condition, Ib = Ic = 0, and Va = Zf Ia. Substituting these constraints into (15) and (18) gives:
I a 0 = I a1 = I a 2 = Ia 3

(56) (57)

Va 0 + Va1 + Va 2 = Z f I a

Combining equation (57) with (55) gives:

( Z 0 I a 0 ) + (Ea Z1 I a1 ) + ( Z 2 I a 2 ) = Z f I a
Ia1 Ea 3Zf Z1

(58)

Z2 Z0

Figure 5 : Interconnection of sequence impedances for single line to ground fault.

ELEC 4100

SYMMETRICAL COMPONENTS

Ia Ea Ec Eb Ib Ic

Zf

Figure 6 : Line to Line Fault connected to a three phase system.

Combining with equation (56) and rearranging gives:

I a 0 = I a1 = I a 2 = Ea (Z 0 + Z1 + Z 2 + 3 Z f

(59)

This result may be represented as the series interconnection of the sequence impedance networks of the faulted power system and three times the fault impedance, as shown in figure 5.

6.2 Line to Line Fault.

Figure 6 shows a three phase system subject to a line to line fault through an impedance Zf. For this fault condition, Ia = 0, Ib = -Ic, Vb = Vc + Zf Ib. Substituting these constraints into (15) and (18) gives:
I a0 = 0 , I a1 = I a 2

(60) (61)

a 2 Va1 + a Va 2 = a Va1 + a 2 Va 2 + Z f I b

Combining equation (61) with (55) gives:

(a

a (Ea Z1I a1 ) = a 2 a ( Z 2 I a 2 ) + Z f a 2 a I a1

(62)

Rearranging gives:

(Ea Z1I a1 ) = Z 2 I a 2 + Z f I a1
Substituting for Ia2 from (60) gives:
I a1 = I a 2 = Ea (Z1 + Z 2 + Z f )

(63)

(64)

This result may be represented as the series interconnection of the positive and negative sequence impedance networks of the faulted power system, and the fault impedance, as shown in figure 7. The zero sequence network is not connected to the supply voltage in this case since there is no zero sequence current flowing under this fault condition.

10

ELEC 4100

SYMMETRICAL COMPONENTS

Ia1 Ea

Z1

Z2 Z0

Zf

Figure 7 : Interconnection of Sequence Impedances for Line to Line fault.

6.3 Double Line to Ground Fault.

Figure 8 shows a three phase system subject to a double line to ground fault through an impedance Zf. For this fault condition, Ia = 0, Ib + Ic = If, and Vb = Vc = Zf If. Substituting these constraints into (15) and (18) gives:
I a 0 = I a 2 I a1 ,

3 Ia0 = I f

(65) (66)

Va 0 + a 2 Va1 + a Va 2 = Va 0 + a Va1 + a 2 Va 2 = Z f I f

Combining equation (66) with (55) gives:

(a

a (Ea Z1 I a1 ) = a 2 a ( Z 2 I a 2 )

(67)

Rearranging:

(Ea Z1 I a1 ) = Z 2 I a 2

(68)

Ia Ea Ec Eb Ib Ic

Z If Z Zf

Figure 8 : Double Line to Ground Fault connected to a three phase system.

11

ELEC 4100

SYMMETRICAL COMPONENTS

Ia1 Ea

Z1

3Zf Z2 Z0
Figure 9 : Interconnection of Sequence Impedances for a Double Line to Ground fault.

Similarly :
a 2 (Ea Z1 I a1 ) a Z 2 I a 2 = Z f I f + Z 0 I a 0 a 2 (Ea Z1 I a1 ) a 2 Z 2 I a 2 = Z f I f + Z 0 I a 0

(69) (70) (71)

Z 2 I a 2 = (3Z f + Z 0 )I a 0

Combining these expressions with (65):

I a1 = Ea (Z1 + Z 2 // (Z0 + 3Z f ))

(71)

I a1 =

Ea Z (Z + 3Z f ) Z1 + 2 0 + + Z Z 3 Z 2 0 f

(72)

This result can be represented as a series/parallel interconnection of the sequence impedance networks of the faulted power system, and the fault impedance, as shown in figure 9.

6.4 Balanced Three Phase Fault.

Figure 10 shows a three phase system subject to a balanced three phase fault. For this condition, Ia + Ib + Ic = 0, and Va = Vb = Vc. Substituting these constraints into (15) and (18) gives: I a0 = 0 Va 0 = Va1 = Va 2 = 0 Then from (55) : (73) (74)

(Ea Z1 I a1 ) = ( Z 2 I a 2 ) = ( Z 0 I a 0 ) = 0
Hence:

(75)

12

ELEC 4100

SYMMETRICAL COMPONENTS

Ia Ea Ec Eb Ib Ic

Figure 10 :Balanced three phase fault connected to a three phase system.

Ia1 Ea

Z1

Z2 Z0
Figure 11 :Interconnection of sequence impedances for three phase fault. I a1 = Ea , Z1 Ia2 = 0
(75)

As expected, this result can be represented as the simple connection of the positive sequence impedance network of the faulted power system, across the Thevenin source voltage as shown in figure 11.

6.5 Other Types of faults.

A similar approach can be taken to solve for other types of fault conditions applied to a three phase network, e.g. single-phase-open-circuit. For each type of fault, the approach is to establish the constraint conditions from basic circuit principles, and to use these constraints to reduce the number of sequence component current and voltage variables, so that a unique solution can be obtained.

13

ELEC 4100

THREE PHASE FAULT CALCULATIONS

ELEC 4100 ELECTRICAL ENERGY SYSTEMS THREE PHASE FAULT CALCULATIONS. 1 Introduction.
Faults occur frequently in a power system, and are caused by a range of problems such as the failure of the insulation of an item of electrical equipment, over-voltage swells caused by lightning strikes and switching surges, flash-over due to insulation contamination, etc. The fault that results from one of these events is often characterised by the placement of a short circuit across the phases or to earth, and this leads to a very large magnitude fault current. If this fault current is not interrupted the result can be equipment failure due to thermal degradation, and/or mechanical breakage caused by the extremely large forces which are applied to equipment during the fault. For these reasons it is necessary to have protection equipment which can break the circuit during a fault, but it is necessary to know the magnitude of the expected fault currents so that the circuit breakers can be designed adequately. Here we consider the development of a methodology for the calculation of fault currents, and we begin the analysis considering the simplest possible fault type the balanced three phase fault.

2 Power System Three-Phase Short Circuits.

In order to calculate the fault current for a balanced three phase short circuit in a power system, we make a number of simplifying assumptions. These include: The fault current is a pure sinusoid we do not consider decaying exponential transients. Transmission lines are represented by their equivalent -section model or just a series reactance term. Synchronous machines are represented by constant voltage sources behind an equivalent series impedance. Effects due to saliency, armature winding resistance and saturation are neglected, and we do not consider the transient Swing effect. Non-rotating impedance loads are neglected. Transformers are represented by their equivalent per unit models.

To determine the fault current at a specific location in a power network, it is necessary to determine the Thevenin equivalent circuit for the network looking in to that point at the network. This is most conveniently done in a Per Unit system since most transformers can be replaced by their equivalent series impedance in that particular Per Unit system. Having obtained the Thevenin equivalent circuit, it is a relatively simple matter to calculate the fault current that flows when a short circuit appears across the terminals of the Thevenin equivalent circuit. 2.1.1 Example: Consider the two bus system shown in figure 1, in which a synchronous generator feeds a synchronous motor through a transmission line and two transformers. A solid three phase fault occurs on Bus 1, and it is necessary to calculate the magnitude of the fault current in this case. To do this it is necessary to determine the Thevenin equivalent network of the power system when viewed from voltage Bus 1, and to then calculate the Thevenin equivalent impedance, and the Thevenin equivalent voltage. The first step is to represent the power system network by its equivalent Per Unit circuit. This is shown in figure 2. The next step is to obtain the fault voltage, and the Thevenin equivalent impedance of the power system network when viewed from voltage bus 1. To obtain the Thevenin equivalent impedance, all voltage sources in the network are set to short circuits, and the equivalent impedance viewed from 1

ELEC 4100

THREE PHASE FAULT CALCULATIONS

Bus 1

Bus 2

T1
100 MVA 13.8 kV Xeq = 0.15p.u 100MVA 13.8kV/138 kV Xeq = 0.1 p.u

20 ohm

T2
100MVA 138kV/13.8 kV Xeq = 0.1 p.u 100 MVA 13.8 kV Xeq = 0.2 p.u

Figure 1 : Single Line Diagram of a Synchronous Generator feeding a Synchronous Motor through a transmission line and two transformers.

jXg If

jXT1

jXline

jXT2

jXm

Eg

Em

Bus 1
Figure 2 : Per Unit Circuit representation of the single line diagram in figure 1. In this case a fault occurs at voltage bus 1. bus 1 is determined. From figure 2, it is clear that when this is done the Thevenin impedance is given by: ZTh = ( jX g )//[ j ( X T 1 + X line + X T 2 + X m )] = ( j 0.15) //[ j (0.1 + 0.105 + 0.1 + 0.2)] = j 0.1156 p.u.

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

ELEC 4100 ELECTRICAL ENERGY SYSTEMS SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS. 1 Introduction.
The theory of symmetrical components is used to represent a set of unbalanced three phase quantities as a summation of three independent three phase systems. These systems are referred to as the positive, negative and zero sequences. Symmetrical component theory is particularly useful when performing fault calculations, because for a three phase system with balanced impedances, the transformation from (abc) frame to the symmetrical component frame (i.e. 012 frame) leads to three uncoupled sequence networks the positive, negative and zero sequence networks. As a result when the system is subjected to an unbalanced fault (e.g. two phases shorted together, or a single phase shorted to earth) it is possible to use the three sequence networks to calculate the magnitude of the fault event. However to do this it is necessary to have the following information: The electrical coupling between the three sequence networks that results for a particular fault event. The actual sequence networks for the system. The Thevenin equivalent network for the system viewed from the location in the network where the fault event occurs.

The first consideration has been dealt with previously in the notes detailing the theory of symmetrical components and how this theory relates to fault events. Now it is necessary to focus on the remaining items the method of deriving the sequence network for the system, and the method of obtaining the Thevenin equivalent network.

2 Sequence Networks of Impedance Loads.

2.1 Balanced Star and Delta connected Loads.
Consider the balanced Y-connected impedance load shown in figure 1. The impedance of each

Va

Ia

Vc Vb

Ic Ib IN ZN

Figure 1 : Balanced Y-connected Load with a Neutral Impedance. 1

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

phase is Z, and there is also a neutral impedance ZN connected between the star point and earth. The phase voltage can be related to the phase and neutral currents according to:

Vag = ZI a + Z N I N

(1) (2) (3)

Vag = ZI a + Z N (I a + I b + I c ) Vag = (Z + Z N )I a + Z N I b + Z N I c
Similarly for the b and c phase voltages we can write:

Vbg = (Z + Z N )I b + Z N I a + Z N I c
and:

(4)

Vag = (Z + Z N )I c + Z N I b + Z N I a
We can write equations (1) through to (5) in matrix form as:

(5)

Vag Z + Z N Vbg = Z N Vcg ZN

Or more compactly:

ZN Z + ZN ZN

Z N I a ZN Ib Z + ZN Ic

(6)

Vabc = Z abc I abc

We can transformer this result into the symmetrical component frame according to:

(7)

AV012 = Z abc AI 012 V012 = A1Z abc AI 012 = Z 012 I 012

So we can calculate the symmetrical component impedance matrix as:\
Z 012 = A1Z abc A 1 1 1 = 1 a 3 2 1 a 1 1 1 = 1 a 3 2 1 a Z + 3Z N = 0 0 1 Z + Z N a2 ZN a ZN 1 Z + 3Z N a2 Z + 3Z N a Z + 3Z N 0 0 Z 0 0 Z ZN Z + ZN ZN Z a Z aZ
2

(8) (9)

Z N 1 1 2 ZN 1 a Z + ZN 1 a Z aZ a2Z

1 a a2

(10)

So for a balanced three phase star connected load the relationship between the symmetrical component currents and voltages is given by:
V0 Z + 3Z N V = 0 1 0 V2 0 Z 0 0 I0 0 I1 Z I2

(11)

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

Ia1 V1

Positive Sequence

Ia2 V2

Negative Sequence

Ia0 V0

3ZN

Zero Sequence

Figure 2 : Symmetrical Component Representation of a Balanced Y-connected Load with a Neutral Impedance.
Here we see that with balanced impedances the three phase star connected load can be represented by three completely decoupled zero, positive and negative sequence networks. The positive sequence network impedance matches the negative sequence impedance, and this is equal to the phase impedance for the star load. The zero sequence impedance is simply the star load phase impedance and three times the neutral to ground impedance. The decoupling ensures that the positive sequence voltage only depends on the positive sequence current, and the same is true for the zero and negative sequence voltages. This means that it is possible to represent the star connected load in the form shown in figure 2. It is important to note that the neutral impedance only contributes to the zero sequence network. This is because the positive and negative sequence currents alone do not contribute to a neutral current (i.e. a zero sequence), and hence they do not flow through the neutral impedance. If there is no neutral wire connected to the star point of the Y-connected load then the neutral impedance is infinite, and the zero sequence network is replaced by an open circuit, and hence there is no zero sequence current in this case. If the star point is solidly grounded through a short circuit then the zero sequence current which flows is caused by imbalance in the applied voltage divided by the phase impedance. Figure 3 shows a balanced -connected load and its equivalent balanced Y-connected load. Since the -connected load does not have a neutral connection, the equivalent Y-connected load has an open circuited neutral. Hence the sequence networks for the -connected load are those shown in figure 4.

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

Va

Va

Ia Z/3

Ia Iab Z Z Z Ibc Ica

Vc Vb Ic Ib Z/3 Z/3

Vb Vc

Ib Ic

Figure 3 : Balanced -connected Load and the equivalent Y-connected Load.

Ia1 V1 Z/3

Positive Sequence

Ia2 V2

Z/3

Negative Sequence

Ia0 V0

Z/3 Zero Sequence

Figure 4 : Symmetrical Component Representation of a Balanced -connected Load.

2.2 Unbalanced Star Connected Load.

Figure 5 shows a general three phase linear impedance load. The general relationship between the line to ground voltages and the line currents can be written as:

Vag Z aa Vbg = Z ab Vcg Z ac

Or:

Z ab Z bb Z bc

Z ac I a Z bc Ib Z cc Ic

(12)

Vabc = Z abc I abc

(13)

where the Vabc are the line to neutral (or phase) voltages, Iabc are the line currents and Zabc are the 3x3 phase and line to line impedances. This assumes a non-rotating load. It is possible to transform the impedance matrix to the symmetrical components frame as:

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

Va Vb Vc

Ia Ib Ic IN

General Three Phase Load

Figure 5 : General Three Phase Load.

Z 012 Z0 = A Z abc A = Z10 Z 20
1

Z 01 Z1 Z 21

Z 02 Z12 Z2

(14)

The diagonal elements of this matrix are the self impedances of the zero, positive and negative sequence networks. The off-diagonal impedances are the mutual impedances between the sequence networks. Hence:
Z 012 Z0 = Z10 Z 20 Z 01 Z1 Z 21 Z 02 1 1 1 Z12 = 1 a 3 2 Z2 1 a 1 Z aa a2 Z ab a Z ac Z ab Z bb Z bc Z ac 1 1 2 Z bc 1 a Z cc 1 a 1 a a2

(15)

Performing the first matrix operation :

Z0 Z 10 Z 20

Z 01 Z1 Z 21

Z 02 1 1 1 Z12 = 1 a 3 2 Z2 1 a

1 a2 a

Z aa + Z ab + Z ac Z ab + Z bb + Z bc Z ac + Z bc + Z cc
After the second matrix operation :

Z aa + a 2 Z ab + aZ ac Z ab + a 2 Z bb + aZ bc Z ac + a 2 Z bc + aZ cc

Z aa + aZ ab + a 2 Z ac Z ab + aZ bb + a 2 Z bc Z ac + aZ bc + a 2 Z cc

(16)

Z 012

Z0 = Z10 Z 20

Z 01 Z1 Z 21

Z 02 Z12 = Z2 Z aa + a 2 Z bb + aZ cc aZ a 2 Z Z ab ac bc + + Z Z Z aa bb cc Z Z Z ac bc ab Z aa + aZ bb + a 2 Z cc 2 + 2a Z + 2aZ + 2 Z ab ac bc Z aa + + aZ bb + a 2 Z cc a 2 Z aZ Z (17) ab ac bc 2 Z aa + a Z bb + aZ cc + 2aZ + 2a 2 Z + 2Z ab ac bc Z aa + Z bb + Z cc Z Z Z ac bc ab

Z aa + Z bb + Z cc + 2 Z ab + 2 Z ac + 2 Z bc Z + aZ + a 2 Z 1 bb cc aa 2 3 a Z ab aZ ac Z bc 2 Z + a Z bb + aZ cc aa aZ a 2 Z Z ab ac bc

Note the following relationships between the elements of the above symmetrical components Impedance matrix:

The diagonal elements Z1 and Z2 are equal.

5

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

The off-diagonal elements Z01 and Z20 are equal. The off-diagonal elements Z02 and Z10 are equal.
This leaves the elements Z0, Z12 and Z21, which are not equal to any other element in the matrix. With this matrix it is possible define the conditions for a symmetrical load. Symmetrical loads have diagonal sequence impedances, which means that all of the off-diagonal elements in Z012 must be zero. So setting this condition in equation (17) requires that:

(Z (Z (Z (Z

aa aa aa aa

+ a 2 Z bb + aZ cc aZ ab a 2 Z ac Z bc = 0 + aZ bb + a 2 Z cc + a Z bb + aZ cc
2 2 ab ac 2 bc

+ aZ bb + a 2 Z cc

) a Z aZ Z ) = 0 + 2aZ + 2a Z + 2Z ) = 0 + 2a Z + 2aZ + 2Z ) = 0
ab ac bc bc 2 ab ac

(18)

Recall the identity :

1 + a + a2 = 0 Then clearly the conditions in (18) are satisfied when : Z aa = Z bb = Z cc Z ab = Z ac = Z bc The symmetrical component impedance matrix then reduces to:
Z 012 Z0 = Z10 Z 20 Z 01 Z1 Z 21 Z 02 Z aa + 2 Z ab Z12 0 = Z2 0 0 Z aa Z ab 0 0 Z aa Z ab 0

(19)

(20)

(21)

Note that the positive and negative sequence networks are identical when the conditions for a symmetrical load are set. This is always the case for non-rotating linear symmetric loads such as transformers and transmission lines. For generators, however, the positive and negative sequence impedances are not equal in general. It should also be noted that the zero-sequence impedance is only ever equal to the positive and negative sequence impedances under the condition that
Ia1 V1

Zaa - Zab

Positive Sequence

Ia2

Zaa - Zab

V2 Negative Sequence

Ia0 V0

Zaa + 2Zab Zero Sequence

Figure 6 : Sequence Networks of a Symmetrical Three Phase Load.

6

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

Ia

Vaa' Za Ib Van Vbn Vbb' Zb Ic Vcn Vcc' Zc

Zab

Zbc

Zac Vb'n Vc'n

Va'n

Figure 7 : Three Phase Series Impedances. Z ab = Z ac = Z bc = 0 , which is equivalent to a solidly grounded star point. The sequence networks for the symmetrical load case are illustrated in figure 6.

3 Sequence Networks of Series Impedance Elements.

Figure 7 shows a series impedance network connected between two different voltage buses, denoted by (a,b,c) and (a,b,c) respectively. The series impedance network contains self impedances for each phase (i.e. the Za, Zb and Zc) and there may also be mutual coupling between the phases (i.e. the Zab, Zbc and Zac). In the (a,b,c) frame the voltage drop across the series impedance elements are given by:
Van Va ' n Vaa ' Z a V V = V = Z b'n bb ' ab bn Z ac Vcc ' Vcn Vc ' n Z ab Zb Z bc Z ac I a Z bc Ib Zc Ic

(22)

The above equation represents non-rotating equipment, and can be used to represent equipment such as transmission lines and transformers as we shall soon see. Note that the (Van,Vbn,Vcn) and the (Van,Vbn,Vcn) represent the voltage at the buses (a,b,c) and (a,b,c) with respect to the neutral. It is possible to transform equation (22) into the symmetrical component frame according to :
V0 V0' A V1 AV1' = V2' V2 V00 ' Z a A V11' = Z ab V22 ' Z ac Z ab Zb Z bc Z ac I 0 Z bc A I1 Zc I2

(23)

Or:
Z a Z ab V00 ' V = A1 Z ab Z b 11' Z ac Z bc V22 ' V00 '11'22 ' = Z 012 I 012 Z ac I 0 Z 0 Z bc A I1 = Z10 Zc Z 20 I2 Z 01 Z1 Z 21 Z 02 I 0 Z12 I1 Z2 I2

(23)

So again we see that :

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

Ia1 V1

Za - Zab V1'

Positive Sequence

Ia2 V2

Za - Zab V2'

Negative Sequence

Ia0 V0

Za+2Zab Zero Sequence V0'

Figure 8 : Three Phase Series Impedances Symmetrical Component Networks.

Z 012 Z0 = A Z abc A = Z10 Z 20
1

Z 01 Z1 Z 21

Z 02 Za 1 Z12 = A Z ab Z2 Z ac

Z ab Zb Z bc

Z ac Z bc A Zc

(24)

From the results of the previous section we know the conditions for a symmetrical series impedance to be: Z a = Zb = Zc Z ab = Z ac = Z bc
Z0 = 0 0
0 Z a + 2 Z ab 0 0 = 0 Z2

(25)

Under this condition the series impedance matrix in symmetrical component form reduces to:
0 0

Z 012

Z1 0

Z a Z ab 0

0 Z c Z ab
0

(26)

Similarly the voltage drop between the two voltage buses in symmetrical component form becomes : V1 V1' = Z1I1 = (Z a Z ab )I1 V0 V0 ' = Z 0 I 0 = (Z a + 2Z ab )I 0 (27)

V2 V2 ' = Z 2 I 2 = (Z a Z ab )I 2

So the symmetrical component representation is three decoupled equations, in which the positive and negative sequence impedances are equal. It is therefore possible to represent the series impedance network with the symmetrical component networks as shown in figure 8. 8

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

Ia Za

Va

Ea Ec Eb Zc ZN IN Zb Ib Ic Vb Vc

Figure 9 :Star Connected Synchronous Generator.

4 Sequence Networks of Rotating Machines.

A star connected synchronous generator is shown in figure 9. This generator is shown to be connected to earth through a neutral impedance ZN. The generator also has self impedance branches (Za,Zb,Zc), and produces terminal voltages (Va,Vb,Vc), while the generator EMF voltages are (Ea,Eb,Ec). The sequence networks for this generator are shown in figure 10. There are a number of points that need to be focused upon: (a) The first point to recognise is that the positive sequence network is the only sequence with a source voltage. The reason for this is that generators are designed to produce

I1 Eg1

Z1 V1

I2

Z2 V2

I0

Z0

3ZN V0

Figure 10 :Star Connected Synchronous Generator Sequence Networks. 9

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS balanced voltages, and hence will produce only a positive sequence EMF voltage.

(b)

The second point to note is that the zero sequence network contains the zero sequence impedance of the generator, and three times the neutral branch impedance. This is because the voltage drop across the neutral impedance is due solely to the presence of a neutral current, which must be a zero sequence component. Thirdly, unlike the non-rotating networks analysed previously, in this case the positive and negative sequence impedances of the generator are not in general equal.

(c)

The third point needs some further explanation. When a synchronous generator stator has balanced three-phase currents under steady-state conditions, the net m.m.f. produced by these currents rotates at synchronous speed to the rotor, and also in the same direction. This ensures that the positive sequence field has a high degree of penetration into the rotor steel, which leads to an effective high value for the positive sequence inductance hence the positive sequence impedance is large. When there are balanced negative sequence currents, the net m.m.f. still rotates at synchronous speed, but this time in the opposite direction to the rotor. This means that the apparent field frequency for the negative frequency is twice the synchronous frequency not DC. This limits the degree of rotor steel penetration which reduces the effective negative sequence inductance hence the negative sequence impedance is small. The zero sequence impedance is the smallest sequence impedance, and this is because it is due to non-ideal effects such as flux leakage, harmonic flux, etc. When a synchronous machine is operated in motoring mode the same sequence networks are applicable, except that the networks are shown to sink rather than source currents. Induction machines can also be represented in the same way, however since induction machines do not have a back-e.m.f. generated by a separate excitation source, the positive sequence voltage is not included in the sequence network.

5 Sequence Networks of Transformers.

Recall that the single phase Per Unit model of a transformer consists of a series impedance (which represents flux leakage effects and the resistance of the windings), and a shunt branch (which represents the magnetization and core loss effects). Generally for fault calculations the shunt branch impedance is much larger than the Thevenin equivalent shunt impedance to earth, viewed from the section of the network where the fault occurs. Therefore it is a reasonable approximation to neglect the excitation branch of the transformer model when considering sequence models for the purposes of fault calculations. The same statement is also true for transmission lines, and so the shunt branches at each end of the transmission line are often neglected in fault calculations. This approximations leads to simpler overall network models. The positive and negative sequence impedance models for a transformer are identical. This is because in contrast to rotational loads, the phase order of the applied voltage set has no effect on the impedance values of the transformer, provided that the applied voltages are balanced. Under the condition that the phase shift of Y- and -Y transformers is neglected, the positive and negative sequence networks of the transformer can be represented as those shown in figure 11. The zero sequence network of a transformer requires more careful consideration.

5.1 Transformer Zero Sequence Networks.

The zero sequence model for a three phase transformer bank depends on the type of connections that are used to form the transformer windings (i.e. Y or ) in both the primary and secondary. The models can be developed easily by keeping in mind the fact that the primary current can be 10

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

Ia1 V1

Z1 V1'

Positive Sequence

Ia2 V2

Z2 V2'

Negative Sequence

Z1 = Z2
Figure 11: Transformer Positive and Negative Sequence Networks. Transformer Phase shifts neglected. calculated from the secondary current and the transformer turns ratio. Therefore there can only be a current flowing in a particular phase of the primary winding, if it possible for the equivalent current to flow in the corresponding winding of the transformer secondary. Figure 12 illustrates the equivalent models for five different transformer winding arrangements. This figure does not represent all possible transformer configurations, but it does illustrate the method for determining the component models. Note that arrows in the diagram indicate the possible zero sequence current paths. The absence of an arrow indicates that it is not possible for a zero sequence current to flow, due to the electrical connection of the transformer. The justification for each model can be summarised in the discussion of the following two cases.
5.1.1 Y- Bank with Grounded Y. This is the third case in figure 12. Since the primary Y is grounded, zero sequence currents have a conduction path to flow in these windings. The existence of this path, however, does not mean that such a current can flow. The zero sequence current can only flow if there exists an equivalent path on the transformer secondary. The secondary is delta connected, and hence the induced zero sequence currents could circulate in the delta. Note however that the zero sequence circulating currents can not leave the delta, and it is left as an exercise for the student to justify this statement. The equivalent model, therefore, must provide a path from the line on the Y side (point P) to the reference node for the flow of zero sequence currents. An open circuit must appear from the line side of the delta side (point Q) to the reference node, since the zero sequence currents can not exist in the lines from the delta connected winding. If the Y is connected to earth through an impedance Zn, then an impedance of 3Zn appears in series with the Z0 of the transformer. 5.1.2 - Bank. Zero sequence currents could circulate inside the transformer windings in both the primary and the secondary but as these currents can not leave the delta, and so there will be no zero sequence currents in the lines. The equivalent circuit must therefore have open circuits from both the points P and Q to the reference node, but the transformer zero sequence impedance does form a loop to the reference node.

5.2 Transformer Phase Shifts.

The positive and negative sequence networks shown in figure 11 assumed that the phase shift that arises in Y- and -Y transformers can be neglected. However in general this is not the case, and so it is necessary to modify these sequence networks so as to include the phase shift. However 11

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

P P Q Z0

P P Q Z0

P P Q Z0

P P Q Z0

P P Q Z0

Figure 12: Transformer Zero Sequence Networks for different Winding Configurations. care is needed to ensure that the standard conventions are employed when calculating the phase shift. Recall that the standard convention states that: The positive sequence voltages and currents for the high voltage side of the Y- transformer lead the corresponding quantities on the low voltage side of the transformer by 300. Hence the negative sequence voltages and currents for the high voltage side of the Y- transformer must lag the corresponding quantities on the low voltage side by 300.

This means it is a relatively simple matter to modify the per unit sequence models of figure 11 for specific winding arrangements, as is illustrated in figure 13. 12

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

Ia1 V1

Z1 V1'
0

Positive Sequence

ej30 : 1 Ia2 V2

Z2 V2'
0

Negative Sequence

Z1 = Z2

e-j30 : 1

Figure 13: Y- Transformer Positive and Negative Sequence Networks. Note that although in practice the zero sequence impedance of a transformer may differ slightly to the positive and negative sequence impedances, often the same numeric value is used for all three impedances.

6 Fault Calculations Using Power System Sequence Networks.

It was previously shown that the fault current for a power system can be calculated by connecting the Thevenin equivalent sequence networks together in the required configuration for that particular fault. For example, in a single line to ground fault the three sequence networks are connected in series, whereas for a balanced three phase fault the positive sequence network is shorted, while the negative and zero sequence networks are left open circuit. To use this technique it remains to determine the Thevenin equivalent sequence networks for the power system at the point in the system where the fault occurs, and this relies on the component sequence models which have just been derived. To illustrate the approach a number of examples will be presented.
6.1.1 Simple Two Bus System. Figure 14 shows the single line diagram of a simple two bus system, in which a synchronous generator supplies a synchronous motor through two transformers and a transmission line. The positive, negative and zero sequence impedances for the system are also given. The neutrals of the generator and the two transformers are solidly earthed. The neutral of the motor is grounded through a reactance of Xn = 0.05 p.u. Pre-fault currents, and transformer phase shifts due to the Y-
Bus 1 Generator Bus 2

T1
X1=X2=20 X0=60

T2

Motor

100 MVA 13.8 kV X1=0.15 p.u X2=0.17 p.u X0=0.05 p.u

100MVA 13.8kV / 138 kV Y X1=X2= 0.1 p.u

100MVA 138 kV Y / 13.8kV X1=X2= 0.1 p.u

100 MVA 13.8 kV X1=0.20 p.u X2=0.21 p.u X0=0.10 p.u Xn=0.05 p.u

Figure 14 : Single Line Diagram of a Synchronous Generator feeding a Synchronous Motor through a transmission line and two transformers. 13

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

transformers are neglected in this example. It is required to : (a) (b) (c) (d) (e) (f) Draw the per unit zero, positive and negative sequence networks of the system on a 100MVA base, with the low voltage base as 13.8kV in the zone of the generator. Determine the Thevenin equivalent sequence networks as viewed from bus 2 in the power system. Calculate the fault current at bus 2 for a solid balanced three phase fault, for the case where the pre-fault voltage is 1.05 p.u. Calculate the fault current at bus 2 for a solid single line to ground fault, for the case where the pre-fault voltage is 1.05 p.u. Calculate the fault current at bus 2 for a solid line to line fault, for the case where the pre-fault voltage is 1.05 p.u. Calculate the fault current at bus 2 for a double line to ground fault, for the case where the pre-fault voltage is 1.05 p.u.

SOLUTION.

(a) The sequence networks for this power system are shown in figure 15. Note that the positive

jXg1
j0.15 p.u

jXT1_1
j0.10 p.u

jXline_1
j0.105 p.u

jXT2_1
j0.10 p.u

jXm1
j0.20 p.u

Bus 1 Eg=1.05 p.u Positive Sequence

Bus 2 Em=1.05 p.u

(a) Positive Sequence Network

jXg2
j0.17 p.u

jXT1_2
j0.10 p.u

jXline_2
j0.105 p.u

jXT2_2
j0.10 p.u

jXm2
j0.21 p.u

Bus 1 Negative Sequence

Bus 2

(b) Negative Sequence Network.

jXg2
j0.05 p.u

jXT1_2
j0.10 p.u

jXline_2
j0.315 p.u

jXT2_2
j0.10 p.u

jXm2
j0.25 p.u

Bus 1 Zero Sequence

Bus 2

(c) Zero Sequence Network Figure 15 : Sequence Networks for the Two Bus Power System. 14

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

sequence network is the only network in which there exists a voltage source. This is due to the fact that generators are designed to produce positive sequence voltages only. The negative sequence network is very similar with the differences being the absence of the voltage sources, and the slightly different sequence impedances for the rotating items of equipment. The zero sequence network is quite different, however, due to the Y- connections of the transformers. The fact that the neutral of the Y connected winding is earthed means that a zero sequence current can flow, and the delta connection means that this current can circulate in the delta, but it can not leave the delta to flow in the lines to the generator and the motor. Therefore at the delta the zero sequence network must be shorted to the reference nodes. Also note that the zero sequence of the motor is given by Zm0 + 3Zn. Similarly the transmission line p.u. values are easily calculated on the 100MVA base and the required 138kV voltage base in the high voltage zone. (b) The Thevenin equivalent sequence networks when viewed from bus 2 are shown in figure 16. The Thevenin equivalent impedance value in the positive sequence network is readily calculated as: Z Th1 = ( jX m1 ) // j (X T 1_1 + X line _1 + X T 2 _1 + X g1 ) = ( j 0.20) //[ j (0.1 + 0.105 + 0.1 + 0.15)] = j 0.13893 p.u.

Similarly the Thevenin equivalent impedance value in the positive sequence network is readily calculated as: Z Th 2 = ( jX m 2 ) // j (X T 1_ 2 + X line _ 2 + X T 2 _ 2 + X g 2 ) = ( j 0.21) //[ j (0.1 + 0.105 + 0.1 + 0.17 )] = j 0.14562 p.u.

Finally the Thevenin equivalent impedance values in the zero sequence network is simply the

I1

ZTh1 = j0.13893 V1

VF = 1.05 p.u. Positive Sequence I2

ZTh2 = j0.14562 V2

Negative Sequence I0

ZTh0 = j0.25 V0

Zero Sequence
Figure 16 : Thevenin Equivalent Sequence Networks for the Two Bus power system when viewed from Bus 2. 15

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

I1

ZTh1 = j0.13893 V1

VF = 1.05 p.u. Positive Sequence I2

ZTh2 = j0.14562 V2

Negative Sequence I0

ZTh0 = j0.25 V0

Zero Sequence

Figure 17 :Sequence Network Connections for a Single Line to Ground Fault.

series connection of the motor zero sequence impedance, and three times the neutral impedance. (c) For a balanced three phase fault, recall that the required circuit configuration is to short the positive sequence network only. This means that there will only be a positive sequence current, and this is given by: I a1 = VF Z Th1 = 1.05 j 0.13893

= j 7.558 p.u. Since there is only a positive sequence current, this means that the fault current for each phase will be 7.558 p.u. (d) For a solid single line to ground fault the three Thevenin equivalent sequence networks are connected in series, as shown in this case in figure 17. So in this case all three sequence currents are identical, and the value of the current is given by: I 0 = I1 = I 2 = VF 1.050 0 = Z Th1 + Z Th 2 + Z Th 0 j (0.13893 + 0.14562 + 0.25)

= j1.96427 p.u. The actual phase currents are then given by:
I a 1 1 1 I 0 1 1 I = 1 a 2 a I = 1 a 2 1 b 2 1 a I 2 1 a a Ic 5.8928 p.u. = 0 0 1 j1.96427 a j1.96427 a2 j1.96427

Not surprisingly the a phase is the only line to carry the current. Now the effect on the system voltages can also be determined at the fault location according to:

16

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

I1

ZTh1 = j0.13893 V1

VF = 1.05 p.u. Positive Sequence I2

ZTh2 = j0.14562 V2

Negative Sequence I0

ZTh0 = j0.25 V0

Zero Sequence

Figure 18 :Sequence Network Connections for a Line to Line Fault.

V012 = ETh _ 012 Z Th _ 012 I 012 V0 0 j 0.25 V = 1.0500 0 1 0 V 2 0 0.49107 p.u. = 0.77710 p.u. 0.28604 p.u. 0 j 0.13893 0 j1.96427 j1.96427 0 1 . 96427 j 0.14562 j 0

The phase to ground voltages are then given by:

Vag ,bg ,cg = AV012 Vag 1 1 2 Vbg = 1 a Vcg 1 a 1 0.49107 a 0.77710 a2 0.28604

0 0 = 1.179 p.u.231.3 0 1.179 p.u.128.7

Unsurprisingly the phase A voltage is zero as defined by the fault condition. (e) For the line to line fault condition recall that the positive and negative sequence networks are connected in series as shown in figure 18. Since the zero sequence network is not connected there can be no zero sequence current. The positive and negative sequence currents can be calculated according to: I1 = I 2 = VF 1.05 = Z Th1 + Z Th 2 j (0.13893 + 0.14562)

= 3.690 90 0 So the phase currents are given by: 17

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

I1

ZTh1 = j0.13893 V1

VF = 1.05 p.u. Positive Sequence I2

ZTh2 = j0.14562 V2

Negative Sequence I0

ZTh0 = j0.25 V0

Zero Sequence

Figure 19 :Sequence Network Connections for a Double Line to Ground Fault. I a 1 1 I = 1 a 2 b 1 a Ic 1 I 0 1 1 2 a I1 = 1 a a2 1 a I 2 1 j 3.690 a j 3.690 a2 0

0 0 = 6.391 p.u. 180 0 6.391 p.u. 0 Not surprisingly the b and c phases carry equal and opposite currents, while the a phase current is zero. (f) Recall that for the Double Line to Ground fault the zero and negative sequences are placed in parallel, and are connected in series with the positive sequence network. This connection of sequence networks is illustrated in figure 19. The positive sequence current is calculated as:
I1 = VF 1.0500 = (0.14562)(0.25) ZTh1 + (ZTh 2 // ZTh 0 ) j 0.13893 + 0.14562 + 0.25 = j 4.5464 p.u.

Then by the current divider rule:

ZTh 0 j 0.25 I 2 = I1 = + j 4.5464 j 0.25 + j 0.14562 ZTh 0 + ZTh 2 = j 2.8730 p.u. ZTh 2 j 0.14562 I 0 = I1 = + j 4.5464 j 0.25 + j 0.14562 ZTh 0 + ZTh 2 = j1.6734 p.u.

Transforming these currents into the (a,b,c) domain gives:

18

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS I a 1 1 1 I 0 1 1 I = 1 a 2 a I = 1 a 2 1 b 2 1 a I 2 1 a a Ic 0 0 = 6.8983 p.u 158.66 0 6.8983 p.u 21.34 1 j1.6734 a j 4.5464 a2 + j 2.8730

Again it is not surprising to see that the a phase current is zero in this case, and the b and c phases which are shorted to earth carry the same magnitude current.

6.2 Effect of -Y Transformer Phase Shifts.

Previously transformer component sequence models have been shown which include the effect of the transformer phase shift. These can now be incorporated into the example power system of figure 14 to determine just what effect transformer phase shifts have on the fault calculation. The sequence networks when the complete transformer models are included are shown in figure 20. Note that the standard convention is applied such that the high voltage positive sequence quantities lead the low voltage equivalent quantities, but the high voltage negative sequence quantities lag the

jXg1
j0.15 p.u

1 : ej30

jXT1_1
j0.10 p.u

jXline_1
j0.105 p.u

jXT2_1
j0.10 p.u

ej30 : 1

jXm1
j0.20 p.u

Bus 1 Eg=1.05
p.u

Bus 2 Positive Sequence Em=1.05

p.u

(d) Positive Sequence Network

jXg2
j0.17 p.u

ej30 : 1

jXT1_2
j0.10 p.u

jXline_2
j0.105 p.u

jXT2_2
j0.10 p.u

1 : ej30

jXm2
j0.21 p.u

Bus 1 Negative Sequence

Bus 2

(e) Negative Sequence Network.

jXg2
j0.05 p.u

jXT1_2
j0.10 p.u

jXline_2
j0.315 p.u

jXT2_2
j0.10 p.u

jXm2
j0.25 p.u

Bus 1 Zero Sequence

Bus 2

(f) Zero Sequence Network Figure 20 : Sequence Networks for the Two Bus Power System, including the effects of transformer phase shifts. 19

ELEC 4100

VOLTAGE CONTROL

ELEC 4100 ELECTRICAL ENERGY SYSTEMS VOLTAGE CONTROL IN POWER SYSTEMS. 1 Introduction.
The frequency of operation of a power system is determined by the speed of rotation of the generators which supply power to the network. The speed depends on the balance between the system load demand and the generator output. The point at which the frequency control is exercised is within the speed governing loop at the generators. At steady state, the frequency throughout the network is the same. This contrasts with voltage control. The voltage can vary greatly from place to place within the network. To investigate the control of voltage within the system, we use load flow analysis which allows us to calculate the voltage magnitude, the voltage angle and the real and reactive power flows in the system. Here we focus on the control of the voltage magnitude and the reactive power in the system. The voltage within the power system is strongly coupled to the reactive power flow balance within the network. As transmission lines are reactive in nature, it is the flow of reactive power through the lines that determines the voltage profile to a large extent. To investigate the voltage profile within a power network we consider the sources and loads of reactive power within the power system.

2 Reactive Power Flow and Voltage Profile.

Consider the transmission line shown in figure 1, connecting two distinct voltage buses : Bus 1 and Bus2. The reactance of the transmission line connecting the buses is X, and the voltages at each bus are identified as V11 and V2 2 respectively. The current in the transmission line is given by : I= V11 V2 2 jX (1)

The apparent power leaving bus 1 is then given by: S1 = (V11 )I V ( 1 ) V2( 2 ) S1 = (V11 ) 1 jX (2) (3)

V12 V1V2(1 2 ) S1 = jX

(4)

V1 10 jX I 0
Bus 1

V2 20

Bus 2

P1+jQ1
Figure 1 : Power Flow on a Transmission Line.
1

P2+jQ2

ELEC 4100 This becomes:

VOLTAGE CONTROL

V1V2 sin (1 2 ) + S1 = P 1 + jQ1 = X

V 2 V1V2 cos(1 2 ) j 1 X

(5)

We can investigate the variation in the real and reactive power flow through this transmission line by forming the partial derivatives of P and Q with respect to the voltage angles and magnitudes. This gives:
P V V cos(1 2 ) 1 = 1 2 (1 2 ) X Q1 V V sin (1 2 ) = 1 2 (1 2 ) X
P V sin (1 2 ) 1 = 2 V1 X Q1 2V1 V2 cos(1 2 ) = V1 X

(6)

(7)

(8)

(9)

Consider the example where the voltages on each bus are V1 = V2 = 1pu, the line impedance becomes X= 0.2pu, and the difference in voltage angles across the transmission line is (1 2 ) = 5 . Then the partial derivatives in (6) to (9) become in this case:

P 1 = 4.981 p.u. (1 2 )
P 1 = 0.436 p.u. V1

Q1 = 0.436 p.u. (1 2 )
Q1 = 5.019 p.u. V1

These derivatives show that the real power is strongly coupled to the difference in voltage angles across the transmission line, but is only weakly coupled to changes in the voltage magnitude. The reactive power on the other hand is strongly coupled to changes in the voltage magnitude, but only weakly affected by differences in the voltage angle. The conclusion is that to control voltage it is necessary to regulate the reactive power flow, but to control the real power flow it is necessary to control the voltage angle difference. Often these relationships are approximated with the application of Taylor series expressions. Recall that the Taylor series expansions of the sine and cosine functions are respectively: sin ( x ) = x cos( x ) = 1 Then: x3 x5 x7 + +" 3! 5! 7! x2 x4 x6 + +" 2! 4! 6! (10) (11)

P 1 = Q1 =

V1V2 sin (1 2 ) V1V2 (1 2 ) = X X V1 V1V2 cos(1 2 ) V1 (V1 V2 ) = X X

2

(12) (13)
2

ELEC 4100

VOLTAGE CONTROL

Now consider the power flow at the second voltage bus. The apparent power flowing at bus 2 is given by: S 2 = (V2 2 )I V ( 1 ) V2( 2 ) S 2 = (V2 2 ) 1 jX V V ( 1 + 2 ) V2 2 S2 = 1 2 jX This becomes: (14) (15)

(16)

V2 2 + V1V2 cos(1 2 ) V1V2 sin (1 2 ) S 2 = P2 + jQ2 = + j X X

(17)

So clearly the real power flowing at bus 2 is the same as at bus 1, and this is expected in this case since the only impedance elements in the transmission line are reactive. The change in the reactive power across the transmission line is given by: + V 2 V1V2 cos(1 2 ) V2 2 + V1V2 cos(1 2 ) Q1 Q2 = 1 X X V 2 + V2 2 2V1V2 cos(1 2 ) = 1 X =I X
2

(18)

So the change in reactive power is due to the line reactance in exactly the same way as real power loss in a transmission line is due to the line resistance. The above analysis was carried out considering only a purely reactive line, and in practice there is also line resistance. Hence although the active component of the power relates to the actual consumption of power (Watts), the supply of reactive power causes active power loss when the resistance is considered.

3 Reactive Power Sources and Loads.

As with the real power flow in a power system and its relationship with the frequency, there exists a balance between the demand and supply of reactive power and this determines the voltage profile within the system. The following is a list of the voltage and reactive power control sources within the power network. 1. Controllable reactive power sources. 2. On-Load Tap Changing (OLTC) transformers. 3. Emergency measures (Load reduction or shedding, power system rearrangement). Unlike the situation which applies with frequency control, the sources of reactive power are not confined to a few locations within the network. The following is list the reactive power sources. 1. Over-Excited Synchronous Machines. 2. Shunt and series capacitors. 3. Controlled static compensators.
3

ELEC 4100

VOLTAGE CONTROL

EXCITER

SYNCHRONOUS GENERATOR

a b c

POWER GRID

VOLTAGE REGULATOR VOLTAGE FEEDBACK

POTENTIAL TRANSFROMER

Figure 2 : Voltage Control on a Synchronous Generator. 4. Changing capacitance of transmission lines and cables. Figure 2 illustrates the components of voltage control associated with a synchronous generator. The potential transformer monitors the output voltage of the generator and that voltage is compared with the voltage reference required for the machine. The current in the exciter is controlled by the regulator to achieve the target voltage. The generation of reactive power by over-excited and under-excited machines is illustrated in figure 3. Figure 4 shows one phase of a static var compensator (SVC). The reactive power is supplied by the parallel combination of the inductors and capacitors . In the configuration shown, the current through the inductance is controlled by the time delay on the firing of the SCR devices. Hence the net reactance of the SVC can be controlled over a range from full capacitance (inductance blocked) to inductive (total inductance less the total capacitance). The switching of the SCRs is controlled by comparing the actual bus voltage with the desired voltage. This is carried out by the controller.

E0

I E0

jXs V
I
(a)

jXsI

(b)

E0

I
(c)

jXsI

E0

jXsI V
I
(d)

Figure 3 : Synchronous Machines as sources of reactive power (a) Machine Model (b) Over-excited Generator, (c) Under-excited Generator (d) Synchronous Condenser. 4

ELEC 4100

VOLTAGE CONTROL

SCR's

Gate Control Potential Transformer Air-Core Reactor Bank

Figure 4 : Static Var Compensator (SVC).

Capacitor Bank

Often, SVCs include filters to remove harmonics generated by the switching of the SCRs. Figure 5 shows a schematic representation of an on line tap changer. The voltage at the terminals of the transformer is monitored and compared against a voltage reference. The error signal is used to change the tapping position (and hence the turns ratio) and control the output voltage. In contrast with the typical situation with real power sources, the sources of reactive power are not confined to a relatively small number of positions. A list of power loads is given below: 1. Magnetising current for iron-cored equipment (transformers, fluorescent lamps, arc furnaces, rectifiers). 2. Induction motors. 3. Transmission losses in transformers and transmission lines. 4. Under-excited synchronous machines. 5. Shunt and series reactances. 6. Controlled static converters. The reactive power demand obviously changes with the voltage level. As the number of source and loads of reactive power are many and distributed widely around the system, the control of reactive power flow, and hence voltage, is difficult. Very often, voltage control tends to be localised rather than system-wide.

POWER TRANSFORMER

Potential Transformer TAP CHANGER TAP CONTROL Reference Voltage

Figure 5 : On Line Tap Changing (OLTC) Transformer. 5

ELEC 4100

VOLTAGE CONTROL

4 Reactive Power Compensation.

Figure 6 shows a load which is supplied through a transmission line of reactance X. The load is a parallel combination of resistance RL and inductance XL. The susceptance of the compensation element which is in shunt with the load is B. The voltage at bus 2 is given by:
V2 = V1 jIX

(19) (20)

V V V2 = V1 jX 2 j 2 + jV2 B XL RL

This can be rearranged as : 1 j V2 1 + jX R X + jB = V1 L L V2 = V1 1 j 1 + jX R X + jB L L (21)

(22)

So the magnitude of the voltage at bus 2 is given by: V2 = V1 X X 1 XB + X + R L L

2 2

(23)

So the reactive voltage drop of the transmission line is compensated for when the following is achieved : X X 1 XB + X + R =1 L L Which requires: X X 1+ 1 R XL L B= X
2 2 2

(24)

(25)

This simple case illustrates some of the complexity of voltage regulation within a large power network.

jX

V1

RL

jXL V2

Figure 6 : Simple Load with a Reactive Compensator. 6

ELEC 4100

SYMMETRICAL COMPONENT MODELS AND FAULT CALCULATIONS

low voltage equivalent quantities. Note the following properties: The zero sequence network is unchanged due to the addition of the phase shifting transformers. Therefore the Thevenin equivalent network for the zero sequence is unchanged. Since the phase shifting transformers are ideal, the referral of a per unit impedance from one side of the transformer to the other is unchanged. Hence it is possible to refer all impedance values to the bus 2 side of the transformers, and so the Thevenin equivalent impedance for both the positive and negative networks viewed from bus 2 is also unchanged. The referral of the Thevenin equivalent voltage to the bus 2 side of the transformer does change the angle of the voltage, but not the magnitude. The conclusion then is that the phase shifting transformers do not affect the magnitude of the fault calculation, justifying the approximation previously taken in which the phase shift is neglected. It is important to recognise that while the phase shift of the -Y transformer does not affect the calculation of the magnitude of the fault current, it does change the phase of the positive and negative sequence components on the other side of the transformer. Hence the magnitude of the phase currents on the other side of the transformer is different since the positive and negative sequence currents are shifted by 300 in opposing directions. Note however that it is not a significant problem to account for these effects once the overall fault current has been calculated.

20

ELEC 4100

TRANSIENT STABILITY

ELEC 4100 ELECTRICAL ENERGY SYSTEMS TRANSIENT STABILITY. 1 Introduction.

The power system is continually being subjected to changes or disturbances of various forms: faults, load changes, connection/disconnection of generators etc. As the power network is a complex electro-mechanical system, these events cause oscillations in the speed and angles of the machines and in power flows along the lines. Transient stability analysis is the study of the system in response to these changes and is used to determine if the system will be stable after a given disturbance. For proper operation of the system, it is essential to ensure that after a given disturbance, the system settles to a new stable condition. Transient power frequency control in response to load/generation area changes, can be addressed via system linearisation about a nominal operating point, enabling the application of linear Laplace transform methods. In contrast, transient stability analysis is generally concerned with the analysis of the effects of major disturbance events, such as line faults. An allied problem is the study of steady-state (or dynamic) stability. This is used to determine if a certain system condition is stable and to look at the response for small fluctuations. As with load/frequency control, linear analysis can be used for steady-state stability analysis. On the other hand, transient stability analysis involves major disturbances. If a three-phase fault occurs close to a machine, then the ability of that machine to transfer power changes from possibly 100% to zero. Hence, we must consider a much wider response range and the analysis is essentially non-linear. For that reason, the transient analysis of a power network in adequate detail is a formidable task and is one of the major studies involved in power system analysis. In the past, considerable simplifying assumptions were used to allow for some analysis of transient stability. This often involved considering one machine connected to an infinite system (constant frequency and voltage) through a transmission line. With the improved simulation tools available, it is now possible to investigate the transient stability of extensive systems consisting of many machines. The significant components for transient stability are: 1. The network before, during and after the transient disturbance. 2. The loads and their characteristics. 3. The parameters if the synchronous generators. 4. The excitation systems of the generators. 5. The turbine and speed governors.

2 Single Machine Connected to an Infinite Bus.

Figure 1 shows a single synchronous machine connected to an infinite bus through parallel transmission lines. The lines have circuit breakers installed at each end. A fault occurs at some point

Infinite Bus Generator Fault

Figure 1 : Single Generator connected to an Infinite Bus. 1

ELEC 4100

TRANSIENT STABILITY

XL1 X'd XT1

Vt E' = 0

XL2 V = 00

Figure 2 : Schematic Representation of the system.

along the second transmission line. The objective is to determine the behaviour of the system before the fault, during the fault, and after the fault has been cleared by opening the faulted transmission line. To do this we need to consider the representation of both the electrical and mechanical aspects of this system. Figure 2 shows the equivalent circuit of the network. The synchronous generator is represented as a constant voltage E behind the transient reactance Xd. The voltage behind the transient reactance makes an angle of with the infinite bus voltage and this is equal to the machine rotor angle. The terminal voltage of the generator is Vt and the transformer reactance is Xt. The reactances of the lines are XL1 and XL2.

3 The Swing Equation.

The swing equation describes the behaviour of the machine to disturbances in the network. Figure 3 shows a generator of inertia J, with an input mechanical torque of Tm, and an output electrical torque Te. The angular position of the machine, , is given by:

= t +
where

(1)

= the angular speed of rotation.

= the rotor angle with respect to the reference angle (i.e. infinite bus)
So considering the angular position:

d d = + dt dt d 2 d 2 = 2 dt 2 dt
Where the angular speed is constant. The acceleration torque on the machine is given by:

(2) (3)

Pm Tm

Pe Te
2

Infinite Bus

Figure 3 : Representation of a single Generator.

ELEC 4100

TRANSIENT STABILITY

Ta = Tm Te = J
In terms of power:

d 2 d 2 = J dt 2 dt 2 d 2 d 2 = M dt 2 dt 2

(4)

Pa = Pm Pe = J

(5)

Where M is the angular momentum, and it is assumed that the change in speed is small. If damping effects are included, then:

d 2 d 2 d Pm Pe = M 2 + Pd = M 2 + K d dt dt dt
Multiplying by /2 gives: 1 1 2 d 2 (Pm Pe ) = J 2 2 2 dt and

(6)

(7)

1 2 J represents the kinetic energy of the generator. 2

If we divide by the base MVA, Sbase, we have:

1 2 J d 2 1 Pm Pe 2 = 2 2 Sbase Sbase Sbase dt 1 d 2 (Pm _ pu Pe _ pu ) = H 2 2 dt (8)

H is the inertia constant, and is equal to the kinetic energy of the generator divided by its power rating. Hence the units of H are MJ/MW or seconds. The typical synchronous generator has an inertia constant between 2 and 10 seconds. If we now drop the p.u. notation (assume from now on that we are dealing with per unit quantities) we have:
Pm Pe = 2 H d 2 H d 2 = dt 2 f dt 2

(9)

With damping:
Pm Pe = H d 2 d + Kd 2 dt f dt

(10)

Transient stability analysis involves the solution of this differential equation the swing equation. There are two significant aspects to transient stability: The input mechanical power Pm. The output electrical power transferred to the load or network Pe.

4 Two Machines Connected in Parallel.

As shown in figure 4, if we have two machines connected in parallel and supplying the same load, the swing equations are given by:

ELEC 4100

TRANSIENT STABILITY

Pm1 Tm1

Pe1 Te1

Infinite Bus

Pm2 Tm2

Pe2 Te2

Figure 4 : Two Machines Connected to a Common Bus.

H1 d 21 = Pm1 Pe1 f dt 2 H 2 d 2 2 = Pm 2 Pe 2 f dt 2

(11a)

(11b)

If the machines swing together, then :

1 = 2 =
So:

(H1 + H 2 ) d 2 = (P + P ) (P + P ) m1 m2 e1 e2 2
f
dt

(12)

This assumption can be used to simplify a system if a number of machines are connected to the same bus.

5 Two Machines Arranged as Generator and Load.

Figure 5 shows a two machine configuration in which one machine is acting as the generator, and the second machine acts as a load. The swing equations are:
H1 d 21 = Pm1 Pe1 f dt 2 H 2 d 2 2 = Pm 2 Pe 2 f dt 2

(13a)

(13b)

These equations can be rearranged as:

Pm1 Tm1

Pe1 Te1
4

Pe2 Te2

Pm2 Tm2

Figure 5 : Two Machines Connected as a Generator and Motor.

ELEC 4100
Pm1 Pe1 Pm 2 Pe 2 1 d 21 d 2 2 = 2 2 H1 H2 dt f dt

TRANSIENT STABILITY

(14)

Now define the difference between the machine angles as: 12 = 1 2 , then: H 2 Pm1 H1Pm 2 H 2 Pe1 H1Pe 2 1 d 212 = H1 H 2 H1 H 2 f dt 2
H 2 Pm1 H1Pm 2 H 2 Pe1 H1Pe 2 1 H1H 2 d 212 = 2 H1 + H 2 H1 + H 2 f H1 + H 2 dt Pm Pe = H d 212 f dt 2

(15)

(16)

(17)

Where:

H=
Pm = Pe =

H1 H 2 H1 + H 2
H 2 Pm1 H1Pm 2 H1 + H 2 H 2 Pe1 H1Pe 2 H1 + H 2

(18) (19) (20)

If there are no losses in the system, then:

Pm1 = Pm 2 Pe1 = Pe 2

(21) (22)

If one machine approximates an infinite bus, then H 2 , so:

Pm1 Pe1 = H1 d 212 f dt 2

(23)

Clearly in assessing transient stability, the angle difference is most important when considering two or more machines.

6 Transmitted Power.
To investigate the swing equation we need an expression which describes the power flow from the machine to the load or the infinite bus. Consider the following simplified power system, shown in figure 6. Here Xe is the Thevenin equivalent reactance of the network. E is the transient excitation voltage behind the transient reactance, and Xd is the transient reactance. Figure 7 shows the phasor diagram for the circuit. The active power being transferred to the infinite bus is given by:
E ' V Pe = Re E ' I = Re E ' j ( X 'd + X e )

(24)

Pe =

E 'V sin X 'd + X e

(25)

ELEC 4100

TRANSIENT STABILITY

X'd

Xe

Vt E' = 0 V = 00

Figure 6 :Equivalent Circuit of a Synchronous Machine Connected to an Infinite Bus.

E'

jI(X'd+Xe) I
Figure 7 : Phasor Representation of the Synchronous Machine Circuit. If we substitute the expression for the electrical power into the swing equation we have: Pm = H d 2 E 'V sin + 2 f dt X 'd + X e (26)

Vt

This non-linear differential equation is solved to calculate the transient behaviour of the machine due to a fault. For simplified analysis, the mechanical power input and the voltage behind the transient reactance are often assumed to be constant.

7 Linearisation of the Swing Equation.

For small changes in the rotor angle , the swing equation can be linearised. For example, if changes to + , then: sin ( + ) = sin ( )cos( ) + cos( )sin ( ) = sin ( )cos( ) + cos( )sin ( ) If the is sufficiently small, then :
cos( ) 1, sin ( )

(27)

So the power transferred to the infinite bus changes to: Pe Pe + Pe = E 'V [sin ( ) + cos( )] X 'd + X e (28)

Comparing the swing equations gives: Pm Pe = H d 2 d + Kd 2 f dt dt

H d 2 ( + ) d ( + ) + Kd Pm (Pe + Pe ) = 2 f dt dt 6

(29)

ELEC 4100 Subtracting one equation from the other gives: d E 'V H d 2 + Kd + cos( ) = 0 2 f dt dt X 'd + X e E 'V cos( ) X 'd + X e

TRANSIENT STABILITY

(30)

This is a second order linear differential equation. Define the synchronising power coefficient as: Sp = (31)

So equation (30) becomes:

H d 2 d + Kd + S p = 0 2 dt f dt

(32)

Applying the Laplace transform and neglecting initial conditions:

(s ) s 2 +

S f K d f s+ p H H

=0

(33)

The roots in s give the system poles as:

2 K d f K d f S pf s1 , s2 = 2H H 2

(34)

Now consider the simplified case where the damping is assumed to be zero. Then: s1 , s2 = S pf H (35)

If the synchronising power coefficient Sp is positive (i.e. 900 900 ) then the roots are imaginary and are given by: s1 , s2 = S pf H = j n (36)

Therefore the system responds with sustained oscillations at the natural frequency n. If the synchronising power coefficient Sp is negative (i.e. 900 2700 ) then the roots are real, with one root in the left hand plane, and the second root in the right hand plane. This is clearly an unstable condition. Hence, the synchronising power coefficient must always be kept positive to

Sp > 0

Sp < 0

Figure 8 : Pole locations for positive and negative synchronising power coefficients.

ELEC 4100
XL X'd XT1 (1-a)XL Vt E' = 0 aXL V = 00

TRANSIENT STABILITY

Figure 9 :Single Machine, Double Transmission Line network during the fault.

XL X'd XT1 (1-a)XL

XEQ

Figure 10 : Application of the star-delta transformation to the network of figure 9.

ensure steady-state stability, requiring that the rotor angle is restricted to 900 900 . Figure 8 illustrates the S-plane locations for the poles of the linearised system for the two different regions defined by the synchronising power coefficient.

8 Analysis of the Transmission Network.

To investigate the transient behaviour of the system, we need to solve the transmission network for conditions before the fault, during the fault and after the fault has been cleared. Consider the system shown in figure 2, consisting of a synchronous machine connected to an infinite bus through two parallel transmission lines. For the pre-fault situation the power flow from the machine to the infinite bus is given by:
Pe _ prefault = E 'V sin X L1 X L 2 X 'd + X t + X L1 + X L 2 (37)

If the impedances of the transmission lines are matched, then:

Pe _ prefault =

E 'V 1 X 'd + X t + X L 2

sin

(38)

Where: X L1 = X L 2 = X L . When a fault occurs on transmission line 2, at a distance a p.u. from the infinite bus, then the circuit can be redrawn as shown in figure 9. Here it can be seen the that faulted line forms an equivalent star circuit, which can be simplified using the star-delta transformation, as shown in figure 10. When this is done, the equivalent reactance is given by:
2 ( X 'd + X t )X L + (1 a )X L + ( X 'd + X t )(1 a )X L X eq = (1 a )X L

(39)

If the fault occurs at the mid-point of the line, then a = 0.5 and:
8

ELEC 4100
X eq = 3( X 'd + X t ) + X L The power flow is then given by: Pe _ fault = E 'V sin 3( X 'd + X t ) + X L

TRANSIENT STABILITY
(40)

(41)

Alternatively, the Thevenin equivalent of the infinite bus voltage as seen from the line terminal of the transformer could be used to produce the same result. The fault is cleared by opening both circuit breakers on the faulted line. In this case the power flow is given by:
Pe _ post fault = E 'V sin X 'd + X t + X L

(42)

Equations (38), (41) and (42) are therefore used to define the three differential equations that predict the transient characteristics of the machine, pre-fault, during the fault, and after the fault has been cleared.

9 The Equal Area Criterion.

One method of investigating the stability behaviour of a single machine/infinite bus system is to apply the Equal Area Criterion. The method does not solve for the rotor angle, but rather it tells us the maximum angle which the machine can advance to before the fault is cleared in order to preserve transient stability. Consider the following swing equation:
Pa = Pm Pe = H d 2 f dt 2

(43)

This may be written as:

d 2 f = Pa dt 2 H

(44)

Also:
d d dt dt
2

d d 2 =2 dt dt 2

(45)

Therefore the swing equation can be written as:

d d 2 f d = Pa 2 dt dt H dt 2 d d d 2f Pa = dt dt dt H

(46)

Integrating both sides of this equation gives: 2f d = H dt

2

Pa d

(47)

Or:
d = dt 2f H

Pa d

(48)

ELEC 4100

TRANSIENT STABILITY

d = 0 , the machine rotor angle is no longer increasing, and has dt reached a maximum (or minimum) and this occurs at a maximum angle m, where:

So, when the derivative

m
0

Pa d = 0

(49)

Separating equation (49) into accelerating and decelerating areas, gives:

1
0

Pa d + Pa d = 0
1

(50)

In equation (50), the first integral may represent the accelerating region, where the accelerating power is positive, and so 1 is the cross-over angle where the accelerating power is zero. The second integral may represent the decelerating area, where the accelerating power is negative. If the d sum of these two integrals is not zero, then by definition 0 , and this means that the rotor dt angle has not reached a maximum or a minimum. Or in other words the rotor angle is either continually increasing or decreasing. This constitutes unstable operation (i.e. continuous poleslipping). d = 0 , which means dt that the rotor angle has achieved a maximum or minimum value. This means that the rotor angle is confined to a particular angular region, which is a necessary condition for stability. Therefore the criterion of equal accelerating and decelerating areas is an immediate indicator of stability. If the accelerating and decelerating areas are equal, then the derivative
9.1.1 Example. Figure 11 shows an application of the equal area criterion to a synchronous machine connected to an infinite bus. The machine is being driven with an initial mechanical power Pm0. The corresponding operating point on the electrical power curve is at an initial rotor angle 0. If a step change in the mechanical power occurs, such that the new value is Pm1, then the machine will begin to accelerate, and the rotor angle will begin to increase. This acceleration continues, until the rotor angle reaches 1 at which point the machine begins to decelerate. By application of the equal area

P (p.u.) Pmax A2 Pm1 Pm0 A1 Pmaxsin()

0 1 2 /2

(rad)

Figure 11 : Application of the Equal Area Criterion.

10

ELEC 4100

TRANSIENT STABILITY

criteria, the integral of the difference between the new mechanical power and the output electrical power in acceleration mode (i.e. Area A1), must equal the same integral in deceleration mode. So the rotor angle continues to increase until it reaches the angle 2 at which point the areas A1 and A2 are equal. Now the machine begins oscillating until it settles at the new operating point 1. If the acceleration phase was such that the rotor angle would have to advance to 3 to achieve equal areas, then this is the stability limit. If the rotor angle would need to advance beyond 3 then it is not possible to satisfy the equal area criterion, and pole-slipping will occur.
9.1.2 Example - Fault Clearing Angles. The equal area criterion can be used to determine the critical rotor angles, by which point a fault must be cleared to ensure that the generator can remain synchronised to the infinite bus. Recall that in section 8 the power curves were determined for the pre-fault, during fault, and post-fault conditions. These curves can be plotted against the rotor angle as shown in figure 12.

Figure 13 illustrates how to determine the critical clearing angle using these curves. For the prefault condition, the input mechanical power and the pre-fault electrical power curve determine the initial operating rotor angle 0. When the fault occurs, the machine changes to the during fault curve, and begins to accelerate. It should be immediately clear for this example that without clearing the fault the machine will lose synchronisation since the acceleration and deceleration areas can not be made equal. However if the fault clears by some critical rotor angle crit, then it is
P (p.u.) Prefault Power Post-fault Power

Pm Fault Power

/2

(rad)

Figure 12 : Electrical Power Curves During a Fault Event.

P (p.u.) Prefault Power Post-fault Power

Pm Fault Power

0 fin

/2

crit m

(rad)

Figure 13 :Application of Equal Area Criterion to a Fault Event.

11

ELEC 4100

TRANSIENT STABILITY

possible to increase the deceleration area using the post-fault power curve, as illustrated in figure 13. After the fault has been cleared by this critical angle, the machine will oscillate and return to the final operating point at fin. The critical clearing angle can be calculate using:
(P
crit 0

Pfault ) d =

crit

(P

post _ fault

Pm ) d

(51)

This can also be written as:

crit
0

Pm E 'V sin ( ) d = m E 'V sin ( ) Pm d crit X post _ fault X fault

(52)

Integrating: cos( m ) cos( 0 ) 1 1 Pm ( m 0 ) + E 'V = E 'V cos( crit ) X fault X post _ fault X post _ fault X fault (53)

This expression allows the direct determination of the critical clearing angle for the synchronous machine.
9.1.3 Example : Fault Clearing Angles - Maximum Swing. In the above example the critical clearing angle for a fault is calculated based on the equal area criterion. An allied problem is the determination of the maximum angle that the machine swings to given that the fault has cleared by a given angle. The challenge with this kind of problem is that a transcendental equation results, and the solution of this expression requires the use of numerical techniques.

Consider the above example, in which a fault occurs, but clears at an angle c, and then the generator swings to a maximum angle max. To determine max apply the equal area criterion:

(P
c 0
c

Pfault ) d =

max

(P

post _ fault

Pm ) d

(54)

This can also be written as: Pm E 'V sin ( ) d = max E 'V sin ( ) Pm d 0 X fault c X post _ fault Integrating: Pm max + cos( c ) cos( c ) cos( 0 ) E 'V cos( max ) = Pm 0 + E 'V + X post _ fault X X X post fault fault fault _ (56) (55)

Here the unknown max appears on the left hand side of the equation, while all the known terms appear on the right hand side of the equation. This equation can be re-written as:

max + cos( max ) =

Where:

(57)

= Pm
=
E 'V X post _ fault

(58) (59)

12

ELEC 4100 cos( c ) cos( c ) cos( 0 ) + X X X fault _ post fault fault

TRANSIENT STABILITY

= Pm 0 + E 'V

(60)

To solve equation (57) for max, apply the Newton-Raphson iterative method. Recall that the Newton-Raphson method solves systems of the form:
f (x ) = y

(61)

Where f is a non-linear function of the unknown x. The iterative technique relies on the first order Taylor series expansion of f(x), such that the (p+1)th estimate of x is related to the pth estimate according to: x p +1 = x p +

[y f (x )] f ' (x )
p p

(62)

So applying this to equation (57) gives:

max p +1 = max p + max p cos max p sin max p

)] [

)]

(63)

It should be pointed out that the solution to this equation does require a good initial estimate to the maximum angle to ensure convergence to the correct solution.

10 Calculation of the Critical Clearing Time.

The equal area criterion allows for the determination of the critical clearing angle, but not the critical clearing time. Since the swing equation is a non-linear second order differential equation, generally we need to use a numerical method to solve it. We can use the Euler, Runge-Kutta, or any other appropriate numerical technique, with the result being the variation of the rotor angle in time.
10.1.1 Example: Consider a synchronous machine connected to an infinite bus, with the following operating parameters:

Inertia Constant H = 5 seconds. Synchronous Frequency f = 50Hz. Mechanical Power Pm = 0.8 p.u. Pre-fault Electrical Power Pe = 2.34 p.u. During-fault Electrical Power Pe = 0.9 p.u. Post-fault Electrical Power Pe = 1.8 p.u. No damping.

Figure 14 shows the response of the synchronous machine to the fault when the fault is cleared after a time interval of 0.7 sec. Here it can be seen that the rotor angle swings to a large angle (1300) but is stable after the first swing. Figure 15 shows the same system, but in this case the clearing time for the fault is in fact 0.8 sec. It is evident that this case is unstable, and pole slipping has occurred.

13

ELEC 4100
140 120 100 80
Rotor Angle (deg)

TRANSIENT STABILITY

60 40 20 0 -20 -40 -60 0 0.2 0.4 0.6 1.0 0.8 Time (sec) 1.2 1.4 1.6 1.8 2.0

Figure 14 : Undamped Swing Characteristic Stable Operation.

1000 900 800 700

Rotor Angle (deg)

600 500 400 300 200 100 0 0 0.2 0.4 0.6 1.0 0.8 Time (sec) 1.2 1.4 1.6 1.8 2.0

Figure 15 : Undamped Swing Characteristic Unstable Operation.

11 Review of Classical Model.

The representation which has been used for the above analysis is called the classical model. Some of the shortcomings of this model are as follows: Transient stability is decided in the first swing. 14

ELEC 4100 Constant generator main field-winding flux linkage. Neglecting the effects of damping. Constant mechanical power. Representing loads by constant passive impedances.

TRANSIENT STABILITY

Possible improvements to this representation include: Study the transient response for more than one second as maximum swing may not be the first swing in a multi-machine or multi-bus system. Include a representation of dynamic response to improve the accuracy of the model. Also represent saturation in iron. Loads may need to be represented by constant P and Q, or have voltage and frequency dependence.

15

ELEC 4100

PROTECTION

ELEC 4100 ELECTRICAL ENERGY SYSTEMS PROTECTION. 1 Introduction.

The operation of a power system means the transmission of power at high voltages over considerable distances, and the transformation of power from one voltage to another. This activity normally proceeds smoothly, but is not performed without some risk of something, somewhere and sometime going wrong. This is most often a result of high voltage conductors being accidentally connected to objects at vastly different potentials (e.g. a single line to ground fault). Such unwanted fault events can be highly destructive to equipment in the power system, and hence there is a need for a protection system which reduces the impact of these events. Power system protection does not prevent fault events, but is concerned with their detection and swift removal, before they have a chance to cause equipment damage or endanger life, and to minimise voltage disturbances to customers. Protection is generally designed to switch out of service the minimum required amount of power system plant to clear the fault. The plant switched out of service must include the faulted section of network, but often other plant not specifically involved with the fault must also be switched out of service. A fault represents the departure of one or more components of the power system from its normal operating condition. Such unwanted departures, however, may also include abnormal operating conditions that are not specifically regarded as faults, for example voltage collapse due to system overload. Faults can be due to many different types of event, including: Over-voltages caused by lightning or switching transients. Mechanical failure, e.g. falling conductors. Natural causes, e.g. trees growing into conductors, or animals climbing onto the line. Insulation degradation and contamination.

Faults usually produce currents that are several orders of magnitude larger than normal load currents. Though in weak systems, fault currents may not be significantly larger than load currents (and this causes protection difficulties in terms of fault detection). The art of system protection therefore involves the ability to discriminate true fault events from the general operational state of the power system. Consequences of faults include: Arcing, leading to high temperatures, and possibly fires and explosion. Sustained overheating due to high currents may reduce the useful life of equipment, e.g. insulation damage and conductor annealing. Destructive mechanical forces on windings and busbars due to the high currents involved. Over-voltages may stress insulation beyond its breakdown value. Remove faults from the power system as quickly as possible. Remove no more of the system than is absolutely necessary to clear the fault. As much of the load as possible should continue to be supplied. 1

The protection system must:

ELEC 4100

PROTECTION

Faults are removed from a system by opening or tripping circuit breakers, or by fuses (typically at low voltage levels). The operation should be highly selective. Only those breakers closest to the fault should operate to remove or clear the fault. The rest of the system should remain intact. Fault conditions are detected by monitoring the voltages and currents at various points in the system (Note that this is done implicitly by fuses). For circuit breakers, abnormal values individually or in combination cause relays to operate, energising the breakers tripping circuits. Faults can occur within the protection system itself, and so there is also a need for some form of backup protection.

1.1 Why is Rapid Fault Clearing necessary?

1.1.1 Voltage Dip Prevention. The duration and extent of voltage dips (or sags) are critical factors, especially for motors, and Programmable Logic Controllers (PLCs). PLCs are in fact particularly susceptible, and can reset if the voltage dip is too long. For batch processes controlled using micro-processors, this can compromise a production run, and so customers are now becoming more demanding with their requirements as the penetration of computer controlled equipment increases. Generally the higher the voltage at which the fault occurs, the more widespread the voltage dip, and therefore the greater the number of customers affected. Higher voltage circuits therefore usually warrant more attention to clearing speed than low voltage circuits. 1.1.2 To Prevent Damage to Electrical Plant. Damage to electrical plant is predominantly due to high magnitude fault currents. Equipment other than the faulted plant may be damaged if it carries large fault currents for longer than it has been designed for. Transformers, isolators, circuit breakers and current transformers may suffer thermal damage (I2t) and mechanical damage due to the large electro-mechanical forces involved. Conductors in transmission lines are also susceptible to high currents, and can anneal if their temperature climbs above a critical threshold. 1.1.3 System Stability. Power system stability is a problem at transmission voltages (i.e. above 220kV in Australia). As a result of a substantial transmission line or bus fault, generators at different sites may lose synchronism with each other. This produces wildly fluctuating power flows, which eventually split the system (often as a result of unintended protection relay operation), causing the wide-spread black-outs. This is sometimes referred to as a Cascaded failure of the power system. Factors affecting the system stability include: The physical location of the fault. The fault type (e.g. single line to ground, bolted three phase etc.) The fault duration. The load flow at the time of the fault (the larger the load flow, the greater the potential for instability). The generation pattern the location and size of on-line machines. The physical inertia of the machines lighter machines have a greater tendency for instability. The voltage level where the fault occurs. 2

ELEC 4100

PROTECTION

These factors all have a large impact on whether the system will remain stable or split at the occurrence of a fault. Protection systems in the vicinity of generators are generally designed to operate quickly if it is considered that the generator in question has a risk of instability when subjected to a nearby fault. 1.1.4 Safety. Since fault currents are generally large compared to normal operational currents, the voltage drop across impedances in the current path can become quite large. For ground faults this can lead to large voltages on equipment and along the ground itself. If personnel are with the vicinity of these potentials (generally referred to as touch and step potentials) then there is a risk of electrocution fatalities are possible. Clearing a fault quickly is therefore critical to minimise the risk of injury due to rising touch and step potentials. 1.1.5 Fault Limiting. In general faults are not self-limiting. Once a fault commences it can generally sustain itself indefinitely. The exceptions are systems earthed with Petersen Coil systems or large Neutral Earth Resistors (NERs) where the fault level is in fact very low. Protection systems are therefore necessary to limit the time duration that a large fault current is sustained.

2 Instrument Transformers.
To be able to clear a fault from a power system it is first necessary to be able to detect that the fault has occurred. This function is performed using instrument transformers, and there are two basic types potential transformers for monitoring voltage, and current transformers for monitoring the line current. The purpose of instrument transformers is to reduce the power system level voltage and currents (recall that fault currents are often 10p.u. or more) to standardised levels that suitable for relays and protective devices.

2.1 Potential Transformers (PTs).

Potential transformers monitor the system voltage, and exploit a simple two winding structure, which provides galvanic isolation from the power system level voltages, and steps the voltage down to manageable levels. Ideally a PT is terminated with an open circuit and can be modelled as an ideal transformer where the secondary side voltage is related to the primary voltage according to:

V2 =

N2 V1 N1

(1)

PTs are often designed to have a secondary side voltage of 120V, and the turns ratio can vary between unity to 4500:1. In practice a PT must be terminated by a high impedance voltage sensitive device commonly referred to as a burden. Typical PT ratings are in the vicinity of 50VA, and the polarity is indicated with the standard dot convention. For high voltage levels it is often necessary to use a capacitive voltage divider to drop the primary voltage of the PT. This is illustrated in figure 1 below. A problem with this approach is that the burden and magnetising current loads the capacitive divider, and this must be compensated for to eliminate any errors introduced. Typically a reactor is placed in series with the PT to achieve this function. This is further illustrated in figure 2 which shows the Thevenin equivalent network of the PT viewed from node B, in which: CT = C1 + C2 XT = (2)

1 CT
3

(3)

ELEC 4100

PROTECTION

C1 C2

Va jX1

Vb PT Load

Figure 1 : PT with a capacitive voltage divider.

Va VT -jXT jX1

Vb Referred Load

Figure 2 : Thevenin equivalent network. The impedance that then appears between the Thevenin equivalent voltage and node b is given by:
ZT = j [ X 1 X T ]

(4)

Clearly if X1 is made equal to XT at the power system frequency, then the effect of loading the capacitive divider is mitigated.

2.2 Current Transformers (CTs).

Current transformers monitor the line current, but unlike PTs a protection CT must be able to tolerate very large overcurrent events without introducing significant errors. The saturation characteristics of a CT therefore become significant, and it is not always possible to model the CT as an ideal transformer. The structure of a CT is usually a single turn primary with a multi-turn secondary winding which is ideally terminated with a short circuit. In practice the termination is a finite impedance burden, and effects such as magnetisation and leakage will introduce measurement errors. The equivalent circuit of a practical CT illustrates this as shown in figure 3. Here the shunt magnetisation branch element is a variable inductance to account for saturation. The measurement

I1

I'2

Z2

I2

E2

Im

jXm

V2

ZB

N1

N2
4

Figure 3 : Practical CT with a shunt magnetisation branch and finite burden.

ELEC 4100 error of the CT is defined as:

CTerr = I '2 I 2 I '2 = Im I '2

PROTECTION

(5)

To calculate this error for a given CT, it becomes necessary to refer to that CTs excitation curve which relates the secondary side excitation voltage and the secondary side excitation current. So for a specific CT output current I2, the secondary side excitation voltage is calculated according to:
E2 = (Z B + Z 2 )I 2

(6)

The magnetisation current Im, is then calculated using the CT excitation data. It is then possible to obtain the secondary referred input current I2, which enables the calculation of the CT error according to (5) for the specific operating condition. For a specific application it is generally necessary to assess the CT error over the reasonable operating range, and for the specific burden. Example :- See Glover and Sarma, Chapter 10, page 455, example 10.1.

3 Protection Relays.
The principle function of relays in a power system is to protect the system and associated equipment from damage caused by abnormal or faulted conditions. The input(s) to a relay include the current and/or voltage measurements from PTs and CTs at different points in the system, and the relay must make a decision as to whether these measurements are within acceptable tolerances. The decision basis can be simple (i.e. threshold detection) or include complex phase and/or magnitude information. Once a relay has made a decision to isolate a section of the network it must somehow feed this information to the circuit breaker itself which is rated for the full line currents and voltage levels of the relevant section of the power system. Relays can be either electromechanical or solid state.

3.1 Overcurrent Relays.

The simplest relay structure is the instantaneous overcurrent relay, as illustrated in figure 4 below. This relay is made up of a magnetic core, an excitation coil wound around this core, a moveable plunger and spring, and a set of electrical contacts. When there is no excitation signal present on the coil, the spring pulls the plunger down into the normal position. Here the normally closed contact pair is made so that there is an electrical connection between the points B and B.
Magnetic Core

Excitation Coil Plunger

A Normally Open Contacts A'

B Normally Closed Contacts B' Spring

Figure 4 : Mechanical structure of a simple instantaneous overcurrent relay.

ELEC 4100

PROTECTION

Similarly the normally open contacts are broken. When the excitation coil is energised the field in the coil produces an electromagnetic force which tries to reduce the length of the air gap in the magnetic circuit. If the applied current is above a sufficient threshold (called the pick up current) to overcome the spring stiffness (and gravity) the plunger moves vertically, breaking the normally closed contacts and making the normally open contacts. When the plunger is in the vertical position (i.e. minimum air gap and hence flux path) the ampere-turn required to hold this position is much less than that required to move the plunger into this position. The current must therefore fall well below the pick up current before the plunger returns to the normal position, and the current at which this occurs is called the drop out current. Instantaneous over-current relays have one major draw-back in that they provide very little ability to discriminate between fault events so as to localise the event. This makes it very hard to design a protection system with high selectivity and reliability.

3.2 Inverse Time Overcurrent Relays.

One way to introduce a degree of selectivity into the overcurrent relay is to add a magnitude dependent time delay function. The way this is normally achieved in practice (for electromechanical relays) is to utilise a freely rotating induction disc in the reluctance path of a relay, as illustrated in figure 5 below. When the excitation coil is energised, a field is set up through the magnetic circuit, which induces eddy currents in the induction disc (which is normally made from aluminium). The interaction between the field in the magnetic core, and the field produced by the eddy currents in the induction disc produces a torque. The induction disc is held in place with a spiral spring, but if the torque is sufficient to overcome the stiffness force of the spring, the disc will rotate. The current required to produce such a torque is called the pick up current. Once the disc has rotated through an angle , then the output contacts are made. Clearly the time required to rotate through this angle depends upon the magnitude of the torque, which depends upon the magnitude of the applied current. In this way, not only does the relay respond to an overcurrent event, but the time required for the relay to operate is magnitude dependent. Figure 6 illustrates a typical inverse time overcurrent relay characteristic for a Westinghouse CO-8 relay. There are two settings for this relay the Current Tap Setting (CTS), and the Time Dial Setting (TDS). The characteristic curves for the CO-8 relay are then given as a function of these two settings, with the x-axis corresponding to multiples of the CTS, the y-axis corresponding to the breaker trip time, and each curve shown corresponds to a specific TDS. Each curve asymptotes to as the x goes to unity (recall this corresponds to the pick up current), and asymptotes to zero as x goes to .

Magnetic Core

Excitation Coil Induction Disk

(a)

(b)

Figure 5 : An electromechanical, inverse time overcurrent relay, using an induction disc. (a) side-on view (b) top view. 6

ELEC 4100

PROTECTION

Figure 6 : Westinghouse CO-8 inverse time overcurrent relay characteristic.

3.3 Protection of Radial Systems.

Inverse time overcurrent relays are well suited to the protection of radial systems. A radial system is characterised by a group of load buses that feed out radially from a central supply point, as illustrated in figure 7. Such structures are very common in rural areas, especially in Single Wire Earth Return (SWER) systems. Clearly then if the central supply is isolated, then the entire system loses supply, and hence there is a strong need to localise fault events, and ensure that the upstream breakers do not respond to downstream faults. However it is also possible to use the upstream breakers to provide back-up protection for the downstream breakers by operating more slowly than the downstream breakers, and this is how inverse time overcurrent relays are used. The application of these relays to radial system protection is best illustrated with an example. Figure 7 shows a three bus radial system, with three time delay overcurrent relays protecting the system. For faults between the breaker B3 and Bus 3, it is clear that the breaker B3 should operate to clear the fault so that Bus 1 and Bus 2 can remain in service. However if breaker B3 failed to operate, B2 could operate as a back-up. To ensure that B2 only operates in back-up mode, the time delays for each relay should be coordinated so as to achieve sequential tripping. Similarly breaker B1 should be coordinated with B2 so as to achieve back-up protection of B2. The coordination time interval is the time interval between the primary and the remote back-up protection devices. Consider the application of inverse time overcurrent relays to the protection of the system shown in
Bus 1 Bus 2 Bus 3

B1

B2

B3

L1

L2
7

L3

Figure 7 : Single Line Diagram of a Radial System.

ELEC 4100 figure 7 with the following parameters [See Glover and Sarma Example 10.4].
Example :

PROTECTION

The system shown in figure 7 has loading, fault, relay and breaker data as shown in tables 1 to 3. It is necessary to select CTS and TDS values for the three breakers so as to achieve a coordination time interval of 0.3s. The line to line voltage of the system is 34.5kV. Table 1 : Maximum Loads for system shown in figure 7. Bus Bus 1 Bus 2 Bus 3 Load MVA 11.0 MVA 4.0 MVA 6.0 MVA Lagging p.f. 0.95 0.95 0.95

Table 2 : Maximum and Minimum Fault Currents for system in figure 7. Bus Bus 1 Bus 2 Bus 3 Maximum Fault Current (3) 3000 A 2000 A 1000 A Minimum Fault Current (L-G) 2200 A 1500 A 700 A

Table 3 : Breaker, CT, and relay data for system in figure 7. Breaker B1 B2 B3
Solution :

Breaker Operating Time 100 ms (5 cycles @50Hz) 100 ms (5 cycles @50Hz) 100 ms (5 cycles @50Hz)

CT Ratio 400 : 5 200 : 5 200 : 5

Relay Type CO-8 CO-8 CO-8

The first step is to select the three CTS values so that the three relays will not operate under normal maximum loading conditions. So using the three maximum loading conditions the normal loading current in each breaker and CT secondaries can be found as: I B3 =
I B2 =

SL3 VLL

6 106 = = 100.4 A 3 ( 34.5 103 ) 3

I 'B 3 =
I 'B 2 = I 'B1 =

I B3 = 2.51A 200 / 5
IB2 = 4.18 A 200 / 5 I B1 = 4.39 A 400 / 5

( 4 + 6 ) 106 = 167.3 A SL3 + SL 2 = VLL 3 ( 34.5 103 ) 3

6 S L 3 + S L 2 + S L1 ( 4 + 6 + 11) 10 I B1 = = = 351.4 A VLL 3 ( 34.5 103 ) 3

ELEC 4100

PROTECTION

Notice that for breakers B1 and B2 the downstream loads must be accounted for when determining the actual current carried by the breaker. It is necessary to ensure that the breakers do not operate for the CT secondary currents given above. Therefore select pick-up currents above these values. There is a wide range of possible values that can be used, but realistically any value that is sufficiently larger than the maximum load current, but also sufficiently smaller than the minimum likely fault current should be used. As a rule, the reliable operation of a time delay relay requires that the ratio of the minimum fault current to the pick-up current should be 2. So the secondary referred minimum fault currents are: I B 3_ min fault = I B 2 _ min fault = I B1_ min fault = 700 = 17.5 A 200 / 5 700 = 17.5 A 200 / 5 1500 = 18.75 A 400 / 5 (in back-up mode for B3) (in back-up mode for B2)

Therefore we can choose: CTS B 3 = 3 A CTS B 2 = 5 A CTS B1 = 5 A It now remains to coordinate the relays for the maximum fault currents. For the breaker B3, the maximum fault current that this relay will see is 2000A, which will correspond to a secondary side current of: I 'B 3_ fault = 2000 = 50 A 200 / 5 and so I 'B 3_ fault CTS B 3 = 50 = 16.7 3

For the fastest possible trip time, use the following time dial setting: TDS B 3 = 1 2

From figure 6, the relay operating time for the maximum fault current is: t B 3 = 0.05s For the breaker B2, the secondary referred fault current and fault to pick up current ratio is: I 'B 2 _ fault = 2000 = 50 A 200 / 5 and so I 'B 2 _ fault CTS B 2 = 50 = 10 5

Then taking into account the required coordination interval and the operation time, B2 should not operate until after a time delay of:

t B 2 = t B 3 + tc + top = 0.05s + 0.1s + 0.3s = 0.45s So from figure 6, this corresponds to a time dial setting of: TDS B 2 = 2 So now B2 is coordinated with B3, and will operate as a back-up protection device for faults at bus 3. It remains to coordinate B1 with B2, for the maximum fault events that can occur at Bus 2.
9

ELEC 4100

PROTECTION

The largest fault at Bus 2 is 3000A. Hence the secondary referred fault current and fault to pick up current ratio is: I 'B 2 _ fault = 3000 = 75 A 200 / 5 and so I 'B 2 _ fault CTS B 2 = 75 = 15 5

From figure 6, the relay operating time for the maximum fault current is: t B 3 = 0.35s For the breaker B1, the secondary referred fault current and fault to pick up current ratio is: I 'B1_ fault = 3000 = 37.5 A 400 / 5 and so I 'B1_ fault CTS B1 = 37.5 = 7.5 5

Then taking into account the required coordination interval and the operation time, B1 should not operate until after a time delay of: t B1 = t B 2 + tc + top = 0.35s + 0.1s + 0.3s = 0.75s So from figure 6, this corresponds to a time dial setting of: TDS B1 = 2.8 The coordinate relay settings are summarised in table 4 below. Table 4 : Coordinated Relay settings for the system in figure 7. Breaker B1 B2 B3 Relay Type CO-8 CO-8 CO-8 CTS 5A 5A 3A TDS 2.8 2 0.5

4 Directional Relays.
Systems that do not have a radial structure are difficult to protect using simple time coordinated delay relays. Consider the system shown in figure 8, which contains two sources and four voltage buses. In the event that there is a fault in the middle of the transmission line between buses 1 and 3, the breakers B31 and B32 will both sense the overcurrent. However to minimise the amount of the system taken out of service ideally the breakers B13 and B32 should operate. We could use time delay relays to coordinate B32 with B31, but in this case B31 would always operate before B32,
Bus 1 B12 B21 Bus 2 Bus 4

B1

B13

B23

B24

B42

B4

B31 Bus 3

B32

Figure 8 : Single Line Diagram of a Two Source System. 10

ELEC 4100
CT

PROTECTION

PT Direction O/C

N/O Relay Contacts

Figure 9 : Electrical connections of a directional relay. which even when there is a fault on the transmission line between buses 2 and 3 which is clearly not acceptable. So the relays need a way to distinguish whether the fault is on the transmission line they are connected to, or behind them on the voltage bus or another transmission line on that bus. In other words the relays need directional information. A way to equip relays with a directional capability is to measure the local system voltage, and compare the system current phase angle to voltage. Such a relay structure is illustrated in figure 9. The directional relay relies on the fact that the transmission line is predominantly reactive. Therefore if there is a fault to the right of the relay the measured voltage must lead the fault current by almost 900. If however the fault current is to the left of the relay, the current phase angle will have changed by 1800, and so the measured voltage will lag the current by almost 900. In figure 9 the directional relay is shown with two distinct sections, a directional section, and a conventional overcurrent relay. However an integrated solution is the induction cup relay shown in figure 10. This relay consists of a voltage and a current coil wound onto a magnetic core. The coil directions are perpendicular to one another. At the centre of the core is a disc which is free to rotate. The fields produced by the voltage and current coils produce a torque on the disc, and the sign of the phase shift between the two signals determines the direction that the induction disc rotates. In one direction the disc rotation causes the relay contacts to close if the magnitude of the current is sufficiently large. In the other direction a mechanical stopper prevents the contacts from being closed, providing the directional capability required.
Contacts

Voltage Coil Normally Open Contacts

Current Coil

Figure 10 : Induction Cup Relay. 11

ELEC 4100
Bus 1 Bus 2 Bus 3

PROTECTION

B1

B12

B21

B23

B32

B3

L1

L2

L3

Figure 11 : Simple Two Source Power System.

4.1 Protection of Two-Source Systems.

It is possible to use a combination of directional and inverse time overcurrent relays to protect a two source systems. The directional relays are used to ensure that each relay only looks in one direction (i.e. downstream) and the time delay relays are coordinated so that the apparent upstream breakers provide back up protection to the downstream breakers. This is illustrated for the twosource system shown in figure 11. To protect this system the breakers B12 and B23 should be made directional, looking only to the right. Similarly breakers B21 and B32 should be made directional, only looking to the left. Breakers B1 and B3 can be simple inverse time overcurrent relays. The CTS and TDS values for the overcurrent relays can be set so that B12 coordinates with B23, and B1 should coordinate with B12. Similarly B32 should coordinate with B21, and B3 should coordinate with B32.

5 Protection Zones.
As soon as the power system becomes more complex than a radial structure, the effective protection of the system using simple directional or inverse time overcurrent relays becomes infeasible. This is because a key protection objective is to take the minimum amount of the system out of service, so as to maintain as much of the supply as possible to the remaining customers. For more general mesh type system structures, the concept of a zone of protection is far more effective for localising the fault, and then removing it. Protection zones are defined for : Voltage buses Transformers Generators Compensators Rotational loads Transmission lines

The zone concept is illustrated in figure 12 for a four bus, two generator power system. This system is shown to have 11 zones of protection. These zones are defined to ensure that each of the above listed elements has an associated protective zone, and the operation of the circuit breakers within that zone will take the element out of service. Zones of protection have the following properties: Neighbouring zones overlap with one another. This overlapping of zones is necessary to ensure that there is no region in the power system that is unprotected in the event of a fault. The only devices placed in the zone overlap regions are circuit breakers. Hence circuit breakers defined the boundaries of all protective zones in the system. This also explains why there is only one element in zone 6 (the voltage bus) but two in zone 1 (the generator and the transformer). 12

ELEC 4100
Bus 1 Zone 2 B12

PROTECTION

Bus 2

Zone 8 Bus 4 Zone 11 B24a B42a B4 B24b B42b Zone 9 Zone 10

P1
Zone 3

B21

P2
B1 Zone 1 Zone 4 B31 Bus 3 Zone 6 B32 B13 B32

Zone 5 Zone 7

Figure 12 : Single Generator connected to an Infinite Bus. The protective action for a fault in a given zone is the operation of all circuit breakers within that zone. This must therefore necessarily involve the operation of circuit breakers which lie in all adjacent zones due to the overlap.

This last point deserves some additional attention. Consider the two faults indicated at points P1 and P2 in figure 12. Clearly the P1 fault lies in zone 3 and so the corrective action to be taken here is the opening of all circuit breakers in zone 3 B21 and B12. However the fault P2 lies in the overlap region of zones 2 and 3. So in this case all breakers in both zones 2 and 3 should operate B1, B12, B21 and B13. It is for this reason it is necessary to minimise the overlap region between zones so that the minimum section of the system is taken out of service.

6 Distance Relays.
While directional relaying improves the selectivity of a protection system for simple two source systems, for systems with many meshes it becomes considerably more challenging to coordinate the relays effectively. One approach to overcome these difficulties is to recognise that the ratio of the voltage to current at a circuit breaker is more sensitive to the location of the fault than the current or voltage alone. Relays which respond to the ratio of the voltage to current are referred to as Distance relays (also called Impedance or Ratio relays). To understand why distance relays provide better selectivity, recall that during a fault event the current through a circuit breaker increases while simultaneously the voltage collapses. So for example if there is a solid line to ground fault right next to a circuit breaker then the local voltage collapses to zero. The current however can increase to several p.u. and is limited by the up-stream impedance. To consider another example, consider the case where the local voltage falls to 25% of normal rated conditions, while the fault current reaches a level of 10 p.u. The distance relay would detect and impedance change to 2.5%, representing a percentage change of 4000%, or four times the change in the current. It is precisely because of this increase in sensitivity that distance relays are attractive for the protection of transmission lines. Figure 13 shows the trip region for a standard distance relay. The relay is configured with a trip impedance level Ztrip, and if the ratio of measured voltage to current is above this threshold then the system is deemed to be operating in a normal loading condition. When the measured ratio falls below this threshold a trip event occurs. Distance relays can be modified to include a directional capability in two distinct ways. The simplest approach is to simply include a directional relay in series with the distance relay. In this way the phase angle of the ratio must be within a specified range to cause a trip event, providing a trip characteristic as shown in figure 14. The second approach is to offset the trip range of the 13

ELEC 4100
Im{Z} Ztrip Fault Re{Z}

PROTECTION

Normal Condition

Normal Condition

Figure 13 : Trip Zones for a standard Distance Relay.

Im{Z} Normal Condition Ztrip Fault Re{Z} Normal Condition

Figure 13 : Trip Zones for a directional Distance Relay.

Im{Z}

Normal Condition

Fault

Ztrip Re{Z}

Normal Condition

Figure 15 : Trip Zones for a modified (Mho) Distance Relay. distance relay from the origin. Such a relay is referred to as a modified impedance (or Mho) relay, and has a trip characteristic as shown in figure 15. The mho relay has selectivity advantages for high power factor loads. The trip impedance of a relay (Ztrip) is also often defined in terms reach. Here the trip impedance is set to be a certain percentage of the transmission line impedance. So for example if the trip impedance is set for a reach of 80% of the line, for a fault that occurs 90% along the length of the line the distance relay would not trip. It is this concept of reach that has given the relay its name. Back-up protection can also be incorporated into a distance relay. Consider the relaying associated with B23 in figure 16. Usually three impedance relays would be installed at B23. The first provides the first zone of protection, and is usually set to a reach of 80% of the associated line. The second relay protecting zone 2 has a longer reach (usually 120% of the associated line) and includes 14

ELEC 4100

PROTECTION

Bus 4 Bus 2 B2 B23 (a) (b) Zone 1 Zone 2 Bus 5 Zone 3 B5 (e) B32 (c) B35 B53 Bus 3 B34 (d) B43 B4b B4a

Figure 16 : Trip Zones for a standard Distance Relay. all of zone 1. A fault at point (d), for example, would cause breaker B23 to open after a time delay T2. This fault at point (d) is in the primary protection zone of B34, but if this breaker failed to operate then the breaker B23 will provide the back up protection. The third impedance relay provides another level of back up protection. This relay is set to a longer reach again (usually 100% of the associated line and 120% of the adjacent line), and is set to trip after a longer time delay T3. In this way B23 can provide back-up protection for faults that occur at point (e). Zone 1 protection is usually set to cover less than 100% of the associated line. The reason for this is illustrated in figure 16. The impedance seen by the relay for the fault at point (c) corresponds to effectively 100% of the line and offers essentially the same impedance as the fault at point (d). If zone 1 was set to 100% reach then for the fault at (d) B23 would trip instantaneously. To provide selectivity it is therefore necessary to reduce the reach of zone 1 to less than the impedance of the overall line. A fault at point (c) would be cleared instantaneously by B32. Often trip signals are sent from one end of the line to the other, so if B32 tripped, then B23 would receive a corresponding instantaneous trip command despite the fact that the fault lied outside its primary zone of protection.

7 Differential Relays.
Differential protection is commonly used for the protection of generators, voltage buses and transformers. This is because these elements are not distributed and so it is relatively easy to obtain access to the terminal voltages and currents. Any difference between the input and output power flow indicates the presence of an internal fault.

7.1 Generators.
The basic concept of differential protection of a generator is shown in figure 17. In the absence of an internal fault, the current I2 = I1. So using identical CTs, I2 = I1. Since the CTs are connected in series, the current flows from one secondary to the other, and there is no operating current in the winding of the relay. Consider though a phase to phase short in the generator winding. Then I2 I1, and so I2 I1. A differential current I2 I1. will flow in the operating winding of the relay, and if this is above a certain threshold a trip signal will result.

7.2 Voltage Buses.

The differential protection of voltage buses is illustrated in figure 18. The secondaries of CT1 and CT2 are in parallel so that their currents add constructively. In the absence of a bus fault, I1 + I 2 = I 3 . Assume that all of the CTs have the same turns ratio. Then I '1 + I '2 = I '3 . Under these 15

ELEC 4100
Generator Winding

PROTECTION

Neutral Breaker

Main Breaker I1 CT1 I2 CT2

I'1

Trip Winding

I'1-I'2

I'2

Restraining Windings

Figure 17 : Differential Protection of a Generator.

I1 CT1 CT3 I'1 I2 CT2 I3

I'2

Trip Winding

I'1+I'2-I'3

I'3

Restraining Windings

Figure 18 : Differential Protection of a Voltage Bus. conditions the differential relay will not operate. However, if there was a fault on the bus, then I '1 + I '2 I 3 . There would be a differential current, and the relay would trip. An actual implementation of this scheme would require a differential relay for each phase. Operation of any of the phase relays would activate the trip coils of all three CBs, isolating the voltage bus. This technique can be extended to situations where there are many lines entering a voltage bus.

7.3 Transformers.
Figure 19 illustrates the application of differential protection to a transformer. The complication here is that the turns ratio of the transformer must be accounted for before a differential signal can be applied to a relay. This can be done by using CTs with different turns ratios n1 and n2. This gives secondary side measured currents as: I '1 = I1 n1 I '2 = I2 n2 (7)

The current in the relay operating coil is then: 16

ELEC 4100
N1 I1 N2 I2

PROTECTION

CT1

CT2

I'1

Trip Winding

I'1-I'2

I'2

Restraining Windings

Figure 19 : Differential Protection of a Transformer. I ' = I '1 I '2 = I1 I 2 n1 n2 (8)

Then assuming an ideal transformer relationship: I2 = N1 I1 N2

I1 N1 N 2 1 n1 n2 n1

(9)

Combining (8) and (9):

I'=

(10)

When there is no internal fault this current should be zero, and this then requires that: n2 N1 = n1 N 2 (11)

17

ELEC 4100

TRANSMISSION LINES

ELEC 4100 ELECTRICAL ENERGY SYSTEMS TRANSMISSION LINES. 1 Introduction.

In electrical power systems transmission lines form the inter-connects between different generators and different parts of the network. Frequently these inter-connects cover large distances and so in terms of analysis it is often necessary to apply distributed models of the transmission lines. Already in this course the steady state characteristics of transmission lines have been examined with respect to load flow through the network. An equally important consideration is how the transient characteristics of transmission lines affect the system, and this most commonly arises when considering surge phenomena. Over-voltage surges arise frequently on transmission lines due to lightning strikes and switching events (eg. Capacitor banks being switched into service and PWM inverters). The amplitude of over-voltage surges on transmission lines can be considerably greater than the normal 50Hz voltage of the line. For system voltages below 66kV the surges can be 5-10 times the rated AC voltage, whereas for systems of 110kV and above a limitation of the surge voltages to less than 2-3 times the rated AC voltage is necessary for practical and economic reasons. To adequately design over-voltage protection systems and appropriately rate the system insulation (including air gaps) on the line it is required to quantify the magnitude of the worst case surge events. This analysis required detailed transient models of transmission lines, where refection and refraction phenomena can be accounted for.

2 Basic Transmission Line Theory.

2.1 Transmission Line Equations.
In the discussion of load flow analysis, the basic transmission line equations were introduced to illustrate the means by which a -section steady state model could be derived. Now the development of these equations is re-visited so as to illustrate how they relate to transient phenomena on the transmission line. A single phase transmission line with uniformly distributed parameters can be represented by a simple equivalent circuit as illustrated in figure 1. The distributed parameters for a short section of line dx are represented by the per unit length resistance R, per unit length inductance L, per unit length shunt capacitance C, and per unit length shunt conductance G. The equations for this system are easily derived as:
v(x, t ) i(x, t ) = ri(x, t ) L x t i(x, t ) v(x, t ) = Gv(x, t ) C x t

(1a) (1b)

i(0,t)

i(x,t) Ldx Rdx

i(x+dx,t)

i(d,t)

v(0,t)

v(x,t) Cdx

Gdx

v(x+dx,t)

v(d,t)

Figure 1 : A Distributed Parameter Transmission Line. 1

ELEC 4100

TRANSMISSION LINES

Previously it was assumed that the time variation could be modelled by with sinusoidal time dependency. When this is not the case, it is still possible to simplify the above partial differential equations by taking the Laplace transform of the equations with respect to the time derivative, as: dV ( x, s ) = ( r + sL ) I ( x, s ) Li ( x, 0 ) dx dI ( x, s ) = ( G + sC ) V ( x, s ) Cv ( x, 0 ) dx (2a) (2b)

For an initially quiescent line the initial conditions along the line are zero, and so equation 2 reduces to the form: dV ( x, s ) = ( r + sL ) I ( x, s ) dx dI ( x, s ) = ( G + sC ) V ( x, s ) dx (3a) (3b)

Using the same technique as illustrated for the sinusoidal steady state case, we can differentiate both expressions in equation 3 with respect to x, to get:

d 2V ( x, s ) dI ( x, s ) = r + sL ( ) dx 2 dx

(4a)

d 2 I ( x, s ) dV ( x, s ) = G + sC ( ) dx 2 dx Combining equations 3 and 4 gives: d 2V ( x, s ) = ( r + sL )( G + sC ) V ( x, s ) = 2V ( x, s ) 2 dx d 2 I ( x, s ) = ( R + sL )( G + sC ) I ( x, s ) = 2 I ( x, s ) 2 dx Where:

(4b)

(5a) (5b)

( R + sL )( G + sC )

(6)

Equation 5 represents a pair of linear second order homogeneous differential equations and have the following general solution:
V ( x, s ) = k1e x + k2 e x

(7a) (7b)

I ( x, s ) = Where:
Z0 =

1 k1e x k2 e x Z0

( R + sL ) ( G + sC )

(8)

Equation 7 represents the general expressions for the behaviour of the transmission line voltage and currents in the Laplace domain, and time domain expressions can be formed by determining the inverse transform of equation 7. However as seen in equations 6 and 8, in the general case the
2

ELEC 4100

TRANSMISSION LINES

parameters and Z0 are considerably complex functions of s, and this introduces a significant challenge when attempting to find the inverse Laplace transform.

2.2 Travelling Wave on a Lossless Line.

A lossless line is one in which the resistance per unit length and the shunt conductance per unit length are zero. Hence:

= s LC
Z0 = L C
+ k2 e s

(9) (10)

Equation 7(a) then becomes:

V ( x, s ) = k1e s
LC x LC x

(11)

Now recall that an exponential function in the Laplace domain becomes a time shift in the time domain. Then defining vf (t) = L -1{k1} and vb (t) = L -1{k2} gives:
x x v ( x, t ) = v f t v t + + b v v 0 0

(12)

Where: v0 = 1 LC (13)

Equation 12 indicates that the voltage on the line consists of two waves, a forward travelling wave vf (x,t), and a backwards travelling wave vb (x,t). Both waves travel at the same velocity v0, and there is no attenuation of the wave shape from the initial condition. The initial wave shape is determined by the boundary conditions applied to the transmission line. This is illustrated in figure 2, in which the forward propagating wave shape is seen to travel forwards the distance x0 in a time x0/v0, and similarly the backward propagating wave shape is seen to travel back the distance x0 in a time x0/v0. As seen in equation 13, the propagation velocity along the line is a function of the capacitance and inductance per unit length, and for overhead lines this velocity is approximately: v0 = 1 1 1 = = = 3 108 ms 1 LC 0 0 (14)

This is the speed of light in free space. For cables the relative permittivity is in the range 2 3, and so the propagation velocity is in the range of 50% to 70% of the speed of light in free space. The same analysis applies to the current waveform. Equation 7(b) becomes: I ( x, s ) = 1 s k1e Z0
LC x

k2 e s

LC x

(15)

The inverse transform of this expression becomes:

x x i ( x, t ) = i f t + v + ib t v 0 0

(16)

Where:

ELEC 4100
Forward Travelling Wave vf (t) vf (t-x0/v0)

TRANSMISSION LINES

t x0/v0 Transmission Line x=0 vb (t) x = x0 vb (t+x0/v0)

t x0/v0 Backward Travelling Wave

Figure 2 : Forward and Backward Propagation of Travelling Waves on a Lossless Line.

Forward Travelling Wave vf (t-x/v) if (t-x/v) t t Backward Travelling Wave vb (t-x/v)

ib (t-x/v)

Figure 3 : Voltage and Current Surges on a transmission line.

x 1 x = if t v t f v Z v 0 0 0 x 1 x + = + ib t v t b v Z v 0 0 0

(17)

Note that for the forward propagating wave the current and the voltage have the same sign, but for the backward propagating wave the current and the voltage have the opposite sign. The following summary can be made: (i) (ii) A surge voltage on a transmission line is equal to the sum of a forward and backward travelling wave, either of which can be zero. The same applies for surge current. The forward travelling wave voltage is directly proportional to the forward current by the factor Z0. The backward travelling wave is directly proportional to the backward current by the factor -Z0. Z0 is the characteristic impedance of the line. The sign difference between the current and voltage waves is illustrated in figure 3. 4

ELEC 4100

TRANSMISSION LINES

2.3 Reflection and Refraction of Travelling Waves.

When a surge voltage travelling along a line arrives at a discontinuity where the characteristic impedance changes, the energy associated with the travelling wave will be distributed to two travelling wave pairs. This includes a reflected wave and its associated current surge, and this travelling wave pair is superimposed over the incident wave. The second travelling wave pair is the transmitted voltage and current surge across the junction, and this is called the refraction pair. This process is illustrated in figure 4 below.
v1f Refracted Wave v2f

Z1 v1b Reflected Wave i1f i1b Reflected Wave Z1

Z2

Refracted Wave i2f

Z2

Figure 4 : Refraction and Reflection on a transmission line. In figure 4, the junction is represented by the point A, and denotes the change in characteristic impedance from Z1 to Z2. A step voltage v1f approaches the junction along the transmission line with an associated current i1 f = v1 f Z1 . After the travelling waves reach the junction A, a pair of voltage and current waves, related by i1b = v1b Z1 , are reflected backwards along the line towards the source. Another pair of travelling waves related by i2 f = v2 f Z 2 penetrate into the cable and constitute the refracted wave pair. Since the voltage and current must be continuous at the junction A, it follows that:

v1 f + v1b = v2 f i1 f + i1b = i2 f
Equation 18(b) can be re-written to show: 1 1 v1 f v1b ) = v2 f ( Z1 Z2 Equations 18(a) and 19 can be combined to show that:

(18a) (18b)

(19)

v1b = v2 f =

Z 2 Z1 v1 f Z 2 + Z1
2Z 2 v1 f Z 2 + Z1

(20a) (20b)

This leads to the definitions of the following quantities:

ELEC 4100

TRANSMISSION LINES

Z 2 Z1 Z 2 + Z1
2Z 2 Z 2 + Z1

(20a) (20b)

The parameter is called the reflection coefficient, and is called the refraction coefficient. Note that the reflection coefficient can be either positive or negative. The refraction coefficient can range from 0 to 2. For a line terminated with an impedance Z to ground, there will also be terminal travelling wave reflections. This is illustrated in figure 5.
v1f Incident Wave Z1 v1b Reflected Wave Z

Figure 5 : Reflection at a transmission line termination.

Since current and voltage continuity must be preserved at the transmission line termination, it can be shown that the voltage across the impedance, and the current through the impedance must satisfy:

v1 f + v1b = vZ i1 f + i1b = iZ
Equation 21(b) can be re-written to show: 1 1 v1 f v1b ) = vz ( Z1 Z Hence:

(21a) (22b)

(23)

v1b = vz =

Z Z1 v1 f = v1 f Z + Z1
2Z v1 f = v1 f Z + Z1

(24a) (24b)

These expressions match those obtained for an interface between transmission line sections with different characteristic impedances. So the following conclusions can be drawn. If the transmission line is terminated by an open circuit, then = 1 , and so the wave is completely reflected. If the line is terminated by a short circuit, then = 1 , and the wave is again completely reflected, but inverted. If the line is terminated with an impedance Z = Z1 , then = 0 and there is no reflection. In this case the line is said to be matched.

ELEC 4100

TRANSMISSION LINES

2.4 Sinusoidal Voltage and Current on the Line.

For a line with per unit length parameters R, L, G and C, the generic solution to the transmission line equations in the Laplace domain are given by:
V ( x, s ) = k1e x + k2 e x

(25) (25)

I ( x, s ) =
Where:

1 k1e x k2 e x Z0

=
Z0 =

( R + sL )( G + sC )

(26) (27)

( R + sL ) ( G + sC )

Now let: A = k1 + k2 and B = k1 k2 then:

V ( x, s ) = A cosh ( x ) + B sinh ( x )

(28) (29)

I ( x, s ) =

1 A sinh ( x ) + B cosh ( x ) Z0
x Sending End Z0 v(x,t) i(x,t) x =0 Receiving End

Now consider the transmission line shown below, with a terminating impedance Z at x = 0.

Figure 6 : Transmission Line Terminated by an impedance Z. The receiving end corresponds to x = 0, and x increases to the left.
The receiving end imposes the following boundary conditions:
V ( 0, s ) = Z I ( 0, s )

(30)

Substituting this constraint into equations 28 and 29 gives:

A = V ( 0, s ) = ZI ( 0, s ) =
Therefore:

Z B Z0

(31)

Z V ( x, s ) = B cosh ( x ) + sinh ( x ) Z0

(32a)

ELEC 4100
I ( x, s ) = B Z0 Z sinh ( x ) + cosh ( x ) Z0

TRANSMISSION LINES

(32b)

So for a transmission line of length d, equation 32 can be used to define the entry impedance into the line according to:

Z (d, s) =

Z cosh ( d ) + Z 0 sinh ( d ) V (d, s) = Z0 I (d, s) Z sinh ( d ) + Z 0 cosh ( d )

(33)

When the line termination is a short circuit:

Z sc ( d , s ) = Z 0 tanh ( d )

(34)

When the line termination is an open circuit:

Z oc ( d , s ) = Z 0 coth ( d )

(35)

When the line termination is matched (i.e. Z = Z0) then:

Zm ( d , s ) = Z0

(36)

It is also possible to determine the relationship between the sending and receiving end voltages by applying equation 32(a) according to:
V ( 0, s ) Z = V ( d , s ) Z cosh ( d ) + Z 0 sinh ( d ) So again when the line termination is a short circuit (Z = 0):
V ( 0, s ) =0 V ( d , s ) SC

(37)

(38)

When the line termination is an open circuit Z = :

V ( 0, s ) 1 = V ( d , s ) OC cosh ( d )

(37)

2.4.1 Special Case Lossless Line These expressions can now be used to determine the characteristics of the line for sinusoidal conditions. If we replace the Laplace variable s by j for the lossless line the above expressions reduce to:

= s LC

Z0 = L

(39)

So equation 32 becomes:

C V ( x, j ) = B Z cosh j LC x + sinh j LC x L

(40a)

I ( x, j ) =

B Z0

C sinh j LC x + cosh j LC x Z L

(40b)

Recall that hyperbolic functions of imaginary quantities reduce to trigonometric functions of the form:

ELEC 4100 C V ( x, j ) = B Z cos LC x + j sin LC x L

TRANSMISSION LINES

(41a)

I ( x, j ) =

B C sin LC x + cos LC x jZ Z0 L

(41b)

So the entry impedance becomes:

L Z cos LCd + j sin LCd L C Z ( d , j ) = C L cos LCd jZ sin LCd + C

(41)

When the line termination is a short circuit:

Z sc ( d , j ) = j L tan LCd C

)
)

(42)

When the line termination is an open circuit:

Z oc ( d , j ) = j L cot LCd C

(43)

When the line termination is matched (i.e. Z = Z0) then:

Z m ( d , j ) = L C

(44)

The relationship between the voltages at the sending and the receiving ends of the transmission lines are given by: V ( 0, j ) = V ( d , j ) Z L Z cos LCd + j sin LCd C

(45)

So again when the line termination is a short circuit (Z = 0):

V ( 0, s ) =0 V ( d , s ) SC

(46)

When the line termination is an open circuit Z = :

V ( 0, j ) 1 = = sec LCd V ( d , j ) OC cos LCd

(47)

Recall that sec ( x ) , as x 900 . Therefore if LCd = 2 simple sinusoidal excitation

of the transmission line can lead to infinitely large excitation voltages at the receiving end of the line. This of course assumes ideal behaviour, but it is worth examining this constraint further. The worst case over-voltage on the line occurs when:

LCd =

or d =

1 4f

v 1 = 0 LC 4 f
9

(48)

ELEC 4100

TRANSMISSION LINES

For an overhead conductor in free space, recall that the propagation speed for the wave is the speed of light in free space, and so for a 50Hz system: d= 3 108 ms 1 = 1.5 106 m = 1500km 4 ( 50 Hz ) (49)

This length corresponds to a wavelength distance for the sine wave on the transmission line. Most transmission lines are well below this figure, but for an open circuited lossless transmission line as long as 1000km (corresponding to 1/6 wavelength distance, and an argument in the sec function of 600), the transmission line will have a double-magnitude over-voltage under normal sinusoidal excitation.

3 Calculations of Transmission Line Transients.

There are various methods for travelling wave calculations. For simple circuits it may be possible to perform the calculations by hand using either the Bewley lattice diagram method, or the Bergeron graphical approach. Closed form solutions may also be found using the Laplace transform, but for most practical systems numerical solution techniques are generally required.

3.1 Bewleys Lattice Diagram method.

If a voltage is applied to a transmission line, reflections will occur at the ends (unless the line is perfectly matched) which will travel up and down the line until finally a steady state equilibrium is reached. The voltage and current at any position along the line will be a superposition of all the travelling waves presented at a specific time under consideration. The Bewley lattice diagram conveniently organises the reflections that occur during line transients.
3.1.1 Example: A step voltage E = 400V is applied to a cable through a resistance R = 40 ohm. The cable has a transit time of , and is open circuited at the receiving end. Figure 7 shows the circuit and its Lattice diagram for the determination of travelling waves on the cable.
=-0.5 40 v(0,t) 400V v(0,t) 300V 300V -150V 450V -150V 375V 75V 412.5V -37.5V 75V Transmission Line Z0=120 300V 2 3 4 5 6 7 450V 375V 300V 600V =1 v(d,t) v(d,t)

Figure 7 : The Bewley Lattice diagram for travelling wave calculations. 10

ELEC 4100

TRANSMISSION LINES

The vertical scale represents time and is scaled in units of the transmission line transit time . The diagonal lines represent the travelling wave fronts, and their magnitudes. Each waveform magnitude is calculated using equation 24(a), and by calculating the reflection coefficients at each end of the line. Note that the voltage source is represented as a short circuit for the determination of the reflection coefficient at the source end, and the travelling wave front entering the line is determined by voltage divider rule with the source impedance and the transmission line characteristic impedance only. The voltage at each end of the line, at any point in time is determined by summing the incident and reflected voltage wavefronts up until that point in time, leading to the transient waveforms shown for v(0,t) and v(d,t).
3.1.2 Example : Consider the circuit shown in figure 8. This system consists of two transmission lines with differing characteristic impedances Z1 and Z2. The combination of the two lines is terminated with an impedance ZL, and a surge voltage of 2kV is applied to the transmission line through a source impedance ZS. Both transmission lines have propagation times from one end of the line to the other of = 0.1 ms . The Bewley lattice method can be used to determine the transient voltage at the sending end of the line, the junction between the two lines, and the termination of the line.

ZS 400 2kV v(0,t)

= 0.1ms Z1=400

= 0.1ms Z2=100
ZL=200

v(d1,t)

v(d1+d2,t)

Figure 8 : Surge applied to two transmission lines through a source impedance. To solve this type of problem it is first necessary to calculate the relevant reflection and refraction coefficients. The reflection coefficient at the sending end of the transmission line is: S = Z s Z1 400 400 = =0 Z s + Z1 400 + 400 Z 2 Z1 100 400 3 = = Z 2 + Z1 100 + 400 5 2Z 2 2(100) 2 = = Z1 + Z 2 100 + 400 5 Z1 Z 2 400 100 3 = = Z 2 + Z1 100 + 400 5 2 Z1 2(400) 8 = = Z1 + Z 2 100 + 400 5 (50)

The reflection and refraction coefficients on the sending side of the junction are: 12 = (51) (52)

12 =

The reflection and refraction coefficients on the receiving side of the junction are: 21 = (53) (54)

12 =

The reflection coefficient at the receiving end of the line is given by:

11

ELEC 4100 R = Z L Z 2 200 100 1 = = Z L + Z1 100 + 200 3

TRANSMISSION LINES

(55)

With these parameters it is now possible to construct the Bewley lattice diagram as shown in figure 9.
x=0 1000V -600V 133.3V 213.3V 26.67V 42.67V 5.3V 16V 80V 400V x = d1 x = d1+d2

2 3 4 5 6 7

14.2V

3.2V

Figure 9 : Lattice diagram for travelling wave reflections on the circuit shown in figure 8. Having determined the travelling wave reflections and refractions, it is now possible to calculate the voltages at the sending end of the line, the receiving end of the line, and the junction between the two different sections, as illustrated in figure 10.
v(d1+d2,t) 661.3V 533.3V 640V

t v(d1,t) 2 3 4 5 656V 6 7

613.3V 400V

t v(0,t) 1000V 400V 613.3V 656V 2 3 4 5 6 7

t 2 3 4 5 6 7

Figure 10 : Voltage waveforms at the sending, junction and receiving sections of the line. 12

ELEC 4100

TRANSMISSION LINES

3.2 Bergeron Graphical Method.

The Bergeron graphical method is similar to the Bewley lattice diagram method, except that rather than being based on reflection and refraction events, it is based on a linear combination of the voltage and current characteristics on a transmission line. Recall that for a lossless line the voltage and current at any position of the line consists of a forward and/or backward propagating travelling wave pair of the form:
x x v ( x, t ) = v f t v + vb t + v 0 0 x x Z 0i ( x, t ) = v f t vb t + v0 v0

(56a)

(56b)

These expressions can be re-written by adding and subtracting 56(b) from 56(a) to yield:
x v ( x , t ) + Z 0 i ( x , t ) = 2v f t v0 x v ( x, t ) Z 0i ( x, t ) = 2vb t + v0

(57a)

(57b)

For the lossless line there is no attenuation of the travelling waves along the line, so equation 57 defines two linear expressions in the voltage-current plane, with slopes of Z0. At any point along the line x, and any time t, current and voltage continuity requires that the forward and backward propagating characteristics intersect at the same voltage and current point. Therefore the operating voltage and current along the line can be found by superimposing the transmission line v-i forward and backward characteristics over the top of the sending and receiving end impedance characteristics in the voltage-current plane. To illustrate the concept, consider the example transmission line in figure 11. Here a surge voltage E is applied to a transmission line with characteristic impedance Z0 through a supply resistance R. The end of the transmission line is terminated by a surge arrester which possesses a non-linear v-i characteristic v = k0i , where k0 and are constants.

Z0

E v(0,t) v(d,t) ZL

Figure 11: Step voltage applied to a transmission line with a non-linear impedance. The volt-amp characteristic equations for the sending end impedance, and the receiving end impedance respectively are given by:
v ( 0, t ) = E Ri ( 0, t )

(58) (59)

v ( d , t ) = k0 i ( d , t )

13

ELEC 4100

TRANSMISSION LINES

Figure 12 plots these characteristics. Initially the voltage along the line is zero, so since the step is applied at the sending end of the line, the backward propagating wave is initially zero, and the transient initiates a forward propagating wave. The forward characteristic passes through the origin, and the initial forward propagating wave is determined by the intersection with the source end v-i characteristic. This intersection then determines the backward propagation v-i characteristic, and its intersection with the non-linear impedance curve determines the initial receiving end voltage after one transmission line transit duration. This process is repeated allowing the source end and receiving end transient voltage profiles to be developed graphically.
V E v(0,0) v(0,2) v(0,4) V = k0I v(d,5) v(d,3) v(d,) V = E-RI 2 3 4 5 6 t v(d,t) v(0,t) V

Figure 12 : Bergeron Graphical method applied to the transmission line of figure 11.

3.3 Bergeron-Dommel Method.

The basis of the Bergeron-Dommel method is to develop discrete-time models of transmission lines and lumped elements, that allow classical nodal analysis to be performed at each time step. This will be illustrated below for the lossless line. Recall that the characteristic for a forward propagating wave on a transmission line is given by:
x v ( x , t ) + Z 0 i ( x , t ) = 2v f t v0

(60)

Clearly when the argument ( t x v0 ) is constant, then equation (60) is constant. So for a transmission line with a propagation time from terminal k to terminal m (as illustrated in figure 13), the following result must hold:
vk ( t ) + Z 0ikm ( t ) = vm ( t ) Z 0imk ( t )

(61)

The negative sign for the current imk(t) denotes the direction change from (k m) to (m k).
ikm(0,t) Z0 imk(0,t)

vk(0,t) vm(d,t)

Figure 13 : Transmission line with nodes k and m at each end. 14

ELEC 4100 Rearranging equation 61 gives: imk ( t ) = Where: Im (t ) = 1 2 vk ( t ) ikm ( t ) = vk ( t ) I k ( t 2 ) Z0 Z0 1 vm ( t ) + I m ( t ) Z0

TRANSMISSION LINES

(62)

(63)

Equations 62 and 63 hold for forward propagating waves only. For the backward propagating waves we can similarly write:
x v ( x, t ) Z 0i ( x, t ) = 2vb t + v0

(64)

Again when the argument ( t + x v0 ) is constant, then equation (60) is constant. So for a wave propagating from node m to node k the following result must hold: vm ( t ) Z 0 imk ( t ) = vk ( t ) Z 0ikm ( t ) Or : ikm ( t ) = Where: Ik (t ) = 1 2 vm ( t ) imk ( t ) = vm ( t ) I m ( t 2 ) Z0 Z0 (67) 1 vk ( t ) + I k ( t ) Z0 (66) (65)

We can represent equations 62 and 63, and 66 and 67, by the equivalent circuit shown in figure 14. The currents sources Im and Ik are given by equations 63 and 67, and represent the past history of the transmission line. Note that in this model the line is simplified to a circuit consisting of lumped impedances only, and there is no direct connection from node k to m, by rather a linked by the current sources which introduce a delay of .

ikm Ik(t-) Im(t-)

imk

vk

Z0

Z0

vm

Figure 14 : Discrete time-domain model of a transmission line.

3.3.1 Lumped Inductance. Recall that the equation for an inductance with nodes k and m is given by:

vL ( t ) = vk ( t ) vm ( t ) = L

dikm ( t ) dt

(68)

Integrating this equation from t to t + t , and using the trapezoidal rule gives: 15

ELEC 4100
ikm ( t ) = ikm ( t t ) + Or: ikm ( t ) = Where: I L ( t t ) = ikm ( t t ) + RL = 2L t 1 vk ( t t ) vm ( t t ) RL 1 vk ( t ) + vm ( t ) + I L ( t t ) RL t vL ( t t ) + vL ( t ) 2L

TRANSMISSION LINES

(69)

(70)

(71) (72)

This shows that in a discrete time sense, the inductor can be modelled by a resistor with a shunt current source, which represents the history of the inductor current. This is illustrated in figure 15(a).
3.3.2 Lumped Capacitance. Recall that the equation for a capacitance with nodes k and m is given by:
ikm ( t ) = C d vk ( t ) vm ( t ) dvC ( t ) =C dt dt

(73)

Following the same procedure as for the inductor, it can be shown that : ikm ( t ) = With: I C ( t t ) = ikm ( t t ) RC = t 2C 1 vk ( t t ) vm ( t t ) RC (75) (76) 1 vk ( t ) vm ( t ) + I C ( t t ) RC (74)

This gives the capacitor model as shown in figure 15(b).

IL(t-t) ikm vk RL
(a)

IC(t-t) ikm

vm

vk RC
(b)

vm

Figure 15 : Discrete time domain model of an inductor and a capacitor.

16

ELEC 4100

TRANSMISSION LINES

3.3.3 Nodal Analysis. A network that consists of both lumped and distributed elements can now be modelled by applying nodal analysis to the network with equivalent discrete time models, and this will now be illustrated. Figure 16(a) shows a transmission line which is energised with a voltage surge E, through a source resistance R and inductance L. The equivalent Bergeron Dommel representation of the line is shown in figure 16(b).

Z0

(a)

1 RL

IL(t-t) 2 I2(t-) Z0

3 Z0

I3(t-)
(b)

Figure 16 : Transmission line and equivalent Bergeron Dommel representation. The nodal equations for the Bergeron Dommel representation of the line can be readily derived by inspection and is given by: 1 1 E R + R L R I L ( t t ) 1 I L ( t t ) I 3 ( t ) = RL I2 (t ) 0 1 RL 0 V1 0 V2 V 3 1 Z0

1 1 + Z 0 RL 0

(77)

The admittance matrix can now be inverted enabling a solution to be obtained for the node voltages at each time step. An appropriate algorithm for the simulation is as follows: At t = 0, initialise all currents to zero. The source current is E/R. At t = 0, compute the nodal voltages by solving the nodal expression in 77. Increment the time step to t + t . Compute the currents I L ( t t ) , I 2 ( t ) and
I3 ( t ) .

Compute the nodal voltages by solving the nodal expression in 77, and go back to the step above. Repeat until the required simulation time has been reached.

17

ELEC 4100

TRANSMISSION LINES

4 Distortion and Attenuation.

Thus far the analysis has focused principally on the lossless transmission line. Another important case is the distortionless line. Recall the general solution for the transmission line equations is of the form:
V ( x, s ) = k1e x + k2 e x

(78)

Where:

( R + sL )( G + sC )
2

(79)

This can also be written as:

= LC ( s + ) 2
Where:

(80)

= + 2L C = 2L C
1 R G

1 R

(81) (82)

The distortionless line is the special case in which R L = G C . Under this constraint = R L and = 0 , so:

= ( s + R L ) LC
So taking the inverse transform of equation (78) gives:
x v ( x , t ) = e x v f t x v + e vb t + x v 0 0

(83)

(84)

Similarly it can be shown that:

x i ( x , t ) = e x i f t x v + e ib t + x v 0 0

(85)

These expressions show that for the distortionless line, the wave-shape for the forward and backward propagating waves is not altered, other than pure attenuation (hence the name). For realistic transmission lines the distributed parameters, especially R and L, are functions of frequency due to the finite resistivity of earth, the variation of flux penetration into the line conductors, and many other second order effects (e.g. skin effect, and corona). This variation in the distributed parameters of the transmission line means that the criteria for distortionless propagation is rarely achieved, and so voltage surges on a line are attenuated and distorted as they propagate along the line. Usually this distortion and attenuation results in slowing of the rise time, in addition to a reduction in the peak of the surge.

5 Three Phase Transmission Lines.

For a three phase transmission line, waves travel in three modes, each with its own wave velocity and surge impedance. For completely transposed lines under lossless conditions the transmission line equations become:

18

ELEC 4100 dV ( x, s ) = sLI ( x, s ) dx dI ( x, s ) = sCV ( x, s ) dx

I a ( x, s ) I ( x, s ) = I b ( x , s ) I c ( x, s )

TRANSMISSION LINES

(86a) (86b)

Where:
Va ( x, s ) V ( x, s ) = Vb ( x, s ) Vc ( x, s ) and (87)

Which represent the line to ground voltages and the line currents respectively. For a completely transposed line the inductance and capacitance matrices are:
LS L= LM LM CS C= CM CM LM LS LM CM CS CM LM LM LS CM CM CS

(88)

(89)

Where LS and LM are the self and mutual inductances of the line respectively, and CS and CM are the self and mutual capacitances of the line respectively. Since the line is completely transposed, the matrices in (88) and (89) are completely symmetrical, and so can be transformed into a diagonal form using:
1 1 1 1 1 1 1 1 T = 1 2 1 and T = 1 1 0 3 1 1 2 1 0 1

(90)

Hence:
LS + 2 LM T 1 LT = 0 0 CS + 2CM T CT = 0 0
1

0 LS LM 0 0 CS CM 0

0 0 LS LM 0 0 CS CM

(91)

(92)

Then: T 1 T 1 dV ( x, s ) = sT 1 LT T 1 I ( x, s ) dx dI ( x, s ) = sT 1CT T 1V ( x, s ) dx (93a) (93b)

Or alternatively:

19

ELEC 4100 dV M ( x, s ) dx dI M ( x, s ) dx = sL M I M ( x, s ) = sC M V M ( x, s )

TRANSMISSION LINES

(94a) (94b)

With L M and C M defined in equations 91 and 92. V0 ( x, s ) V M ( x, s ) = T V ( x, s ) = V1 ( x, s ) V2 ( x, s )

1

(95)

I 0 ( x, s ) I M ( x, s ) = T I ( x, s ) = I1 ( x, s ) I 2 ( x, s )
1

(96)

These are the mode voltages and currents, and consist of zero, positive and negative sequence terms. This transformation into sequence components makes it possible to write three separate and decoupled differential equations for the three sequence terms. It should be noted that the positive and negative sequence inductances and capacitances are identical, but that they are not the same as the zero sequence terms. This means that the zero sequence propagation velocity is different to the positive and negative sequence propagation speed, and the characteristic impedances are also different, with:
v0 = 1

( LS + 2 LM )( CS + 2CM )

(97a)

Z0 =

( LS + 2 LM ) ( CS + 2CM )
1

(97a)

And:
v1 = v2 =

( LS LM )( CS CM )
( LS LM ) ( CS CM )

(98a)

Z1 = Z 2 =

(98a)

Finally note that the transformation used in equation (90) is not the only transformation that can be used for this decoupling purpose. Other suitable transformations include the Clark and the sequence transformation which was defined in the symmetrical components and fault calculation section of the course.

20

ELEC 4100

SUPPLY QUALITY

ELEC 4100 ELECTRICAL ENERGY SYSTEMS QUALITY OF SUPPLY. 1 Introduction.

The quality of the electrical supply deals with the departure of the voltage, frequency and current from the ideal supply conditions. An ideal three phase supply would have a fixed magnitude, constant frequency, sinusoidal supply voltages, with 1200 angular displacement, and would operate with 100% reliability. It is not possible to achieve this ideal, primarily because the nature of the loads connected to the supply can not be controlled by the utility, and also external disturbances such as lightning strikes can not be predicted nor controlled. Small deviations from ideal conditions may not have an impact for the utility or the customer, however power quality problems do arise when the deviations become large. Dugan et al. 1 defines a power quality problem as: any power problem manifested in voltage, current, or frequency deviations that results in failure or misoperation of customer equipment. Such power problems might include waveform distortion, transient events, flicker, unbalance, frequency and voltage variations. These problems generally result in three different types of issue for the customer, viz.: Damage or destruction to items of electrical plant. This can be caused by thermal stress (due to increased losses in the plant from harmonics etc.), insulation stress (due to overvoltage), and other sources such as excessive vibration (especially in motors). Nuisance misbehaviour of equipment. These effects can include light flicker due to low frequency disturbances, tripping and resetting of microprocessor based systems, and increased electrical noise in appliances. Economic loss through lost production time or replacement of equipment, and also increased maintenance scheduling for equipment subjected to the power quality problem.

For these reasons there is now a drive to better understand the nature of power quality problems, and to develop mitigation methods for them. However it should be noted that the level of supply disturbance that is accepted is primarily decided upon according to economic criteria.

2 Nominal Supply
Utilities have statutory obligations to provide customers with a supply that is constrained to lie within a specified operating band. The supply voltage and frequency are ideally within the centre of these operating regions, with the long term average values defining the nominal supply quantities. For example in NSW the long term average single phase domestic supply is 240V at 50Hz. The statutory bodies define the limits to which the supply is allowed to deviate from the nominal values, and in Australia these regulations specify: Supply voltage : Limits = 6%. Grid frequency : Limits = 0.1% (some regulators allow 1.5% in emergency cases).

A further regulatory constraint is that the supply voltages must be balanced, such that the negative sequence is limited to be less than 2% of the positive sequence component. Typically this is controlled by distributing single phase loads equally across the three phases, but the 2% limit can be particularly challenging to meet.

R. Dugan, M. McGranaghan, W. Beatty, Electrical Power Systems Quality, McGraw Hill, First edition, 1996.

ELEC 4100

SUPPLY QUALITY

While waveform distortion must be minimised to eliminate power quality problems, currently the level of disturbance that is allowed is not regulated. A number of international bodies (i.e. IEEE, CIGRE, IEC, SEMI, ANSI) have prepared standards dealing with supply waveform distortion, however the recommendations of these standards are not law. The decision to invest in infrastructure to mitigate the power quality problem would then be made based on an economic rationale, and the expenditure may be made by either the customer, the utility, or shared by both.

3 Classification and Nature of Disturbances.

There are seven generally accepted categories of power quality disturbance, viz.: Type 1 : Transient disturbances: o Impulsive transients. o Oscillatory transients. Type 2 : Long-Duration Voltage Variations: o Over-voltages. o Under-voltages. o Sustained supply interruptions. Type 3 : Short-Duration Voltage Variations: o Short lived supply interruptions. o Voltage sag events (i.e. dips). o Voltage swell events. Type 4 : Voltage Unbalance i.e. the presence of a negative and/or zero sequence. Type 5 : Waveform Disturbance : o DC offset. o Harmonic distortion. o Inter-harmonics. o Waveform Notching. o Noise. Type 6 : Voltage Fluctuations : o Voltage flicker (i.e. low frequency sub-harmonic oscillation) Type 7 : Power frequency variations. Figure 1 illustrates examples of these power quality problems. Many of these disturbances are caused by short-lived events such as lightning strikes, falling trees and other such accidents, sudden large changes in load (eg. DOL Motors), system faults, capacitor bank switching. Other disturbances are caused by non-linear loads, such as rectifiers (including phase controlled rectifiers and cyclo-converters), variable speed drives, static VAR compensators, and fluctuating loads (eg. arc furnaces, welders, etc.).

ELEC 4100

SUPPLY QUALITY

1.0 p.u.

1.0 p.u.

-1.0 p.u. 0 10 ms 20 ms 30 ms 40 ms 50 ms 60 ms 70 ms 80 ms

-1.0 p.u. 0 10 ms 20 ms 30 ms 40 ms 50 ms 60 ms 70 ms 80 ms

(a) Ideal Supply

1.0 p.u.

(b) Type 1 Disturbance : Impulse Transient.

1.0 p.u.

-1.0 p.u. 0 10 ms 20 ms 30 ms 40 ms 50 ms 60 ms 70 ms 80 ms

-1.0 p.u. 0 10 ms 20 ms 30 ms 40 ms 50 ms 60 ms 70 ms 80 ms

(c) Type 2 Disturbance : Under Voltage.

1.0 p.u. 1.0 p.u.

(d) Type 3 Disturbance : Voltage Sag.

-1.0 p.u. 0 5 ms 10 ms 15 ms 20 ms 25 ms 30 ms 35 ms 40 ms

-1.0 p.u. 0 5 ms 10 ms 15 ms 20 ms 25 ms 30 ms 35 ms 40 ms

(e) Type 4 Disturbance : Voltage Unbalance

1.0 p.u.

(f) Type 5 Disturbance : Harmonic Distortion

1.0 p.u.

-1.0 p.u. 0 0.02s 0.04s 0.06s 0.08s 0.1s 0.12s 0.14s 0.16s 0.18s 0.2 s

-1.0 p.u. 0 10 ms 20 ms 30 ms 40 ms 50 ms 60 ms 70 ms 80 ms

(g) Type 6 Disturbance : Voltage Flicker

(h) Type 7 Disturbance : Frequency Variation.

Figure 1 : Summary of common power quality disturbances.

ELEC 4100

SUPPLY QUALITY

4 Mitigation of Power Quality Problems.

The mitigation of power quality problems is a complex task, and each type of disturbance outlined above is suited to specific solution methods, viz.: Type 1 disturbances : Impulse notches can not easily be compensated for, but since transient over-voltages potentially destructive they must be mitigated. Since these events are short lived, fast acting protective devices are required, such as varistors and transorbs. are the As these are short lived events, they can only be mitigated by fast acting passive elements, such as