10

Distributed Generation

Juan Carlos Gomez Targarona and Medhat M. Morcos*

Instituto de Protecciones de Sistemas Eléctricos de Potencia,

Departamento de Electricidad y Electrónica,
Universidad Nacional de Río Cuarto,
Ruta 8, Km. 603,
(5800) Río Cuarto, Córdoba, Argentine.
Email: jcgomez@[Link]

Department of Electrical and Computing Engineering,*

Kansas State University,
289 Rathbone Hall,
Manhattan, KS 66506, USA.
Email: morcos@[Link]

10.1 Introduction
The growing concern on climate changes and the ever-present oil crisis have
recently increased the interest in distributed generation (DG) from renewable and
traditional (but highly efficient) sources. Deregulation and growth of competitive
supply markets also magnify this interest. In general, distributed resources are
defined as sources of electrical power that are not directly connected to a bulk
power-transmission system. They include both generators and energy-storage
technologies, with a power rating of 10 MW or less [1].

Currently, distribution-system engineers are divided between DG advocates and

adversaries, each having their own valid reasons. One of the main reasons for this
conflict is just “fear of what could happen” due to the lack of practical knowledge
on traditional power systems having a high level of DG penetration. Still fresh in
the minds of many engineers is the wrong approach that has been taken in the past
by large computer manufacturers when they underestimated the growing market of
personal computers; an analogy with the present DG situation that is easy to make.

The impact of DG on distribution systems depends on the penetration level, a

presence of approximately 20% on energy resources for the years 2010–2020 is
predicted. During the first part of the current decade fewer industrial and
commercial businesses have expressed an interest in adopting onsite DG
equipment [2]. After the widespread blackouts that took place in the USA and
294 J. C. Gomez Targarona and M. M. Morcos

several European countries by the end of 2003 and their related huge costs,
industries and utilities started to change their philosophy.

The widespread use of DG or distributed resources has created a series of new

problems, some of them closely related with power quality such as selective
coordination of overcurrent protection and its effects on continuity (service
quality), in addition to control of the voltage-sag magnitude and duration by using
overcurrent protective devices (PD) considered as included in product quality. The
most feared effect is the islanding that can be highly risky for user equipment,
utility elements and personnel, but can be desirable from the power-quality point of
view. Interaction and impact of DG on network operation, fault detection, fault
clearing, and reclosing operations are also of great concern [3].

It has been suggested that the widespread use of DG defines old and new
scenarios, calling for highly sophisticated equipment (using communications,
electronics, and microprocessors) as the only way to cope with the current
distribution-system scheme. Several protection concepts have been presented
recently, many are so contradictory that protection studies become even harder.
The doubts and contradictions have delayed the approval of the IEEE P1547 Draft,
“Standard for Interconnecting Distributed Resources with Electric Power
Systems,” which has been under study for several years and has been recently
issued [4].

Most of the utilities have established new connection rules for DG, also
changing their own distribution systems. One of the changes is the avoidance of
any single-phase protective device positioned between the utility transformer and
the three-phase DG. Also, it has been realized that DG protection needs a longer
time (up to 1 s) to detect faults and operate, thus longer reclosing intervals are used
in utilities.

The ANSI/IEEE Standard 1001 presents the following out-of-range trip settings
[5]:

voltage higher than 137%, maximum trip time of 2 cycles;

voltage higher than 106%, and lower than 88%, maximum trip time of 6
cycles;
voltage lower than 50%, maximum trip time of 6 cycles;
frequency is considered normal between 59.3–60.5 Hz (60 Hz base), with
15 cycles of trip delay when out of tolerance.

It has been reported that the use of novel fast-clearance protection systems to
protect DG would improve transient stability and also improve power quality (by
reducing voltage-sag duration). Several issues are currently being considered, for
example:

1. Whether to disconnect the DG very rapidly or keep it connected as long as

possible.
 Distributed Generation 295

2. Impairment or improvement of the traditional overcurrent coordination with

DG addition.
3. Islanding advantages or disadvantages from the voltage-sag ride-through
point of view.

The answers to these issues are different, depending on what portion of the
power system is considered. In the authors’ opinion more research is needed with
regard to classical protection coordination, as fuses are still widely used in
distribution systems. One possible reason is based on the fuse simplicity and
reliability, being able to operate even in extremely hard environments. New
protection systems need to be studied or restudied for this application, until all the
traditional protection possibilities have been exhausted.

The application of DG is very useful for deferring investments, improving the

operation of the existent distribution networks, and increasing the customer’s
reliability (perhaps only for the DG owner and not for the other customers
connected to the same system) but it creates new challenges for the engineers
involved. When the original idea of DG application came out, utilities were very
sceptical about their capability, especially resistant to introduce this technology in
their own systems, and refused to allow DG installation on customer circuits.

Besides, distribution systems have been mainly designed for energy flowing in
just one direction. If the DG equipment does not belong to the utility, the customer
decides about its connection to or disconnection from the system, based on its own
needs. If the DG belongs to a cogeneration scheme, in the case of a blackout the
customer could keep the equipment feeding the critical loads operating as an
isolated circuit or island, and needing resynchronisation when the external supply
is restored. This scheme created more than one problem to the utility when it is
applied during voltage sags caused by nearby faults [6].

The effect of DG greatly depends on the technology used and on the network
characteristics. The main aspect is that now the energy source is not unique and
then the energy flow could be changed or reversed under normal or abnormal
conditions.

There are several types of power generators that can be used as DG that, in
order of importance, are:

• wind turbines;
• fuel cells;
• photovoltaic cells;
• small and micro-turbines;
• internal combustion engines.

Each of these power generators has its own advantages and disadvantages, such
as easy installation based on its small size, fast start, generates voltage distortion,
and uncertain power availability. In spite of the distributed-generation promotion
296 J. C. Gomez Targarona and M. M. Morcos

carried out by utilities, they have been expressing concerns related with potential
distribution-system problems.

10.2 General Impact of Distributed Generation on Power Quality

to Strong or Weak Electrical Systems
The strength of the existent distribution system where DG is connected is very
important from the point of view of the impact on power quality. Some power-
quality phenomena are very noticeable when they take place in weak systems and
pass practically unnoticed when they happen in strong distribution networks. In
other words, the importance of the events is closely related with the system
regulation.

10.3 Dissimilar Effect of the Different Distributed Resources

Technologies
Depending on the available energy source, DG can be of dissimilar technologies,
such as: wind turbines, photovoltaic cells, fuel engines, gas turbines, vapour
turbines, fuel cells, etc., the first mentioned being the currently more developed,
reaching today a unit power of 2 to 3 MW. Each technology has its own
advantages and drawbacks. The DG deployment depends among other factors on
geographic situations, weather conditions, electric-energy cost and availability,
utility tariff policy, existence of waste having usable calorific power, etc.

10.4 Coordination between Overcurrent Protection and Sensitive

Equipment Voltage-sag Immunity
Distribution-system protection interaction with power quality, especially related to
“product quality”, is one of the aspects where the DG has higher impact. Besides,
most of the voltage sags are generated by faults whose duration is ruled by the
protective-device operation speed. This speed is modified when the DG is
supplying part of the fault current and is also supporting the voltage system. Thus,
voltage-sag studies are widely affected by any DG connected to the network.
One of the important issues that the new scenario will introduce is overcurrent
protection, where there will be changes in system behaviour and flow of power
under short-circuit conditions. Normally, there is no concern with the DG when its
power is less than 10% of the minimum load demand in the feeder [6].
The most immediate consequence is the need for verification of the protective-
device breaking capacity, which might not be enough due to the increase of the
available short-circuit power. The short-circuit current and transient behaviour of
the generator that provides power through inverters – such as photovoltaic cells,
fuel cells, wind turbines, and induction generators – are completely different from
 Distributed Generation 297

the synchronous generator response. Induction generators directly connected to the

supply will show a special behaviour when a short-circuit takes place. Due to the
fact that this type of generator gets its excitation from the mains, it is not able to
maintain the short-circuit currents for a relatively long time.

The situation when the mains supply is intentionally or unintentionally

disconnected from the power system having at least one DG, which continues
operating with this single source, is called islanded-mode operation.

Due to the increase of power-electronics applications in industry, residences

and elsewhere, users have become very sensitive to voltage perturbations, among
which voltage sag is the event of most concern. An aspect that requires special
treatment is the coordination between overcurrent protection and the voltage-sag
ride-through capability of sensitive equipment.

Voltage sag is almost universally considered as “a nonpermanent voltage

reduction with values between 10% and 90% of the rated voltage and duration
between ½ cycle and a few minutes”. The ability of sensitive equipment (SE) to
withstand voltage sags without dropout is called the ride-through capability.
Reductions in voltage between 0% and 10% are considered as there is no
perturbation since the voltage is included in the normal zone of voltage variation in
distribution systems. Voltage reduction of more than 90% or applied voltage less
than 10% are considered as an interruption. The zero voltages with durations
shorter than three minutes correspond to the phenomenon known as
microinterruption [7].

Many typical ride-through capability curves have been proposed as guidelines,

presented by institutions such as the U.S. Department of Commerce-Federal
Information Processing Standards Publication No. 94, the Computer Business
Equipment Manufacturing Association (CBEMA), SEMI F47 presented by
semiconductor manufacturers together with EPRI, and the Information Technology
Industry Council (ITIC). The current version of the ITIC curve, revised in 1996,
has reached prominent widespread use. These guidelines are normally given in
graphical form, having two well-bounded parts corresponding to the overvoltage
and undervoltage withstand values.

These curves represent an important tool for manufacturers, users, and system
designers. The present work is confined to only the lower part of the voltage-sag
ride-through capability curves, adopting the Computer Business Equipment
Manufacturing Association (CBEMA) curve as a guideline. Fig. 10.1 shows the
lower legs of the CBEMA and ITIC curves [7].
298 J. C. Gomez Targarona and M. M. Morcos

100

80
Voltage (%)

60

40

20

0
0,01 0,1 1 10 100 1000 10000

Time (cycles)

CBEMA ITIC Anaytic

Figure 10.1. CBEMA and ITIC sensitive-equipment-immunity curves

10.4.1 Application of Specific Energy Concept

The sensitive-equipment ride-through capability can be considered as directly

related to the energy stored in kinetic, electrostatic, or electromagnetic form that is
the source used to keep the apparatus working while the supply voltage is below
the device needs. Due to cost and competitive reasons, the manufacturer supplies
as low a ride-through capability as possible, which causes users to need to add
auxiliary equipment in order to withstand the frequent voltage sags. Many
technical articles and papers on the auxiliary equipment are readily available,
however, only a few are related to the sensitive equipment incorporations tending
to improve its immunity.
The main exception is related to one of the most sensitive applications, the
adjustable-speed drives (ASD), where the consequences of equipment dropout (or
shutdown) are extremely expensive. Besides, an extensive study on personal
computers has been published where it is estimated that 80 to 90% of voltage-sag
problems have durations shorter than one second. The increase of the ride-through
capability to this time duration will produce a considerable improvement in the
computer behaviour [8].
The methodology of increasing the energy-storage ability was adopted by a few
manufacturers of household appliances, like digital clocks, VCRs, phone
answering machines, TVs, etc. A few sensitive-equipment customers are starting to
specify the ride-through capability when ordering new equipment, for instance
whether the equipment follows the CBEMA-ITIC curve, then doing the
corresponding ride-through coordination study. The CBEMA curve in Fig. 10.1,
 Distributed Generation 299

shows the existence of a nearly smooth voltage – duration correlation. The ITIC
curve is given in a stepwise shape; however its average really follows the CBEMA
curve [9]. It would be of great interest if the manufacturer provided information
about the ride-through capability in a way that is compatible with CBEMA-ITIC,
such as a voltage – time curve instead of giving one point only (actual
methodology) related to this curve.
IEEE Standard 1346 gives three values of voltage tolerance for six sensitive
equipments, which correspond to the typical upper, average, and lower ranges
available in the market. The cited sensitive equipment are: programmable logic
controller (PLC), PLC card input, adjustable-speed drive (ASD), control relay,
motor starter, and personal computer [10].
The SE ride-through capability is affected by the pre-sag voltage, which could
confirm the concept that a voltage over 87% is used for storage, where this energy
is to be released in the case of voltage-sag occurrence. The value 87% of the rated
voltage is the minimum voltage specified by the CBEMA curve, for which the
equipment should work indefinitely without any performance limitations [9].
The stored-energy concept leads to the presence of a variation law related to the
product of the squared voltage and time. The third curve in Fig. 10.1, identified as
analytic, represents the relation:

V2 t = constant, (10.1)

which shows a very good agreement with the CBEMA curve. In this case, the
voltage is expressed in per cent and the time in cycles, resulting in a constant value
of 4400.
The previously defined methodology (CBEMA-ITIC curves) agreed on
considering the value of maximum voltage drop as the magnitude of nominal
voltage sag. The difficulty in determining the magnitude of nominal voltage sag is
in assigning the duration, since in the real world the voltage variation is usually not
so sharp (square sag); its waveform depends on the system and equipment
characteristics.
Another difficulty in the voltage-sag characterization is the presence of a
progressive or evolved fault, which starts as single phase and after a short time
becomes a multiple-phase fault, or modifies the current without changing the fault
type [11].

The application of the “energy concept” will lead to the change of Eq. 10.1 to:

∫ v2 dt = k, (10.2)

where ∫ v2 dt will be called the specific energy in analogy to I2 t which is used

extensively in overcurrent-protection studies. The introduction of the integral for
each sensitive equipment would solve the ambiguity issue, giving a full voltage-sag
characterization. The proposed methodology is able to consider the value of the
initial voltage sag (or small step), which is normally present due to the voltage
developed across an arcing fault. This effect is also noticeable on double-supply
300 J. C. Gomez Targarona and M. M. Morcos

systems, which is precisely the situation when a DG exists, after the first operation
of the circuit breaker takes place.
The ride-through capability of three-phase equipment is not the same for single-
phase voltage sags as for three-phase voltage sags. In the case of single-phase
faults, the nonfaulted phases usually present an overvoltage, which overfeeds two
phases of the three-phase equipment. The application of the specific-energy
concept provides an understanding of why a three-phase equipment suffering 18%
voltage sag during 6 cycles remains undisturbed. During this event the other two
phases were increased by 16%, which gives a specific energy fall of only 5%, a
value that can be withstood by the equipment without any difficulty during the
circuit-breaker operating time [9].
It has been reported that one phase-to-phase voltage must sag to below 75% of
nominal and stay below 95% for six or more cycles to disrupt operations. Also,
three-phase faults are more devastating to the plant because the phase-to-phase
voltage does not have to drop as much on all three phases as one event affecting
just one phase-to-phase voltage. The application of the specific-energy concept, as
expressed by Eq. 10.2, will clarify and simplify the analysis.
It is not difficult to understand the concept of specific energy for the SE ride-
through capability, however, this value should be correlated with something similar
to the probability of occurrence in the system under analysis, and the system-
performance indices. Besides, the three-phase SE will have a unique value of ride-
through specific energy, which should be compared with the single-phase value or
three-phase composite value, depending on the characteristics of the event under
study (voltage sag). Furthermore, it should be noted that the probability of three-
phase voltage sags is much less than the presence of single-phase dips. Also, the
type of transformer connection modifies the voltage-sag effect. For instance, a
single-phase fault in the utility system is seen as a two-phase voltage sag on the
customer side of a delta-wye utility transformer.
The specific-energy concept has been proposed for the analysis of SE
immunity, showing its application not only to individual SE but also to sensitive
industrial plants, production lines, and parts of distribution systems [6]. The effect
of voltage sag on some specific sensitive equipment is not only related to the
magnitude and duration, but also to the phase-angle jump (unusual voltage zero-
crossing position), such as ASDs. The phase-angle jump effect is not considered in
this work.

10.4.2 Transformation of Protective Device Time–Current Curves into Time–

Voltage Curves, for Voltage-Sag Coordination Studies

The present study is initially done on a simple distribution system where the effect
of DG presence can be easily analysed. The system is typical of rural, industrial, or
low-power distribution schemes. Most of the distribution systems are operated
radially due to the simple and cheap overcurrent protection. The change of the
whole distribution protection in order to allow the unlimited DG penetration is a
very expensive, and perhaps unrealistic, idea. The operation would be optimised
under the present situation in order to do only a certain corrective work.
 Distributed Generation 301

It should be taken into account that the type of fault is strongly influenced by
the generator and isolation transformer connections. Transformers with an
ungrounded primary would not contribute significantly to single line-to-ground
faults, thus the utility protection would delay its trip signal. In this case, a
temporary fault can become permanent with the disadvantage that a momentary
interruption turns into a sustained one and the voltage sag is now followed by an
outage [6]. The present study is carried out from two points of view, voltage sags
and overcurrent coordination.

A. Radial system from the voltage-sag coordination point of view

The classical study of overcurrent protection–voltage sag coordination is given
using the single-phase equivalent circuit shown in Fig. 10.2, where PCC indicates
the point of common coupling and SE specifies the sensitive equipment
(equipment having low voltage-sag ride-through capability). Basically, the PCC
will be defined as the point of the circuit where the SE current is generally
separated from the distorted or too high current path. In this case it is the short-
circuit main flow path as can be seen in Fig. 10.2 [11].

Steps of the coordination methodology have been reported in [9].

The voltage applied to SE during the three-phase fault is:

VSE = 1 – If Zs [pu], (10.3)

which can be rewritten as,

VSE = Zf / (Zs + Zf) [pu]. (10.4)

Figure 10.2. Simple radial circuit for voltage-sag studies

302 J. C. Gomez Targarona and M. M. Morcos

When DG is introduced into the system, the circuit of Fig. 10.2 is transformed
into the complex form of Fig. 10.3, where the classical concept of PCC, which
considers that the energy-flow direction is unique, is no longer valid. The circuit
now includes the DG impedance (Zd) and the interconnection impedance (Zi),
where Zd is a nonconstant value, usually represented by the DG transient
impedance. Also, each of the two branches (where SE and the faulted branch are
connected) could be located anywhere between the DG and the power-system
supply points.

Figure 10.3. Circuit with DG for voltage-sag studies

Depending on the locations, four main cases can be analysed.

Case I – with the faulted branch connected at the power system side and the SE
at the DG side. Case II – both branches are at the system side. Case III – both
branches at the DG side. Case IV – the SE connected at the power-system side and
the faulted branch at the DG side.
For each of the four extreme cases the circuit behaviour and equations would be
different. From a quick analysis of the cited cases, it can be deduced that:

1. Cases I and IV: DG power-supply combination determines a minimum

voltage-sag magnitude at the SE connection point. These minimum values,
expressed in pu, are Zi / (Zd + Zi) for case I, and Zi / (Zs + Zi) for case IV.
2. Cases II and III: the power supply is reinforced by the DG collaboration,
increasing the short-circuit current at the PCC, with Zi in series with Zd in the first
case, and with Zs alone in the second case. Depending on the power system and DG
relative sizes, impedance Zi can represent a substantial short-circuit current
limitation. For these two cases there are no minimum voltage-sag values.

In general, DG can be considered as mitigating voltage sags in two ways:

increasing the fault level (reducing system impedance Zs to Zs in parallel with Zd),
 Distributed Generation 303

and keeping up the voltage at the neighbouring load (VSE) by feeding the fault. For
the circuit shown in Fig. 10.2, the coordination study (carried out by converting the
protective device time–current characteristic curve into a time–voltage curve) is
relatively easy.
When both types of data, protective device and SE-immunity characteristics are
overlapped in the voltage–time graph, the interception (if any) between the ride-
through-capability curve and the protective device curve indicates the SE dropout
possibility. The comparison is not direct, because the fuse TCC shall be
transformed in the voltage–time curve, which can be easily done by subtracting the
voltage drop caused by the fault current in the PCC impedance (Zs) from the
system voltage, shown in Fig. 10.2. In order to fill the graph, as the work is done in
percentile values, the only necessary information from the circuit is the short-
circuit power at the PCC. In this way the voltage-sag depth is determined for a few
selected time durations, with eight values being enough due to the regularity of the
fuse TCC curve. For very high short-circuit currents (or faults near the PCC), the
voltage sag is practically zero, needing a very fast protection that interrupts the
current in less than half a cycle in order to avoid sensitive-equipment disruption.
Due to device characteristics, the only protectors with sufficient speed are current-
limiting fuses and in some extension CL circuit breakers. If for some reason (for
instance the existence of limiting inductances) the current does not reach this high
value, any device of the non current-limiting type can be used, directly comparing
the TCC and voltage–time curves, as shown in Fig. 10.4. In the figure, the motor
start and load-pickup characteristics have also been drawn, in order to determine
the mutual effect of the protective device, sensitive equipment and overcurrents
due to faults, motor starts or load pickups [12].

Figure 10.4. TV characteristics of fuses, circuit breakers, sensitive equipment, motor start
and load pickup
304 J. C. Gomez Targarona and M. M. Morcos

Now, besides the circuit complexities, an additional uncertainty is provided at

the instant when the DG circuit breaker opens, changing the circuit behaviour and
altering both fault-current and voltage-sag magnitudes. The study of this situation
is given below, using the specific-energy concept.

B. Radial system from the overcurrent coordination point of view

Due to its simplicity and uniformity from the time–current characteristic point of
view, the study will be carried out for fuses, explaining the main differences if
circuit breakers are used instead. The DG presence in the circuit can cause fuse
blowing, since the fuse current now has two components, one coming from the
supply (already considered during the distribution system design), and the other
from the DG, which creates an acceleration of the fuse operation. Figure 10.5,
which is derived from Fig. 10.3, shows a typical radial circuit with five possible
fault locations and subsequent fuse operations that require a special analysis.

Figure 10.5. Typical rural radial system with distributed generation

1,2
Current and voltage

1
0,8
(pu)

0,6
0,4
0,2
0
-0,1 0 0,1 0,2 0,3 0,4 0,5 0,6
Time (s)
Faulted branch current SE voltage

Figure 10.6. Faulted-branch current and SE voltage for a circuit having a double supply
 Distributed Generation 305

1. Fault SC1: current mainly supplied by the power system (generator G) would
cause fuse F1 to operate, and if the DG circuit breaker is not fast enough it also
would cause fuse F2 to operate. The operation of fuse F2 would cause some
confusion on the task aimed at locating the fault, but, would allow that part of the
circuit to continue to be supplied by the DG (islanding). It must be remembered
that islanding is not always allowed by the local authorities but sometimes is
desirable from the users’ point of view.
2. Fault SC2: current supplied by the power system would cause fuse F2
operation. The DG circuit breaker would also trip (in a shorter or longer time).
3. Fault SC3: current supplied by the power system as well as by the DG would
cause fuse F3 operation. If the fuses F3 and F2 have been properly coordinated
without considering the DG, the new situation would enhance the coordination by
accelerating fuse F3 fault interruption. The DG circuit breaker can also open
depending on its coordination with F3. When the power system can not supply the
loads without the DG or if this generation is for some reason necessary, it would be
desirable for the DG breaker not to open.
4. Fault SC4: the only fuse that needs to be operated is F4. Fuse F1, if properly
coordinated, would not operate. The inconvenience comes from fuse F2 operation
caused by the DG supplied current (when fuse F2 rated current is similar to or
lower than that of F4), leaving part of the system operating as an island. This
situation is possible when an important load is supplied from the feeder but its
frequency of occurrence is very low. The possible existence of this situation needs
to be analysed further.
5. Fault SC5: the only fuse that needs to operate is F5. Fuse F2, if properly
coordinated would not operate. Also, it is desirable that CB remains closed.

From the cases mentioned above, the most conflictive situation is presented by
fault SC4, which requires special consideration.

The coordination analysis is done using the graph shown in Fig. 10.6. Now, the
specific energy of current, I2t(t), needs to be applied instead of the classical time–
current characteristic I(t), due mainly to the large variation of the DG current
supplied during the early part of the short-circuit process. In this conflictive
situation, the specific energy of G and DG currents through F4 (prearcing + arcing)
needs to be lower than the F1 prearcing energy due to G, and also lower than F2
prearcing energy due to DG. The new problematic application, when DG is
connected in the system, is analysed below.

C. SE Voltage-sag Ride-through Capability

The study is carried out considering the faulted branch being protected by fuses
and that the DG possesses a circuit breaker commanded by a definite-time relay.
The analysis is directed to determine if the voltage sag caused by the fault current,
with the consequent DG disconnection, would cause SE dropout. The specific-
energy concept can be applied to this study based upon its capability to analyse PD
and SE behaviour as time functions [9]. Figure 10.6 shows the voltage-sag and
current magnitudes due to a change in source impedance when the DG circuit
breaker opens. The fault impedance was assumed to initially produce a fault
306 J. C. Gomez Targarona and M. M. Morcos

current (1 pu) five times the prefault load, and a voltage sag to 0.45 pu. It was also
assumed that the DG breaker opens after 200 ms, reducing the current to 0.6 pu
and the voltage sag to 0.3 pu. The PD operation ─ 490 ms after the fault started ─
has restored the voltage to 1 pu. The comparison between the voltage-sag specific
energy and the SE ride-through-capability curve would indicate the SE withstand
or dropout conditions. In other words, the SE would drop out if the energy deficit
(due to 0.45 pu voltage for 200 ms and 0.30 pu voltage for 290 ms) was higher
than the SE withstand-energy deficit.
A generic circuit ─ derived from that shown in Fig. 10.3 is presented in Fig.
10.7, where the interconnection impedance has been divided into three components
in order to allow the branch locations to be considered. The components are
derived from the original impedance Zi multiplied by three factors, whose sum is
always equal to one. The three factors transforming Zid, Zim and Zis allow the
representation of any particular case. Table 10.1 shows the values for the
previously described cases, I through IV.

Figure 10.7. Generic circuit with DG for voltage-sag studies

Table 10.1. Impedance factors for the four extreme cases

Cases
Impedances I II III IV*
Zid 0 1 0 0
Zim 1 0 0 1
Zis 0 0 1 0

*Note: for this case Zs needs to be replaced by Zd

The formula below, developed from the circuit in Fig. 10.7, is the general
expression for the SE voltage calculation:
 Distributed Generation 307

VSE [pu] = 1–(Zs+Zis)*(Zd+Zid)/(Zf*Z+(Zs+Zis)*(Zd+Zid+Zim)) (10.5)

where, Z = Zs + Zi +Zd.

The situation after the DG breaker opens can be easily calculated using the
above expression and considering Zd as infinite. In conclusion, the coordination
task to be carried out consists of the estimation of the specific energy of the fuse
current as a function of time, together with the calculation of the specific energy of
the SE voltage, also as a function of time. The new situation needs to be considered
after the DG breaker opening, if it happens. If the SE specific-energy value is
reached before fuse opening, the SE would suffer a dropout. The aim is to have the
fuse open before the SE stored specific energy is exhausted. The SE lack of
specific energy is greatly influenced by the operating time of the DG breaker due
to the fault backup given by the DG, as can be concluded from Fig. 10.6.
The procedure can be applied also to determine the improvement in the SE
ride-through capability when a DG of a given power is connected somewhere to a
simple radial circuit. Figure 10.8 shows an example of the procedure (the values
correspond to Fig. 10.6); the thick curve represents the increase in the specific
energy (of current) as a function of time, whereas the thin curves, dotted and solid,
represent the defective specific energy (of voltage), respectively. The fuse would
operate when its curve reaches the prearcing + arcing specific energy, for our case
0.066 pu2 s (of current, pointed by the upper arrow), and the SE would drop out for
a value of 0.205 pu2 s (of voltage shown by two collinear arrows). It can be seen
that the DG breaker tripping at 200 ms (point from which dotted and solid lines
depart) causes the SE dropout at 420 ms (left, lower arrow), before the fuse
operation, which takes place at 490 ms (shown by upper arrow). The dropout could
be avoided by delaying the breaker operation to 380 ms (indicated by a slope
change of the thin solid line). The acceptability of the circuit-breaker delay needs
to be carefully considered due to the possible damage to equipment and hazard to
personnel, in spite of being a weak distribution system.

Figure 10.8. Comparison of fuse and SE specific energies

308 J. C. Gomez Targarona and M. M. Morcos

D. Voltage-sag and overcurrent protection coordination using time ─ voltage

characteristics
The electric fuse, or simply “fuse”, is the most commonly used protective device in
distribution systems, being manufactured in a series called a homogeneous series,
defined by ANSI/IEEE C37.41 and IEC 60282. The homogeneity signifies that any
fuse is similar to the one having higher or lower rated current, with the only
difference based on the load-current capability. The characteristic curves of
dissimilar fuse ratings are parallel to each other due to the use of logarithmic scales
on both axes [13]. The same homogeneity can be found in circuit breakers, having
much more tailoring possibilities. As a result of this homogeneity the time–current
characteristic curves for all PD (of the same series) can be approximated by one
curve only on the graph, expressing the horizontal axis in multiples of the current
rating instead of A or kA. Applying the selected PD individual curve can later
solve this approximation.
In order to clarify the idea, the study would begin with the analysis of the
simple radial circuit shown in Fig. 10.2, from which the expression for the voltage
applied to faulted SE was obtained. When a fuse is used as a protective device
connected on the faulted branch, the fault current can be expressed as a multiple of
the fuse current rating, with Irf being the fuse current rating and kf the multiplier.
For example, in Fig. 10.9 the fuse under study (rated current 200 A) having a given
current value will operate in a time of 1 cycle, and after doing the corresponding
conversion from the time–current characteristic into the time–voltage
characteristic, the voltage drop due to this current will create a voltage sag to 80%.
Sensitive equipment would withstand during this time period a voltage sag as deep
as 20%, thus the equipment will withstand this event. As the entire fuse
characteristic is to the upper left of the SE curve, the equipment will be protected
by fuse operation.

Figure 10.9. Example of fuse time–voltage characteristic and the CBEMA curve
 Distributed Generation 309

A similar procedure can be followed when circuit breakers are applied, using
the methodology for the I(t) characteristic transformation into the V(t)
characteristic using Zs as the conversion factor [9].

Equation 10.3 can be rewritten as,

VSE = 1 – kf Irf Zs [pu] (10.6)

where the product IrfZs represents the system impedance voltage drop, considering
the faulted-branch as the only circuit load. If there are other loads connected, the
faulted-branch load can be represented as a pu value of the total load, expressed by
kl, which is always less than one.

Equation 10.6 now becomes:

VSE = 1 – kf kl Irf Zs [pu]. (10.7)

This expression can be represented on a time–voltage graph, together with the

SE immunity data given either by CBEMA-ITIC or IEEE standards, as shown in
Fig. 10.9. For this work, the CBEMA curve was selected based upon its worldwide
application and also due to its regular form. The fuse curve has been determined
for both Zs and Irf values.
A generalization can be obtained by changing the vertical axis scale into
logarithmic, then the curves representing fuse ratings, source impedances, and load
percentage would shift horizontally, taking the point defined by a 100% voltage
and 0.01 cycle as the new origin. In other words, once the study has been done
under basic conditions, it is very easy to improve the results by changing any of the
constants in Eq. 10.7, which leads to the shifting of the fuse–time curve. The
changeable constants are kf, kl, Irf, and Zs, which implies modifications on the short-
circuit current, load sharing, fuse current rating, and source impedance,
respectively. Any of the four constants has the same weight on the curve shifting,
besides the final curve location can be obtained by a combination of changes in
some, or all, of the four constants. Figure 10.10 shows the change obtained by the
modification of the source impedance.
When the conditions are changed, for instance the new fuse characteristic has
moved to the right and down as shown in Fig. 10.10, the point corresponding to 1
cycle will be withstood by the SE, but as some crossing points are shown, some
voltage-sag events will cause the SE dropout.
310 J. C. Gomez Targarona and M. M. Morcos

Figure 10.10. Similar to Fig. 10.9 with source impedance modified

Following the same analysis, when DG is used, the SE and the faulted branch
have two energy sources, the mains and the DG. When the fault starts, PD of both
sources would detect the overcurrent and one of the devices would be open at first
producing a sharp change in the source impedance. The proposed methodology
would allow the SE ride-through analysis to be carried out, as the device opening
produces a change in the source impedance, as described above. For instance, if a
short-circuit takes place, the voltage sag detected by the SE would be 80%. If after
1 cycle the source impedance is changed, the voltage would fall to 40%. As the
fuse curve is still to the left of the immunity curve, the protective device would
clear the fault avoiding SE dropout. If the circuit modification produces a slightly
lower curve, the SE would not be able to withstand the voltage sag due to the
intersection of the two curves, Fig. 10.10.
The methodology can be applied, based on the specific-energy concept, which
signifies that the stored energy is slowly exhausted at first during one cycle
(voltage sag 80%), then quickly consumed, since the voltage sag is now 40%. As
the total specific-energy deficit is still lower than the dropout curve, the SE would
ride through the event.
In the case of using more sophisticated PD, data management is not as easy as
with fuses, due to the possible time–current characteristic changes. Modern devices
allow the modification to be done with both instantaneous and delayed trips, as
well as with multiples of the pickup current. Figure 10.11 shows both the CBEMA
immunity curve and one of the available characteristics after being transformed
into a time–voltage curve. The characteristic curve is formed by several horizontal
and vertical segments that can be moved either vertically or horizontally, so each
location would have its own constants.
 Distributed Generation 311

100

Voltage (%)

10
0,01 0,1 1 10 100 1000 10000
Time (cycles)
CBEMA Modern PD

Figure 10.11. Coordination of CBEMA and modern PD curves

It would be very helpful if the PD manufacturers provided the basic

characteristics and constants for allowing any modifications – related to source
impedance, branch load, and short-circuit current – to be done. The basic curve can
be completely defined by two figures, the specific energy and the minimum
voltage for steady-state operation.
The different coordination methodologies described above present a short guide
for the improvement of the protection–voltage-sag coordination. By using the
general expression represented by Eq. 10.5, it is possible to consider the faulted
branch and DG connection point together with the supply/DG impedance ratio. The
temporal behaviour of the applied protection scheme having dissimilar operation
characteristics can be easily calculated using the specific energy instead of the
time–current characteristic approach. In the case of having several SE connected to
the same branch, the envelope of the combined immunity curves would be used
instead of the individual SE V(t) characteristic. In this way, both the instant at
which the protective device operates and the SE dropout time can be calculated,
and by comparing the two values it can be determined whether the SE would ride
through the event or would drop out. The methodology provides the distribution-
system engineer with the necessary tools in order to do the protection
coordination/power quality tradeoff.

E. Coordination study with variable-supply scheme

As was previously explained, the coordination study between the overcurrent
protection device (PD) and the SE voltage immunity located in parallel branches or
feeders, is done in graphical form, comparing the adapted time–current
characteristic (TCC) of the protective device with the previously cited CBEMA
curve. An adapted protective device TCC is a curve transformed into a time–
voltage characteristic (TVC), that represents the voltage sag that the protective
device allows to be applied to the SE under study. Voltage sag applied to SE is
312 J. C. Gomez Targarona and M. M. Morcos

directly the voltage at the PCC or the supply voltage minus the voltage drop across
Z1 due to the short-circuit current [12].
Figure 10.12 shows the new scheme of the system with utility and distributed
generation, connected by an islanding circuit breaker (ICB), where DG may be
representing more than one device. When the islanding circuit breaker is closed the
source impedance is approximately the parallel combination of the utility (area
EPS) and DG impedances. When ICB is opened the source impedance jumps to a
larger value, thus in the presence of a fault, the short-circuit fault current will be
greatly attenuated and the operating time of the protective device will be increased
subsequently. Thus the coordination study needs to be done only for the two
branches on the DG side of the ICB, with maximum and minimum rated currents
of the protective device.

PCC DG

ZDR

SE

PD

ICB

EPS

Z1

Figure 10.12. System with utility and distributed resource

Considering that the utility and DG have nearly the same voltage-regulation
value, the voltage-sag magnitude at the SE connection point (or PCC) is given by,

Vs = VEPS – (Z1 // ZDR) Isc [pu], (10.8)

where,
Vs = voltage-sag value
VEPS = electric power system voltage
Z1 = utility impedance
ZDR = distributed resource impedance
Isc = short-circuit fault current

The product ZIsc in Eq. 10.8 can be identified. During the protective-device
TVC calculation, once the rated current has been adopted, each selected time value
will correspond to one value of the voltage-sag magnitude. Isc is expressed as a
 Distributed Generation 313

function of the PD rated current, thus the fault current will be a number of times
the rated current.
Normally, fuses are built in a homogeneous way with the same structure and
with characteristic curves approximately parallel, thus, increasing the number of
fuse elements will increase the rated current. Hence, the Isc value expressed as a
multiple of the fuse rated current has a multiplier that increases with the decrease
in the fuse rated current and vice versa. A short-circuit current of 1000 A will
trigger the operation of a fuse of 100-A rated current in nearly 1.4 cycles and the
same current will melt the 200-A rated fuse in approximately 20 cycles. In
conclusion, the product Z Isc can be considered as a function of the source
impedance and PD rated current. Thus, a reduction in the short-circuit power
(islanding mode) is equivalent to an increase in the rated current of the protective
device and vice versa.
The previous analysis, although applied to fuses, is applicable also to circuit
breakers and to any other type of low-voltage protective device.
The effect of a protective-device rated current can be seen in Fig. 10.2, where
the TVCs of the two fuses mentioned above where plotted. To clarify the
procedure, the methodology for obtaining a point on each curve is as follows.
Considering an application with a load rated current of 1000 A and a source
impedance of 4% (source impedance voltage drop due to rated current is of the
same value as the voltage regulation), with two feeders coming out from the bus-
bar. They are protected by fuses with rated currents of 100 A and 200 A,
respectively.
An arbitrary time of 10 cycles (200 ms) was selected for the explanation, for
which the fuses will need currents (to melt) of 600 A and 1200 A, respectively.
The fuse-rated currents expressed in pu of the circuit-rated current are 0.1 and 0.2,
respectively, then the base current for 10 cycles is 6000 A. Voltage sags that the
fuses allow to be applied to SE connected to the parallel feeders are:

• For the 100-A fuse; Vs (%) = 100 – (0.04 0.1 6000) = 97.6%;
• For the 200-A fuse, Vs (%) = 100 – (0.04 0.2 6000) = 95.2%.

The same procedure allows the calculation of the complete fuse TVC curves.
The protecting-device rated current is crucial for the coordination between the
protecting device and the SE immunity curve. This is due to the fact that, as the
rating current increases the curves approach each other, and when they intersect
there are points of miscoordination.
If, during parallel operation of utility and DG, due to any cause, the IB opens,
the circuit suffers a source-impedance increase, modifying the situation. Figure
10.13 was drawn for a similar situation as before, but changing the source
impedance from 0.04 to 0.06 pu and maintaining similar fuse-rated currents. It can
be seen that the protection given by the 100-A fuse is still satisfactory, but the
curve representing the 200-A fuse now intersects the immunity curve of the
sensitive equipment, which might signify SE dropout.
314 J. C. Gomez Targarona and M. M. Morcos

100

80
Voltage (%)

60

40

20

0
0,01 0,1 1 10 100 1000 10000
Time (cycles)

CBEMA Fuse 100 A Fuse 200 A

Figure 10.13. Fuse time–voltage characteristics and CBEMA curve for DG alone

It may be deduced that the protective-device TVC has been transformed into a
zone bordered by the curve locations when the supply is the utility plus DG and the
new curve location having DG as only the supply. In the present case, it is the zone
enclosed by the positions of the 200-A fuse curves for both source impedances.

10.4.3 Mitigation of Distant Voltage-sag Penetration into Industrial Premises

by Using Semirigid Connection

As was mentioned above, the ability of sensitive equipment to withstand or

ridethrough these voltage sags without dropout is a function of the magnitude of
voltage reduction and its duration. The curve proposed initially by the Computer
Business Equipment Manufacturer Association (CBEMA) has been adopted as the
graph of immunity. This means SE will withstand voltage sags represented by
points given by their depth and duration as coordinates, which are to the left and
above the graph line. The points below the curve and to the right will represent
voltage sags that produce SE dropouts.
The concept of having points that represent perturbation depth, duration, and
location on the CBEMA graph is only applicable if the voltage-sag magnitude is
constant during the event duration. Voltage sag caused by a short-circuit, possesses
a depth that depends on the fault current and on the event location, and the time
taken for the protective device to disconnect the faulted area. In this case there is
no control on the voltage-sag depth but on the duration that depends on the delay in
the protective-device response.
For the many cases where the voltage magnitude is far from being constant, the
analysis related to SE dropout can be done by using the energy integral during the
perturbation and comparing it with the energy necessary for SE energy. If the
 Distributed Generation 315

available energy is higher than that required by the SE, the equipment will
withstand the event without dropout [9].
The concept is based on the energy deficit and operating voltage; the ride-
through (or dropout) occurrence is determined by comparing the SE prefault stored
energy with the energy deficit during the voltage-sag event. Based on the above
explanation, the possibility of SE ride-through will be increased if the depth of the
voltage sag can be reduced during part of the perturbation duration. For the case of
distribution systems with DG, the extra generators can be kept connected in order
to support SE, thus reducing the energy supplied to the fault.
Using a connection between DG equipment and the supply system, whose
rigidity is variable depending of the working conditions, can carry out this task. If
the system is under steady-state conditions, the connection is “rigid” (or of very
low impedance). When the presence of an external fault is detected, the
interconnection must become nonrigid – having a significant impedance – in order
to support the SE and reduce the fault-current contribution.
Figure 10.14 shows a simple distribution system having a DG with a variable
interconnection. Zv&CB represent the combination of limiting impedance and
bypass circuit breaker, supplying a rigid connection when the circuit breaker is
closed, switching to nonrigid connection when open.

Gs Zs Zf fault

DG ZDG Zv & CB

Load
Figure 10.14. Simplified distribution system with embedded DG

Figure 10.15 shows the voltage applied to the SE for a situation similar to that
presented in Fig. 10.14 when a fault takes place, assuming impedances and faults
that result in voltage sags to 30% (0.3 pu) and 80% (0.8 pu), followed by a voltage
swell to 110% (1.1 pu). The duration was assumed by considering the circuit
breaker needs 40 ms to detect and transfer the fault current and that the fault is
cleared in 160 ms after the fault start. As a portion of the system was severed from
it, the assumption was that the voltage would be increased to 110%. It was also
assumed, as in the normal situation, that the system power is at least 20 times
higher than the DG power in order to neglect the change in fault duration due to the
316 J. C. Gomez Targarona and M. M. Morcos

current supplied by the DG. As this hypothetical voltage sag is not of constant
depth, the CBEMA curve cannot be directly applied; the use of the specific-energy
concept was necessary.

The energy deficit suffered by the customer’s SE, without using the nonrigid
connection, would be:
EDr = (87– 30%)2 0.16 s = 520%2 s.

Using the nonrigid connection:

EDnr = (87– 30%)2 0.04 s + (87– 80%)2 0.12 s = 136%2 s.

In spite of the unconventional unit, the energy deficit was reduced from 520 to
136. In other words the voltage-sag effect is reduced to 26% by using a nonrigid
connection, thus the possibility that SE would withstand the perturbation is greatly
increased.

1,5

0,5
Voltage (V)

-0,5

-1

-1,5
0 0,05 0,1 0,15 0,2 0,25 0,3
Time (s)

Figure 10.15. Voltage applied to SE during a hypothetical fault

– Numerical example:
A typical distribution system with a customer having his own DG was
modelled using MATLAB. The system characteristics are: rated voltage 13.2 kV,
rated power 30 MVA, DG rated power 3 MVA. A three-phase fault was assumed
10 km downstream of the DG location. Voltage calculations for different fault-
currents and limiting-impedance values on two circuit locations were done. The
circuit locations were assumed where the DG owner has the SE and where another
customer is connected to the main 13.2 kV feeder. For all the calculations the CB
detection plus commutation time was kept constant at 40 ms. The comparison is
made between the specific energy of SE that belongs to the customer with and
without DG and Zv&CB.
Table 10.2 shows the analytical results where the comparison can be made for
different Zv and fault-current values. The current values are expressed as a
 Distributed Generation 317

percentage of the three-phase short-circuit current located in the part of the

distribution system mentioned above.

Table 10.2. Specific-energy deficit depending on Zv and short-circuit currents

Impedance Ifault Specific-energy lack Specific-energy lack

with DG and Zv&CB without DG and
Zv&CB
Ohm % s-c %2 s %2 s
j 15.7 3 19.22 438.9
j 15.7 6 30.60 276.4
j 15.7 6 26.04 388.7
j 15.7 10 33.31 757.1
j 6.28 50 47.87 1062.6
j 3.14 100 60.83 955.5
j 31.4 100 64.80 952.2

The table shows the remarkable effect on the reduction of the specific-energy
deficit caused by the application of a nonrigid connection.

10.4.4. New Overcurrent-protection Schemes using Intelligence

There is a strong trend aimed to add flexibility to the traditional distribution fuse.
The cited flexibility can also be called intelligence, where an intelligent fuse is one
that is able to make decisions, for instance doing opening and reclosing operations,
discriminating direction and detecting phase currents from earth currents.
The current widespread application of distribution fuses is based upon their low
cost, high reliability that arises mainly from their simplicity, and from the
application of well-known physical principles. Since the earliest fuse, the first
scientific reference was by Sir Edward Nairne during 1773 and at first official US
fuse patent granted to Thomas Edison by 1880, the main improvements have been
aimed at the use of better materials, to extend the current and voltage application
ranges, and towards the development of faster and cheaper construction techniques
[14, 15].
After these early improvements many changes of the original fuse design have
been presented, in order to extend the low current interruption capability having
different acceptation or success degrees.
During more than 120 years of fuse application many improvements have been
published at conferences and magazines, but just a few have been readily available
and of stable market position. The phrase “New fuse design or type for specific
applications …” has been groundlessly expressed many times. No critical changes
towards adding intelligence to the traditional fuse have been produced lately.
A literature survey ─ of the innovative additions to the early fuse designs ─
was conducted to analyse the presented DG problem, just two original ideas have
arised as the application of chemical charges and the use of saturable transformers
[15]. The two ideas can be summarised as follows.
318 J. C. Gomez Targarona and M. M. Morcos

a) Application of chemical charges:

The idea was originally presented by Muth and Zimmermann in 1938, based
upon the fuse blowing due to the ignition of a chemical charge caused by any
externally controlled heater, the device described as the combination of the
accuracy of a relay with the cheapness of the fuse. Later, the same idea was
pursued and developed, especially on the ignition control system, introducing a
device called “limiter” in 1963, which has been on the market ever since. The same
concept was further developed and applied to the Automatic Seccionalizer, coming
out in the 1980s. In the same decade many other designs using this principle were
presented, developed especially for medium-voltage distribution systems, having
dissimilar market success. One of them, called the electronic power fuse loaded
with up-to-date or last-generation electronics has a relatively good foothold in the
market, in spite of its high cost. By 1990 a technical paper was published that
presented a new design applying the concept to low-voltage DC systems, called a
Smart Fuse. Five years ago, limitations on the application of this chemically
charged device in distribution systems were emphasised giving recommendations
about how to avoid frequent misapplications [16, 17].
b) Use of saturable transformers:
During the 1970s an interesting idea was proposed, related to the availability in
a single fuse cutout of a double fuse time–current characteristic. This double TCC
was obtained by using a current transformer whose working zone included the
saturated and nonsaturated areas, changing the two paths current sharing depending
on the overcurrent level. The design was proposed for the solution of coordination
problems between the main feeder and branch protections. Initially, the current
transformer was located by the detachable fuse link, lately being attached to the
fixed cutout part [18].
One of the most important needs is the availability of medium-voltage fuses
designed especially for distribution systems having embedded generation, where
particular requirements shall be considered. Figure 10.6 shows a typical system
having embedded distributed generation. In Fig. 10.6 the sensitive-equipment and
faulted-branch locations are shown in distributed generation and main supply sides
respectively. Currently, there is a big movement in the field of distribution
generation having DG mainly due to the present strict power-quality requirement,
voltage-sag presence is one of the biggest issues [6].
From the voltage-sag viewpoint, it is very important to keep the DG connected
and feeding the system during fault events, due to its capability to back up the
system voltage, increasing the possibility that the sensitive equipment would
remain operative. On the other hand, there is a serious risk of keeping part of the
system working under “islanding” conditions. In addition, there are difficulties in
getting selective coordination for all possible operating conditions, having the fault
energy coming from either the main supply or from the DG, the DG presence in the
system is the decision of the DG owner and not of the main utility.
The solution to the listed drawbacks can be done by using complex and
expensive protection schemes, as many of the needed tasks are beyond the reach of
a traditional fuse. Among the intelligence factors that a current fuse needs to have,
are directionality and discrimination if ground is included in the fault path. In
 Distributed Generation 319

addition, the solution must be of low cost so as to keep the complex fuse as an
attractive possibility if compared with other choices.
Both ideas have been combined in a design that analytical studies have shown
to be very auspicious. As the design is entering the patenting process, no further
details can be provided here [15].

10.5 Impact of DG on Recloser–Fuse Coordination

The recloser is an invaluable tool for the reliability of rural and low-power
distribution systems. The success of its application is based on the fact that most
faults are of a temporary nature or start as temporary (approximately 85 to 90% of
faults), and if momentarily interrupted they are self-eliminated during the reclosing
interval. In the case of a recurrent fault (or a permanent one), another fast (or
delayed) operation is provided in order to allow the operation of the down-stream
fuse, isolating the faulted-system sector in this way. At each reclosing operation
the adjacent customers are subject to a new voltage sag. Some utilities especially
those with a high percentage of urban and suburban loads, have changed their
reclosing-protection practices, eliminating the second instantaneous trip (fuse-
saving operation), as can be seen in Fig. 10.16 [19].

Figure 10.16. Voltage sags generated by a recloser without the second instantaneous
operation

Generally, the first fast operation eliminates 85% of the faults, 4% are cleared
during the second reclosing interval, and only 1% are eliminated in the third
interval. Ninety percent of faults are removed without fuse operation or
maintenance calls [20]. Unfortunately recloser and DG are irreconcilable; if DG is
left connected during the recloser fault-clearing operation, fault clearing can be
incorrect and there is a high risk of DG damage when reconnected.
320 J. C. Gomez Targarona and M. M. Morcos

One of the main problems is the possible lack of coordination during the fast
operation of the first recloser when DG is connected at any point between the
recloser and the fuse nearest, and upstream, to the fault. The current flowing
through this fuse is supplied by the power system as well as by the DG, both can
cause fuse operation during the first fast recloser operation, making the first
reclosing interval useless for fault self-clearing. During the reclosing time the DG,
if it remains connected, will supply current to the transient fault avoiding its
deionisation. Experimental studies have indicated that a current of 5 A supplied by
the DG during the reclosing interval would be large enough to sustain the fault,
preventing the needed self-clearing from taking place [21]. Thus, the recloser´s
main advantage is eliminated, and the protected equipment will suffer unnecessary
“through faults”, shortening the equipment life span and increasing repair and
maintenance cost. Also, customers will experience several long interruptions,
losing the benefit supplied by recloser applications. The temporary fault is then
considered permanent and the recloser would move to delayed operation and
possibly to a lockout position. In this way the advantage of clearing nearly 85% of
the faults at the first operation is eliminated.
The recloser user can normally adjust the reclosing interval in order to meet its
own requirements. A reclosing interval of 2 s is normally used; a value that
traditionally was able to solve most of the existing problems. At present there is a
strong tendency to reduce this value to 0.5 s in order to improve the power quality
of customers having digital clocks, reducing customers´ complaints due to the
nuisance of digital-clock blinking. The use of instantaneous reclosing when DG
remains connected worsened the previously described problem.
However, if DG is connected to the circuit it is necessary to increase this
interval in order to be sure that the DG breaker has separated the generation before
the recloser reconnection. Some utilities are requiring a 5-s reclosing interval on
feeders having DG [6].
The problem of out-of-phase connection could cause physical damage to the
DG. This is an old issue that has been studied for induction motors ─ when
suffering short interruptions or voltage sags ─ that run as induction generators
during the event. The study normally concludes with the resolution of keeping the
motor connected for at least 500 ms in order to keep the production line in
operation as long as possible [22]. The main difference from DG is that the motor
slows down and perhaps stalls, subsequently reducing its emf. In the case of DG,
the generator excitation maintains the emf and the machine is kept running by the
prime mover, causing a speed increase [3]. It can be concluded that there is a need
for an engineering tradeoff between keeping DG connected and then improving the
continuity of part of the system (islanding and its own risks), or quickly
disconnecting it in order to increase the reliability of the whole distribution system,
which is detrimental to the users’ quality of power.

10.6 Harmonics Generated by Distributed Generators

Synchronous generators, when well designed and built, do not generate any
harmonic voltage. On the other hand, generators using modern switched inverters
 Distributed Generation 321

(such as induction, photovoltaic, fuel cell, etc.) should theoretically have a low
level of low-order harmonics that the distribution engineer is normally concerned
with. By controlling the switching, a relatively clean current waveform can be
achieved, which does not mean that there are no harmonics problems.
Underground cables seem to create natural resonance on the utility distribution
system in the 25th to 35th harmonic range. Some discussions have been started in
the USA with respect to the agreement with the stringent IEEE Standards 519
limitations on generators of some popular inverters. It is argued that this equipment
can be considered as a negative load, in which case the limits change from 5% to
20%. Besides, there is no problem to introduce in the inverter the filtering function
in order to reach the specified limits, but the cost would increase [23].

10.7 Flicker Due to Wind Gusts and Tower Shadow

According to the IEEE Standard Dictionary of Electrical and Electronics Terms,
flicker is the “impression of fluctuating brightness or colour, occurring when the
frequency of an observed variation lies between a few hertz and the fusion
frequency of images”. If the voltage variation is large enough or in a certain critical
frequency range, the equipment performance can be affected. The main problem of
voltage fluctuations is usually the effect on lighting loads, mainly when the
illumination of the lamp varies with frequencies between 1 and 10 Hz.
Additionally, the phenomenon can slightly affect other types of loads such as
motors, electronic devices and process controllers. Voltage fluctuations may
originate in the power system, but most frequently in the equipment or load
connected to it. The main voltage-fluctuation generators are arc furnaces, welders,
alternators, and motors [24, 25].
The application of wind turbines in isolated form or in a wind farm is becoming
widespread, thus the quality of the generated voltages and their impact on the
power grid, especially the high-power fluctuations resulting in flicker, are of great
concern to utilities and customers.
The main cause of flicker is the frequent switching on and off when the wind
powers the generator to a speed around the cut-in speed. The utility to which the
generators are connected usually limits the switching operation number to no more
than 3 to 4 times per hour. One of the possible solutions is to keep the rotor at
standstill until the wind reaches a stable speed beyond the cut-in speed, not letting
it free until the speed is exceeded by at least 1 m/s, and similarly to the switching-
off criterion. In this way, a certain hysteresis in the switching is introduced, which
reduces the flicker issue. This problem is very noticeable when many individual
generators form the wind farm, and obviously the wind energy is not uniformly
distributed among them.
Another utility compulsory limits is related with the network minimum short-
circuit level, which should be at least 20 times the wind-generator rating. This limit
is not easy to fulfill due to the fact that strong winds are available in the non-highly
urbanised areas, where the supply systems are usually very weak.
322 J. C. Gomez Targarona and M. M. Morcos

The wind turbine generator is also characterised by a low-frequency mode of

oscillation. Random wind fluctuations, wind shear and tower shadow effects (blade
passage along the tower), may excite this mode producing large ripple on the drive-
train torque as well as on the generated electric power, being noticeable as a
voltage flicker at approximately 0.5% and a frequency between 2 and 4 Hz.

A flicker coefficient for the wind turbine, a function of the specific wind
turbine and average wind speed, has been proposed in Denmark (one of the most
developed countries with respect to wind turbines), which should be between
certain limits in order to get installation approval. This coefficient is closely related
to the short-term flicker emission Pst previously defined and allows its calculation
taking into account the system short-circuit power [24].

Pst = [cos ϕsc + dQ/dP sin ϕsc ] Swn / Ssc Iw fw(W) (10.9)

where:
ϕsc = short-circuit angle
Q = reactive power
P = active power
Swn = wind turbine rated power
Ssc = supply system short-circuit power
Iw = average value of wind turbulence intensity
fw (W) = flicker coefficient (wind speed function)

The joint effect of several wind turbines on the Pst value can be assessed by the
application of a methodology analogous to the procedure designed for the flicker
generated by arc furnaces.
One of the most popular wind-energy conversion systems uses slip power
recovery. These systems employ a double-output induction generator, which is
connected directly to the grid by stator and through a static converter by a rotor.
The existence of ripple in the generated torque may damage the drive-train
components; besides fluctuations of the generated electric power may cause a
considerable flicker in weak grids. Several control strategies of the static converter
have been introduced in order to provide damping to the oscillation mode. One of
the proposed control systems follows the wind-speed variations resulting in a large
reduction of torque ripple and consequently control the voltage fluctuations.

10.8 Ferroresonant Overvoltages

There are some reports available that indicate that the application of DG through a
service transformer could lead to the presence of ferroresonant voltages. This
ferroresonance phenomenon, not exclusive to DG application, is presented when a
grounded delta-wye transformer is used as DG service equipment, having the DG
on the wye side and the delta side is supplied by an underground cable. If one of
the phase voltages from the supply is missing, the DG protection disconnects the
equipment, then a new series circuit through the transformer magnetising
 Distributed Generation 323

inductance and the cable capacitance is formed that leads to high voltages and
currents. If the phenomenon is not detected for a relatively long time, the
overvoltages can damage the transformer and its arresters [26].

10.9 Conclusions
The new scenario including distributed generation, changes the well-known
protection concepts. Sensitive-equipment protection against voltage sags can be
given for overcurrent protective devices, provided that careful coordination has
been carried out. The readily available protective device TVC now moves into a
zone that shall be above the SE immunity curve if effective coordination is desired.
A simple methodology has been presented, and its fundamentals and difficulties
are explained. The analysis is simple when fuses are used as protective devices
when using more sophisticated equipment having variable TCC. The methodology
has been specifically explained when applied to circuits having distributed
generation, which produces a sharp impedance change. The application of a
nonrigid connection between the customer’s DG and the system supply can be used
as an able tool to reduce the effect of voltage sags caused by faults located outside
the customer circuit. Coordination studies could be facilitated if PD manufacturers
would provide a new set of data for their products. Distribution systems is still a
field where fuses are strongly suited, thus any effort towards sustaining and
increasing their presence would be worthwhile. Medium-voltage distribution fuse
technology is urgently in need of new and fresh ideas. The information needed to
carry out the proper study under the new situation is available. By using the
traditional PD and wise judgment of the distribution engineer, a reasonably good
coordination can be reached at a reasonable cost.

References
[1] Borbely, A. M., Kreide, J. F. (2001), Distributed Generation, The Power Paradigm for
the New Millennium, CRC Press, Boca Raton, FL
[2] Gómez, J. C., Morcos, M. M. (2005), Coordination Analysis of Voltage Sag and
Overcurrent Protection in Electrical Systems with Distributed Generation, IEEE
Transactions on Power Delivery, 20: 214–218
[3] Gómez, J. C., Morcos, M. M. (2002), Coordination analysis of voltage sag - fuse
characteristics in electrical systems having embedded generation, Second
International Symposium on Distributed Generation: Power System and Market
Aspects, October, Stockholm, Sweden.
[4] IEEE Standard 1547 (2003), IEEE Standard for Interconnecting Distributed
Resources with Electric Power Systems
[5] ANSI/IEEE Standard 1001 (1988), IEEE Guide for Interfacing Dispersed Storage and
Generation Facilities with Electric Utility Systems
[6] Dugan, R. C., (2002) Impact of DG on Reliability, Transmission & Distribution
World, October: 50–55
[7] IEEE Standard 446 (1995), IEEE Recommended Practice for Emergency and Standby
Power Systems for Industrial and Commercial Applications
324 J. C. Gomez Targarona and M. M. Morcos

[8] Djokic, S., Desmet, J., Vanalme, G., Milanovic, J., Stockman, K., (2005) Sensitivity
of personal computers to voltage sags and short interruptions, IEEE Transactions on
Power Delivery, 20: 375–383
[9] Gómez, J. C., Morcos, M. M. (2002), Voltage sag and recovery time in repetitive
events, IEEE Transactions on Power Delivery, 17: 1037–1043
[10] IEEE Standard 1346 (1998), IEEE Recommended Practice for Evaluating Electric
Power System Compatibility with Electronic Process Equipment
[11] Gómez, J.C., Morcos, M.M.(2000); Effect of distribution system protection on voltage
sags; IEEE Power Engineering Review, 20, 66–68
[12] Gomez, J. C., Campetelli, G. N. (2000), Voltage sag mitigation by current limiting
fuses; IEEE IAS Annual Meeting, October, Rome, 8–12, 5: 3202–3207
[13] International Standard IEC 60282 (2005), High-Voltage Fuses, 6th edn
[14] Wright, A., Newberry, P.G. (1982), Electric Fuses, Peter Peregrinus Ltd., IEE,
London
[15] Gómez, J. C. (2003), Intelligent fuse for M.V. distribution systems: a current need,
7th International Conference on Electric Fuses and their Applications, Gdansk,
Poland, 50–54
[16] Lapple, H. (1952), Electric Fuses, Butterworths Scientific Publications, London
[17] S&C (1991), S&C Fault Fiter: Electronic Power Fuse, Descriptive Bulletin 441–30
[18] Aubrey, D. (1974), New 11 kV expulsion fuses for overhead lines, Electrical Times,
August 1
[19] Dugan, R. C., McGranaghan, M., Beaty, H. W. (1996), Electrical Power Systems
Quality, McGraw-Hill, New York
[20] Electrical distribution system protection (1990), Cooper Power Systems Bulletin
90020
[21] Gómez, J. C., Tourn, D. H., Amatti, J. C. (2003), Experimental determination of the
reclosing time self-extinguish current for its application to distributed generation -
reclosers coordination studies, CIRED, May, Barcelona, Spain
[22] Gómez, J. C. (2006), Efecto del Aumento de la Velocidad de la Protección contra
Sobrecorriente en la Calidad de Potencia, VIII Simposio Iberoamericano sobre
Protección de Sistemas Eléctricos de Potencia, May, Monterrey, Mexico
[23] Dugan, R. C., McDermott, T. E. (2000), Distributed Generation and Power Quality,
PQA 2000 North America, May, Memphis, USA
[24] Morcos, M. M., Gómez, J. C. (2002), Flicker Sources and Mitigation, IEEE Power
Engineering Review, 22: 5-10
[25] Sorensen, P., Tande, J. O. Sondergaard, L. M., Kledal, J. D. (1996), “Flicker emission
levels from wind turbines,” Wind Engineering, 20: 39–46
[26] Dugan, R. C., McDermott, T. E. (2002), Distributed Generation, IEEE Industry
Applications, March / April, 19–25