CONCEPTUAL

CLARIFICATIONS IN
ELECTRICAL POWER
ENGINEERING
Part-1
Dr K Rajamani,
Reliance Infrastructure Ltd., MUMBAI
(August 2016, IEEMA Journal, Page 69 to 80)
 Conceptual Clarifications in Electrical Power Engineering – Part 1
K Rajamani, Reliance infrastructure Ltd
1.0 Introduction

     "Life is really simple, but we insist on making it complicated." - Confucius

Field engineers encounter many problems during testing, commissioning,

operation and maintenance. Most of the problems are solved either by trial and
error or seeking advice from ‘experienced’ persons who might have encountered
similar problems in the past. But a sounder approach will be to understand the
basic concepts of ‘whys of things’ that will eliminate substantial uncertainty in
resolving issues. In short, to recognize the elephant in the room is the first step.
In this series of articles, basic conceptual confusions that confront power
engineers in field are stated with their resolution. Minimum theoretical concepts
required to explain field problems are included.

2.0 Concept of Reactive Power and Reactive power loss

The concept of reactive power is brilliantly explained in Ref [1,2]. Any electric
circuit is always a combination of resistance, inductance and capacitance. Refer
Figure 1 and Table 1.

Fig 1

v = Vm Cos ωt; i = Ιm Cos(ωt-θ); pf = cos θ

Instantaneous Power = Vm Ιm Cos ωt Cos(ωt-θ)
= (1/2) Vm Ιm Cos θ (1+Cos 2ωt) + (1/2) Vm Ιm Sin θ Sin 2ωt
Instantaneous Active Power Instantaneous Reactive Power
(1/2) Vm Ιm Cos θ (1+Cos 2ωt) (1/2) Vm Ιm Sin θ Sin 2ωt
Average Active Power Average Rective Power
P = V Ι Cos θ 0
Called simply Active Power Usually Ignored
Amplitude Instantaneous Active Power Amplitude Instantaneous Reactive Power
P = V Ι Cos θ Q = V Ι Sin θ
Usually ignored as it is simply P Called simply as Reactive Power
Table 1

1 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
The instantaneous power waveform is illustrated with a numerical example in
Fig 2.

Time (ms)
Fig 2

There are two components in instantaneous power. One is called the “Active
power”. Averaging of instantaneous active power is called Active Power =V Ι
cosθ. This is always positive (0,8 in Fig 2) and instantaneous real power
oscillates around this value. It does the useful work.

Another component is called the “Reactive power”. Reactive power = V Ι sinθ is

the maximum value of instantaneous reactive power (0.6 in Fig 2). Thus, though
V and Ι are RMS values, VΙsinθ is not average value but instantaneous value at
its maximum (minimum) peak. Average of instantaneous reactive power is zero
as it oscillates around X axis. It is the energy stored in the circuit inductance and
capacitance. Physically it implies what the system delivers in one quarter cycle,
inductance / capacitance delivers back to system in next quarter cycle.

2 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
If this is the case, what is meant by reactive power loss? Refer Figure 3.
Instantaneous reactive power waveforms at sending end and receiving end are
shown. The decrease in amplitude while delivering reactive power is termed as
reactive power loss.

Fig 3

The under lying concept is further illustrated with an example. Consider a single
phase network with only reactive elements as shown in Fig 4. Let VS = 230V;

X1 = 1.26Ω; X2 = 3.77Ω.

Fig 4

The results of simulation are given in Fig 5. There is a drop in amplitude of

receiving end reactive power (QR) compared to sending end reactive power

(QS). But the average values of QS and QR are still zero as they oscillate around
X axis.

Fig 5

3 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
From simulation, QS = 10.52KVAR and QR = 7.89KVAR. The reactive power
loss = 10.52-7.89 = 2.63KVAR.
2
Reactive power loss is also given by Ι X1.

Ι = 230 / (1.26+3.77) = 45.73A.rms

2
QLOSS = 45.73 x 1.26 /1000 = 2.63KVAR
This is same as obtained from simulation.

The interesting point to note is that Ι is RMS while Q is amplitude of

instantaneous quantity. This peculiarity is due to way Reactive Power is
defined.

The sending end voltage (VS) and receiving end voltage (VR) are shown in Fig
5. The reduction in voltage at receiving end is due to voltage drop in reactor.
Sonu Karekar’s help in PDCAD simulation is acknowledged.

3.0 Regulation and transformer impedance

Regulation refers to change in terminal voltage when a network element like

transformer carries load. At the outset, it should be emphasized that %Regulation
is not %Impedance. 10% impedance does not imply that 10% change in voltage
under full load conditions. This is true only when full load is drawn at ZPF. As an
illustration, consider 20MVA transformer with 13% impedance. Refer Fig 6.

Fig 6

4 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
Variation of sending end voltage for a specified (100%) receiving end voltage at rated
MVA for different pf is shown in Fig 7. If load is drawn at 0.9 pf or better, even at full
load, the regulation is less than 5%. Below 0.8 pf, regulation increases rapidly and
reaches 13% (impedance value) at ZPF. It is important to note that magnitude of
2
current drawn (Ι) is same and hence reactive loss (Ι X) in transformer is same in all
cases.

Fig 7
The vector diagram for two extreme cases, UPF and ZPF, are shown in Fig 8. If
current is at UPF, the voltage drop (ΙX) is in quadrature with VR and resultant VS

is close to VR (regulation – 0.84%). If current is at ZPF, voltage drop adds

algebraically with VR resulting in large VS (regulation – 13%).

Fig 8

5 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
Consider another example shown in Fig 9. Transformer parameters are as below:
400 / 11.5 / 11.5 kV; 90 / 45 / 45 MVA

Fig 9
Impedance on 90 MVA base:
HV- MV1: 21.76%
HV-MV2: 21.61%
MV1-MV2: 41.12%
Load at each MV bus = 45 MVA at 0.8 pf = 36 + j 27
The resulting voltages for above loading are given below:
VH = 100%; VMV1 = VMV2 = 96.2%

On no load, VH = 100%; VMV1 = VMV2 = 104.6%

Regulation = [(104.6-96.2) / 104.6] x 100 = 8%
Even though HV to MV impedances are about 22%, the regulation is only 8%
since the load power factor is not low (0.8).
In case of large EHV power transformers (600 MVA and above), impedances in
the range of 15% to 25% are common. These large values are chosen to limit the
fault level within available breaker capacity. But large impedance values per se
do not cause voltage regulation problems as long as load is at good power factor.
At EHV level, the load flow power factor is generally above 0.95.
The raison d’etre of reactive compensation in distribution and transmission
networks follows from the above discussions. Major network elements like
transformer, overhead line, cable, etc are almost reactive (X/R >>1). By reactive
compensation at different voltage levels, power factor of current flowing through
network elements is made near to unity which leads to low regulation and near
normal voltage profile. For deeper insight into reactive compensation details refer
[3].

6 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
4.0 Effectively grounded system
Power supply to Mumbai is derived through multiple voltage transformations.
The bulk power is stepped down at Transmission Stations. A typical
Transmission Station (T/S) has a number of 220/33 kV, Star – Zig Zag
transformers. The Star neutral is solidly grounded whilst Zig Zag neutral is
grounded through NGR (Neutral Grounding Reactor). Each transformer feeds
5 to 6 Receiving Stations (Refer Fig 10).

Fig 10
At the Receiving Station (R/S), step down transformer has the following
rating: 33/11 kV, 20MVA, Delta – Zig Zag. Secondary neutral is solidly
grounded. Each transformer feeds 5 to 6 Ring Mains. Each Ring Main serves
5 to 10 Sub-Stations. (Refer Fig 11). At each Sub-station, 11/0.44 kV
Distribution Transformers (DT) step down power and feed LT distribution
system.
At Transmission Stations, secondary of transformer is ‘effectively grounded’.
At Receiving Stations, secondary of transformer is ‘solidly grounded’. The
meaning of ‘solidly grounded’ is that there is no intentional intervening
impedance present between the transformer neutral and ground. It may be
noted that ‘solidly grounded’ system is a subset of ‘effectively grounded
system’. A ‘solidly grounded’ system is ‘effectively grounded’ but an
‘effectively grounded’ system need not be ‘solidly grounded’.
The two relationships generally used for characterizing effectively grounded
system are given below. Refer Cl 5.0 [4].
a) KF ≥ 0.6.

KF = (Single Phase to ground fault current) ⁄ (3 Phase Fault current)

= Ι1P / Ι3P

7 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
This is frequently used by field engineers as it is easy to understand and
implement.
For solidly grounded system, KF ≥ 1.0.

For ungrounded system, KF ≅ 0

b) EFF (Earth Fault Factor) ≤ 1.4

EFF = Maximum Line to ground voltage on healthy phase during fault ⁄

Rated Line to ground voltage
For solidly grounded system, EFF ≤ 1.0.
For ungrounded system, EFF ≅ 1.732
Case 1
Refer Fig 10 (T/S). Rating of transformer is 125 MVA, 220/33 kV, Star – Zig Zag,
ZP = ZN = 15%; Z0 = 2.5%.

Three phase fault current at 33 kV Bus1 = Ι3P = (125/0.15) / (√3X33) = 14.6 kA

Rated phase voltage = 33 / √3 = 19.05 kV

The secondary neutral is earthed through NGR to limit the ground fault current to
a desired value. Assume XR = 1Ω.

Ground fault is simulated on Phase R very near to Bus1 (F1 in Fig 10). From
results of simulation,
Ι1P = ΙR = 10.18 kA
KF = 10.18 / 14.6 = 0.7

Since KF > 0.6, the system for this fault is effectively grounded.
This can be reconfirmed from voltage rise on healthy phases during fault.
VR = 0

VY = VB = 22 kV (116%)
EFF = 1.16 < 1.4

8 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
Case 2
The same example is repeated with ground fault (R Phase) on cable at 3.6 KM
away from Bus1 (F2 in Fig 10). The cable parameters used for simulation are [5]:
2
Cable size: 3C x 400mm Al

ZP = ZN = 0.08 + j 0.117 Ω/KM

ZO = 0.646 + j 0.644 Ω/KM

From results of simulation:
Ι1P = ΙR = 5.88 kA

The voltages at Bus1 for far end fault are given below:
VR = 9.39 kV (49%)

VY = 22.13 kV (116%)

VB = 19.91 kV (105%)
EFF = 1.16 < 1.4

This brings out an important fact that even though ground fault current is only
40% of three phase fault current at Bus1 (5.88 / 14.6 = 0.4), the voltage rise at
the Bus1 is still within limits (<1.4 pu). Hence all other feeders connected to Bus1
do not experience over voltage.

Case 3
Refer Fig 11 (R/S). Transformer rating is 20 MVA, 33/11 kV, Delta – Zig Zag,
ZP = ZN = 12.5%; Z0 = 3%.

Fig 11

9 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
Three phase fault current at Bus1 = Ι3P = (20/0.125) / (√3X11) = 8.4 kA

Rated phase voltage = 11 / √3 = 6.35 kV

Secondary neutral is solidly grounded.

Case 3.1
Ground fault is simulated on Phase R very near to Bus1. From results of
simulation,
Ι1P = ΙR = 11.25 kA
KF = 11.25 / 8.4 = 1.3

The reasons for KF much greater than 1 are: (i) primary is delta connected (ii)
secondary is solidly grounded and (iii) zero sequence impedance is much smaller
(3%) as secondary is Zig Zag connected.

The phase voltages at Receiving Station bus are:

VR = 0

VY = VB = 5.6 kV (88%)
EFF = 0.88 < 1.4
The system is effectively grounded for fault very near to Receiving Station bus.
None of the connected feeders will experience over voltage.

Case 3.2
In Fig 11, a sample ring main is considered for simulation. It has 7 substations
and distance between substations is 500 meters. The cable parameters used for
simulation are [5]:
2
Cable size: 3C x 300mm Al

ZP = ZN = 0.123 + j 0.102 Ω/KM

ZO = 1.173 + j 0.427 Ω/KM

For a ground fault on R phase at remote Bus8,
Ι1P = ΙR = 2.184 kA
The ground fault current is only 26% of three phase fault current at Bus1 (2.184 /
8.4 = 0.26),

10 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
The phase voltages at faulted Bus 8, intermediate Bus 5 and Receiving Station
Bus 1 are shown in Table 2.

Bus 8 Bus 5 Bus 1

Voltage kV EFF Voltage kV EFF Voltage kV EFF
VR 0 2.513 0.40 5.865 0.92
VY 8.134 1.28 6.961 1.10 5.933 0.93
VB 9.951 1.57 8.409 1.32 6.380 1.00
Table 2

(i) At the remote Bus 8, EFF > 1.4, hence locally it is ‘non-effectively grounded’.
(ii) At intermediate Bus 5, EFF is marginally less than 1.4, just managing to be
categorized as ‘effectively grounded’.
(iii) At the Receiving Station Bus 1, EFF is ≤ 1 and it is ‘effectively grounded’.

From results of above case studies, the following observations are made:
a) Irrespective of fault location, EFF at Receiving Station is ≤ 1.0. This has
important implication that at Receiving Station, voltages of un-faulted phases do
not rise above normal phase voltage. Hence voltage of other feeders (Ring
Mains) connected to the bus will not experience over voltage.

b) As the fault location is moved away from Receiving Station, EFF at remote
location is higher. It can cross the threshold limit of 1.4. At the remote locations it
is no longer effectively grounded system. But in substations closer to Receiving
Station even on the faulted feeder, EFF < 1.4. Thus over voltage is limited to
local area near to faulted point.

The standards recognize this fact. Even in case of solidly grounded system, some
parts of system may not be effectively grounded for particular fault location. The
aim of solid grounding is to limit over voltages to local areas and over voltages
are not felt globally over entire system for fault in any one location.
In this context, the relevant extract (Cl 3.3) from the IEEE Guide [6] is reproduced
below:

11 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
“The overvoltage on un-faulted phases is also of concern because it is applied to
the equipment of customers served from distribution transformers connected from
phase to neutral on four-wire systems. Thus, even if arrester application is not a
limiting factor, the EFF must not be allowed to increase to a level that can impose
intolerable over voltages on customer equipment. As a rule of thumb, EFF at the
substation should not exceed 1.25, which is obtained approximately when X0/X1

= 2. Preferably EFF should not exceed 1.1, which requires an X0/X1 of 1.3 or
less. At locations remote from the substation, the EFF will exceed these values
because of the effects of line impedance. However, the lower values at the
substation are desirable to mitigate the effect of the line impedance and to
localize the over voltages near the fault location rather than requiring the whole
system to withstand them. It is realized however, that higher X0/X1 ratios have
been used satisfactorily”.
It is possible to choose NGR value so that KF = 0.4 to 0.5, with EFF nearly equal
to 1.4 for faults very near to source transformer, anticipating lower ground fault
current. But in this case, no margin is available in EFF. For any fault even slightly
away from transformer, voltage at local substation will rise resulting in EFF > 1.4.
This is the reason why the standards recommend that for effectively grounded
system, NGR is sized such that KF ≥ 0.6.
For academically oriented, a more precise definition for effectively grounded
system is that (X0 / X1) ≤ 3 and (R0 / X1) ≤ 1. The definitions given above for KF
and EFF will suffice for use by practicing engineers.

Summarising:
(i) Size NGR based on KF ≥ 0.6 for a ground fault on terminal of transformer
(ii) Grounding effectiveness at remote locations is based on evaluating EFF at
these locations
(iii) Irrespective of type of grounding, use 100% arrestor for voltages 33kV and
below.

12 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
More than 70% faults are single phase to earth faults. It is important to positively
identify and isolate these faults. Current based earth fault protections are more
sensitive and selective than voltage based system.
In solidly grounded system high magnitude of earth fault current is always
ensured for faults anywhere in the system. It is easy to design sensitive earth
fault detection system. However the damage at fault point could be severe. Also
equipment which experiences the let through current, undergoes higher dynamic
stress.
If we restrict the earth fault current below a certain level by introducing an
impedance in the neutral, the healthy phase voltages rise to L-L values thereby
stressing the insulation of all equipment connected to the system. This is also
detrimental to the health of the equipment particularly in a network with aging
equipment.
Effectively earthed system is balance between the two. We get sufficiently large
current ensuring positive relay operation; at the same time the healthy phase
voltages do not rise to dangerous levels.
The results presented here are outcome of simulation studies done by Sonu
Karekar, Amol Salunke and Ashutosh Pailwan.

5.0 Mirror Image concepts

For want of a better term, the title for this section has been chosen as above. It
also covers concepts which are ‘close and inverted’. It could also be termed as
description of ‘twins’.

5.1 Capacitor and Reactor

5.1.1 Voltage across capacitor can’t change instantly.
Ι = C dV/dt.

13 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
But the current can reach very high values immediately after switching. Refer Fig
12. It could be very large multiples (> 100) of rated current. But it dies down very
rapidly as time constant (CR) is in µsec. The inrush appears as a pulse of very
large magnitude.

Fig 12

5.1.2 Current through inductor (reactor) can’t change instantly.

E = L dΙ/dt.
The voltage across reactor can reach supply voltage immediately after switching.
Refer Fig 13. The time constant is in msec.

Fig 13
5.1.3 Inductor is connected in series with capacitor (Fig 14) to limit peak inrush
current during switching on capacitor banks.

Fig 14

14 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
5.1.4 Capacitor is connected in parallel with inductor
(Fig 15) to limit steepness of incoming surge voltage.

Fig 15
Stator windings of large alternator and motor are basically large inductance coils.
Any very steep front voltage wave entering the stator coils will damage the first
few turns of the windings. To flatten out the steep wave front, capacitor is placed
ahead of alternator or motor (Fig 16). For this reason, it is termed as ‘surge
capacitor’.

Fig 16
Thus reactor and capacitor are natural twins in power system components.

5.2 CBCT and Open Delta PT

5.2.1 CBCT
In solidly grounded system, the earth fault current magnitude is high and
comparable to three phase fault current. In this case, residually connected CT
connection (also termed Holmgreen connection) is used for connection to earth
fault relaying element. Secondary reflected phase currents are physically
summated. Refer Fig 17.

Fig 17
ΙN = ΙRS + ΙYS + ΙBS.

15 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
In low resistance grounded system, where the earth fault current magnitude is
limited to, say 200A to 400A, Core Balance Current Transformer (CBCT) is used
for connection to earth fault relaying element improving sensitivity of fault
detection. CBCT has a torriodal core on which secondary is wound. It encircles a
cable with all three conductors (R,Y,B). Output from secondary is proportional to
net flux produced by sum of three phase currents. Refer Fig 18.

Fig 18
Under healthy conditions, vector sum of the three phase currents is zero.
ΙR + ΙY + ΙB = 0
The resulting flux in the core is zero and current output from CBCT is nil. Since
CBCT output is zero under healthy conditions, its turns ratio is not chosen based
on maximum line current magnitude but on desired value of minimum primary
ground fault current to be detected. Typically it is 50/1.
During line to ground fault,
Ι R + Ι Y + Ι B = Ι N = 3 Ι 0.
To detect small earth fault currents (say 20A), in low resistance grounded system,
CBCT is employed. Numerical relays give an option to connect CBCT output to
relay as direct input rather than summating three phase currents through
software.
Generally CBCT output is wired to a DMT element (50N/2).
5.2.2 Open Delta PT
In ungrounded or very high resistance grounded system, ground fault current is
too low (less than 10 to 15A) for current based protection to pick up. Ground fault
detection is achieved using open delta PT connection.

16 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
Refer Fig 19.

Fig 19

Under healthy conditions, vector sum of the three phase voltages is zero.
VR + VY + VB = 0
During line to ground fault,

VR + VY + VB = VΔ = 3V0 = 3VP

PT connected in open delta measures zero sequence voltage.

For example, consider a 600 MW unit with rated voltage of 20kV.
Phase voltage, VP = 20/√3 = 11.55 kV.
The unit is very high resistance grounded. In case of ground fault on 20 kV side,
voltage sensed by open delta PT on primary side:

VΔ = VR + VY + VB = 3 x 11.55 = 34.65 kV

If the single phase PT ratio is (20/√3) kV / (110/3) kV,

TR = Turns Ratio = (20,000/√3) / (110/3) = 315

Open delta voltage on secondary side = VΔ / TR = 110V

The relay connected across open delta PT can sense the over voltage and
initiate alarm / tripping.
It is interesting to point out that open delta voltage is obtained by physically
connecting three PT outputs in series (Fig 19). In case of residually connected CT
connection, the relay current is obtained by physically connecting three CT
outputs in parallel (Fig 17).
The thing common in CBCT (Fig 18) and Open delta PT (Fig 19) functioning is
that both work on the principle of “Resultant” magnitude.
Thus, CBCT and open delta PT are twins for ground fault detection.
17 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
5.3 Phase voltage and Zero Sequence voltage during ground fault
5.3.1 Phase Voltage
Phase voltage is high at source and almost zero at the fault point. Under voltage
relay located near the fault location responds.
5.3.2 Zero Sequence voltage
Source (generator) does not intentionally produce any zero sequence voltage
and hence zero sequence voltage at source is nearly zero. At the point of ground
fault, phase voltage at faulted point collapses but zero sequence voltage is high
[7]. Refer Fig 20. Under voltage relay connected to phase PT and over voltage
relay connected to open delta PT respond. Thus in both cases, voltage relays
close to fault only respond.

Fig 20

Also to be noted is that the phase voltage at faulted point is nearly zero
irrespective of type of grounding of source. However zero sequence voltage at
faulted point varies widely depending on type of grounding. It is high in
ungrounded system and low in solidly grounded system. For illustration, zero
sequence voltage V0 is evaluated at the faulted point F2, Fig 10 considered in Cl
4.0, Case 2. Values for three types of source grounding obtained from simulation
are given below:
Ungrounded source, V0 = 19 kV

Effectively grounded source (XR = 1Ω), V0 = 12.3 kV

Solidly grounded source = V0 = 10.5 kV

18 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
It is myth to assume that neutral shift does not occur in solidly grounded system,
only its magnitude is less.
Sonu Karekar helped in simulating the above case.

5.4 Line to ground fault reflection in transformer

5.4.1 Delta – Star transformer
Line to ground fault on star side of transformer gets reflected as Line to Line fault
on delta side of transformer. Refer Fig 21.

Fig 21
The current distribution follows two cardinal principles: (i) KCL (Kirchhoff’s
Current Law) (ii) AT (Ampere Turn) balance of windings on same limb of
transformer
5.4.2 Star – Delta transformer with NGT
Line to ground fault on delta side of transformer grounded through Neutral
Grounding Transformer (NGT) gets reflected as Line to Line fault on star side of
transformer. Refer Fig 22.

Fig 22

19 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
5.4.3 Star – Zig Zag transformer
Conceptually it is same as (5.4.2) in which zero sequence isolation between
primary and secondary is obtained. Secondary neutral is available for grounding.
Here also, Line to ground fault on Zig Zag side of transformer is reflected as Line
to Line fault on star side of transformer [8]. Refer Fig 23.

Fig 23
5.4.4 Remarks on vector group selection
In an EHV transformer with HV side voltage of 132 KV and above, it is preferred
to have the HV side as Star to have a commercially cheaper transformer, as
graded insulation can be used.

One of the basic principles of ground fault relay coordination is to achieve zero
sequence isolation between LV and HV side of transformer. In this context, the
least preferred is Star-Star vector group, especially if both the primary and
secondary neutrals are solidly grounded. It is mitigated to a large extent if LV Star
neutral is grounded through resistance to limit the ground fault current to less
than a few hundred amperes, as in Station Transformer in power plant
applications. The reflected fault current on HV side in this case is negligible.

The next choice is to Star-Delta vector group which offers zero sequence
isolation between secondary and primary. However, if we want to have a
sensitive and selective earth fault protection system on the LV side, then we need
to use a NGT (Neutral Grounding Transformer) to create a grounded neutral and
provide a return path for the earth fault current.

20 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
Zig Zag on LV side of transformer combines the benefit of both the system. The
neutral of the Zig Zag winding can be grounded like a Star system, thereby
enabling provision of sensitive and selective earth fault protection. Also zero
sequence isolation is ‘naturally’ obtained as earth fault on Zig Zag side gets
reflected as line to line fault on the HV side.

5.5 Disposition of conductor and other metal parts – Single core cable,
IPBD and ACSR conductor

5.5.1 Single Core Cable

In single core cable, the conductor in the middle is either copper or Aluminum and
the armour surrounding conductor and XLPE/PVC insulation is non-magnetic,
usually Aluminum. Armour is provided for following reasons:
(a) It provides mechanical protection for insulation against external intrusion.
(b) It provides metallic return path for earth fault current. This results in lower
touch and step potentials.

5.5.1.1 Aluminum vs Steel armour in Single Core Cable

In the case of three core cable, under normal operating condition, ΙR+ΙY+ΙB = 0.
Hence, the net flux coming out of cable is zero. In this case steel armour can be
used. In the case of single core cable, the net flux coming out is proportional to
current in conductor and it is not zero. Hence only non magnetic metal like
Aluminum is used as metallic shield in single core cable. FEM analysis is done to
evaluate the eddy current loss with Steel and Aluminum armour. 11kV, 1000
2
mm , Al conductor, single core cable with maximum current carrying capacity of
1000 A is considered for simulation.

21 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
Refer Fig 24 for major cross sectional details. Results are summarized in Table 3.

Fig 24

Skin depth Eddy current

Armour Resistivity ρ Relative
-7 (mm) @ loss –
Material 10 Ω-Met Permeability
50Hz Watt/meter
Aluminum 0.35 1 13.2 33
Steel 1.43 1000 2.69 151
Table 3
Thickness of armour is 2.5mm. If it is steel it is almost equal to skin depth. If it is
Aluminum, it is much less than skin depth. Eddy current loss in Steel is nearly 4.6
times that of Aluminum. Also hysteresis loss is absent in case of Aluminum as it
is non-magnetic whereas in steel it is appreciable. Thus the heat generated due
to eddy current and hysteresis loss in steel armour is significantly higher
compared to Aluminum armour of same thickness which will result in derating of
cable. Hence Aluminum is preferred as armour for single core cables.
FEM analysis was done by Sairam under guidance of Prof S V Kulkarni. Talande
furnished cable parameters and participated in analysis.

5.5.1.2 Solid bonded system

If both ends are bonded, circulating current almost equal to conductor current will
flow in armour. This current is independent of cable length. The additional power
loss due to circulating current in armour will increase the temperature further.
Derating of cable has to be done to limit the temperature to allowable limits as per
type of insulation used (XLPE = 90°C, PVC = 70°C). For this reason, solid
bonding is rarely used in single core cable.

22 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
5.5.1.3 Single point bonded system
Only one end of armour (usually sending end) is earthed and the other end is
insulated. This is called single point bonding. Refer Fig 25.

Fig 25
This prevents circulating current flow in armour. However, in this method, the free
end of the armour (insulated) would develop induced voltage VΙ. Indian Electricity
Rules permit 65 volts as the limit of such induced voltage. Voltage induced in
armour is determined by armour diameter, spacing between cables (trefoil or flat
formation) and phase currents. For LV and MV cables, induced voltage in armour
is approximately given by VΙ ≅ 55mV / Amp / KM.
For example, for a current of 750A and cable length of 0.5KM, induced voltage in
armour = 0.055 x 750 x 0.5 = 23V.

Unlike solid bonding, single point bonding creates discontinuity in armour circuit
and inhibits flow of fault current returning back to source via a metal. In these
cases, it is mandatory to provide additional grounding conductor between two
distribution boards connected by single core cables. Refer Section 5.4.3 of IEEE
Std 575[9].

23 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
In case of an earth fault in any outgoing feeder of the receiving end distribution
board, the separate ground conductor facilitates return of the earth fault current
through the metal to the upstream source, as shown in Fig 26.

Fig 26
5.5.2 Isolated Phase Bus Duct
Isolated phase duct consists of tubular conductor of either Aluminum or copper.
Insulation medium is air. The protective enclosure is a tubular conductor of either
Aluminum or Steel. Typical sectional view of 24kV, 12kA IPBD is shown in Fig 27.

Fig 27

24 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
The major dimensional details are as follows:
Outer diameter of conductor: 500 mm
Thickness of conductor: 12 mm
Outer diameter of enclosure: 1000 mm
Thickness of enclosure: 8 mm
C/S area of conductor = (π/4) (5002 – 4762) = 18,398 sq.mm

C/S area of enclosure / Sheath = (π/4) (10002 – 9842) = 24,932 sq.mm

Bonded housing arrangement is used in majority of applications. The enclosure
for each phase is continuously bonded physically and electrically throughout its
run. The enclosures for the three phases are shorted at the extreme ends. The
situation is akin to solid bonded system in case of single core cable. The
magnitude of current flowing in enclosure is almost same as that of main
conductor (75 to 90%) but in opposite direction. For the above example, cross
section of enclosure is even greater than that of main conductor resulting in
reduced resistance and lower enclosure losses. The magnetic field due to current
in enclosure opposes the field due to current in main conductor at every instant
and the resultant flux is very small. The force on conductor or enclosure is
proportional to current flowing through it and the flux density of field in which it is
embedded. Since the resultant flux is very small, forces on conductor and
enclosure are less. Especially during short circuit condition, mechanical stresses
on conductor, enclosure and support insulators are minimal. Since the resultant
flux is small, IPBD can be supported and routed besides steel structures without
fear of excessive heating due to hysteresis and induced eddy currents within
steel members. As a measure of ‘abundant caution’, earthing conductors
(e.g.65x10mm GI strip) are run in parallel with IPBD enclosures and bonded at
the ends and at intermediate points. Compared to the enclosure resistance,
earthing conductor resistance is too high. Most of the current is carried by
enclosure and very little by parallel earthing conductor.
Summarising, the main functions of enclosure for IPBD are as follows:
(a) It provides mechanical protection for conductor against external intrusion.
(b) It acts as a magnetic shield allowing only very little flux to escape outside
enclosure. Nearby steel structures do not experience hot spots due to
induced currents.
(c) Mechanical stresses on conductor, enclosure and support insulators under
normal and short circuit conditions are considerably reduced.

25 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
Sandeep Lodh’s query was the trigger that prompted the author to study
similarities between single core cable and IPBD. The author acknowledges D
Guha’s contribution towards not only clarifying finer points on comparison
between single core cable and IPBD but also offering critiques on different topics
covered in this article.

5.5.3 ACSR conductor

It is an inverted version of the two cases discussed above. The outer core is
made of Aluminum strands and is the conductor. Aluminum has good
conductivity, low weight and lower cost compared to copper. The inner core is
made of strands of steel. Refer Fig 28. It is akin to a messenger wire over which
conductor is wrapped. Steel core is provided to increase the tensile strength of
cable. Tensile strength to weight ratio of ACSR is almost twice that of only
Aluminum conductor. Hence with ACSR, span length can be much higher without
increasing sag.

Fig 28

Since relative permeability μR of steel is very high (1000) compared to Aluminum

(1), the reactance of steel wire is much higher. It offers high impedance to flow of
AC current. Most of the current is carried by Aluminum wire and very little by
steel. Hence higher resistance of steel does not add to significant increase in
power losses as the current itself in steel wire is low.

26 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
Thus we have two examples where sheath or
enclosure surrounds the conductor (single core cable and IPBD) and another
example where conductor surrounds the steel wire (ACSR).

5.6 Transposition – EHV Overhead line and EHV Cable

5.6.1 EHV Overhead line

In case of EHV lines of very long length (more than 300 KM), the conductors are
transposed to minimize voltage unbalance. Let (1), (2) and (3) be three points in
space with respect to centre line of tower (Fig 29). R phase conductor occupies
position (1) in first section, position (2) in second section and position (3) in last
section. Y phase and B phase conductors are similarly transposed. Two figures of
merit are used to judge effectiveness of transposition.

Fig 29

Ι1, Ι2, Ι0: Positive, negative and Zero sequence currents

Zero sequence unbalance factor = M0 = Ι0 / Ι1

Negative sequence unbalance factor = M2 = Ι2 / Ι1

Ideally if positive sequence voltage is applied to line, only positive sequence
current should flow, i.e., M0 = M2 = 0.
However due to unsymmetrical conductor geometry in space with respect to
tower, in un-transposed line, M0 is about 1% and M2 is 3 to 20%. Refer Section
4.8 [10]. In this case, for application of positive sequence voltage, 1% zero
sequence current and 3 to 20% negative sequence current can flow which is not
desirable. In case of perfectly transposed lines, M0 = M2 = 0.

27 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
5.6.2 EHV Cable
In case of EHV cables, the usual practice is to ‘transpose’ the sheath of individual
EHV cables. The correct terminology used for cables is ‘cross bonding’. Single
core EHV cable has a central conductor of Copper with XLPE insulation over the
conductor. Over the insulation, metallic sheath either of Aluminum or Lead is
provided. When the conductors carry current, voltage induced due to mutual
induction on metallic sheath could be excessive.
If the sheaths are bonded at both the ends (solid bonding), the circulating current
in sheath is high (almost equal to conductor current) resulting in continuous
dissipation of heat. In this case, cable has to be derated to a lower value so that
temperature rise in conductor is within limits applicable for XLPE insulation. By
cross bonding the sheath, voltage induced and the resulting circulating current in
2
sheath is reduced to a minimum. Refer Fig 30. For 220 kV, 1200mm Cu cable,
laid in trefoil, sheath (corrugated Aluminum) cross bonded, carrying a current of
840A, maximum sheath voltage is 25V and sheath current is negligible. Amol
Salunkhe did the simulation using PSCAD to obtain these figures.

Fig 30

Thus, the conductor is transposed in EHV over head lines while the sheath is
transposed in case of EHV cables.

28 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
6.0 Conclusion
In this article, we have paraded a few cases in power engineering that practicing
engineers find it difficult to comprehend. The underlying concepts behind the
cases are explained. Also from the vast pool of information available, there is a
pattern to be unearthed and dots to be connected. These are presented under
the section ‘mirror image concept’. We will elaborate on other difficult to
comprehend cases in future articles.

7.0 References
[1] ‘Reactive Power: A Strange Concept?’ - R Fetea and A Petroianu, University
Of Cape Town, South Africa.

[2] ‘A monograph on reactive power’, M Ramamoorty, ERDA, India, April 2005

[3] ‘Reactive Compensation Fundamentals for Distribution Networks’ - K

Rajamani and Bodhlal Prasad, IEEMA Journal, Aug 2009, pp 112- 115.

[4] ‘Grounding of Electrical System – Part II’, K Rajamani, IEEMA Journal, June
2006, pp 51 to 58.

[5] ‘Cable sequence impedance measurement at site’, K Rajamani and Bina

Mitra, IEEMA Journal, August 2013, pp 84 to 86.

[6] ‘IEEE C62.92.4-1991 – IEEE Guide for the Application of Neutral Grounding in
Electrical Utility Systems, Part IV—Distribution’

[7] ‘Symmetrical components for Power Systems Engineering’, J Lewis

Blackburn, Marcel Dekker, 1993

[8] ‘Zig Zag Transformer - Fault Current Distribution, Short Circuit testing and
Single Phase loading’, K Rajamani and Bina Mitra, IEEMA Journal, July 2013,
pp 84 to 91.

[9] ‘IEEE Std 575 - IEEE Guide for the Application of Sheath-Bonding Methods
for Single-Conductor Cables and the Calculation of Induced Voltages and
Currents in Cable Sheaths’

[10] ‘Analysis of faulted power systems’, P M Anderson, IEEE Press, 1995

29 of 29
August 2016, IEEMA Journal, Page No. 69 to 80
 Conceptual
Clarifications in
Electrical Power
Engineering
Part-2
Dr K Rajamani,
Reliance Infrastructure Ltd., Mumbai
(March 2017, IEEMA Journal, Page 65 to 76)
 Conceptual Clarifications in Electrical Power Engineering – Part 2
K Rajamani, Reliance infrastructure Ltd

"It is important to simplify the explanation, not the subject”-

YouTube video on Physics – Henry Reich

1.0 Introduction

As in Part 1, an attempt has been made in this part to unravel the underlying concepts in selected
topics in power engineering. The topics covered are:
(a) Effect of phase shift introduced by transformer on angle stability
(b) Paralleling and synchronizing of transformer
(c) Ampere Turn Balance in Transformer
(d) Percentage impedance for three phase and equivalent three single phase transformers
(e) Voltage dip experienced at LV side of transformer for faults on HV side
(f) Effect of LV side Unbalance current on HV side reflected current

2.0 Phase Shift and Stability

We will prove that phase shift introduced by (Y-Δ) transformer cannot influence power transfer
magnitude.

2.1 In a power network, transformers that introduce phase shift are present. The most popular vector
group in this category is (Y-Δ) transformer which creates 300 phase shift in voltage and current
between two sides of transformer. When balanced power flow analysis is done for networks having
transformers with different vector groups, it is advantageous to work in pu for following reasons:

a) Per unit impedance (or % impedance) is same whether referred to primary or secondary.

b) The current magnitude in pu is same on primary and secondary side. However, a phase shift
of 300 is introduced between primary and secondary currents.

c) The voltage in pu on primary and secondary side are same if the transformer is unloaded
(current is zero). If the transformer is loaded, voltage on primary side is affected only to the
extent of regulation (X drop) compared to secondary side. A phase shift of 300 is further
introduced between primary and secondary voltages.

March 2017, IEEMA Journal, Page No. 65 to 76

Page 2 of 27
 For simplicity sake, assume that (Y-Δ) transformer is on nominal tap. This does not affect conclusion
even if tap is off nominal. The hand calculations presented in sequel are easier to understand with
nominal tap.

For illustration, consider the network shown in Fig.1

Fig 1

On 100 MVA base,

100
= 0.15 = 0.0476
315
100
= 0.15 = 0.12
125
220
= = 484Ω
100
20
= = 0.0413
484

2.2 First, power transfer is computed in per unit without considering phase shift across transformer.
Choose V4 as reference.
= 1 0

Since load is 100 MW at UPF,

Ι = 1 0

Further it may be noted that in pu

Ι =Ι = Ι = Ι
V =V + Ι jX
= 1 0 + 1 0 x 0.12 90
= 1.0072  6.8428

March 2017, IEEMA Journal, Page No. 65 to 76

Page 3 of 27
 Ι = Ι = 10
V =V + Ι jX
= 1.0072 6.8428 + 1 0 0.041390
= 1.0129  9.1629

Ι = Ι = 10
V =V + Ι jX
= 1.01299.1629 + 1 0 x 0.0476 90
= 1.0216  11.7994

Power transfer can be calculated in two ways:

(a) P = R.P. [V11*]

= R.P. [1.021611.79940 x 10]
= 1 pu

(b)
| || |
= sin( − )
( + + )

1.0216 1.0
= sin(11.7994 − 0)
(0.0476 + 0.0413 + 0.12)

= 1pu … (1)

The above matches with assumed load of 1pu (100MW).

2.3 Now, the same exercise will be carried out considering phase shift due to (Y-Δ) transformation.
Assume vector group of Transformer T2 is Yd1. In this case, voltage and current on star side (220
kV) lead voltage and current on delta side (33 kV) by 300.

On 33kV side,

= 1 0
Ι = 1 0

March 2017, IEEMA Journal, Page No. 65 to 76

Page 4 of 27
On 220kV side,

V = 1.0072  6.8428 + 30
= 1.0072  36.8428
Ι = 1  30

V3 in pu accounts for X drop across transformer T2.

V =V + Ι jX

= 1.0072 36.8428 + 1 30 0.041390

= 1.013  39.1589

Ι = Ι = 130
V =V + Ι jX
= 1.01339.1589 + 1 30 0.0476 90
= 1.0216  41.7994

On 220kV side,

V = 1.013  39.1589
Ι = 1  30

On 16kV side,

V = 1.0216  41.7994 + 30
= 1.0216  71.7994
Ι = 130 + 30 = 160

P = R.P. [V11*]
0
= R.P. [1.021671.79940 x 1-60 ]
= 1 pu

This is same as obtained without taking into account phase shift across (Y-Δ) transformer.

March 2017, IEEMA Journal, Page No. 65 to 76

Page 5 of 27
 Thus, for balanced (positive sequence) load flow calculations, phase shift due to vector group of
transformer will not influence power transfer calculations.

The above conclusion is also in line with common sense reasoning. Input and output power (MW) of
ideal transformers is same (neglecting losses) and this is true irrespective of vector group of
transformer.

Another way to look at the problem is to consider a generator connected to a resistive load through
Yd9 (rare vector group, given here for just illustration) transformer which introduces 90 phase shift.
In this case, will the resistor look like an inductor as seen from generator? This is not possible as
both voltage and current are shifted by 90 and the generator will still see the load as resistor only.

2.4 The power transfer relation used in stability analysis is given by (Fig.2)

= sin … (2)

Fig 2

Torque angle  = 1 - 2

The stability limit is reached when is 900.

March 2017, IEEMA Journal, Page No. 65 to 76

Page 6 of 27
 It must be emphasised that Eqn. (2) used for checking stability limit implicitly ignores phase shift
across transformer due to different vector groups. This is reaffirmed by the straightforward
application demonstrated in Eqn. (1) of Cl 2.2.

2.5 Remarks on PMU data analysis

PMU (Phasor Measurement Units) are deployed in EHV (765 / 400 / 220 kV) networks at different
locations for Wide Area Monitoring. There are many transformers present in EHV level but all of
them are either autotransformers or star-star transformers which do not create phase shift in either
voltage or current between secondary and primary. Thus, comparison of voltage phase angle of
different buses of the network based on PMU data is feasible.

Since PMU measures actual angle of phases VR, VY and VB, any phase shift introduced by vector
group [(Y-Δ), (Y-Z), etc] or phase shifting transformer will be reflected in measurement set. Hence,
stability limits in these cases cannot be assessed by direct comparison of raw data of phase angles
between different buses unless phase shifts introduced by transformers are accounted for.

The author benefitted immensely from the discussions with Prof M V Hariharan and Prof Anil Kulkarni
on this topic.

3.0 Paralleling and Synchronizing

3.1 Parallel Operation of Transformers
In Fig 3, two transformers to be paralleled are shown.

Fig. 3

March 2017, IEEMA Journal, Page No. 65 to 76

Page 7 of 27
Assume the vector group of TR1 is Yd11 and TR2 is Yd1. For common primary voltage, voltages on
secondary side on either side of switch (S) will be V11 and V22. For the selected vector group,
1 = +30 and 2 = -30. Choosing V1 as reference, voltages on either side of switch are V10 and
V2-60. Differential voltage ( − ) appearing across switch (Fig 4):

V = sqrt [V12+ V22 – 2 V1 V2 Cos]

Let V1 = V2 = 1.0 pu and  = -60.

V = 1.0 pu
Fig.4
Assume XT1 = XT2 = 0.1 pu (10%)
After closing the switch without any external load, circulating current:
C = V / (XT1 + XT2) = 5 pu
500% of rated current will circulate between the transformers which will damage the equipment.

Hence for paralleling transformers, it is essential that vector group of transformers by clock position
must match to avoid circulating current. In Mumbai Transmission, Yd11 and Yz11 transformers
operate in parallel without any problem as the clock positions of both transformers are same, though
the secondary winding connections are delta and zig zag. Similarly, it is possible to operate Yd1 and
Dy1 in parallel as the clock position is same in both transformers

Next, selection of vector group of transformers in power plant is discussed. Typical SLD is shown in
Fig 5. Generated power is evacuated to system (Bus2) through GT (Generator Transformer).

Fig. 5

During start-up of a unit, there is no power at the generator terminals. During this time, the unit Bus3
is fed from the station transformer through station Bus4 and station to unit tie by closing tie Breaker
March 2017, IEEMA Journal, Page No. 65 to 76
Page 8 of 27
 (Bkr B). Post synchronization, when the unit picked up load, the supply to unit bus is switched over
to UAT without interruption. UAT and ST are momentarily paralleled by closing Bkr A and then Bkr
B is tripped. But for safe momentary paralleling, secondary voltages of UAT and ST must be in
phase. Usually vector group of ST is fixed as Yy0. Assume vector group of GT is Yd1. To match
phase voltages on secondary side, vector group of UAT must be chosen as Dy11. If vector group of
GT is chosen as Yd11, vector group of UAT have to be Dy1 to match phase voltages on secondary
side.

If the vector group of GT is Yd1 and ST is Yyo, theoretically it is possible to select vector group of
UAT as Yd11 or Yz11 as the clock position is same as conventional Dy11. Selection of star / delta /
zig zag winding is based on techno-commercial reasons like type of grounding, size and cost.

3.2 Parallel Operation of Generators through GT (Synchronising)

Parallel operation of generators implicitly assumes all generators are connected to a common bus
without any intervening impedance between generator and bus. But the case under discussion is
about generators connected to a common bus but through respective GTs. Refer Fig 6. No other
tapping is taken from generator terminal except for GT. In this case, it is not necessary to have GTs
with identical clock positions.

Theoretically, GT1 can be Dy5 and GT2 can be Dz10. The reason is that transformers are not really
paralleled as discussed in previous sections, but controllable sources are connected to one side of
transformer through a process called ‘synchronisation’.

Fig. 6
March 2017, IEEMA Journal, Page No. 65 to 76
Page 9 of 27
 Assume, Generator 1 has started, and supply is extended to Bus2 and Bus3 through GT1 and GT2.
Consider, vector group of GT1 as Yd1 and that of GT2 as Yd11. Though voltages of Bus1 and Bus3
are phase shifted by 60, it has no impact as Bus1 and Bus3 are not directly tied. At Bus3 any
arbitrary voltage phasor can appear. At some time t1 the generator voltage phasor is shown. Both
the magnitude and phase angle of ‘incoming’ voltage are very different from ‘running’ voltage on
Bus3. By adjusting power output from turbine, the machine speed can be changed to modify phase
angle. By adjusting the excitation, voltage magnitude can be changed. Generator phase voltage at
time t2 is shown which is closer to ‘running’ voltage on Bus3. Either by manual or auto-synchroniser,
the ‘incoming’ voltage is brought almost in line with ‘running’ voltage. At time t3, the switch is closed,
and the generator voltage locks onto ‘running’ voltage. Since controllable voltage source is
connected to Bus3, there is no restriction on running voltage phasor of Bus3.

The situation is akin to docking of unmanned supply spaceship with manned International Space
Station (ISS). Under remote control, supply spaceship ‘chases’ ISS and docks with ISS at the proper
moment.

Thus, the clock position of GTs connected to a common bus can be different. Theoretically, there is
no limitation on choice of primary and secondary winding connection (star, delta or zig zag).
The author is indebted to D Guha for his substantial contribution on the above topic.

4.0 AT Balance Principle in transformer

4.1 AT balance & KCL
Two fundamental principles of transformer operation are AT (Ampere-turn) balance and KCL. If both
cannot be satisfied due to any constraints, no current flows in transformer. In Fig 7, Delta-Star
transformer is shown.

Fig. 7

March 2017, IEEMA Journal, Page No. 65 to 76

Page 10 of 27
 Primary and secondary windings in identical alignment are wound on the same limb of transformer
(Fig 8).

Fig. 8

For successful transformation following shall be satisfied:

Primary Winding Current x Primary Turns = Secondary Winding Current x Secondary Turns
RTR = rTr ; YTY = yTy ; BTB = bTb

4.2 Experimental verification

The above can be verified by a simple experiment which can be done in any college laboratory.
Experiments were carried out on 11/0.433 kV, 400KVA transformer. Results are shown in Fig 9 to
Fig 14 and Table 1. In all cases, transformer is energized from 11kV side using 240V single phase
supply and LV side is shorted through shorting link.

HV side (11kV) LV side (0.433kV)

Fig Voltage Measured Measured Shorting
rn (A) yn (A) bn (A)
No across Volts (1) Current (A) Link
9 RY 231 4.8 bn - - 215
10 YB 231 4.8 yn - 215 -
11 BR 231 4.8 rn 215 - -
12 RY 231 - yn - - -
13 RY 231 - ry - - -
14 RY 231 - yb - - -

Table 1

March 2017, IEEMA Journal, Page No. 65 to 76

Page 11 of 27
In Fig 9, voltage is applied across RY and bn is shorted on LV side. Current through secondary
windings which are open is zero and corresponding primary windings also will not carry any current
to satisfy AT balance principle. The measured secondary and primary currents are 215A and 4.8A.
The current ratio is 45 (215/4.8) which matches with Turns Ratio (11/{0.433/3}).

Fig 9

In Fig 10, voltage is applied across YB and yn is shorted on LV side voltage. In Fig 11, voltage is
applied across BR and rn is shorted on LV side voltage. The measured secondary and primary
currents are 215A and 4.8A, same as in Fig 9.

Fig 10

Fig 11

March 2017, IEEMA Journal, Page No. 65 to 76

Page 12 of 27
 In Fig 12, voltage is applied across RY and yn is shorted on LV side. Since secondary windings r
and b are open, the currents in corresponding primary windings are forced to zero. The only
possible current distribution is shown in Fig 12, but this cannot happen as B phase on primary side
is open and there is no return path for current. Current does not flow either on primary or secondary
windings as confirmed from Table 1.

Fig 12

In Fig 13, the possible current distribution is shown when voltage is applied across RY and ry is
shorted on LV side (phase to phase short). But this cannot happen as B phase on primary side is
open and there is no return path for current. Current does not flow either on primary or secondary
windings as seen from Table 1.

Fig 13

March 2017, IEEMA Journal, Page No. 65 to 76

Page 13 of 27
 In Fig 14, phase to phase short circuit between y and b phase is created. In this case also, B phase
on primary side is open and there is no return path for current. Absence of current flow on both
secondary and primary windings is confirmed from results of experiments given in Table 1.
Avinash Gawde performed the above experiments and his contribution is acknowledged.

Fig 14

Analysis of current distribution in Zig Zag connected transformers reveals interesting results as
windings on the same limb of transformer carry current from two different phases. For more details,
refer [1].

4.3 Constant Flux Operation

An important consequence of AT balance is that it results in constant flux apparatus. A single phase
transformer is considered for illustration. Let
Primary Turns TP = 100
Secondary turns TS = 10.
No load primary current = 1A
Full load secondary current = 990A
Corresponding full load primary current = 100A
Under no load condition, net flux in the core (Fig 9) corresponds to 100AT (1 x 100).

Fig 15
March 2017, IEEMA Journal, Page No. 65 to 76
Page 14 of 27
 Under full load condition, Primary AT (100 x 100) is almost nullified by secondary AT (990 x 10) so
that net flux () in the core again corresponds to 100AT (Fig 16). The flux in the core almost remains
same from no load to full load operation. For chosen flux density B (say 1.7T), the cross section of
core can be fixed ( / B).

Fig 16

4.4 Return Fault Current Distribution

Thus, flow of current in transformer happens only when both principles (AT balance and KCL) are
satisfied. In Fig 17, fault occurs at ‘F’. Fault current cannot return to neutral of any arbitrary
transformer (e.g. A, B or D) but will return to C which alone satisfies both the principles stated above.
In fact, concepts in neutral grounding are basically based on above two principles.

Fig 17

March 2017, IEEMA Journal, Page No. 65 to 76

Page 15 of 27
4.5 Spatial AT balance

It is not enough to have AT balance for the winding ‘as a whole’. AT balance shall also be achieved
‘spatially’, i.e. at every ℓ height of winding Primary and Secondary AT shall be balanced. Refer
Fig 18. Otherwise when the transformer feeds external short circuit current carried by both Primary
and secondary windings, the dynamic short circuit forces at the place where spatial AT balance is
not obtained, may lead to winding deformation if supporting and clamping structures are not
adequately designed.

Fig 18

4.5.1 Distribution Transformers

But maintaining perfect spatial AT balance over the entire height for different main windings and tap
winding dispositions is not practical in many cases. For example, consider the ubiquitous Distribution
Transformers (DTs). Vector group of most of the DTs (11kV/433V or 6.6kV/433V) are Delta – Star
with off load taps on HV side. Off circuit taps are provided in the middle of main winding itself (Fig
19). Perfect AT balance between HV and LV windings may not be obtained in the tap region. The
designer calculates resulting short circuit forces when the transformer feeds external short circuit
current and provides the necessary support and clamping structures to minimize winding
deformation.

Fig 19
March 2017, IEEMA Journal, Page No. 65 to 76
Page 16 of 27
 In passing, it may be remarked that foil wound transformers for LV winding of DTs have superior
short circuit withstand capability compared to conventional wire or strip wound transformers. Foil
width covers almost entire height of winding. AT unbalance created by taps on HV side etc is
automatically compensated by an appropriate internal current distribution in LV foil winding. This
reduces axial forces due to short circuit current flow to a negligible value. Elaborate coil clamping
arrangement is not necessary [2]. Aluminium foil wound Transformers are deployed in greater
numbers by utilities at distribution level. The number of faults in LT distribution system is very high
and foil wound transformers have excellent through fault short circuit withstand strength.

In the case of conventional EHV transformers with OLTC on HV side, tap winding is mostly a
separate one (outermost from core) and occupies shorter height corresponding to HV and LV
windings. In this case also spatial AT balance over the entire height of winding is reasonably
achieved.

4.5.2 Split winding Transformers

It is pertinent to make a remark on winding dispositions in a three winding transformer here. For
illustration, 400/11.5/11.5kV transformer is considered. The HV side is made of two windings
connected electrically in parallel and physically placed one above the other (Referred as Top
Winding (TW) and Bottom Winding (BW) in Fig 20). The two secondaries LV1 and LV2 are linked to
Top and Bottom windings of HV respectively. This arrangement, called ‘split winding’, is much
cheaper compared to having separate two double winding transformers. But this economy comes
with a drawback. When both LV1 and LV2 carry normal current, AT balance is maintained spatially.
However, if LV2 feeds a through fault, though BW carries majority of reflected fault current, not so
insignificant current (about 5% of reflected current) flows also in TW due to coupling between TW
and LV2. There is no counter balancing current in LV1. This creates spatial AT unbalance. The
situation is accentuated if taps are present on HV side. The outer tapping winding cannot be of full
height since the line lead coming out from mid-height of the HV must be cleared by the tap winding.
This also creates spatial AT unbalance. The design of support and clamping structures to withstand
short circuit forces is a challenging task.

Fig 20

March 2017, IEEMA Journal, Page No. 65 to 76

Page 17 of 27
4.5.3 Inrush Current

When a transformer is switched on, inrush current (up to 6 to 8 times rated current) flows. The
magnitude of inrush current depends on what point of voltage waveform the transformer is switched
on and polarity and magnitude of residual magnetism present before switching. If transformer is
switched from HV side, Inrush current flows only in HV winding whilst the LV windings do not carry
any current. Thus, AT unbalance occurs every time the transformer is switched in. The winding that
carries the inrush current is subject to mechanical stresses. Transformers subjected to repeated
switching (from same side HV or LV) can suffer winding deformation if they are not designed to
withstand the forces with sufficient safety margin.

In very large power transformers, the situation is mitigated to a large extent by using CSD (Controlled
Switching Device) in which each pole of breaker is closed at the most favourable instant on voltage
waveform that will cause least inrush current. For obvious reasons CSD is not applicable for gang
operated breakers. An example of inrush current waveform, captured from numerical relay records,
is shown in Fig 21. The auto-transformer is rated for 765/400/33 kV, 1000 MVA (3x333), and is
switched from 765kV side using CSD. The maximum inrush current observed is only 11% of rated
current, substantially lower than 200% to 800% expected when switching without CSD. The winding
does not practically experience any dynamic forces.

Fig 21

March 2017, IEEMA Journal, Page No. 65 to 76

Page 18 of 27
4.5.4 Design Approach

The designer must ensure that the transformer withstands resulting dynamic forces in all above
cases and this involves:
(a) Precise calculation of the magnetic flux distribution in the windings
(b) Proper design of support structure
(c) Judicial choice of materials (work hardened and/or epoxy bonded conductors, well stabilized
insulating materials, high strength structural steel etc)
(d) Correct processing and clamping of the transformer winding system.

Sophisticated software tools (e.g. SMC ELDINST from Ukraine, COMSOL from Sweden) to calculate
flux distribution and dynamic forces are available to aid the designer to achieve the above. For more
in-depth analysis on this subject, Chapter 6 of Ref [3] can be consulted.

The author acknowledges the clarifications provided by Vikrant Joshi and P Ramachandran on
various aspects of spatial AT unbalance.

5.0 Impedance specification for 3 Single phase transformers vs Three phase transformer

For a 600MW unit, typical parameters of Generator Transformer are as follows: 750MVA,
20kV/420kV, YNd1, XT = 15% impedance at principal tap. Generator Transformers of large units
(600MW and above) are generally made up of 3 single phase units due to transport limitations. Delta
on LV side and Star on HV side are formed externally. Refer Fig 22.

Fig. 22

March 2017, IEEMA Journal, Page No. 65 to 76

Page 19 of 27
MVA and voltage rating of single phase transformer are specified, without ambiguity, as 250MVA
and 20kV/420/3kV. However, a lingering doubt arises in the mind of young design engineer when
specifying impedance for single phase unit (5%, 15% or 45%?). The real advantage of working in
per unit system is that the same percentage impedance required for three phase unit can be specified
for single phase unit. This will be clarified with a numerical example.
Single Phase unit: 250MVA, 20kV/(420/3)kV and XT = 15%.

Rated Current R = 250/20 = 12.5kA

Base Impedance ZB = 202/250 = 1.6Ω

Impedance ZT = 0.15 x 1.6 = 0.24Ω

Impedance Volts V = 20 x 0.15 = 3kV

By definition, if impedance volt is applied on LV side of transformer with HV side shorted, rated
current will flow. For single phase circuit (Fig 23),

Fig.23
 = 3/0.24 = 12.5kA
This matches with rated current.
Three Phase unit: 750MVA, 20kV/420kV and XT = 15%. (assumed same as for single phase unit)

Rated Current R = 750/(3 x 20) = 21.6513kA

Base Impedance ZB = 202/750 = 0.5333Ω

Impedance ZT / phase = 0.15 x 0.5333 = 0.08Ω

Impedance Volts V = 20 x 0.15 = 3kV

By definition, if impedance volt is applied on LV side of transformer with HV side shorted, rated
current will flow. For three phase circuit (Fig 24),

Fig.24

 = (3/3) / 0.08 = 21.6513kA

March 2017, IEEMA Journal, Page No. 65 to 76
Page 20 of 27
 This matches with rated current.
Thus, percentage impedance for single phase units will be same as that for equivalent three phase
unit.

6.0 Voltage dip for fault on primary

Effect of upstream fault on downstream voltages

Voltage dip is defined as the difference between reference voltage (usually steady state pre-fault
voltage) and residual voltage during fault expressed as percentage. For example, if the pre-fault
voltage of bus is 100% and voltage of bus during fault is 55%, voltage dip is 45%. This is as per IEC
definition [4].

Voltage dips on upstream side due to a fault is felt on downstream levels. Power Quality issues due
to upstream voltage dips are discussed in detail in [5]. The levels to which downstream bus voltages
dip depend on vector group of intervening transformers present. The results of simulation for isolated
transformer are shown in Table 2 and are in line with values given in Table 1 of [4]. Following
assumptions are made in simulation:
(a) Source on primary side is solidly grounded.
(b) Line to ground fault on primary side without fault impedance, i.e. voltage of faulted phase on
primary side is zero.

Secondary side voltages for (L-G) fault on primary side

Vector Phase Voltages Line Voltages Phase
Group RN YN BN RY YB BR Unbalance
YNyn 0 1 1 0.58 1 0.58 1.0
Dyn 0.58 1 0.58 0.88 0.88 0.33 0.39
YNd - - - 0.33 0.88 0.88 -
YNzn 0.58 1 0.58 0.88 0.88 0.33 0.39
Dzn 0.88 0.88 0.33 1 0.58 0.58 0.53

Table 2

Unbalance indicated in the last column of Table -2 is evaluated as follows:

Phase quantities - VR, VY and VB

Average value VAVE = (VR + VY + VB) / 3

VUNB = Max { VAVE - VR , (VAVE - VY , (VAVE - VB  } / VAVE

March 2017, IEEMA Journal, Page No. 65 to 76

Page 21 of 27
In all cases, unbalance is 1pu (100%) on primary side as one of the phase voltage is zero (faulted
phase). Vector groups other than Yy reduce the unbalance substantially. The steep voltage dip in
one phase in primary is ‘distributed’ across three phases in secondary. Thus, a step down
transformer achieves the following:

(a) Steps down the voltage

(b) Reduces the fault level
(c) Reduces the voltage unbalance for dip in upstream voltage.

The first two are well known but the third is significant from power quality point of view in transmission
/ distribution systems.
In previous discussion, only an isolated transformer is considered. In practical power systems, series
of step down transformers are involved from EHV system to consumer substation. For analysis,
transmission, and distribution system of author’s company in Mumbai is shown in Fig 25.

Fig. 25

March 2017, IEEMA Journal, Page No. 65 to 76

Page 22 of 27
 Salient details (typical) are given below:
(a) 220kV system - solidly grounded
(b) TR1 - 220/33kV, 125MVA, YNzn11 transformer. 33kV system is ‘effectively grounded’
through NGR (Neutral Grounding Reactor) such that (1P / 3P) > 0.6
(c) TR2 - 33/11kV, 25MVA, Dzn10. 11 kV system is solidly grounded.
(d) TR3 – 11/0.415kV, 1MVA, Dyn1. LV system is solidly grounded.

(L-G) fault is simulated on 220kV, 33kV and 11kV systems and the results are shown in Fig 25. The
three phase voltages shown are in kV. Values in pu are shown within brackets. Following
observations are made:

(a) LV (415V) side can experience voltage dip from 31% to as high as 67%.
(b) 11kV side can experience voltage dip for upstream faults to the extent of 45%.

The above results are based on extreme case of fault with zero impedance on upstream side. If fault
impedance is present, dips will be correspondingly less.

Faults in utility system occur due to atmospheric conditions, equipment failure or external intrusion
damaging the equipment. In these cases, voltage dip at consumer end can’t be avoided. The
consumer must design his equipment which are sensitive to voltage dips to have ‘ride through
capability’ to override upstream transient faults. Typical over ride time of at least 300 msec is
recommended before which upstream fault is expected to be cleared [5]. Another method to prevent
loss of drive during transient dips is to employ ‘Reacceleration schemes’.

Contributions of Sonu Karekar and Amol Salunkhe in doing the simulation using PSCAD and PSSE
are acknowledged.

7.0 Current Unbalance reduction in Delta - Star transformer

In Cl 6 and Table 2, we discussed how a steep dip in upstream voltage of a phase is more evenly
distributed in downstream buses due to presence of Delta – Star transformer resulting in reduction
in ‘voltage unbalance’. Here we will demonstrate how current unbalance in downstream side is
mitigated on upstream side of Delta –Star transformer.

March 2017, IEEMA Journal, Page No. 65 to 76

Page 23 of 27
Consider 11/0.44kV, Distribution Transformer (DT). Vector group of DTs used in utilities and power
plant / industrial plant auxiliary systems is almost universally Delta – Star. On the LV side (Star),
single phase loads are present. Perfect balancing of loads in three phases is difficult to achieve
especially in utility distribution systems. Sample spot readings taken on DTs in author’s utility are
given in Table 3

KVA R (A) Y (A) B (A) Unbalance (%)

1000 1440 1580 1380 7.7

1000 1270 1611 1231 17.5
630 1005 875 970 7.9
630 963 724 965 18.1

Table 3

Unbalance indicated in the last column of Table 3 is evaluated as follows:

Phase quantities - R, Y and B

Average value AVE = (R + Y + B) / 3

UNB = Max { AVE - R , (AVE - Y , (AVE - B  } / AVE

For illustrating the unbalance mitigation offered by Delta – Star vector group, let the current on
secondary (star) side (in pu) be as follows:

RS = 10 ; YS = 0.9-140 ; ; BS = 1.1100

Average value AVES = (1+0.9+1.1) / 3 = 1.0

UNBS = 10%

Using sequence components, zero, positive and negative sequence components are evaluated as
follows:
 0 S = [  R S +  YS +  BS ] / 3
= [10 + 0.9-140 + 1.1100] / 3 = 0.172976.7

1S = [ RS + a YS + a2 BS ] / 3 = 0.9865-13.4

2S = [ RS + a2 YS + a BS ] / 3 = 0.059789.6

March 2017, IEEMA Journal, Page No. 65 to 76

Page 24 of 27
Following Stevenson Convention [6] for sequence component transformation across Delta – Star
transformer,

1P = j 1S
= 0.986576.6

2 P = - j 2 S
= 0.0597-0.4

0 P = 0

Current on primary side are worked out as follows:

RP = 0P + 1P + 2P = 1.001673.3

YP = 0P + a2 1P + a 2P = 0.9295-42.3

BP = 0P + a 1P + a2 2P = 1.031-161.1

Average value AVEP = (1.0016 + 0.9295 + 1.031) / 3 = 0.9874

UNBP = 5.9%

Almost 40% reduction in current unbalance (10% to 5.9%) is obtained on delta side primarily
because the zero sequence component is trapped within delta. Refer Fig 26 for current distribution.

Fig 26

March 2017, IEEMA Journal, Page No. 65 to 76

Page 25 of 27
 We can now summarize the main reasons for choosing vector group of Distribution Transformers as
Delta – Star are as follows:

(a) Zero sequence isolation between primary and secondary is obtained. Ground fault on LV side
will be reflected only as phase to phase fault on HV side. Refer Cl 5.4.1 of [7]. Ground fault
relays even with sensitive setting on H V side will not operate inadvertently for faults on LV
side. This is very essential as faults on LV side in distribution system are large.

(b) Steep voltage dip on one phase of HV side is more evenly distributed among phases on LV
side. For example, for line to ground fault on HV side, voltage unbalance on HV side is 100%
and corresponding unbalance on LV side is 39%. Voltage dip experienced at consumer end
for upstream faults is less severe.

(c) In distribution system, ideal phase balancing is not possible. The unbalance in current on LV
side is reduced on reflected current on HV side due to presence of delta winding.

8.0 Conclusions

In this article, we have concentrated on transformer and its influence on power system under normal
and fault conditions. The major observations are as follows:
(a) Phase shift introduced by (Y-Δ) transformer cannot influence power transfer magnitude.
Otherwise just three intermediate (Y-Δ) transformers can introduce in sum 90 shift, thus
reaching stability limit which is untrue.
(b) Subtle difference between paralleling and synchronizing is explained. Critical remarks are made
on vector group selection.
(c) Fundamental concepts of AT balance of transformer have been explained supported by
experimental results. Significance of spatial AT unbalance is discussed in detail especially with
respect to withstand capability of transformer against dynamic forces.
(d) Confusion regarding specifying percentage impedance of 3 phase transformer and equivalent 3
x1 phase transformer has been clarified.
(e) The transformer acts like a ‘smoothing’ element against steep voltage dips on primary side and
unbalance currents on secondary side.

March 2017, IEEMA Journal, Page No. 65 to 76

Page 26 of 27
9.0 Reference

[1] “Zig Zag Transformer – Fault Current Distribution, Short Circuit testing and Single Phase
Loading”, K Rajamani and Bina Mitra, IEEEMA Journal, July 2013, pp 84 - 91.
[2] “Construction of Distribution Transformer – winding”, Crompton Greaves Brouchure
[3] “Transformer Engineering – design and Practice”, S V Kulkarni and S A Khaparde, Marcel
Dekker, 2004
[4] “Environment – Voltage dips and short interruptions on public electric power supply systems
with statistical measurement results”, IEC 61000-2-8, 2002
[5] “Power Quality Overview – Practical aspects”, K Rajamani, IEEEMA Journal, May 2016,
pp 73-78.
[6] “Elements of Power Systems Analysis “, Stevenson, W.D., McGraw Hill, New York, 1982.
[7] “Conceptual clarifications in Electrical Power Engineering – Part 1”, K Rajamani, IEEEMA
Journal, Aug 2016, pp 69 - 80.

March 2017, IEEMA Journal, Page No. 65 to 76

Page 27 of 27
 Conceptual
Clarifications in
Electrical Power
Engineering
Part-3
Dr K Rajamani,
Reliance Infrastructure Ltd., MUMBAI
(June 2017, IEEMA Journal, Page 80 to 92)
 Conceptual Clarifications in Electrical Power Engineering – Part 3
K Rajamani, Reliance infrastructure Ltd

“Simplicity is an exact medium between too little and too much. “ —

Joshua Reynold, Discourses on Art

1.0 Introduction

In Part 2, some aspects of transformer operation were clarified. The present article is focused on
induction motor. The practicing engineer is faced with curse of ‘information over-load’ on this topic
with conflicting suggestions in some cases. The aim of this article is cut through the plethora of
information and offer easy to understand resolution of problems faced by design and field
engineers. Wherever required, actual field measurements are presented to support the theory.

The topics covered in this article are:

(a) Voltage dip during HT and LT motor starting
(b) Locked rotor protection
(c) High impedance differential protection
(d) Application of surge arrestor for VCB switched motor
(e) Service Factor
(f) Special protection for wound rotor motor

2.0 Voltage Dip During HT Motor Starting

For estimating the voltage dip during HT motor starting, simple hand calculations as illustrated
below will suffice in most of the cases. Only in case the voltage dips by hand calculation exceeds
15%, detailed motor starting studies using software are warranted. Hand calculations give feel for
results which can be cross checked using software in critical cases if required.

June 2017, IEEMA Journal, Page No.80 to 92

Page 2 of 35
2.1 Approximate evaluation of voltage dip during motor starting

Refer Fig 1.

Fig. 1

Expressing on common Base MVA

Ι =
+

= Ι =
+

1
=
1+

1
=
1+

= =

June 2017, IEEMA Journal, Page No.80 to 92

Page 3 of 35
2.1.1 Case Studies

Case 1:

UAT (Unit Auxiliary Transformer): 25MVA, 20 / 6.6 kV, 10.57%

BFP Motor: 5.6MW, 6.6kV
Full Load Efficiency = 0.968
Full Load PF = 0.85
Starting current = 450%

25
= = 236.5184
0.1057
5.6
= = 6.806
0.968 0.85

Starting MVA = 4.5 x 6.806 = 30.6271 MVA

1
=
1+
V+ = Bus Voltage after motor switching
VO = Bus voltage before motor is switched.
C = Factor

30.6271
= = 0.1295
236.5184

Assuming VO = 1.0pu
1
= 1. 0 = 0.8853
1 + 0.1295

The estimated dip is 11.5%

HT motors are designed to start with 80% voltage. Since estimated dip is much lower than
allowable dip of 20%, further studies using software are not required.

June 2017, IEEMA Journal, Page No.80 to 92

Page 4 of 35
 Case 2:

UAT: 50MVA, 21 / 11 KV, 10%

BFP Motor: 17MW, 11KV
= 0.95
PF = 0.9
Starting current = 550%

17
= = 19.9
0.95 0.9

Starting MVA = 5.5 x 19.9

= 109MVA

50
= = 500
0.1

109
= = 0.22
500

Assuming, pre-switching voltage as 100%

1
= 1. 0 = 0.82
1 + 0.22

The estimated dip is 18% which is close to permissible limit of 20%. In this case, verification using
software is desirable to confirm adequacy of system design.

2.2 Remarks on starting current and bus voltage dip

Current, bus voltage and speed during motor acceleration obtained using simulation are shown in
Fig 2. Stator transients are ignored in the simulation. It must be emphasised that starting current
(550%) remains at the high value and sharply falls down to normal value (below100%) only after
the motor speed has attained 90% speed. Consequently, the bus voltage also dips during the
entire starting period. If the starting current gradually decreases from high value to low value as the
motor accelerates, as depicted in dashed curve, bus voltage also would have recovered quickly as
the motor picks up speed.
But this does not happen as starting current characteristics does not follow the dashed line
resulting in prolonged voltage dip.

June 2017, IEEMA Journal, Page No.80 to 92

Page 5 of 35
 Fig. 2

2.3 Site measurements of voltage dip during motor starting

Starting current and bus voltage wave forms were captured at site during starting of largest motor
(BFP) in power plants. Motor details are shown in Table 1 and captured waveforms are shown in
Fig 3 to 5. Voltage and current in Fig 3 are instantaneous values whilst in Fig 4 & Fig 5 are RMS
values. In Fig 4 records up to 1.6 seconds registered in the relay are only shown though the actual
starting time is 7 seconds. The transients in starting current last for about 200 msec after
switching. All the records clearly show that the voltage dip exists during the entire starting period.
The author is indebted to Mahesh Bhadoria, Gouni Reddy, Alok Uppal and Kini Venkatesh for
sharing BFP starting characteristics recorded at different sites.

Unit Size Motor Data Voltage Starting

Fig No.
MW VRAT (kV) PRAT (MW) RAT (A) Dip (%) time (Sec)

250 6.6 9 922 17 4.9 Fig 3

660 11 17 1005 18 7.0 Fig 4
800 11 18.1 1088 17 8.5 Fig 5

Table 1

June 2017, IEEMA Journal, Page No.80 to 92

Page 6 of 35
 Fig. 3

Fig. 4

June 2017, IEEMA Journal, Page No.80 to 92

Page 7 of 35
 Fig. 5

2.4 Software Simulation.

In the software (NEPLAN, ETAP, PSSE, etc), complete auxiliary system including generators,
transformers, motors already running on the bus and motor to be started, cable between the bus
and motor, etc. are represented. When starting current is given as 500% in motor data sheet, it is
on the assumption that motor is started on ‘infinite bus’. This implies that terminal voltage of bus is
unaffected by motor starting MVA (mostly reactive) and starting current of 500% is maintained
throughout the starting period. In practice, the fault level of bus is finite (500 MVA in case 2 of Cl
2.1.1). Immediately after switching in motor, if bus voltage dips to 82%, the starting current reduces
almost proportionately, i.e. 5 x 0.82 = 4.1pu. This will result in slight recovery of bus voltage.
Simultaneously, those motors which are already running on the bus will try to draw increased
current at reduced voltage to maintain same power. This will try to depress the voltage. All these
effects are captured in software which has motor starting dynamics module. But as stated
previously, it is recommended to do ‘order of magnitude’ calculations by hand and proceed for
software simulation only in critical cases. Only if dip by hand calculation exceeds a critical value
detailed simulation using software is warranted. Doing motor starting dynamics simulation using
software for trivial cases (like starting 2MW motor on 25MVA transformer) gives a ‘perceived sense
of accuracy’ but does not have much practical value addition.

June 2017, IEEMA Journal, Page No.80 to 92

Page 8 of 35
3.0 Voltage Drop during LT Motor Starting

Previous section dealt with voltage dip during starting of big MV motors. Voltage dip during starting
of large LV (415V) motors connected by long cable is covered in this section. If voltage dip during
starting is excessive, two options are available to reduce the dip: increase the cable size or
increase the number of runs. With an example, we will illustrate why later option is preferred.

3.1 Analytical expression for voltage drop

Consider an LT motor fed from MCC. Refer Fig 6.

Fig 6

The phasor diagram is shown in Fig 7.

Fig 7

June 2017, IEEMA Journal, Page No.80 to 92

Page 9 of 35
 Sending end voltage (MCC) - VS

Receiving end motor terminal voltage - VR

Cable impedance - R + jX
Current -  -θ

VS2 = (VR + Rcosθ + Xsinθ)2 + (Xcosθ – Rsin θ)2 … (1)

After simplification,
VR2 + 2 (R2+ X2) + 2VR (Rcosθ + Xsinθ) - VS2 = 0 … (2)

Solving for VR in the above quadratic equation,

2
VR = -  (Rcosθ + Xsinθ) + V S2 - I 2(Rsinθ - Xcosθ) … (3)

3.2 Case Studies

The system parameters for the example are given below:

MCC bus voltage: Line Voltage 415V; Phase Voltage VS = 240V

Running load current of motor RUN = 166A at 0.85 pf

Starting current of motor STA = 1000A at 0.2 pf

Cable length – 250M from MCC to motor

Case 1: Size of cable connecting MCC to motor – 1 x 3C x 120mm2 Al

The cable parameters are: R = 0.323Ω/KM; X = 0.0712Ω/KM

The motor terminal voltage for running and starting conditions, evaluated using Eqn (3),
is given in Table 2. The terminal voltage during starting is 80.9%, almost 19% dip in
voltage.

Case 2: To improve the motor terminal voltage during starting conditions, let the cable size be
doubled.

Cable Size – 1 x 3C x 240mm2 Al

The cable parameters are: R = 0.161Ω/KM; X = 0.0710Ω/KM
The terminal voltage during starting is 88.3%, though better than Case 1, the
improvement is not substantial.

June 2017, IEEMA Journal, Page No.80 to 92

Page 10 of 35
 Case 3: Instead of doubling the size, consider two runs of cable.
Cable Size – 2 x 3C x 120mm2 Al
The terminal voltage during starting is 96.2%, much higher than in Case 2. The
resulting dip is only 4%.

Voltage at Motor end (%)

Case No Cable size
Starting Running Voltage drop
Case 1 1 x 3C x 120 80.9 94.6  (R + jX)

R
Case 2 1 x 3C x 240 88.3 97.0 ( + jX)
2

Case 3 2 x 3C x 120 96.2 98.6 (R + jX)
2

Table 2

The value of R reduces almost linearly with increase in size. When size (A) is doubled, R
decreases by half (R = ℓ/A). However, X almost remains the same for wide range in size from
25mm2 to 500mm2. Hence X drop is nearly same irrespective of cable size. Moreover X drop
adds almost algebraically with voltage in case current drawn is at poor power factor like during
motor starting condition. Refer Fig 8 of [1].

In case of multi-run cables, X drop decreases linearly as per number of runs resulting in significant
improvement in voltage profile.

Summarizing, if voltage drop is excessive during motor starting conditions, increase the number of
runs rather than the cable size.

4.0 Locked rotor condition

This is also referred as stalling condition. Motor stalls during running or unable to accelerate
because of excessive load, under voltage, single phasing, mechanical jamming, etc. When the
motor stalls, the stalling current or locked rotor current is almost equal to starting current. Even
though current during starting and stalling are almost same, there is a subtle difference between
the two conditions from thermal stress point of view. In case of starting, once the motor has picked
up speed, cooling fan is on whilst under stalled condition there is no cooling as summarized in
Table 3.

June 2017, IEEMA Journal, Page No.80 to 92

Page 11 of 35
 Description Starting Stalling (Locked Rotor)
Current 6 N 6 N
Motor Cooling Fan Yes No

Table 3

During starting, temperature rise is about 3C per second as heat loss is proportional to 2R. i.e.,

36R.If starting time is 10 sec, even with cooling fan on, the temperature rise will be nearly 30C.
Under locked rotor condition, with no cooling, temperature rise will be unacceptable if prolonged
beyond a time. One of the data furnished by motor manufacturer is the ‘Locked Rotor Withstand
Time’, also referred as ‘Hot Safe Stall Withstand Time’ (THSST). If actual stall time is more than

THSST, the motor must be tripped to prevent thermal damage.

4.1 Pick up setting for locked rotor protection

Current magnitudes during different operating conditions are shown in Fig 2.

(a)   1.0pu – Normal loading condition

(b) 1.0pu <  < 1.5pu – Over load region

(c) 5.0pu <  < 6.0pu – During stalling condition; During starting condition, current practically
remains in this region till the speed reaches above 0.9pu
(d) 1.5pu <  < 5pu – Only during transients. Sustained motor operation in this region is not
practical. Either motor operates near normal speed or stalls.

Assume starting current is 6pu (600%). It is unwise to set the current pickup for Locked Rotor
Protection close to 6pu, say 5.5pu. Also, if stalling occurs under single phasing condition, the
stalling current is (3/2) times ‘normal’ stalling current, i.e., 5.2 pu (6 x 0.866). In this case if pickup
is set at 5.5pu, relay will not operate.

It is recommended to set current pickup as, say, 2pu (200%). Under starting or stalling condition,
the relay will positively pickup as the setting is well below the starting or stalling current of 6pu.If
the current is above 2pu for sustained period it is abnormal condition

June 2017, IEEMA Journal, Page No.80 to 92

Page 12 of 35
4.2 Handle to distinguish stalling condition from starting

The challenge is how to distinguish between ‘normal’ starting condition and stalling condition, since
current drawn is same order of magnitude in both cases. The situation is similar to that faced in
differential protection of transformer to distinguish between transformer inrush and internal fault. In
case of transformer differential protection, the handle used to distinguish between inrush and
internal fault is second harmonic component of current, which is high for inrush and low for internal
fault. In case of locked rotor protection for motor, one of the handles used to distinguish between
starting condition and stalling condition is time.

4.3 Locked rotor protection based on current and time

For example, assume hot safe stall withstand time of motor is 20 sec and starting time of motor is
10 sec. Set the stall element time delay above starting time of motor but less than safe stall
withstand time. In this example, time delay can be set at 11 sec. Every time motor starts, the
current element set at 2pu will pick up as starting current is 6pu. Under successful start condition,
the current falls below 1pu when speed reaches 0.9pu after about 8 to 9 sec and the current
element drops off. As soon as motor is switched on stall unit picks up but drops off after motor has
successful started.

In case motor speed does not rise and crawls even after expected starting of 10 sec, the stall unit
trips the motor after set time delay of 11 sec. If feasible, thermal element is set as backup to
Locked rotor protection. Refer Fig 8.

TST < TLR < TTH < THSST

TST : Motor Starting time at 80 % UN : 10 Sec

TLR : Stalling protection time delay : 11 Sec

TTH : Relay thermal element operating time at starting current corresponding to 100% UN : 17 Sec
(Back up to stalling protection)
THSST : Hot safe stall withstand time of motor : 20 Sec

Fig 8

June 2017, IEEMA Journal, Page No.80 to 92

Page 13 of 35
4.4 Locked rotor protection based on thermal stress measurement

Most of the numerical relays offer this protection. The principle of operation is based on thermal
stress calculation during starting / stalling condition. Instead of fixed current and time setting as in
Cl 4.3., here both current and time can vary and the rise in temperature is proportional to 2t.
Depending on fault level and motor rating, the voltage of bus during starting can vary which in turn
modifies starting current (Refer Cl 2.4). If starting current is high, staring time will be less and if
starting current is less, starting time will be high. Thermal stress under all scenarios is correctly
captured by monitoring 2t (Fig 9). Relay operates when thermal content set in the relay is

exceeded. The thermal content is set in terms of starting current (S) and starting time (TS).

Fig 9

The setting concept is explained with an example. The relevant motor data are given in Columns A
to D of Table 4. Column E gives 2t consumed during starting. Column F gives 2t thermal

withstand capacity under stalling condition. Set 2t trip setting as 626 (599+653/2).

The set 2t (626) is greater than maximum of Colum E. Thus, the relay does not trip during starting,
thereby permitting successful start at all voltage levels.

The set 2t (626) is less than minimum of Colum F. Thus, the relay protects the motor during
stalling at all voltage levels.

Final Locked rotor setting: S = 6 and TS = 17 resulting in S2TS = 612, close to desired value.
Bina Mitra was instrumental in formalizing this approach and implementing the same at various
sites.

June 2017, IEEMA Journal, Page No.80 to 92

Page 14 of 35
 A B C D E F
Manufacturer’s Data
2 x t1 2 x t2
Starting current Stall withstand consumed during withstand
Voltage Starting
in terms of time starting during stalling
Time
multiple of FLC, in hot condition (CoI B)2 x Col C (CoI B)2 x Col D
t1 in secs
 t2 in secs
80% 4.8 26 32 599 737

100% 6 14 19 504 684

110% 6.6 11 15 479 653

Table 4

5.0 High Impedance Differential Protection of MV motors

Differential protection is a high-speed protection provided for clearing internal faults in stator. It
does not respond to faults in rotor. It is typically provided for MV motors (3.3, 6.6, 11kV) rated
above 2MW. To implement this protection, windings on neutral side must be brought outside to
neutral side terminal box. The conceptual differences between differential protection of motor and
transformer / generator are elaborated in Cl 3.2 of Ref [2]. In any general differential protection
scheme, the major concern is inadvertent operation of scheme during through fault or energization.
Through fault stability is not applicable for motor. In case of motor, the KPV (Knee Point Voltage) of
CTs used for differential protection and Stabilizing Resistor value are based on starting current of
motor rather than system fault current which is much higher. In case of transformer energization,
the inrush current flows on only one side of protected object. In case of motor the starting current
flows on both sides of protected object ensuring stability. Non-operation of any differential scheme
for internal fault has never been an issue.

5.1 Sample Calculations

This is illustrated with a detailed workout for two motors (one a very large motor and the other a
relatively small motor). Refer Table 5 and Fig 10.

Fig.10

June 2017, IEEMA Journal, Page No.80 to 92

Page 15 of 35
Item Name BFP PA Fan

1 Output (MW) 17 3.4

2 Rated Voltage (kV) 11 11

3 Rated Current (A) 1000 233

4 Starting Current (A) - S 4500 1398

5 CTR - CT Ratio (N) 1500/1 300/1

6 VK - KPV (Volts) 410 165

7 RCT () 10 2.5

8 2RL () 3 3

9 Voltage developed during motor (4500/1500) x (10+3) = 39 (1398/300) x (2.5+3) = 26

starting - VST

10 P - Pickup 0.05 N 0.1 N

11 RSTAB () VST / P = 39 / 0.05 = 782 VST / P = 26 / 0.1 = 260

12 FP - 3 Fault Current(kA) 26 26

13 FS – Reflected Fault Current = 17.4 87

FP / CTR

14 VF - Voltage developed across 17.4 x (10 + 3 + 782) = 87 x (2.5+3 +260)

CT during 3 phase internal fault 13,827
= 22,916

15 Peak voltage developed across 2 x Sqrt[2VK(VF – VK)] 2 x Sqrt[2VK(VF – VK)]

CT considering saturation
= 6639 = 7012

16 Permissible current to limit 3000/(RSTAB + RCT + 2RL) = 3000/(RSTAB + RCT + 2RL)

voltage below 3kV 3.8 (21% of FS) = 11.4 (13% of FS)

Table - 5

Most of the expressions are self-explanatory. A few remarks are made:

Items (1) to (7): As per manufacturer data

Item (8): Based on 200M to & fro lead length, 2.5mm2 Cu wires.

June 2017, IEEMA Journal, Page No.80 to 92

Page 16 of 35
 Item (9): KPV requirement is evaluated using starting current value.
Voltage developed across CT during motor starting = VST = (S / CTR) x (RCT + 2RL)
Item (10): The ground fault current in MV system is typically limited to 400A using NGR. The
pickup setting in case of BFP is 75A and PA Fan is 30A, much lower than ground fault level of
400A.
Item (14): In case of internal three phase fault, voltage developed across CT-
VF = FS x (RCT + 2RL + RSTAB)
Item (15): As per Alstom Application Guide Cl 16.19.2.3 of Ref [3]

Item (16): For relay circuit on CT secondary side, limiting voltage is fixed as 3kV. In case of internal
fault, current is forced into the relay branch through stabilising resistor. For PA Fan, even if 87% of
reflected fault current is consumed by CT due to saturation and only 13% is fed into burden (relay
branch), the voltage across relay branch will reach 3kV. The corresponding figures for BFP are
79% and 21%. In practice, CT output to burden is expected to be higher than the limiting value of
just 13% and 21%. Also, CT takes some time to saturate and before this time the CT output to
burden will be even higher.

Considering the above points, it is accepted practice in industry to provide metrosil (non-linear
resistor) to limit the voltage across relay branch for all motor feeders that employ high impedance
scheme for differential protection.

5.2 Summary of procedure

Following are salient points to be considered for high impedance differential protection for HT
motors in Auxiliary System of power plants:

Step 1: Find Voltage developed across CT during motor starting

VST = (S / CTR) x (RCT + 2RL)

Step 2: The calculated value of VST will be rather small. Hence select Minimum Knee Point Voltage
of CT liberally, say VK > 5 to 10 times VST

Step 3: Set the pickup for relay (P). Since the system is usually resistance grounded to limit
ground fault current to, say 400A, pickup value can be 5 to 10% of N. Usually BFP is the
largest motor with rated current of 500A to 1000A. In this case, pick up is about 50A
compared to earth fault current of 400A. Achieved sensitivity is acceptable. For other
motors of lesser rating, sensitivity is not an issue as the CT ratio is much less.

June 2017, IEEMA Journal, Page No.80 to 92

Page 17 of 35
 Step 4: Find value of stabilizing resistor
RSTAB = VST / P.

Step 5: During internal three phase fault, high voltage (above 3kV) will develop irrespective of
motor size. In case of smaller motors, CT ratio is small, reflected fault current is high and
stabilizing resistor value will be less. In case of bigger motors, CT ratio is high, reflected
fault current is not high but stabilizing resistor value will be large. Hence in all cases, as a
routine practice, it is recommended to provide metrosil.
For theoretical completeness, following calculations are done:
Voltage developed across CT during 3 phase internal fault,
VF = (F / CTR) x (RCT + 2RL + RSTAB)

Peak voltage developed across CT considering saturation

= 2 2 ( − )

The above value will be generally higher than 3kV. Metrosil is provided across the
stabilizing resistor and relay to limit the voltage to within 3 kV.

Step 6: To avoid spurious tripping, time delay of 50msec is recommended.

S N Misal contributed to make the above perspicuous explanation possible.

5.4 Backup to Differential Protection

The details are given in Table 6. It may be noted that for ground fault, the fault current is too low
(400A) for phase over current element (1 >) to pick up.

Fault Type Differential Phase O/C (1>) Ground O/C (0>)

Phase Fault Primary Back up ____

Earth Fault Primary Does not pick up Back up

Table 6

June 2017, IEEMA Journal, Page No.80 to 92

Page 18 of 35
5.5 Remarks on Phase side and Neutral side CTs

Phase side CTs and Motor Protection Relay that includes differential protection are located in MV
Switchgear. Neutral side CTs are located in Neutral Terminal Box of motor. Many times, it is over-
emphasized that neutral side CT and phase side CT shall have identical excitation characteristic
‘point by point’. In extreme, motor manufacturer is forced to procure neutral side CT from same
vendor who has supplied phase side CT. This over emphasis is not called for as explained below:
KPV (VK) is relevant during fault conditions so that CT develops sufficient voltage in presence of
saturation to drive the current through connected burden. Excitation Current (EX) is relevant during
normal operating condition. In current comparison scheme like differential protection, the errors
from CTs on both sides of object should not exceed pick up setting of differential relay during
normal operating condition. Typically, EX < 30 mA at VK/2.

From Cl 5.2 – Step 2, assuming S = 6RAT

VK = KPV  5 (S / CTR) x (RCT + 2RL)
= 30 (RAT / CTR) x (RCT + 2RL)

Voltage developed across CT during normal operating condition,

VNOR = (RAT / CTR) x (RCT + 2RL)
= VK / 30

If we assume the actual tested value of EX is nearly 30 mA at VK/2, the excitation current will be
very small at V K/30 (Fig 11). Even if values of EX are slightly different for the phase side CT and
neutral side CT, they are too small to have any adverse effect on operation of differential relay
under normal operating condition.

Fig. 11

June 2017, IEEMA Journal, Page No.80 to 92

Page 19 of 35
 In conclusion, it is sufficient to specify excitation current in conventional way, say EX < 30 mA at
VK/2. It is not mandatory that phase side CT and neutral side CT shall have ‘identical excitation
characteristics’ and also need not be procured from same vendor.

5.6 Differential Protection of MV motors using CBCT

Some manufacturers (e.g. Hitachi) offer this feature. The winding from neutral side is again brought
towards phase side and neutral is formed in Terminal Box on phase side (Fig 12). CBCT encloses
phase side and neutral side stator conductor. Under normal or starting conditions, currents in two
conductors within CBCT flow in opposite direction and net flux is zero. CBCT output is nil. In case
of internal fault CBCT output is nonzero and DMT relay connected to CBCT picks up. Typical
CBCT ratio is 50/1 irrespective of motor size. Thus, the scheme is akin to differential protection.
This requires special design of Terminal Box and agreement between user and vendor is required
in the design stage itself.

In passing, it may be mentioned that this terminal box arrangement is ideally suited for installing
High Sensitivity differential Current Transformer (HSCT) used for measuring C and tan of winding
as part of on-line health monitoring [4].

Fig. 12

6 Protection of MV induction motors against switching surges

6.1 Motor Insulation characteristics

The rated voltages of motors under discussion are 3.3kV, 6.6kV and 11kV and controlled by VCBs
(Vacuum Circuit Breaker). The impulse voltage withstand characteristics of rotating equipment like
motor is compared against other equipment in Table 7. Refer [5] & [6]. Since the motor winding
must be placed within the confined slot space, its BIL is lower compared to other equipment. This

June 2017, IEEMA Journal, Page No.80 to 92

Page 20 of 35
is an important difference to be noticed. For motor, front time of 1.2sec is termed as LIWV
(Lightning Impulse Withstand Voltage) and front time of 0.2sec is termed as SFIWV (Steep Front
Impulse Withstand Voltage). Here lightning is used in generic sense and does not mean the origin
of surge has to be lightning but refers to any surge with a front time close to 1.2sec.

Insulation withstand kVpeak (pu)

Rated
Others Motor
Voltage
LIWV LIWV SFIWV
(kVRMS)
Front Time: 1.2sec Front Time: 1.2sec Front Time: 0.2sec
3.3 20 (7.4) 18 (6.7) 12 (4.5)
6.6 60 (11.1) 31 (5.8) 20 ((3.7)
11 75 (8.4) 49 (5.5) 32 (3.6)

Table 7

The stator winding of each phase is made up a number of formed coils connected in series. Each
formed coil is made up a number of turns of conductor usually rectangular in shape. Typical
number of coils per phase is 20 and turns per coil could be between 5 to 20 depending on voltage
rating.

Two terms are frequently used when specifying insulation withstand strength of stator winding –
Ground wall insulation and Inter-turn Insulation.

Ground wall insulation refers to withstand strength between conductor and steel slot in which
conductor rests. The deciding factor is BIL corresponding to LIWV. Usually this is easily met for all
modern motors.

BIL corresponding to SFIWV corresponds to inter turn insulation. When a fast front surge
approaches the motor, the maximum stress appears on the first few turns of entry coil near phase
terminal. Under this condition, turn to turn insulation failure should not occur. Surge protection
device, if employed, is mainly for limiting the surge voltage within SFIWV. Most of the discussions
in the sequel center around limiting the fast front surge.

The tail time (e.g. time to reach 50% of specified amplitude) for surge is omitted in the above
discussions. It must be emphasized that amplitude and front time are deciding factors and large
variation in tail time does not have much impact.

If inter turn fault occurs, it is very difficult to identify by monitoring quantities from motor terminal.
Locally the current within shorted turn can be very high but may not lead to noticeable change in
terminal current. The local heating gradually damages the insulation and will finally lead to ground

June 2017, IEEMA Journal, Page No.80 to 92

Page 21 of 35
 wall insulation failure. The situation is very similar to inter turn fault in transformer where the only
clue for identification can be either Buchholz operation due to gas formation because of local
heating or changes in online DGA parameters monitored if available.

6.2 VCB Switching and Surge Arrestor Requirement

The source of steep front surge in motor application is VCB switching operations. Modern MV
switchgears at 3.3kV, 6.6kV and 11kV mostly employ VCBs. The current chopping level of modern
VCB using copper – chromium contact material is less than 5A. Of course, the level of chopping
current is dependent on load or fault current flowing through VCB. In case of high load or fault
current, the chopping current is practically zero. In case of breaking low currents, the chopping
current is higher due to instability of arc [7].

Consider the case when VCB breaks the current of a normally running induction motor in say 100
msec. The back emf of running motor during this time is substantial as open circuit time constant of
motor is of the order of couple of seconds. Refer Cl 8.2 of [8]. Thus, when VCB contacts open the
voltage across the breaker contacts is minimum due to presence of significant voltage on load
side. Under this condition, probability of restrike is practically nil.

Consider another case when VCB trips either during staring or under stalled condition. Under both
the conditions, back emf of motor is very low. Since the load side voltage is very low, voltage
across break contacts (TRV) can be substantial to initiate multiple restrikes. This generates steep
front over voltages that can endanger inter-turn insulation of first coil of motor.
As per industry experience cut off current is 600A. Refer Cl 9.7 of [9] and [10]. If the breaking
current is less than 600A, there is a possibility of multiple restrikes. If the breaking current is more
than 600A VCB can satisfactorily break without restrike.

Assume START = STALL = 5.5 RAT (550%)

If cut off current limit is 600A (starting or stalling current),
RAT = 600 / 5.5 = 109A
Assume  (efficiency) = 0.95 and pf (power factor) = 0.9
Cut off power rating  1.732 x V x 109 x 0.95 x 0.9 = 161 x V
For 3.3kV motor, P  531KW
For 6.6kV motor, P  1062KW
For 11kV motor, P  1771KW
Rounding off, following cut off values are suggested:
For 3.3kV motor, P  600KW

June 2017, IEEMA Journal, Page No.80 to 92

Page 22 of 35
 For 6.6kV motor, P  1000KW
For 11kV motor, P  2000KW

For motors rated above cut off value, no additional surge protection equipment is required, and
inherent motor insulation is adequate to protect against steep front surges. For motors rated below
cut off value, surge arrestor is recommended.
In this context, it is pertinent to discuss about surge impedance of motor. It is given (approximately)
by following formula, Eqn A.2 of Ref [11]:

ZM = 200 x (kV)0.32 X (kHP) -0..64

Surge impedance against Rating for the three voltage levels are shown in Fig 13. The surge
impedance is very low for motors of higher rating and is substantially higher for motors of smaller
ratings.

Fig.13

Assume a steep front surge enters from VCB into the connecting cable to motor. The Surge
impedance of cable (ZC) is typically 30. The magnitude of surge entering the motor is given by
(Fig.14):

Fig.14

June 2017, IEEMA Journal, Page No.80 to 92

Page 23 of 35
 2
=
1+

For high capacity motors, ZM is small and amplitude of transmitted wave is less.
For small capacity motors, ZM is higher and amplitude of transmitted wave is also higher and can
reach almost twice that of incoming surge.

This is another reason the surge arrestor is required only for motors of smaller capacity.

6.3 Recommendation for practical implementation at site

In the last thirty years, technology of VCB manufacture has dramatically improved with superior
contact materials. Also, there is concomitant improvement in insulation systems of stator coils of
motors. The old apprehensions that existed when VCBs were introduced for motor duty
applications are carried for too long and surge arrestors are specified as de-facto standard
irrespective of motor size. But in majority of motor applications surge arrestors may not be required
and if provided only increases unreliability. Surge arrestor failure under normal running condition is
not uncommon and this creates bus fault resulting in flow of large fault current. Also, that particular
feeder is temporarily out of service even though connecting cable and motor are healthy. Instead of
eliminating surge arrestor altogether, we however suggest a more moderate approach in
application of surge arrestors when motors are controlled by VCBs.

Our recommendations are summarized below:

1. Surge arrestors are recommended for following ratings switched by VCBs:

For 3.3kV motor, P  600KW

For 6.6kV motor, P  1000KW
For 11kV motor, P  2000KW

2. Surge arrestors are not needed for motors switched by Vacuum Contactors.

3. The insulation system of stator coils shall strictly conform to [6]. Two main tests to be performed
on sample coil are (i) impulse test on inter-turn insulation as per SFIWV in Table 7 and (ii)
impulse test for ground wall insulation as per LIWV in Table 7. Though standards allow power
frequency withstand test as an alternative for (ii), user should prefer only LIWV.

June 2017, IEEMA Journal, Page No.80 to 92

Page 24 of 35
 4. The wound stator before impregnation must undergo surge comparison test to positively confirm
absence of turn to turn fault. Details of surge test and nuances in interpreting the results are
given in [12].

5. If 3C cables are used, the armour shall be bonded at both the ends (switchgear end and motor
end). This is irrespective of motor size.

6. If single core cable is used, armour shall be bonded only at motor end. This can substantially
reduce magnitude of steep front surge impinged on motor. Refer Cl 6.2(f) of [11]. This is
irrespective of motor size. The ‘conventional wisdom’ is to earth the armour of single core cable
at switchgear end but in case of motor, it is preferred to earth only at motor end.

7. The ideal location for surge arrestor will be very near to motor terminal. However, in majority of
cases the arrestor is bought as part of switchgear and located at switchgear end. Thus, the
location of arrestor itself casts some doubt about the effectiveness of arrestor to limit the surge
voltage at motor terminal to the desired extent. But having decided to locate the arrestor at
switchgear end, it is desirable to select the arrestor that will give adequate protective margin
against steep front voltages. The deciding criterion is the residual voltage offered by surge
arrestor for steep front impulse voltage. Steep front surge is the most onerous one that leads to
inter-turn fault. When selecting surge arrestor, residual voltage for conventional 8/20sec
discharge current of 5kA shall be less than LIWV of motor to give adequate protective margin.
This is easily satisfied and corresponds to ground wall insulation. In addition, residual voltage
for steep front current of 5kA with 1sec front time shall not be more than SFIWV of motor. This
will hopefully minimize probability of inter-turn failure.
Interactions with Rahul Gosain greatly benefitted the author in understanding insulation
characteristic of HT motors. Amol Salunkhe provided clarifications on many aspects of VCB
switching transients and Surge Arrestor characteristics.

6.4 Illustration for surge arrestor selection

Motor Rating: 6.6 kV

From Table 7,

Lightning Impulse Withstand Voltage (LIWV) = 31kV

Steep Front Impulse Withstand Voltage (SFIWV) = 20kV

June 2017, IEEMA Journal, Page No.80 to 92

Page 25 of 35
 Broad details of Surge Arrestor chosen:

1. Make: ABB
2. Type: MWK 06.
3. Rated Voltage - 7.5 kV RMS
4. COV - Continuous Operating Voltage - 6 kV RMS
5. Residual voltage for 8/20sec at 5 kA – 17.4 kV (<LIWV )
6. Residual voltage for steep front at 5 kA - 19.2 kV (< SFIWV )
7. Discharge class - 2

7.0 Loading, Insulation Class and Service Factor (SF)

7.1 Current Loading

The sizing of motors generally follows the following sequence:

(a) During design stage, process group estimates load requirement, adds 10 to 15% margin and
passes on the data to electrical group.

(b) Electrical group selects next higher standard size taking into account ambient conditions.

(c) Generally, this results in actual load current at site being on average about 80% or lower of
rated current. This is in broad agreement with actual measurements done at two different
power plant sites when the units were generating maximum rated power. Refer Tables 8 and 9
for sample readings.

(d) Thus, margin is already built in design stage as far as current loading is concerned.

June 2017, IEEMA Journal, Page No.80 to 92

Page 26 of 35
 Site 1: HT Motor loading – Unit Generation 300 MW

Rating Site measurement % Current

Sr. No Motor
KW KV Amps Current in Amps loading

1 BFP 5600 6.6 565 417 74

2 ID FAN 3050 6.6 313 187 60

3 PA FAN 2300 6.6 231 136 59

4 CW PUMP 1500 6.6 162 142 87

5 FD FAN 1120 6.6 112 31 28

6 CEP 1000 6.6 100 60 60

7 COAL MILL 560 6.6 61 37 61

8 ACW 350 6.6 39 36 92

Average 65

Table 8

June 2017, IEEMA Journal, Page No.80 to 92

Page 27 of 35
 Site 2: HT Motor loading – Unit Generation 660 MW

Rating Site measurement % Current

Sr. No Motor
KW KV Amps Current in Amps loading

1 MD BFP 17000 11 1005 914 91

2 ID FAN 4950 11 298 254 85

3 PA FAN 3300 11 199 150 75

4 CW PUMP 3600 11 252 233 92

5 FD FAN 1850 3.3 379 205 54

6 CEP 1250 3.3 252 216 86

7 COAL MILL 950 3.3 206 148 72

8 AIR COMPRESSOR 770 3.3 158 115 73

Average 78

Table 9

June 2017, IEEMA Journal, Page No.80 to 92

Page 28 of 35
7.2 Insulation Class
Both HT and LT motors are procured with Class F insulation (155C) but temperature rise is limited
as per Class B insulation (130V). This is usually termed as ‘F/B’. To understand implication of this
choice, refer Fig 15 which shows relationship between temperature and insulation life. Insulation
life is defined with base of 20,000 hours and tensile strength reducing to half its original virgin value
at specified temperature. Tensile strength will reduce by half if Class F material is maintained at
155C and Class B material is maintained at 130C for 20,000 hours. Also it can be observed that
life reduces by half for every 10C rise in temperature.

Fig. 15

Assume the cooling system is designed, with 10C margin, to limit the temperature to 120C. For
Class F insulation, expected life at 120C is 200,000 hours. At 6500 hours per year of operation,
Operating life = 200,000 / 6,500 = 31 years

Also, the margin obtained by choosing “F/B” instead of ‘B/B’ is illustrated here. At 120C, with class
B insulation, life is 40,000 hours. With Class F insulation, life is 200,000 hours. Thus, insulation life
is five times more with “F/B’ compared to ‘B/B’.

June 2017, IEEMA Journal, Page No.80 to 92

Page 29 of 35
7.3 Service Factor (SF)

 It specifies capacity of motor to withstand periodic over load conditions. It is legacy from NEMA
standards. A motor with SF of 1.0 operating for a prolonged period above full load at rated
ambient temperature will suffer insulation damage that will shorten operational life. A motor
with SF of 1.15 can work at 15% above rated power without immediate failure and for extended
and repeated periods (hours) but may suffer damage which shortens motor life.

 Specifying SF more than 1.0 is conceptually same as specifying ‘F/B’. With rated current,
cooling system is designed to limit temperature within 130C as per Class B insulation. With
over load of say 15%, corresponding to SF of 1.15, temperature will be limited within 155C as
per Class F insulation.

 In brief, if motor is designed for ‘F/B’ (Class F insulation with Class B temperature rise), there is
no need to assign any Service Factor and default value of 1.0 will suffice. Design margins in
current (Cl 7.1) and cooling / insulation life (Cl 7.2) ensure longer operating life of motor.

8.0 Wound Rotor Induction Motor

Wound rotor motors are used when high starting torque and reduced starting current are required.
External resistors in the three phases of rotor circuit cut in during starting. The resistor is gradually
cut out once the motor picks up speed. At full speed, external resistor is shorted. Typically starting
current (ST) is limited to 300% which would have been 600% without the resistor in rotor circuit. In
this section two aspect peculiar to wound rotor motor namely stalling protection and rotor open
circuit protection are discussed.

8.1 Difference between cage rotor and wound rotor characteristics

In case of cage rotor motor, sustained operation in current range of 150% to 500% is not practical
(Refer Cl 4.1). For wound rotor motor, operation in this range for significant time is possible. If
starting current is limited to 300%, current during the entire starting period will be nearly 300%. But
once the motor has started and rotor resistance shorted, sustained operation in current range of
150% to 500% is again not feasible.

June 2017, IEEMA Journal, Page No.80 to 92

Page 30 of 35
 During starting if stalling occurs, stalling current (LRS) will be limited to 300%. However, if stalling

occurs under running condition with external rotor resistor shorted, stalling current (LRR) will be
600% and not 300%.

a) In case of cage rotor, ST = LRS = LRR = 6pu (600%)

Transient Reactance X’ = 1 / ST = 0.16 pu

For terminal fault, motor contribution to fault = 1 / X’ = ST = 6pu (600%)

Negative sequence reactance X2 = 1 / ST = 0.16pu

b) In case of wound rotor, ST = 3pu (300%)

Transient Reactance X’  1 / ST  0.33pu

= 1 / LRR = 1 / 6 = 0.16pu

For terminal fault, motor contribution to fault = 1 / X’ = LRR = 6pu (600%)

Negative sequence reactance X2 = 1 / LRR = 0.16 pu

8.2 Stalling protection

In case of cage rotor, pick up for stalling protection (PU) is set at 200%. In case of wound rotor with
rotor resistance start, PU is set at, say 350%. If stalling occurs during running condition only,
stalling protection picks up. During starting, if stalling occurs, thermal element offers protection. It is
not that onerous as current is limited within 300%. Example of typical setting adopted is given
below:

a) Motor data
Starting current = 300%
Starting time = 15 sec
Locked rotor (Stalling) current = 600%
Hot Safe stall withstand time = 7 sec

b) Stall unit setting

Current pick up PU = 350%
Time delay = 6.5 sec
Refer Cl 4.3 for comparison with cage rotor.

June 2017, IEEMA Journal, Page No.80 to 92

Page 31 of 35
8.3 Open Circuited Rotor Phase

From stator side it appears as line to line fault (Fig 16). The stator current will have significant
negative sequence component.

Fig 16

If rotor is open at start, motor cannot accelerate beyond 50 % speed. Near 50% speed the
electrical torque developed by motor collapses (Fig 17) and this is called ‘Goerges Phenomenon’
[13].

Fig 17

June 2017, IEEMA Journal, Page No.80 to 92

Page 32 of 35
During running, if one of the rotor phases open, stator current oscillates (Fig 18). This makes it
difficult for the relay to issue trip command positively. Some incidents of motor damage have been
reported from sites where the relay has failed to pick up for this condition.

Fig 18

Two suggestions to improve positive tripping are given below:

a) Since current oscillates, conventional over current element (>) with DMT characteristics will
pick up and drop off periodically without initiating tripping. If reset time is instantaneous when
current falls below pickup value, the relay will reset immediately. Pick up and drop off will
occur continuously till fault becomes permanent during which time motor may be damaged.
To overcome this problem, numerical relays (e.g. MiCOM P141 to 145) now offer ‘timer hold’
facility. With the proper reset timer settings, it is possible to accumulate the current excursion
times and issue the trip command after the cumulative time has elapsed. For example, with
reference to Fig 18, the reset time can be set at 3 sec, and current pickup can be 130% with
time delay of 15 sec. The current pulse duration when current magnitude exceeds 130% is
integrated and when the accumulated value reaches 15sec, trip command is issued. To
prevent tripping during starting, either this element shall be bypassed during starting through
logic or time delay shall be more than starting time of motor. Some users prefer to wire this
protection for only alarm so that ordered manual shut down can be initiated from process
point of view.

b) Thermal element may act as a back up to over current element with ‘timer hold’ facility but
operating time is very uncertain. Current seen by thermal element is given by:

Ι = (Ι + KΙ )

June 2017, IEEMA Journal, Page No.80 to 92

Page 33 of 35
 Since negative sequence component of stator current is significant under rotor open circuit
condition, chances of TH pickup can be improved by choosing higher value of K, say 6 to 8 instead
of 3.

The author greatly benefited from suggestions offered by Bina Mitra on the above topic.

9.0 Conclusions

The major observations are as follows:

(a) Formula for ‘back of envelop’ calculations for estimating dip during starting of HT motor are
given. Only in case the simple hand calculations indicate dip above 15%, it is necessary to go
in for simulation using advanced software.

(b) In case of LT motors connected by long cable, voltage dip during motor starting can be high
under certain conditions. In these cases, to reduce starting voltage dip, it is recommended to
increase number of runs rather than increasing the cable size.

(c) Procedure for setting Locked rotor protection based on thermal stress evaluation is explained
with a practical example.

(d) Metrosil shall be provided in high impedance differential protection schemes irrespective of
motor size to limit secondary voltage within limits during an internal fault. It is not mandatory to
have identical excitation characteristics for phase side CT and neutral side CT. Differential
protection using CBCT is also possible.

(e) Recommendations for protection of MV motors controlled by VCB against steep front surges
are listed in Cl 6.3.

(f) There is no need to specify Service Factor for motors designed for ‘F/B’ (Class F insulation with
Class B temperature rise).

(g) Differences in locked rotor protection philosophy between cage rotor and wound rotor are
brought out. Rotor open circuit in case of wound rotor motor can go undetected due to
oscillating nature of current resulting in motor damage. By enabling ‘timer hold facility’, positive
pickup can be ensured.

June 2017, IEEMA Journal, Page No.80 to 92

Page 34 of 35
10.0 Reference

[1] “Conceptual Clarifications in Electrical Power Engineering – Part 1” K Rajamani, IEEMA

Journal, Aug 2016, pp 69-80

[2] “Selection of current transformer parameters for optimum design – User perspective”, K
Rajamani and Bina Mitra, Second International Conference on Instrument Transformers,
Jan 2010, IEEMA, Mumbai, Page II-8 to 13

[3] Alstom Network Protection and Automation Guide, 2011

[4] “Online Health Monitoring of Motor Insulation”, Prabhakar Neti and Brant Wilhelm,
General Electric Document, 2013

[5] “Insulation co-ordination – Part 1: Definitions, principles and rules”, IEC 60071-1, 2011

[6] “Impulse voltage withstand levels of rotating ac machines with form wound stator coils”,
IEC 60034-15, 1995

[7] “A Statistical Vacuum Circuit Breaker Model for Simulation of Transient Overvoltages”,
Janko Kosmac and Peter Zunko, IEEE Trans on Power delivery, Jan 1995, pp 294 – 300

[8] “Auto Changeover in Power Plants and Induction motor performance”, ”, K Rajamani and
Bina Mitra, IEEMA Journal, December 2016, pp 76 – 83

[9] “Application Guidelines - Dimensioning, testing and application of metal oxide surge
arresters in medium voltage systems”, ABB, 2009

[10] “Surge limiter application recommendations for metal-clad switchgear up to 15 kV“,

Siemens, 2009

[11] “IEEE Guide for the Application of Surge Voltage Protective Equipment on AC Rotating
Machinery 1000 V and Greater”, IEEE Std C62.21 – 2003

[12] “Current state of surge testing induction machines”, John Wilson, Baker Instrument
Company, Iris Rotating Machine Conference, June 2003, Santa Monica, CA

[13] “The Goerges Phenomenon – Induction Motors with Unbalanced Rotor Impedances”, H L
Garbarino and E T B Gross, AIEE Transactions, 1950, Vol 69, pp 1569 – 1575.

June 2017, IEEMA Journal, Page No.80 to 92

Page 35 of 35