Voltage Regulator and Parallel Operation

Generator sets are operated in parallel to improve fuel economy and reliability of the power
supply. Economy is improved with multiple paralleled generators by selecting only sufficient generators to carry the load demand at any given time. By operating each generator
near its full capacity, fuel is utilized efficiently.
Power system reliability is improved by availability of generators not in use as backup for
units on line. In addition, protective systems can be designed to detect a faulty element
and isolate it from the healthy part while maintaining power to the remaining system. With
multiple generators carrying the load, a generator that develops a problem can be shut
down, allowing the remaining generators to carry the load.
Because of advantages of parallel operation, multiple generator installation has become
common in applications for standby and prime power, portable and stationary power, commercial and military power; and they continue to grow.
Operating generators in parallel requires attention to the two control systems of the generator set - the voltage regulator and the speed governor. This discussion is limited to the
control of the voltage regulator.
Parallel Operation and the Voltage Regulator
An example to illustrate the effect of the voltage regulator on the generator system can be
seen using two batteries and a load. In Figure 1, two batteries with voltages exactly equal
(open circuit) could divide load equally, with each battery pulling its share. If the two voltages are not exactly equal, the load will not be shared equally, and the benefit of two batteries will be lost.

Figure 1: Paralleled Batteries - Equal Voltage

Further, the load division is very sensitive to voltage unbalance. In the following examples,
a small resistance is placed in series with the voltage source to represent its inherent internal impedance.
In Figure 2, the battery with an output voltage of 102V (B2) will determine the load voltage,
and the load sees a proportionally higher current. Battery B1 is being charged by B2 at a
9.5 ampere rate because its terminal voltage is lower. This charging current is obviously
not proportional to the voltage level and results in a load current on B2 of 10.5 Amperes.

Figure 2: Paralleled Batteries - Unequal Voltage

In Figure 3, B1 is higher in voltage than B2 by 2 volts, causing B1 to supply the load and
charge B2. In this scenario, B2 supplies 10.5A of load current.

Figure 3: Paralleled Batteries - Unequal Voltage

Please note that the voltage referred to in the diagrams is the open-circuit voltage. As
soon as the batteries are paralleled, the voltage at the terminals of the batteries becomes
identical.
Two generators operating in parallel to supply a common load operate in a similar manner.
If voltage (open-circuit) is exactly the same on both generators, they will divide the load
equally between them. See Figure 4. Any small difference between generator voltages will
result in unbalanced load division (Figure 5) or, in the extreme, circulating current (Figure 6).

Figure 4: Paralleled Generators Balanced Voltage

Figure 5: Paralleled Generators Voltage Unbalanced

Figure 6: Paralleled Generators - Voltage Unbalanced

In practice, the precise matching of voltages is not possible. Some means must be
provided to make load sharing between paralleled generators simple to control.
Because circulating current or load unbalance described above is the result of voltage
mismatch, you might look to the voltage regulator for some help. The regulator can provide
this help by using paralleling compensation circuits called reactive droop compensation or
reactive crosscurrent compensation. Both types will be described in detail in later papers.
The principle of operation of the reactive droop circuit, which is the simplest and most commonly used circuit, can be described by the curve in Figure 7.

Figure 7: Reactive Droop Curves

Using a regulator designed to maintain precise voltage regulation, a circuit is added that
accepts a current signal from the generators output. This current signal is combined with the
generators sensing voltage signal to develop a vector summed voltage proportional to
reactive load.
For example, if voltage decreases from 480 volts down to 458 voltage from no load to rated
reactive (kVar) load, the voltage droop is -4.3%. If two generators are operated in parallel,
with their droop curves set the same and their voltage set points adjusted to proportionally
share the reactive load, any imbalance that would increase the load on one machine and
decrease the load on the other would cause the droop circuits to change the voltage set
points in a direction to bring the load back into balance. With droop compensation, the bus
voltage will droop with changing reactive load.
3

Figure 8: AVR with Droop

By building this characteristic into the regulator of each generator operating in parallel,
sharing of the load is controllable and uncomplicated.
To parallel two generators, voltage on each machine should be matched prior to closing the
breaker to minimize the current surge at breaker closing.

Figure 9: Paralleled Generators - Balanced Voltage

With voltages balanced and 100 Amperes load, each generator will supply its share of the
load. If G2 voltage is increased, its output current will try to increase. This increase will
cause a voltage droop, which counteracts the voltage increase. G1 will see its output current decrease, resulting in a droop circuit action to increase voltage. The result is a load
balance control action that works to hold loading balanced when two or more generators
are operating in parallel.
Why Control Two Variables?
KW, KVaR, KVa
In the dc analogy with batteries in parallel, only the voltage had to be controlled to allow
for load sharing. When paralleling ac generators, torque and excitation both must be
controlled properly. Torque applied to the generator must be controlled in order to divide
real power (kilowatts); excitation to the generator must be controlled to divide the reactive
power.

Real power is work done by the electrical energy of the generator. This power is supplied
by the prime mover in the form of torque. The generator converts mechanical torque to
electrical energy. This energy is supplied to a load to be converted to the desired form of
energy, such as heat, light, mechanical energy (motor), etc.
Reactive power is that power required by loads with inductance or capacitance to store
energy on each half cycle. In the example of Figure 10, the load is purely resistive, such as
a heating element. Because no reactive power is required at any point on the sine wave,
current is directly proportional to the voltage at that point. The real power can be calculated
using Ohms Law.

Figure 10: Resistive Load

If an inductance (Figure 11) is connected to the generator output, the relationship between
voltage and current is different. The inductor will not allow current to flow without first building up flux in its magnetic circuit. Thus, the current through the inductor is proportional to
voltage, but the current lags the voltage by 90 electrical degrees (Figure 12).

Figure 11: Inductive Load

Figure 12: Current Lags Voltage

The capacitor, on the other hand, will not allow voltage across its terminals until some
charge has been deposited on its plates (Figure 13). For this reason, the current must flow
before a voltage can exist. Thus the capacitor current leads the voltage by 90 electrical
degrees (Figure 14).

Figure 13: Capacitive Load

Figure 14: Current Leads Voltage

For a load containing all three elements in parallel (Figure 15), it is convenient to calculate
the real or reactive power in each element separately. The Theory of Superposition (Figure
16) allows the addition of the three separate calculations vectorially to determine the total
load.

Figure 15: Combined RLC Load

Figure 16: Theory of Superposition

A look at I1 and I22 indicates that these currents are always opposite in polarity. If the currents
are equal, the net result is zero! For this reason, inductive loads are said to accept reactive
power, while capacitors are said to supply reactive power. Power Factor correction capacitors
use this principle to compensate for inductive loading. The reactive power flow is illustrated
in Figure 17.
6

Figure 17: LC Loads

Thus, the reactive power, expressed in Vars, can be computed by:
VARs =

V2
XL-XC

And the real power, expressed in watts, can be computed by:

V 2 = Watts
R
A third parameter can be computed by using the voltage of the generator and the current
IT in Figure 18. This parameter is known as apparent power.

Figure 18: Combined RLC Load

These three parameters, real, reactive, and apparent power, are related. Knowing any two,
the third can be calculated using the Pythagorean Theorem:

Or they may be calculated using vector arithmetic (Figure 19):

Figure 19A: Power Triangle

Figure 19B: Apparent Power - kVA

7

Metering is commonly provided to measure volts, amps and frequency for generators
operating alone. For parallel application, monitoring current gives an indication of loading,
but there is no indication of the type of current flow, making load sharing adjustment
impossible. A meter to monitor real power in addition to current will enable power to be
balanced by the governor adjustment, followed by balancing of current using voltage
regulator adjustment. A better scheme adds an additional meter for reactive power, the Var
meter. With a kW and kVar meter on each generator, optimal adjustment of the governor
and the voltage regulator is easily accomplished, and load sharing is easily monitored.
Another metering option is the power factor meter, often used instead of the kVar meter.
Power factor is the ratio of real power to apparent power:

P = REAL POWER, WATTS

S = APPARENT POWER, VOLT-AMPERES

Figure 19C: Reactive Power - kVar

Power factor, because it is affected by both real power and reactive power levels, is more
difficult to interpret for monitoring load sharing performance. Providing good instrumentation is very helpful for operating systems of paralleled generators.
The Voltage Regulator and Parallel Generator Operation
The function of the voltage regulator is to provide precise regulated generator voltage at no
load and with changing loads. When generators are connected together in parallel operation, a parallel compensation circuit is required to assist the voltage regulators in controlling the generator reactive loads.
Reactive loads between generators can become unbalanced when the voltage regulator
varies the excitation to the generator exciter field due to load changes, prime mover speed
variation, thermal drift, etc. This change in excitation may cause large circulating currents
to flow between generators (Figure 20).

Figure 20: Circulating Currents

This causes the generator with the higher field excitation to try and power the generator
with the lower field excitation in an effort to force the generator to have the same output
voltage as the generator with the higher excitation.
8

The parallel compensation circuit will cause the voltage regulator to increase the field excitation on the generator with the lower field excitation and decrease the field excitation on
the generator with the higher field excitation. By controlling the reactive load, the parallel
compensation circuit can eliminate the undesired circulating currents.

Figure 21: Equal Pressure - A

Unequal Pressure - B

An analogy of how one generator tries to power another generator with circulating currents
can be compared to two water pipes of equal diameter feeding into one pipe. When the two
water pipes have the same water pressure, both water pipes will be supplying the same
amount of water to the common water pipe (Figure 21A). If one water pipe suddenly lost a
small amount of pressure, the second water pipe would begin to supply more water to the
common pipe to help maintain the water flow (Figure 21B). Also, because the water pressure in the second water pipe is now greater than the first water pipe, water will begin to flow
from the second water pipe to the first water pipe in effort to force the water pressure to be
the same as the second pipe.
Reactive Droop Compensation and Reactive Differential Compensation
There are two forms of parallel compensation circuit. The most often used type of paralleling compensation is the parallel droop compensation or using the IEEE designation,
reactive droop compensation. The other type is crosscurrent compensation or, again using
IEEE terminology, reactive differential compensation.

Figure 22: Reactive Droop Compensation

9

When reactive droop compensation is used to parallel two or more generators, each
parallel droop circuit is independent of the other (Figure 22). A typical parallel droop circuit
is made up of a current transformer and paralleling module. The paralleling module consists
of a burden resistor and a switch connected across the primary of a transformer (Fig. 23).

Figure 23: Paralleling Module

A switch located on the primary of the transformer in the paralleling module is used to
short the secondary of the current transformer and the burden resistor to allow the
generator to operate independently of the paralleling generating system.
The secondary of the current transformer is connected to the paralleling circuit. Connection to the paralleling circuit places a burden resistor across the output terminals of the
current transformer (Figure 24). The secondary current of the current transformer induces
a voltage across the burden resistor that is vectorially added to the line voltage to produce
an error signal to the voltage regulator. Voltage across the burden resistor is proportional in
magnitude and has the same phase as the line current through the primary of the current
transformer.

Figure 24: Burden Resistor

10

Figure 25: Current Transformer Error Signal Development

The error signal produced across the current transformer burden resistor must be applied
so that when the load is a unity power factor load, no corrective signal will be produced and
no excitation change will occur because of the load. By arranging the burden resistor
voltage so that it will be 90 electrical degrees out of phase with the system voltage when
the power factor of the load on the generator is unity, an appropriate error signal can be
produced. Figure 25 shows the vector representation of system voltage and burden resistor voltage needed to produce the error signal.
The three phase voltages (line to neutral), regardless of their internal connections, produce
output voltages that are displaced from each other by 120 electrical degrees. By taking
advantage of the phase displacement, it is possible to sense the system voltage from line
to line and produce a line voltage that is displaced from the phase voltage by 30 electrical
degrees. The voltage signal from the current transformer burden resistor at unity power
factor (resistive loads) will be displaced by 90 electrical degrees from the system or line
voltage (Figure 26A).

11

Figure 26: Phase Displacement

Under unity power factor, the vector diagram (Figure 26B) representation of the burden
resistor voltage (V
) and the sensing voltage (V ) can be seen to be 90 electrical dePhase B
AC
grees apart. When a reactive load is applied to the generator, the burden resistor voltage
arrow will swing either clockwise or counterclockwise depending upon the type of load,
capacitive or inductive (Figure 26C).

Figure 27: Vector Diagram of Capacitive Load

If the generator load has a leading power factor (capacitive load), the vector diagram of the
burden resistor voltage will rotate counterclockwise from its unity power factor position
(Figure 27). The phase angle between the line voltage and burden resistor voltage will become more out of phase, which will decrease the regulator sensing voltage. The regulator
will receive a smaller sensing signal and thus increase generator excitation.

12

Figure 28: Vector Diagram of Inductive Load

When the load is a lagging power factor or inductive load, the vector diagram of the burden resistor voltage will be rotated clockwise from its unity power factor position (Figure
28). The phase angle between the line voltage and burden resistor voltage will become
more in phase and the voltage sensing signal to the regulator will be increased. The regulator will receive the larger signal and respond by decreasing generator excitation.
The parallel module will supply an error signal to the voltage regulator that will control the
excitation level of the generator. When the generators are paralleled, the regulator
responds to this small increase in sensing voltage by reducing excitation to the generator
field, which will produce a droop in the generator output voltage.
The amount a generator system will droop can be adjusted by the burden resistance and
the current transformer ratio. A typical burden resistor has a resistance of one ohm and
has an adjustable slide tap contact so that the voltage developed across the resistor can
be varied. The amount of error signal sent to the voltage regulator is proportional to the
magnitude of the voltage across the resistor and the vector angle of that voltage. The
amount of voltage across the burden resistor is determined by the secondary or output
current of the current transformer. Typical current transformers are designed for a 5 ampere secondary current with a 25VA maximum burden rating. Ohms Law (Voltage = Current X Resistance) will show the maximum voltage across the one ohm burden resistor to
be 5 volts. It can then be seen that the relative magnitude of the voltage developed across
the burden resistor with full load on the generator is typically 5 percent of the system output voltage (120/208-240/416-480/600) to minimize circulating currents.
When generators are paralleled with reactive droop compensation, most generators are
set to operate with maximum droop. The burden resistor is adjusted for maximum resistance or maximum voltage across the resistor. Allowing the generating system to operate
at maximum droop allows for the best control of circulating currents. Setting the burden
resistor for a system voltage droop of less than three percent may result in the regulator
and paralleling circuit being unable to control circulating currents satisfactorily.
Voltage regulators with single phase sensing provide approximately 8% maximum droop
while three phase sensing regulators provide approximately 6% droop. Single phase sensing provides greater droop because the average value of the error signal is proportionally
greater compared to the average value of the single phase sensing voltage than to the
average value of the three phase sensing voltage. When generators are paralleled on the
same bus and have different type sensing, care must be taken to compensate for sensing
differences using the adjustable burden resistor.
13

Because the burden resistance voltage is dependent on the line current of the generator
through the current transformer, any changes in power factor due to the load will be reflected on the burden resistor voltage. Consequently, when a reactive lagging power factor
load is increased, the bus voltage will droop by an increased amount. If a capacitive leading power factor load is increased, the bus voltage will increase. The magnitude of the
change depends upon the magnitude of the load and power factor.
In order to prevent the voltage from increasing or decreasing with the power factor of the
load, another circuit can be used where the current transformers of the individual regulators are interconnected. Crosscurrent compensation (reactive differential compensation)
allows operation in parallel without voltage droop caused by the error signal.
Figure 29 shows two generators paralleled with reactive differential compensation. Interconnection of the current transformers can be seen. On generator number one, the current
transformer (CT1) terminal with the polarity mark is connected to the current transformer
(CT2) on generator number two at the terminal with no polarity mark. On generator number two, the current transformer (CT2) terminal with the polarity mark is connected to the
current transformer (CT1) on generator number one at the terminal with no polarity mark.

Figure 29: Two Generators Paralleled with Reactive Differential Compensation

Even though the voltage involved is ac, a better understanding of the operation of the
closed crosscurrent loop can be obtained by using a dc voltage analysis. First of all, the
polarity marks on the generator line and current transformer indicate the current direction.
The line current always flows into the polarity mark on the generator line and the current
flows out of the polarity mark on the current transformer (Figure 30).

14

Figure 30: Burden Resistor

The current leaving the polarity marks of the current transformer is divided into two current paths (Figure 29). One circuit path flows through the burden resistor in the paralleling module. The second circuit path flows through the crosscurrent loop. The current that
flows in the crosscurrent loop will enter the burden resistor in the adjoining parallel circuit
and will oppose the current through the burden resistor set up by the paralleling circuits
own current transformer. The resulting current flow through the burden resistor in each
paralleling circuit will be zero because the opposing currents will cancel each other out.
Thus, there will be no voltage developed across the burden resistor and no droop associated with the line voltage.
If one generator begins to export more reactive power than the other generator, the line
current will increase and the current from the secondary of the current transformer will also
increase. The result will cause a greater voltage across the burden resistor of the paralleling circuit, which will cause the voltage regulator to reduce the excitation in that particular
generator, thus decreasing line current. An increase in current through the crosscurrent
connect loop caused by the imbalance of the first generator will develop a voltage across
the second generators paralleling burden resistor that is opposite in polarity to the normal
voltage developed by the second generators own current transformer. Instead of causing a droop in line voltage, the opposite polarity will cause an increase in line voltage. The
resulting increase in one generator and a decrease in the other generator will cause the
parallel generating system to balance itself out.
For the reactive differential compensation to perform properly, all of the paralleling current
transformers on all of the generators delivering power to the bus must be connected into
the crosscurrent loop. The current transformer connected in the loop must have the same
ratios so that each current transformer supplies the same amount of current to properly
cancel the voltage across the burden resistor. In the case where different size generators
are paralleled, current transformer ratios must be changed to give approximately the same
secondary current as the other current transformer(s). Otherwise, cancellation of the currents in the crosscurrent loop will not occur, and the imbalance of current will force the
generators to have circulating currents between them.
In addition to having the same current transformer ratios, all generators must have the
same burden resistor setting in the voltage regulator paralleling circuit. Having the same
burden resistor setting ensures that when an imbalance exists in the generators, the current that flows in the crosscurrent loop will setup a proportional voltage across each burden resistor that will balance the generator systems reactive load.

15

Figure 31: Reactive Differential Compensation Connection

In order to perform properly, the crosscurrent loop must have all the generators tied into
the loop. It is virtually impossible to parallel with an infinite bus (utility). If the crosscurrent
connected generating system was paralleled with the utility outside the crosscurrent loop,
a switch contact installed at any point in the loop may be opened, and all generators will
operate in droop mode.
For compliance with most safety codes, a ground connection is placed on the CT secondary. Only one ground can be installed on the loop, the current transformers will be shorted
out (Figure 31).
Some means of isolation between the regulator sensing circuit and the current transformer
circuit must be used with crosscurrent compensation circuits. Basler regulators and paralleling circuits include isolation transformers used to effectively isolate the crosscurrent loop
from the input voltage circuit of the regulator.
The Current Transformer Unit-Parallel Switch
A switch is placed on the secondary of the current transformer to short circuit the current
transformer and the burden resistor, which will negate any signal to the generator. The
switch in the Unit position allows the generator to operate independently of the parallel
generating system without the effect of the droop circuit.

16

When a generator is operating independently in a parallel droop system and the short
circuit switch is not in the Unit position, the generator will have an unwanted droop in the
generator output voltage. The same effect will happen in a crosscurrent system only the
droop will be smaller because other burden resistors in series act as a voltage divider to
decrease the amount of voltage proportionally.
When generators are operating in a crosscurrent loop and one generator is taken out
of parallel with the other generators to operate independently, the unit parallel switch is
important to maintaining the stability of the remaining generators paralleled. If the non-paralleled generator is disconnected from the line with the current transformer and the burden
resistor not shorted, the parallel system voltage will fluctuate. The generator not paralleled
will be rotating at a different speed and frequency compared to the paralleled generators. The current through the burden resistor of the nonparalleled generator will have a
constantly varying phase angle compared to the current through the paralleled generator
burden resistors. The constantly changing phase angle across the burden resistor current
of the non-paralleled generator will cause the non-paralleled regulator to alternately raise
and lower the excitation of the non-paralleled generator. This will produce a small periodic
change in excitation, which will increase and decrease that generators output voltage. The
rate at which the voltage will fluctuate is equal to the difference in frequency from the nonparalleled generator and the paralleled generators.

Figure 32: Unit - Parallel Switch

If a unit-parallel switch is to be added externally to a paralleling circuit, attention must be
paid to the distance the switch is to be placed from the current transformer secondary
terminals (Figure 32). If a considerable distance is required for connection of the switch,
resistance from the wire to and from the switch might be great enough to allow some of the
current to pass through the burden resistor. If a small current is allowed to flow through the
burden resistor, an equally small voltage will be produced across the burden resistor and
produce a small droop in the system voltage, if the system is paralleled with reactive droop
compensation. If so, reactive differential compensation is used to produce an effect similar
to that just described for the crosscurrent loop in the preceding paragraph. Connections to
the shorting switch should be made as short as possible.
17

Note: When operating with the reactive differential compensation circuit, it is

desirable to have the unit parallel switch be an auxiliary contact on the main
generator breaker. The auxiliary contact should be closed and the CT
secondary shorted when the main generator breaker is open. The instant the
main generator breaker closes connecting the generator in parallel, the
auxiliary contact should open removing the short from the CT secondary
allowing it to give the reactive load signal. This prevents any voltage droop
from being introduced and eliminates the fluctuating voltage of the
oncoming generator described earlier.
The auxiliary contact input is applied differently when the application involves a digital AVR.
With analog AVRs, the CT input is applied to the AVR and a voltage is developed across
the burden resistor as discussed previously in this presentation. A digital AVR typically does
not contain a burden resistor. The CT signal is applied to the AVR and it is measured using
an internal shunt, much like a metering device would utilize a shunt. Then, the digital AVR
calculates the phase angle difference between the voltage and current being applied to the
AVR.
This shunt has very little resistance, so attempting to disable the paralleling provisions by
shorting out the CT will not yield satisfactory results.
On digital AVRs, a contact input is utilized to enable paralleling when the contact is open
and to disable paralleling provisions when the contact is closed.
Another difference with digital AVRs is that, when the CTs are connected in reactive
differential compensation, the AVRs ability to measure AC current, Watts, VArs, and phase
angle as well as the under excitation limiter function is compromised because the current
flow in the CTs is effectively zero when the reactive differential is employed.
To alleviate this, most digital AVRs have a dedicated reactive differential or cross current
input only for that purpose with a separate CT input for metering.

18

Figure 33: Typical DECS-250 Connections

19

Troubleshooting Paralleling Issues

Abnormal indications

Immediate increase in current

Difficult to adjust voltage adjust for null

Changing loads cause unbalance

For immediate high current

Check paralleling in droop only

Verify that kW load is properly shared

Check SENSING connections to AVR

CT in correct phase??

Reverse CT secondary polarity

Try to parallel again

If all above is OK, try to close cross current loop

Correct Ct polarity??

Place CT on correct phase

- No sensing connection, or
- E2 sensing connection

If CT is on correct phase, polarity is:

- Correct, and closing breaker works okay, or
- Reversed, and closing breaker causes immediate high current

Wrong phase, possibilities MULTIPLY!!!

Finding the problems

Check for small voltage across droop input. CT may be shorted somewhere.

Check droop adjust. Max or min??

Further testing

OK to parallel without any droop on one machine, with careful attention.

Check to see that two machines can share load if only one of the droop CTs is
not shorted.

If loading is possible, check for direction of droop by increasing drooping AVR

voltage adjust.
Another

test
Connect oscilloscope to E1 and E3, channel 1
Connect channel 2 to Droop input, 1 and 2
With some resistive load current, observe 90 phase shift from voltage to current

20

Utilizing Communication for Load Sharing versus Cross-Current Compensation

This system was designed to improve on the traditional cross-current compensation (CCC) loop by
removing the extra hardware associated with it, such as the cross-current current transformer (CT)
and corresponding burden resistor. This new approach, called network load sharing (NLS), is the use
of Ethernet over traditional wiring. With traditional wiring, there is a theoretical limit to the distance
between two units in a given cross-current due to impedance of the wiring. But with Ethernet
switches and repeaters, that distance is almost indefinite. This is highly advantageous for marine
applications where one generator may be at the bow while the other is at the stern. In addition to the
removal of distance limits, Ethernet can reduce the susceptibility of the system due to noise that
could be picked up on otherwise unshielded wiring.
NLS was designed to have the same, if not better, performance than traditional reactive cross-current
compensation loops. Shown below is the comparison of the two load sharing methods between a
75kW and a 100kW generator. Both of the generators were diesel generators with appropriately
sized CTs. One was a 75kW generator with 250:5A CT ratio and the other was 100kW with 200:5A
CT ratio. With that, a load of approximately 66kW and 32kVAR was applied to the system and then
removed. Figures 34 and 35 show this load application and removal on a traditional CCC loop and
the responses of the DECS-250s, while Figures 34 and 35 show the same scenario, but with NLS.
The graphs below present the data collected in Per-Unit on the Y-Axis, while time is given in seconds
on the X-Axis. The figures below are given in Per-Unit using the follow numbers as base values:

Parameter

Rated Value

Average Voltage (Line to Line)

Real Power
Reactive Power
Apparent Power

208Vac
75kW
56.25kVAR
93.75kVA

1.2
1
0.8
V avg L-L

0.6

Real Power (PU)

Reactive Power (PU)

0.4

Apparent Power (PU)

0.2

Iaux CrossCurr

0
0

10

20

30

40

50

60

70

-0.2
-0.4

Figure 34: Load application and rejection with CCC on the 75kW generator
21

1.2
1
0.8
V avg L-L
Real Power (PU)

0.6

Reactive Power (PU)

0.4

Apparent Power (PU)

Iaux CrossCurr

0.2
0
0

10

20

30

40

50

60

70

-0.2

Figure 35: Load application and rejection with CCC on the 100kW generator

1.2
1
0.8
V avg L-L

0.6

Real Power (PU)

Reactive Power (PU)

0.4

Apparent Power (PU)

0.2

RCC

0
0

10

20

30

40

50

60

70

-0.2
-0.4

Figure 36: Load application and rejection using NLS on the 75kW generator

22

1.2
1
0.8
V avg L-L
Real Power (PU)

0.6

Reactive Power (PU)

0.4

Apparent Power (PU)

RCC

0.2
0
0

10

20

30

40

50

60

70

-0.2

Figure 37: Load application and rejection using NLS on the 100kW generator
These results show that NLS generators have superior response because they are able to settle
10 seconds before the CCC generators on both load application and rejection, which overall, will
lead to a more stable and reactive system. NLS also tends to be more accurate with the amount
of VARs on each system. Data in Tables 1 and 2 show that the percentages of error due to
differences in reactive power on each generator.
Table 1 Shared VARs in Cross Current Compensation
CCC Data
Reactive/Real
50/50 Split
kVAR on
kVAR on
Differential
Loads at 240Vac
(w/o losses)
100kW
75kW GEN
GEN
75kW/25kVAR
9.3875
7.42
10.2
2.78
75kW/50kVAR
18.775
16.5
18.7
2.2
75kW/75kVAR
28.1625
26.1
27.5
1.4
100kW/75kVAR
28.1625
26.15
27.5
1.35
35kW/25kVAR
9.3875
7.5
9.44
1.94
Table 2 VARhared VARs in Network Load Sharing
NLS Data
Reactive/Real
50/50 Split
kVAR on
kVAR on
Differential
Loads at 240Vac
(w/o losses)
100kW
75kW GEN
GEN
75kW/25kVAR
9.3875
8.2
8.9
0.7
75kW/50kVAR
18.775
17.5
18
0.5
75kW/75kVAR
28.1625
26.5
26.8
0.3
100kW/75kVAR
28.1625
26.5
27.1
0.6
35kW/25kVAR
9.3875
8.2
8.7
0.5
23

Error

2.96%
2.3%
1.49%
1.4%
2.0%

Error

0.74%
0.53%
0.32%
0.64%
0.5%

A concern with using communications for load sharing is how a given controller will perform if, for
example, another unit in the communications circuit fails or a cable falls out. If this were to happen,
the controller will revert to the programmed NLS Droop setting until normal communications are
restored or until the malfunctioning unit is removed properly (e.g. Unit ID is disabled on the other
units). In Figure 38, both generators share together in NLS; then, when the Ethernet connection is
removed, both units transition to droop and remain in that state. Finally, the Ethernet connection is
restored and both units resume load sharing in NLS.
1.2
1
V avg L-L

0.8

Real Power (PU)

Reactive Power (PU)

0.6

Apparent Power (PU)

0.4

PF

0.2

RCC

Ethernet and
NLS Restored

0
Ethernet Removed,
unit begins to0 Droop 5

10

15

20

25

30

Figure 38: 75kW unit handling Ethernet loss and restoration with Sequence of Events

1.2
1
V avg L-L

0.8

Real Power (PU)

Reactive Power (PU)

0.6

Apparent Power (PU)

PF

0.4

RCC
0.2

Ethernet and NLS

Restored

Ethernet Removed,
unit begins
0 to Droop
0

10

15

20

25

30

Figure 39: 100kW unit handling Ethernet loss and restoration with Sequence of Events

24

NLS also allows users to configure controllers for unique applications such as ring bus systems or
selective island configurations. The settings for the Network Load Sharing are relatively easy to
configure and a typical implementation is as follows. The settings and logic below demonstrate ring
bus functionality in one of several possible ways. In addition to ring bus functionality, the controller
can be configured for a variety of systems including multiple islanded bus configurations. In Figure
40, the majority of the relevant settings for the load sharing capabilities are shown and those that are
not shown are logic elements. The load sharing is enabled by selecting the checkbox next to the
enable and having settings in place for the gain and droop percentage. For NLS, the droop and
washout settings will govern how the controller reacts during transients while VAR sharing. The other
setting of concern is the gain, which should be adjusted based on the system. If the gain is left at
zero, the unit will probably not share load well, so it is recommended to modify this before attempting
to load share.
When configuring controllers for NLS, it is important to ensure the IP addresses are assigned
correctly.

Figure 40: Load Share Settings

(Found in Settings Explorer->Operating Settings->Parallel/Line Drop Compensation)

25

Parallel Operation with a Network System

The purpose of this discussion is to address the concerns of and the techniques required
to successfully parallel synchronous generators to a network. This session is expected to
give you an understanding of:
Generator operating configurations
Control of Generator voltage and reactive power
In simple form, a generator set connected to some load is an islanded power system, and
the operation of such an islanded system has been discussed. Voltage regulation maintains constant bus voltage, and speed governing maintains constant frequency. Load on
the generator set is totally determined by the load demand. With only one source of power,
there are no issues with load sharing. For voltage and frequency control, the simple
functions of an automatic voltage regulator (AVR) and a speed governor will suffice to keep
voltage and frequency of the power system on target, accommodating changes in load
demand as needed. Increases in real or reactive power will cause changes in voltage and
frequency, and the control circuits will adjust fuel and excitation to restore the nominal
values. There is no need for any kind of load sharing equipment such as droop
compensation.

Figure 41: Islanded Power System

Next up in complexity is a system of two or more generators connected to a common bus.
This configuration was covered with regard to excitation control when we spoke earlier in
the course about generators operating in parallel. In this configuration, voltage droop or
cross current compensation were methods used to share the reactive load. In this session,
we will briefly discuss the application of governor controls that will provide for droop or
isochronous load sharing of the real power.

26

Figure 42: Multiple Paralleled Generators

Finally, the operation of one or more generators in parallel with a network, such as a utility
power system, is often considered the most complex generator operation mode. This mode
will be the focus of this session.

Networks or BusWhats the difference??

A network consists of two or more buses connected together by one or more power lines
whose impedance compared to the capacity of the network generating capacity cannot be
neglected. Contrast this definition with that of a bus whose impedance is so low that it can
be neglected. This difference makes analyzing the networks flow of current and the voltage drop calculations much more difficult than the paralleled generators with loads all
connected to a common bus. When we are able to neglect the bus resistance and reactance, the solution to current flow and voltage drops is very simple. Large electric power
systems typical of our electric utility can only be analyzed by considering them to be network systems. It would not be economically feasible to transport power the way utilities
currently do if the connections between loads and generators were required to have resistance and reactance negligible compared to the system capacity. With the capacity of
computers to perform complex calculations very quickly, however, the analysis of power
system load flow is greatly simplified compared to the slide rule days.
One way to simplify the analysis of network operation is to adopt the concept of infinite
bus for the network. This assumption can be used if the capacity of the network at a point
of interconnection is at least 10 times greater than the capacity of equipment connecting
to the network. The concept holds true when the network voltage and frequency are not
changed when real and reactive currents change at the point of connection. An example is
a generator connected to a network, with generator size of 100kW and the connection
point being an industrial plant served by a 10MVA transformer. With the transformer being
fed from a line having a capacity of 150MVA, any changes in load flow caused by fuel or
excitation changes at the 100kW generator will have no measurable effect on the voltage
and frequency of the transformer secondary. This is also referred to as the stiffness of the
grid.
27

Figure 43: Typical Paralleled System

Parallel Operation of the Excitation System

To control the reactive power from the 100kW generator, a similar control loop to the
speed governor kW regulator is used to measure one of two parameters of the generator;
the first is reactive power flow, measured by current and voltage transformers connected in
quadra- ture exactly like the arrangement in the droop circuit. By comparing the measured
kVAr load with the set point, adjustable for leading or lagging reactive load, the generator
will supply a constant reactive current to the bus regardless of load demand and variation
in bus voltage.
To help understand what we are discussing, the power triangle is a good tool to use.
Across the bottom axis is the real power expressed in Watts or Kilowatts (W or kW). Along
the hypotenuse of the triangle is the apparent power expressed in Volt Amp or Kilovolt
Amp (VA or kVA). The reactive power is the third leg of the triangle and is expressed as
Volt Amperes reactive or Kilovolt Amperes Reactive (VArs or kVArs).
The cosine of the angle is referred to as a Power Factor, a measure of the ratio of reactive
power to resistive power. For power factor of 1.00, the current is in phase with the voltage,
and the reactive load is zero. Power factor of 0.8 is the rating of most small to medium size
generators, indicating that it is designed to deliver 80% of rated current as resistive load,
and 60% of load current as inductive reactive load current (0.8 is cosine of 36.9 degrees,
sine of 36.9 degrees is 0.6). This rating of the machine indicates the thermal capability of
the stator windings on a continuous basis at maximum ambient temperature. This rating
also indicates the thermal capability of the rotor windings on a continuous basis at maximum ambient temperature. The machine voltage and frequency rating give the thermal
rating of the core of the machine at maximum voltage and nominal speed.

28

Figure 44: Power Triangle

Two parameters, reactive power and power factor, are the ones that the excitation system
can control. The question is then raised as to how this is done. One approach is to treat
the infinite source as another generator to be paralleled to and connect the voltage
regulator in the droop mode.

Figure 45: Typical Droop Connection

In this method, the droop adjustment is made just as if the generator were paralleled with
another generator. If there is also a step up transformer between the generator and the
grid, this approach works even better due to the impedance of the transformer. Adding
droop to the voltage regulator is, in a sense, like adding impedance to the line, so a large
transformer helps in this type of paralleling. If the grid is not very stiff, then paralleling in
the droop mode may be satisfactory. This way, an operator can monitor and make
adjustments as required to maintain bus voltage and reactive load sharing.
Now the question is asked that if generators can be paralleled with the regulators in the
droop mode, can I parallel with the regulators connected in reactive differential or cross
current. The answer to this is that it is not recommended. If we review what a cross
current loop looks like to the voltage regulators, we can see why. Since the regulators
connected in cross current expect that a change in its compensation circuit will cause a
change in the other circuits, there will not be a change in the infinite grid no matter what
the regulator does.
29

Figure 46: Cross Current Compensation

Because many generator power systems must support load in a mode not paralleled to
the grid and at some point will also parallel with the grid, how can the systems be
connected in cross current? The answer is that the cross current connections may be
disabled when the main tie breaker is closed. Breaking the loop in any one spot will
disable the cross current compensation.
It may seem that connecting the regulators in droop to parallel to the grid is the best way
to operate. However, compared to the frequency, bus voltage is subject to much greater
variation, and the source of the variation may be local changes in load or network voltage
variations which occur because of line losses changing with load or changes in network
operating voltage from daytime to nighttime levels. Regardless of the cause, the effect on
a droop compensated generator would be changing reactive load levels. Most regulators
have a maximum 6-8% droop setting. If the grid voltage goes higher than the droop
setting, the regulator decreases its output and a generator is paralleled and must start
importing VArs to operate at the high bus voltage level. This could cause the generator to
become underexcited, start slipping poles, and become damaged.
Conversely, if the voltage on the grid decreases, the regulator tries to drive the voltage
back to its set point. Since by definition the grid is infinite, the generator cannot restore
the voltage. As the excitation increases the generator starts exporting VArs into the grid.
This situation can cause excess heating of the generator rotor windings as well as the
distribution transformer. This is a potentially damaging situation for the generator.

Automatic VAr/Power Factor Regulation

Adding a var regulator to the excitation system and operating in the var regulation mode
allows maximum utilization of the generator reactive load capability independent of real
power load. Alternatively, the var regulator may be set to maintain the generator reactive
load at zero, if the vars are not bringing in revenue, to keep the generator as cool as possible while supplying the revenue-producing real power to the load. With VAr regulation,
the changes in bus voltage which can cause VAr load variations are compensated by
adjustment to the avr set point voltage. The result is a constant VAr load on the generator
without any operating intervention.
30

Figure 47: Alternative Control Strategy

The second type of control loop measures the power factor or the angle of generator current with relation to the voltage. In this control mode, the reactive load is regulated as a
percentage of the real power load, tracking any changes in real power to keep the percentage constant. This form of control is often preferred by operating personnel trained to keep
the power factor constant in their manual control practices. Many generators are equipped
with a power factor meter and a kW meter in addition to volts, amps and frequency. In this
configuration, a var regulator can produce some readings on a pf meter which will alarm
some operators trained to keep power factor to 0.8 or higher. However, use of var regulation does not add any risk to the generator if operating at rated kVar and 10% of rated kW.
The power factor of this load is 0.54 pf at 84% of rated generator current. The generator is
able to operate safely under at this condition. Power factor regulation may be selected as a
control option if desired, but by replacing the traditional power factor meter with a var meter
and regulating the reactive load, better use of machine capacity may result.
There are three basic ways to add VAr/power factor regulation to an excitation system.
They are:
1)
2)
3)

Add additional component to the system,

Integrate VAr/Power factor control into the regulator itself, and
Programmable Logic Controller.

No matter which approach is used, the technique is doing one of the following:

31

As angle changes, the power factor changes.

Figure 48: Constant kVAR Level

As the kW changes, the angle stays the same, the kVAR level changes. Since we are
regu- lating the angle, the cosine of the angle stays the same.

Figure 49: Constant Power factor Load

Adding an additional component to the excitation system is relatively easy. A basic block
diagram is shown in Figure 50.

Figure 50: VAR/PF Controller with Static Exciter

32

Figure 51 shows the faceplate of the Basler Electric SCP-250. Setting on the SCP-250
is from .6 power factor leading to .6 power factor lagging when in the power factor
control mode.

Figure 51: SCP-250 Faceplate

In the VAr control mode the adjustment is either to produce or absorb VArs. The VAr
range adjustment sets up the limits at either end of the potentiometer travel.
Figure 52 illustrates a generator that is equipped with a solid state voltage regulator
having reactive voltage droop compensation. The graph illustrates the effect of bus
voltage changes on the reactive/ampere load on the generator. If the bus voltage drops by
6%, the reactive/ampere generator load will change from 0 - 75%. A further decrease in
bus volt- age exceeding 4% would overload the generator, causing excessive heating in
the field winding as well as the power semiconductors of the automatic voltage regulator.
The VAr/ PF controller regulates at a programmed operating point and is insensitive to
changes in the bus voltage. Field heating increases (higher excitation current) as lagging
reactive load increases. Keeping the lagging VAr load under control protects the generator
field from overheating. This function can be performed by an operator, or the VAr/PF
controller auto- mates the control function.

33

Figure 52: Voltage Regulator Droop versus Var/PF Control Regulation

Figure 52 also illustrates the condition where the bus voltage may increase, causing a
leading power factor condition on the generator. Here, the voltage regulator will
decrease excitation following the characteristic slope of the reactive compensation
circuit. This will keep the system in synchronism. If the bus voltage rises excessively,
however, leading VAr load will increase, leading to a reduction in the field excitation
and causing possible loss of machine synchronism.
The other two techniques used to regulate VArs or power factor follow the same basic
connections of the voltage and current sensing as well as control. There is a voltage input
and a current input at some phase angle difference as well as a means to turn the control
on/off. As with the SCP-250, this is typically a 52b control off the main tie breaker.
In PLC (Programmable Logic Controller) control, the excitation is sometimes controlled by
a DC input into the regulator similar to the SCP-250. However, sometimes the PLC will
have output contacts to control a motor operated potentiometer or reference adjuster to
change the regulator's set point to change excitation. The only potential drawbacks to this
approach are the coordination of the overall control loop so stability is achieved and the
wear and tear of a constantly moving M.O.P. or reference adjuster.
Power Factor Correction
The other use is for power factor correction. Using an unloaded motor as a
synchronous condenser and producing only VArs, it is possible to regulate a plant
power factor compared to the utility input.

34

Figure 53: Power Factor Correction

It is also possible to regulate the power factor of your plant with your in-house generation.

Figure 54: In-house Generation

This configuration allows the user to regulate the VArs at the point where the CT is connected. Since the power factor/VAr regulator is controlling the excitation output, there is a
need to limit its range of control.

35

EXCITATION LIMITERS
What is Generator Synchronism?
The synchronous generator can be represented by a capability curve which shows the
electrical watts and VAr limit of the generator. These quantities are related to the permissible temperature rise of the generator windings and the mechanical limits of the system.
The electrical watts are limited by the horsepower of the prime mover and by the heating
of the rotor and the stator windings.
Under varying conditions, the automatic voltage regulator could command abnormally low
field current due to higher than normal infinite bus voltage. If this occurs, the synchronizing
torque is reduced, allowing the rotor of the synchronous machine to advance beyond a
critical power angle (90 degrees) resulting in loss of generator synchronism.
To better understand the term generator synchronism, imagine a rubber clutch connecting
the shaft of two engines. See Figure 55. As long as the speed remains constant on both
engines, the rubber clutch will lock the two shafts without distorting its own natural shape.
But if one machine begins increasing speed, causing the other to lag, the clutch will be
forced to stretch.

Figure 55: Two Engines Connected by a Rubber Clutch

If the speed continues to increase, causing a greater speed difference, the rubber clutch
will eventually stretch beyond its limits and break. The change in the relative shape of the
clutch is synonymous with the power angle changing between the generator rotor and
stator. In the generator at the time the power angle has advanced beyond the critical angle
of 90 degrees, the system has lost synchronism. See Figure 56. The rubber clutch represents the magnetic flux between the generator rotor and stator which enables the power to
be transferred from the prime mover to the distribution system.

36

Figure 56: Power Relationship versus Displacement Angle

The maximum allowed elasticity of the rubber clutch is defined as the stability limit of the
generator, and each generator has the capability curve which defines this limit. As long as
sufficient excitation is maintained in the field for the generator load, synchronism is assured.

Figure 57: Generator Reactive Capability Curve

37

Figure 58: Static Exciter Excitation Limiting

Overexcitation is designed to protect the machine during operation at lagging power factor
beyond the machine rating, but generally the machine limit of overexcitation is rotor overheating (high field current). To prevent rotor overheating, measuring the excitation current to
the rotating field of the generator and cutting back the excitation current after some time
delay is the most widely accepted way. A higher instantaneous or hard limit may also be a
function included in the limiter. If excitation with very high forcing voltage is used, an instantaneous limiter can keep the excitation current during forcing to a lower level than the
forcing voltage would otherwise drive it, gaining faster generator response from high
forcing, both positive and negative, without need for more cost in the rotor, brushes, and
slip rings to handle high forcing currents.

Figure 59: Overexcitation38

Limiter Limits Max. Field Current

To sufficiently protect the rotor, it is necessary to know the thermal capability of the rotor
under worst case conditions. If the rotor is operating at maximum continuous current and
the load demand moves the excitation higher, then you must know how long the field will
take before temperature of the rotor exceeds its upper limit. With a curve plotting thermal
limit based on full load operation prior to the higher current, the overexcitation scheme
may be coordinated with the generator capability. When an older machine is upgraded
with new excitation, it may not be possible to obtain the rotor thermal capacity curve to
allow accurate setting of the relay. Standard practice is to choose a conservative setting to
make sure limiter operation keeps the field temperature within safe limits.

Figure 60: Rotor Angle or Underexcitation Limiter

Underexcitation is designed to protect the machine during operation at leading power factor.
To perform its function, the underexcitation limiter measures stator voltage, current, and
phase angle. The primary concern for this limiter is the possible loss of synchronization with
the stator resulting from a magnetic field strength too low to keep the rotor from slipping
poles. Machines operating in parallel with an electrical network are most susceptible to this
possibility, mainly instigated by some fault on the network being cleared by a breaker, and
the resultant fast recovery of the system voltage when low magnetic field is available to
maintain the rotor angle less than 90 degrees. Internally, the rotor angle of the machine is a
function of the excitation current and real power load. The rotor angle increases from zero
as load increases and decreases with increased excitation current. If the rotor angle is too
close to the hypothetical 90 degree angle at which pole slip is certain, a transient load
change can cause rotor torque to exceed the synchronizing torque existing between the
rotor and stator. By limiting excitation to give a smaller rotor angle, the possibility of pole slip
is reduced. This area of limiting can contribute greatly to the ability of the machine to be
used in either leading or lagging power factor operation without concern for the possibility of
pole slip. Avoiding pole slip prevents very high mechanical stresses and high currents from
doing severe damage to the machine. The preferred characteristic for underexcitation
39

limiting is referred to by some manufacturers as Rotor Angle Limiting. This simply means
that the limiter monitors stator volts and amps, and will provide limiting to allow full stator
current at leading power factors near 1.0, increasing excitation and decreasing stator
current along a circular curve as the power factor moves further into the leading power
factor area.

Figure 61: Stator Current Limiting

To complete the machine protection, add the stator current limiter. This function accepts
current from generator current transformers, and begins to reduce excitation if the stator
current exceeds the maximum for too long. The time is a function of the generator thermal
capacity, thus the stator current may be instantaneously limited, inverse time delay limited,
or both. Setting the instantaneous limit to protect the machine for fault currents and setting
the time delay for thermal protection gives a high level of protection for the machine.

40

Figure 62: Combined Limiters

41