Sie sind auf Seite 1von 25

Technical Report

DC Protection
Prepared by the
Electric Ship Research and Development Consortium
for the Office of Naval Research
Editor: R. Hebner
December 19, 2011

Mississippi State University

Naval Postgraduate School

Florida State University

University of South Carolina

University of Texas at Austin

Purdue University

Massachusetts Institute of Technology

United States Naval Academy

Table of Contents

Page

Executive Summary ............................................................................................................................. 4


1

Introduction .................................................................................................................................. 5
1.1

MVDC Fault Mitigation Overview ....................................................................................... 5

1.2

Two Competing Protection Strategies ................................................................................... 5

1.2.1

Strategy 1 No DC Breakers ......................................................................................... 5

1.2.2

Strategy 2 Use DC Breakers ........................................................................................ 7

DC Distribution Overview and Challenges .................................................................................. 7


2.1

AC Distribution Disadvantages ............................................................................................. 7

2.2

DC Distribution Advantages ................................................................................................. 8

2.3

DC Distribution Disadvantages ............................................................................................. 8

2.4

General DC Distribution Practices ........................................................................................ 9

2.5

Summary of Typical Loads On-Ship ..................................................................................... 9

2.6

Faults are Challenges for DC Distribution ............................................................................ 9

2.6.1

Challenge 1 DC Breakers .......................................................................................... 10

2.6.1.1

General Circuit Breaker Requirements ............................................................ 10

2.6.1.2

Mechanical AC Breakers in Shipboard MVDC Applications ......................... 11

2.6.1.3

Semiconductor DC Breakers ............................................................................ 12

2.6.1.4

Hybrid DC Breakers ........................................................................................ 12

2.6.1.5

Semiconductor DC Breakers as a Key Enabling Technology ......................... 12

2.6.2

Challenge 2 Bus Capacitor Protection ...................................................................... 13

2.6.3

Challenge 3 Converter Protection of Self and Downstream Components ................ 14

2.6.3.1

Types of Converter Faults Using DC Distribution........................................... 14

2.6.3.2

DC Side Line-to-Line Fault Sequence ............................................................. 14

2.6.3.3

Anti-Parallel Diode Issues ............................................................................... 15

2.6.3.4

DC Side Line-to-Ground Fault ........................................................................ 16

2.6.3.5

DC Side Open Circuit Fault ............................................................................. 16

2.6.3.6

Using Power Electronic Converters as Breakers ............................................. 17

2.6.3.7

Anti-Reversal DC Bus Diodes ......................................................................... 17

2.6.4

Challenge 4 Negative Impedance Instability Faults .................................................. 17

Fault Detection and Clearing Issues and Alternative Techniques .............................................. 18


3.1

Fault Coordination ............................................................................................................... 18

3.2

Fault Detection Methods ..................................................................................................... 18

3.3

Alternative Fault Clearing on Multi-Load DC Buses without Using DC Breakers ............ 18

3.4

Fault Ride Through Energy Storage .................................................................................... 19

3.5

Grounding Issues ................................................................................................................. 20

Manufactured Medium Voltage Drive Review........................................................................... 20

Recommendations ...................................................................................................................... 21

Conclusion .................................................................................................................................. 21

References .......................................................................................................................................... 23

Executive Summary
Shipboard power systems need to be sufficient for all needs as auxiliary power is not available at
sea. In addition, they need to be efficient since inefficiencies in the power system are systematic
losses over the life of the ship, adding to cost of ownership. Finally, they need to be as small as
possible to provide as much payload as possible for mission critical operations. These requirements
have stimulated interest in all-electric ships operating at frequencies other than 60 Hz.
Higher frequency ac offers the potential advantage of smaller size for transformers and generators.
The limitation, however, is that losses tend to increase with increasing frequency. Interest is
growing in dc distribution as an increasing number of loads are inherently dc. DC can be more
efficient than ac. A 2008 study, for example, showed that data centers can be a few percent to
nearly 30 percent more efficient with a dc distribution. While data centers are not good surrogates
for ships, the results suggest careful examination of the advantages of dc on shipboard power
systems is appropriate. The anticipated advantages make it appropriate that dc distribution is on the
Navys electric ship roadmap.
This report examines one important aspect of dc distribution use aboard ships: electrical protection
with particular emphasis on the need (or lack thereof) for dc breakers. Electrical protection is the
safe and effective design and implementation of a fault mitigation strategy in an efficient and cost
effective manner. Such an examination is warranted because, although technology exists for dc
protection, it does not have the breadth of operating experience that has been accumulated for ac
protection schemes.
Two approaches are possible concerning dc breakers in MVDC systems they are either required or
not required. Both strategies yield safe and functional distribution systems. The strategy of not
using dc breakers, however, imposes more limitations upon the distribution designer. Additionally,
hybrids of the two have been considered, particularly using other means of control for high power
and an electromechanical breaker only after the voltage and current have been reduced to acceptable
levels.
The result of this investigation is that multiple approaches can be made to work. But there are two
major issues remaining. First, there is no performance data base available for any of the above
approaches that would permit a protection engineer to specify a particular system with confidence.
Second, the non-hybrid approaches are likely to yield significantly larger than optimal systems.
A possible approach to minimizing both of these problems is to use a hybrid power system that
generates power at a higher ac frequency than 60 Hz, e.g., 240 Hz. Zonal loads that can be deenergized to clear faults will be powered by dc. This will likely yield a smaller system than a pure
dc system. Moreover, anecdotal information suggests most of the breaker experience gathered at 60
Hz will be applicable at 240 Hz.
Design data and system validation can be done using a relatively inexpensive combination of
modeling and experimentation at a reasonable power level, e.g., 0.5-1 MW. Initial modeling has
been done to show the feasibility of some dc protection approaches. In addition, measurements of
dc series faults in a dc microgrid showed that the faults are unlikely to induce voltage spikes that
damage power electronics. This means that power electronic solutions are likely to be robust. A
focused program to combine experiments on the microgrid with comprehensive modeling and
simulation should provide the confidence needed to commit to the smaller hybrid system.
4

Introduction

Shipboard power systems need to be sufficient for all needs as auxiliary power is not available at
sea. In addition, they need to be efficient since inefficiencies in the power system are systematic
losses over the life of the ship, adding to cost of ownership. Finally, they need to be as small as
possible to provide as much volume as possible for mission critical operations. These requirements
have stimulated interest in all electric ships operating at frequencies other than 60 Hz.
Higher frequency ac offers the potential advantage of smaller size for transformers and generators.
The limitation, however, is that losses tend to increase with increasing frequency. Interest is
growing in dc distribution as an increasing number of loads are inherently dc. DC can be more
efficient than ac. A 2008 study, for example, showed that computer centers can be a few percent to
nearly 30 percent more efficient with a dc distribution. While data centers are not good surrogates
for ships, the results suggest careful examination of the advantages of dc on shipboard power
systems is appropriate. The anticipated advantages make it appropriate that dc distribution is on the
Navy's electric ship roadmap.
This report examines one important aspect of dc use aboard ships: electrical protection. Electrical
protection is the safe and effective design and implementation of a fault mitigation strategy in an
efficient and cost effective manner. Such an examination is warranted because, although technology
exists for dc protection, it does not have the breadth of operating experience that has been
accumulated for ac protection schemes.
1.1

MVDC Fault Mitigation Overview

The ability to identify and interrupt faults on a dc bus is of paramount importance for the safe
operation of an MVDC distribution system and is the focus of discussion between the naval and
research communities. At the center of this discussion is the need for and construction of an
MVDC circuit breaker. Currently, no such breaker which meets the Navys demands for MVDC
distribution in the 5-20 kV range is available commercially. Compared to the more familiar ac
breakers, such MVDC breakers are expected to be expensive and complex and thus the question put
forth is: are MVDC breakers needed when using MVDC distribution? This report presents an
overview of MVDC distribution challenges and techniques with regard to fault mitigation including
a review of commercially available MVDC motor drive systems to provide a basis for determining
under what conditions dc breakers are required when using MVDC distribution.
1.2

Two Competing Protection Strategies

Two approaches are possible concerning dc breakers in MVDC systems they are either required or
not required. Both strategies yield safe and functional distribution systems. The strategy of not
using dc breakers, however, imposes more limitations upon the distribution designer. Additionally,
hybrids of the two have been considered, particularly using other means of control for high power
and an electromechanical breaker only after the voltage and current have been reduced to acceptable
levels.
1.2.1

Strategy 1 No DC Breakers

In this strategy dc breakers are not used, thus dc side faults must be handled by the nearest upstream
power electronic converter. The converter can reduce or eliminate faults due to short circuits by
reducing the on time of its internal switches or simply turning them off entirely.
5

If the nearest upstream converter is an uncontrolled device such as a passive rectifier creating a dc
bus from the ac side generation, there is no means to eliminate the fault except for ac side traditional
breakers or dc side fuses. In general, dc side fuses are discouraged due to limited lifetimes, inability
to automatically reset, and expensive replacement costs. Additionally, dc distribution systems
which do not employ dc breakers are typically composed of controlled converters which are able to
act to inhibit faults, and therefore can be designed to work without fuses (see section 4). Multiple
layers of converters in the path from the prime power source to the load provide backup protection
for the primary fault interrupter.
The advantages of this approach include the fact that high interrupting speeds can be achieved, that
the approach can also provide a current-limiting function, and that because this strategy uses power
electronics, it can be anticipated that there will be future performance improvement and cost
reduction. There are of course disadvantages to relying on power converters as primary fault
interrupters, which include the following:
1. Power converter topologies which do not include a solid state switch in series with the
distribution bus (e.g. boost converters) will be ineffective as fault interrupters. These
converters would require an additional series switch to accomplish this goal thus incurring
extra losses.
2. Converters may have low withstand capability in case of transient over-voltages leading to
cascading failures.
3. There is no intrinsic galvanic isolation it will have to be provided by conventional
contactors.
4. If control of the switches within the converter providing the dc power is exercised in
response to a fault, not only is the faulted section of the circuit affected, but all sections fed
by the dc bus are affected.
Number 4 is potentially the most serious of the disadvantages listed above. When dc breakers are
not used, a faulted branch on a multi-branch dc bus cannot be cleared without de-energizing all
loads on the bus. Permanent faults such as a load or bus connection must be cleared before
reenergizing the dc bus. Without the use of a breaker on the dc bus to clear a permanent fault, dc
fuses could potentially be used, or the fault could be manually cleared once the system was fully deenergized. Alternatively the fault clearing variation presented in below has been commonly
proposed.
When a fault occurs on a dc bus in which breakers or fuses are not used, the entire bus must be deenergized in order to clear the fault. A permanent fault would require some sort of manual
intervention without additional hardware. To this end, it has been proposed that instead of dc
circuit breakers, less expensive mechanical non-load-breaking switches are employed in the same
locations a dc breaker would have been needed [28]. The operation of an MVDC system employing
such switches during a dc side fault proceeds as follows:
1. De-energize the entire dc bus upon detecting a dc side fault.
2. Remove the faulted load or bus section by opening the appropriate no-load breaking switch.
3. Re-energize the dc bus.
The disadvantages of this technique are threefold. First, the time to de-energize the bus can be
substantial; this is especially true if regenerative loads are present. During the time taken to deenergize the bus, the fault damage can continue to grow even if the fault current is decreasing.

Second, it is a nontrivial task to identify the location of the fault and then select which switch to
open. This task is aggravated by the fact that all portions of the dc distribution system may not
reach a fully de-energized state at the same time; therefore each switch would require some form of
local sensing to identify at what moment the opening operation could commence safely. Third,
during the time the dc bus is de-energized, there is a lack of power continuity to the loads. Some
solutions have been presented to attempt to overcome this continuity issue unfortunately with
some additional disadvantages as will be discussed in later sections.
1.2.2

Strategy 2 Use DC Breakers

With this strategy, dc breakers are deployed onto the dc bus at strategic locations using similar
reasoning as that behind the placement of ac breakers on an ac bus. While the addition of dc
breakers adds cost and complexity to the MVDC system, it also provides a high degree of fault
protection while enabling flexibility in power management.
Multiple solutions for dc breakers have been proposed in the literature, including those where the
main pass element is entirely semiconductor-based, those that use a mechanical pass element, and
those that use a hybrid semiconductor/mechanical pass element. When used, a mechanical pass
element affords lower losses when closed and galvanic isolation when open.
DC breakers can be designed to be unidirectional or bidirectional. Bidirectional breakers would be
needed if a particular section of a dc bus could be used to route power in either direction as a result
of using the distribution systems reconfigurability. Breakers can also be designed to be fault
current limiting, though no references in the literature have been identified which describe
functional bi-directional fault current limiting breakers.
2
2.1

DC Distribution Overview and Challenges


AC Distribution Disadvantages

In considering dc distribution, it is important to understand the disadvantages of ac distribution. For


example, ac distribution systems require all prime generators to be synchronized in terms of
voltage, frequency, and phase. This presents challenges when distributed generation becomes
islanded and then later attempts to connect to an already powered bus. In general rerouting power
and the ability to select the optimal use of prime generators is less flexible with ac distribution due
to the synchronization need [1].
Other issues with ac distribution arise concerning the presence of harmonics and the numerous
required filters. Additionally, during a short circuit event, an ac breaker must delay the circuit
opening until a zero crossing is reached. This time delay can be several milliseconds for a 60 Hz
system, during which time the fault damage can increase.
Finally, it is important to note that the ease of converting ac voltages with transformers is offset by
their heavy weight and significant bulk; this is particularly true for a 60 Hz power distribution
system [2]. The presence of reactive power in an ac distribution system tends to further increase the
required size of distribution components since reactive power increases the demanded line current
without performing useful work.
Furthermore, in ungrounded or high impedance grounded systems (which are preferred on Navy
ships for the ability to continue operation with a single line-to-ground fault), ground currents in ac
7

systems are governed by the stray capacitances to ground. In cable systems, these capacitances can
be substantial and may cause substantial ground currents. The effect of stray capacitances causes
the amplitude of such ground currents to increase linearly with the ac frequency.
2.2

DC Distribution Advantages

DC distribution counters many of the ac distribution disadvantages [3, 4]. A typical dc distribution
system is shown in Figure 1. For example with dc distribution, a reduced cabling demand exists
due to increased power and energy density, this benefit is made more prominent due to the absence
of reactive power in such a system. Power management is more flexible largely due to the absence
of the synchronization requirement. The plethora of required converters (which could be viewed as
a disadvantage) are able to actively adapt to changing system conditions such as load demand, line
disturbances, and faults in a faster and far more effective way than is possible with traditional ac
distribution systems.
A noticeable advantage of dc systems is that the frequency of the prime movers is decoupled from
the dc buses. Each prime mover can rotate at its most effective frequency and achieve higher power
factors.
High Voltage Bus
5000 V dc
Alternator

Alternator

DC
Load

Low Voltage Bus


300 V dc
Intermediate Voltage Bus
1000 V dc

AC
Load

DC
Load

DC
Load

DC
Load

Figure 1 Typical dc distribution system

2.3

DC Distribution Disadvantages

DC distribution does have some disadvantages; chief among them is the lack of a natural current
zero crossing which would allow the use of traditional ac breakers. DC breakers for medium
voltage are still in the development stages and tend to be complex and expensive and are not widely
available.
DC distribution requires numerous converters (ac-dc, dc-dc, dc-ac) to power a given load. This
incurs efficiency penalties and increases the complexity of the power transmission path for those
loads which can be powered directly by constant frequency ac distribution. For example, to power
low voltage single phase ac loads (e.g., 120 V, 220 V loads) multiple converters are needed to
transport power from the main generators to the loads via dc distribution, resulting in a rather

inefficient system in comparison to a single transformer which could be otherwise used with ac
systems. Electromagnetic interference can also be generated by such converters. Furthermore,
ship-to-shore power transfer complexity increases, but at the same time it can support multiple
frequencies (e.g., 50 Hz).
2.4

General DC Distribution Practices

On ships which use the main power source for propulsion, the main bus (which may be ac or dc)
should be split into at least two parts connected by circuit breakers or other approved means to
ensure redundancy in the event of a single fault [5]. Furthermore, generating sets and rectifiers
should be divided equally between the bus parts so that one can be taken out for maintenance during
light loads [5, 6].
2.5

Summary of Typical Loads On-Ship

Table 1 shows the load summary for a notional all-electric U.S. Navy ship with dc distribution
according to [7]. The powers consumed are shown for battle mode; furthermore, not all loads are
possible at the same time. Note, if propulsion is removed, the power consumption is dominated by
ac loads. It is apparent that numerous ac-dc, dc-dc, dc-ac converters will be needed to service these
diverse loads [4, 8].
A key benefit cited for the use of dc distribution is the ease with which power can be rerouted
around the ship. The use of a ring bus is much discussed in the literature concerning the future dc
distribution architecture. According to Table 1, in order to have the ability to reroute peak
propulsion power for a single motor throughout the ship (perhaps to bypass a battle damaged
portion of the bus) each portion of the bus that is eligible for carrying propulsion power would need
to have the capacity to transport 73 MW while being insulated for more than 5000 V. To avoid the
great expense and volume required for a lengthy high power bus and commensurate switches and
breakers, ideally the propulsion motor drives would be located near the main ac alternators.
Table 1 Load summary on U.S. navy ship with dc distribution

2.6

Load

Voltage (V)

Power (MW)

Zone 1 Zone 4 dc

800, dc

0.941

Radar

300, dc

3.75

Zone 1 Zone 4 ac

450/240/208/120, ac

11.945

Propulsion
(Total for 2 Motors)

5000, dc

73

Faults are Challenges for DC Distribution

Faults present significant challenges for the successful use of dc distribution. Four key points are
summarized below and discussed in detail in later sections.
1. DC breaker needed for multi-load connections: a dc breaker must be very fast acting since,
unlike with ac distribution, the fault current peak is limited only by the line resistance. In
general, a dc distribution design could forgo the use of dc breakers if only one load per dc
9

bus was installed, or if the requirement of de-energizing the dc bus to clear a fault was
acceptable.
2. DC bus capacitor protection: during short circuit faults, capacitors on the dc bus will
discharge into the fault with very short time constants. This in turn results in very large fault
currents which can both damage the capacitor and worsen the fault damage. Such bus
capacitors require protection in the form of a switch that can be opened during a fault
(essentially a dc breaker without galvanic isolation).
3. Converter protection: power electronic converters are typically self protecting, however
anti-parallel diodes within the converter can lead to undesirable passive rectification during
a dc bus voltage collapse, aggravating the fault condition. Furthermore the anti-parallel
diodes can be damaged after the dc bus capacitors discharge since the bus inductance has
been charged by the large capacitor discharge current which ultimately freewheels through
the anti-parallel diodes.
4. Negative impedance instabilities: a properly designed dc distribution system must remain
stable during a fault in the presence of multiple constant power converter loads. Such
converters appear as a negative resistance and tend to collapse an already sagging dc bus
voltage further by drawing ever more current as the voltage falls in an effort to maintain a
constant output power.
2.6.1

Challenge 1 DC Breakers

As with ac distribution, dc distribution systems must be designed to safely handle faulted cables and
buses. Line-to-ground and especially line-to-line faults can lead to very high current discharges and
commensurate fault damage [3, 9]. Cable faults typically occur due to insulation deterioration and
aging resulting from physical damage, environmental stress such as moisture penetration
(particularly at cable ends), and electrical stress particularly at elevated temperature.
When discussing the application of an MVDC distribution system on a naval vessel, it is important
to consider the compact nature of a ship compared to a terrestrial power distribution system [1]. For
instance, with a shipboard application, cable lengths are shorter, load coupling is higher, and the
prime generator stiffness is lower. These conditions ensure that fault currents increase more
quickly, the bus distribution voltage falls more quickly, and the chance of a fault affecting other
loads is higher with dc distribution systems than with a comparable ac system [10]. (The reduced
generator stiffness of a naval application does have one advantage however; it tends to limit the
peak fault current to comparably lower values than would be found with traditional distribution.)
Thus there is a great need for an effective dc breaker to aid in the safe operation of a dc distribution
system, particularly an isolated one such as is found with a microgrid or naval application.
2.6.1.1 General Circuit Breaker Requirements
Circuit breakers for use with naval MVDC distribution systems should have sufficient current (7.515 kA for propulsion related buses) and voltage capabilities (10-20 kV for 5-10 kV buses). In
addition, the breaker should have a low conduction loss, fast opening action (ideally in the
microsecond range) to prevent the growth of large fault currents and corresponding bus voltage
droop, and the ability to absorb the inductive energy stored in the distribution line during a fault
[11].
To interrupt the fault current flow, a breaker must bring the current to zero by developing a counter
voltage. This is accomplished in a mechanical circuit breaker by the arc voltage or by the

10

increasing impedance of a semiconductor breaker as it is gated off. The energy stored in the line
inductance must be dissipated. In mechanical breakers, the arc dissipates most of the energy; thus
the arc is often stretched in an arc chute to increase its length and, therefore, voltage drop.
However, this technique is not economically viable for voltages significantly above 3 kV. A
semiconductor breaker absorbs some energy in the switching element during the turn off event;
however both breaker types often employ voltage-limiting dissipation devices such as varistors
(MOVs) to handle the bulk of the energy absorbing needs. Note, if the lines inductive energy is
not dissipated in a controlled manner, large overvoltages can be developed with destructive
consequences.
Once the fault current is interrupted, a breaker must provide a dielectric barrier to prevent the flow
from restarting [12, 13]. For a mechanical breaker, this requires that the gas previously ionized by
the arc be allowed to cool to prevent reignition which limits the rate of voltage rise across a
mechanical breaker to around 80 V/s [14, 15]. Semiconductor breakers can switch many times
faster and are thus often preferred for dc breaker applications at the medium voltage level.
2.6.1.2 Mechanical AC Breakers in Shipboard MVDC Applications
Mechanical ac breakers have been used in shipboard dc applications. The breakers are de-rated and
used with dc buses as high as 270 V. Fuses are generally incorporated to protect higher dc voltage
distribution over 700 V. Some specialized dc breakers do exist for very particular applications [12].
There are a number of reasons mechanical ac circuit breakers are not typically used as current
interrupters in MVDC distribution systems. The principal reason is that dc current has no natural
zero crossing to allow the breaker to operate normally [14, 16]. Moreover, mechanical breakers
with magnetic trip devices (e.g. low voltage units below about 1 kV) typically require the fault
current to reach many multiples of (typically 10-20 times) the nominal current in order to open at
their fastest speed (on the order of a few milliseconds). This is due to the fact that the magnetic trip
mechanism aids in the opening of the breaker and is more forceful at higher currents. Additionally,
breakers using magnetic blow out coils more quickly move the arc into the arc chute with larger
currents therefore yielding a faster current interrupting time. Ultimately where such breakers are
used, the relatively low available fault current from a closed generation system such as a naval ship
results in relatively slow circuit breaker action [12].
AC breakers with higher voltage ratings (several kV to 10s of kV) typically require 3-10 times the
nominal current to activate their trip relays; however this threshold is often adjustable. Regardless
of the relay setting, the actual opening and arcing time of such breakers is many tens of
milliseconds during a short circuit fault (40-100 ms), during this delay the fault current on stiff
buses can increase to many times its nominal value (e.g. 10 times or more is possible) on weak
buses the voltage will sag. (The opening delay of mechanical breakers is largely responsible for the
requirement that such devices have extremely large interrupting ratings compared to the trip current
value 100 kA and higher interrupting ratings are common for medium voltage ac breakers.) Such
an overload range is well beyond the overload capability of most semiconductor converters which
would be required to support this fault current until the breaker cleared the fault. A chief reason for
the slow mechanical opening time of such breakers is the need to keep impact forces of the springloaded opening mechanism within appropriate limits to achieve long life and effective operation.
The reaction time of medium voltage breakers is also lengthened by the time required to move the
arc into the arc chute [11].

11

Of note is the proposal to form a dc breaker by using a traditional ac breaker and added components
to force a current zero crossing on the dc bus [17]. Artificial zero crossings can be achieved by
inserting an inductor in series with the filter capacitor and fault resistance to form an underdamped
series RLC circuit. As soon as the fault current exceeds a pre-determined threshold, the firing angle
of an upstream thyristor-controlled rectifier is set to 90 and the ac circuit breaker interrupts the
oscillating fault current at its first zero crossing. In this way, standard ac breakers can be used on dc
systems providing the advantages of galvanic isolation and the ability to use conventional control
methods in response to a fault; however the added components include a significant inductance
which offsets some of the advantages of this method. Other disadvantages of this approach include
1. Speed of operation is limited by that of conventional breakers
2. Needs additional components that can induce a zero crossing for the current
3. May prove impervious to significant cost reduction
2.6.1.3 Semiconductor DC Breakers
Semiconductor dc breakers interrupt current without arcing and thus do not require a zero current
crossing [2]. However, breakers in which the semiconductor is the main pass element can have
excessive losses. To accommodate this conduction loss which can be on the order of several
kilowatts, some experimental semiconductor dc breaker designs require water cooling and complex
deionization systems to ensure the cooling water retains its insulative properties which may reduce
overall long-term reliability [2, 18].
Semiconductor breakers typically employ either an IGBT or a GCT as the main pass element [ 19].
The IGBT has simpler controls and has a superior switching speed, however the GCT has lower
conduction losses and is often preferred in a wholly semiconductor-breaker application for this
reason [14].
For semiconductor breakers, some form of voltage clamping is needed to ensure the energy stored
in the line inductance does not lead to a switch overvoltage during a fault turn off event. Typical
clamping devices include capacitive clamps, capacitive-diode clamps, and MOVs.
2.6.1.4 Hybrid DC Breakers
Hybrid breakers employ both a semiconductor and a mechanical switch. The mechanical switch
can be used in shunt or series with the semiconductor switch. When placed in series, the
mechanical switch serves to provide galvanic isolation when the breaker is open. In shunt, the
mechanical switch aids in producing lower conduction losses since its on-state resistance can be
more than an order of magnitude lower than that of the semiconductor switch. Shunt hybrid
breakers require the mechanical switch current to be commutated to the semiconductor to prevent
arc formation; the circuitry to do this is bulky and expensive [14]. A pair of mechanical switches
can be used in a series/shunt arrangement to combine the advantages of both features above.
2.6.1.5 Semiconductor DC Breakers as a Key Enabling Technology
Solid state circuit breakers are a key enabling technology for MVDC power distribution [2]. Their
fast response and opening times ensure that fault currents can be interrupted in microseconds with
resulting fault current peaks of only a few times nominal (two times is typical). Furthermore,
semiconductor dc breakers can be employed similarly to ac breakers, requiring no special

12

installation requirements and largely suitable for placement in locations similar to those of ac
breakers.
Figure 2 shows a particularly interesting case benefiting from the use of a fast-acting
semiconductor-based dc breaker. Consider a line-to-line fault on an MVDC bus lateral. An
installed semiconductor dc breaker will be able to quickly disconnect the faulted bus section
allowing the remainder to operate normally due to experiencing very little voltage sag. Had dc
breakers not been used, the fault could only be cleared by bringing down the entire MVDC bus,
which would result in every load being disabled.
High Voltage Bus
5000 V dc
Alternator

Alternator

Fault
DC
Load

Low Voltage Bus


300 V dc
Intermediate Voltage Bus
1000 V dc

DC Breaker

DC
Load

AC
Load

DC
Load

DC
Load

Figure 2 Fault on a multi-load MVDC bus

2.6.2

Challenge 2 Bus Capacitor Protection

During a dc bus line-to-line fault (or line-to-ground fault depending on the capacitor connection) a
capacitor located across the fault will discharge a large current with a very small time constant onto
the bus, resulting in potentially significant damage to the capacitor and increased damage severity
of the faulted components, as shown in Figure 3 [16]. The discharge current can also charge the
line inductance, which may in turn cause damage to power electronic converters as discussed in
section 2.6.3.2. (Additional infeed current from adjacent sources is possible.) To prevent the above
problems, various researchers have proposed installing what amounts to a dc circuit breaker on each
capacitor where destructive discharges could occur [16, 20]. Alternatively, inductive snubbers can
be fit to each capacitor to limit the fault discharge currents rate of rise. Doing so, however, reduces
the effectiveness of the capacitor during normal operation and incurs extra loss due to the snubber.

Fault

Alternator

Capacitor
Discharge
Current

Figure 3 Capacitor discharge during a dc side fault

13

2.6.3

Challenge 3 Converter Protection of Self and Downstream Components

In a dc distribution system, many power electronic converters are employed. It is desirable that
each converter be able to protect itself from internal faults and serve as either the primary protection
for downstream faults or as backup to another converter or dc breaker.
Protection against overcurrent and short circuit faults requires that the converters possess significant
short-term overload capability. This is made possible by the wide safe operating area of modern
power semiconductors. Current overloads of four times nominal for 10 s are typical; note this
value is much lower than would be needed to trip a mechanical breaker at its fastest speed.
Fortunately, the semiconductor switches composing the converters can both detect and switch off
fault currents in the microsecond range. Current turn off rates for modern devices exceed 1 kA/s
[14].
It is also important that converters be designed to tolerate a large number of possible faults over
their lifetimes; however, faults within a converter are less common than those occurring with
interconnecting cables and bus work which are exposed to the environment [9].
2.6.3.1 Types of Converter Faults Using DC Distribution
When using dc distribution, several types of faults are possible that affect the power electronic
converters. The most severe fault is a line-to-line fault on the dc side of the converter. Other faults
include line-to-ground faults and open-circuit faults. All of these faults can be located within the
converter enclosure or on the adjoining dc bus.
Failures internal to the converter ultimately reduce to one of these three faults. For instance, a nonredundant failed power semiconductor switch will lead to a line-to-line or line-to-ground fault if the
switch fails shorted or an open fault if the switch fails open. While traditional current source
rectifiers and inverters are highly tolerant of dc side short circuit faults due to their large input
inductance, in modern naval dc distribution systems the voltage source converter (VSC) is expected
to be the primary converter of choice due to a number of advantages, such as reduced size and
weight and better dynamic performance [21].
A VSC is depicted in Figure 4, illustrating the various types of dc side converter faults. Details
regarding common VSC faults will be discussed in the sections below.
Open Fault
Line-Line
Fault

Alternator

Line-Ground
Fault

Figure 4 Types of dc side converter faults

2.6.3.2 DC Side Line-to-Line Fault Sequence


Upon a line-to-line fault on the dc side of a converter, the dc bus capacitor will rapidly discharge
into the fault. The resulting large discharge current will potentially damage the capacitor and
14

increase fault damage. The cable inductance will be charged by the capacitor current, which will
then discharge through the anti-parallel diodes of the upstream voltage source rectifier (VSR),
which can result in diode damage due to the large current overload experienced [9]. Note that at
this point in the fault sequence, an ac-side breaker would be of little use. If the dc bus sags far
enough due to a persistently low fault impedance, the ac side of the VSR will begin to feed current
into the dc fault due to the anti-parallel diodes allowing passive rectification. In this case, it is
imperative that either a dc breaker or fuse interrupts the fault current or the dc bus is de-energized.
The above fault sequence can be slowed somewhat to allow more time for fault protection
equipment to operate if additional dc bus inductance is incorporated, as it will limit the fault current
rate of rise [15]. Disadvantages of this method are increased weight, cost, size, and losses,
decreased bus regulation, and poorer transient response at the various loads without increased local
energy storage. Furthermore, increasing the bus inductance is not a panacea; there remain locations
on the dc bus before the inductor, however small, where a line-to-line fault can occur without being
limited by the inductor.
2.6.3.3 Anti-Parallel Diode Issues
As mentioned previously, a VSRs anti-parallel diodes allow passive rectification during dc bus
collapse. The fault current during this rectification will be limited only by the ac side impedance
and that of the fault itself; the large resulting currents are likely to damage the diodes [13, 22]. Ac
side breakers are helpful to interrupt the fault current in this situation, and should be installed
according to standard design practices.
One means of protecting the diodes during a dc side fault is to replace them with controlled devices
such as IGBTs or GCTs to prevent passive rectification from feeding a fault if dc breakers are not
employed [16]. However, in doing so, a difficulty arises with dissipating the stored energy in the
line inductance. With traditional anti-parallel diodes, the line inductance is able to discharge
relatively slowly through the freewheeling path incorporating the diodes, thereby preventing
overvoltages. If the diodes are replaced with controlled devices and rapidly switched off during a
dc fault, an overvoltage may result due to the stored line inductance energy.
One possible solution is to use dc breakers on bus capacitors to prevent the discharge current from
charging the line inductance to large values, and employing dc breakers with the required energy
absorbing components on the dc bus to interrupt the fault current and absorb the inductive energy.
In this manner, standard anti-parallel diodes could be use within the converter and the fault current
could be interrupted before the dc bus falls far enough to initiate passive rectification.
With modern power semiconductor switches, the most commonly available switch type is the
reverse conducting arrangement. In this topology a semiconductor switch is integrated with an antiparallel diode in the same package. Some converter designs, such as buck converters, do not need
the anti-parallel diode to function correctly [16]. Indeed, the presence of this diode can exacerbate
the fault scenario by permitting a discharge path for capacitive energy storage, as shown in Figure 5.
Thus, while reverse conducting power semiconductors are the most common, fault prevention
design considerations require the use of symmetric switches where appropriate. A symmetric
switch does not include an anti-parallel diode; furthermore, it is equally capable of blocking voltage
in either direction.

15

Unneeded copackaged diode

Fault current
from capacitor

Line-to-line fault

Figure 5 Reverse conducting buck converter switch allows capacitor to discharge during a line-to-line fault

2.6.3.4 DC Side Line-to-Ground Fault


U.S. naval ships are required to continue operation of the respective portion of a power system
under the condition of a single line to ground faults. Thus, electrical systems are often highresistance grounded. As such, the first ground fault is of little consequence to equipment operation.
However, a potential risk to personnel does exist, so the fault should be indicated. During a line-toground fault, the VSRs anti-parallel diodes will not conduct the resulting large discharge current of
the nearby dc bus capacitors if a split dc bus is used with a grounded midpoint [9]. The ac side of
the VSR will, however, feed current into the line-to-ground fault if the ac side neutral is grounded.
The proper grounding choice is nontrivial and is examined more in section 3.5.
2.6.3.5 DC Side Open Circuit Fault
Open circuit faults are problematic on a dc bus for three reasons. First, an open circuit fault can be
described as a series (in-line) load in which an arc may form, this is particularly true if the open
circuit occurs while the line is energized. With an arcing open circuit fault the loads downstream of
the arc continue to receive power (at a reduced voltage), and make such faults difficult to detect;
additionally, the arc will persist due to the absence of a natural current zero crossing [23, 24]. The
arc is an extreme fire hazard and can jump chaotically to various nearby equipment creating
additional damage and possible line-to-line and line-to-ground faults.
Second, if the arc does extinguish by some means (series arcing faults do occasionally self
extinguish due to the forces of local air currents, magnetic forces from other nearby circuits,
accruing electrode damage, increasing physical separation, and the general chaotic nature of the
arc), the load will be shed rapidly, potentially resulting in an overvoltage event at the upstream
converter, depending on the response time of the converter and the portion of the total converter
load that was shed. Dc bus overvoltage clamps are helpful in preventing damage during such
events.
Third, dc side open circuit faults can lead to the loss of control of an ac side load. This is
particularly problematic when the converter was performing a braking or regenerative action on a
load such as a motor. Without the ability to regenerate energy, the ac side load could overspeed,
and the dc side voltage could increase to unsafe levels. In critical applications, each converter with
a regenerative load requires an ac side load dump, a dc side chopper, or an energy storage system.
Backup mechanical braking on regenerative inertial loads is also possible.

16

2.6.3.6 Using Power Electronic Converters as Breakers


Due to the unavailability and expense of dc breakers at the MVDC level, it is reasonable to question
whether the MVDC converters themselves could function as dc breakers. Certainly voltage-sourcetype converters are able to actively limit and interrupt fault currents via their microsecond turn-off
capability; this ability can be leveraged to protect entire zones in zone-based architectures such as
those planned for future all-electric naval ships [4, 12]. However, operating a converter as a
primary fault interrupter is viable only when there is a single load per bus, or it is acceptable to deenergize the entire bus to clear the fault [14, 16].
When converters feed multi-load buses, fast acting dc breakers are needed to prevent de-energizing
all bus loads when clearing a fault [12]. Regenerative loads may require bi-directional dc breakers
unless some form of local energy dissipation or storage is provided. Bi-directional dc breakers are
also required on distribution buses that can be reconfigured to transmit power in either direction
[25]. When a converter powers a multi-load bus and it is not acceptable to de-energize the entire
bus to clear a fault on that bus, a converter cannot be used as a dc breaker replacement; however, in
such instances, converters can fill the role of backup protection for dc side breakers for multi-load
buses [16].
2.6.3.7 Anti-Reversal DC Bus Diodes
Voltage source inverters which power regenerative loads on the ac side require anti-reversal diodes
on their dc bus inputs [16]. An anti-reversal diode is a diode placed in series with the dc bus
feeding the VSI and is intended to prevent regenerative energy from feeding a fault on the dc side.
Such diodes can only be used when the regenerated energy is absorbed by a local and dedicated
dissipation or storage mechanism. Conversely, if the regenerated energy is normally intended to be
delivered back into the dc bus, the anti-reversal diode would then need to be replaced with a
controlled turn off device such as an IGBT with a co-packaged anti-parallel diode.
Another use of anti-reversal diodes is found with dc buses which are fed by more than one voltage
source rectifier. In such a situation, anti-reversal diodes can be positioned at the dc outputs of each
rectifier to ensure that an ac side fault at the input of one rectifier is not fed by the remaining VSRs.
2.6.4

Challenge 4 Negative Impedance Instability Faults

The power electronic loads installed on dc buses such as VSIs and dc-dc converters are often
constant power loads. These converters are able to maintain a constant power input and output in
spite of significant changes in input voltage. The disadvantage of this excellent regulation
characteristic is that constant power loads appear as negative impedances to the dc bus to which
they are connected. This observation is easily confirmed by noting that, as the dc input voltage to a
constant power converter falls, the input current must increase. This behavior is identical to that of
negative impedance.
The difficulty presented by a negative impedance load becomes apparent during a fault on the dc
bus. If a dc bus fault or transient occurs that causes the bus voltage to sag, all constant power loads
will draw more current in an effort to maintain constant power at their respective outputs. The
increased current demand on the weakened dc bus can cause it to sag further [3]. This trend can
continue until the low input voltage limits are reached by the load converters, at which point they
will begin to trip off. Quickly identifying and removing the faulted load or bus section that is
causing the dc voltage sag can help mitigate negative impedance instability. Another solution is to
17

use fault current limiting at the input of each load to alter the negative impedance characteristic
during a severe dc bus voltage sag [3, 26]. A disadvantage of this method is that the loads will no
longer receive constant power during such a sag. To overcome this difficulty, it may be possible to
fit local energy storage at critical loads to provide fault ride-through time.
3
3.1

Fault Detection and Clearing Issues and Alternative Techniques


Fault Coordination

A well-designed fault protection system should employ fault coordination, meaning that protection
circuitry should be designed in layers, with a primary fault interrupter intended to eliminate local
faults and a secondary fault interrupter serving as a backup in case the primary interrupter fails.
Layered fault protection can extend beyond two levels depending on the structure of the power
distribution system and the attached loads. The primary fault protection for a given location should
function as quickly as possible to minimize fault damage and dc bus voltage sag. Each successive
layer serves as backup protection for the former layer with an increase in response time. In this way
the backup protection will not operate before the primary protection has had a chance to clear the
fault and thus, the number of de-energized buses and loads will be kept to a minimum [9].
Fault coordination or selectivity can be achieved through the use of time delays programmed into
measuring relays or as part of the intrinsic construction of fuses and mechanical breakers.
Mechanical breakers installed on the ac load side of a VSI need particular attention, as they may not
trip during a fault since many multiples of normal load current are needed to do so; even then their
response time is typically long enough to risk damage to the semiconductors within the converter
[27]. If a VSI is powering only a single ac load, it may be preferable to allow the converter to
function as the primary fault interrupter for the load; this is especially true when the load is nonregenerative. Fuses can be used where fast response for protection is needed but automatic
restarting is not [9, 16].
Depending on the fault location, however, it may be difficult not to have two or more sets of
protection breakers open simultaneously [22]. This can occur, for example, when a line-to-line fault
occurs on a bus branch immediately next to the main bus. In this situation, the line-to-line fault
appears from a breakers perspective to be located on both the main bus and the branch circuit.
Therefore, the main bus and branch bus breakers will likely trip together.
3.2

Fault Detection Methods

The chosen fault detection method depends on the load or bus being protected. Some examples
relevant for dc distribution include monitoring bus current amplitude, bus current rate of change,
and dc voltage amplitude. Large time-constant faults may be difficult to discern from a step load
change. In this case, it is better to use the fault current amplitude to trigger the fault protection.
Conversely, small time-constant faults, such as capacitor discharges, are best sensed via a
combination of the fault current derivative and amplitude [22].
3.3

Alternative Fault Clearing on Multi-Load DC Buses without Using DC Breakers

It is possible to clear a line-to-line or line-to-ground fault on a dc distribution bus without the use of
dc breakers by de-energizing the entire dc bus [3, 28, 29]. Loads can be diode isolated from the dc
bus with local energy storage for fault ride through time. If the fault is non-permanent, the dc bus
can be re-energized; permanent faults require the faulted load or bus section be removed from the
18

distribution network. This can be accomplished by first locating the fault and then isolating the
fault via a non-load breaking mechanical switch, only then can the bus be re-energized.
Although possible, performing such a dc bus power down has a number of significant disadvantages
compared to using dc breakers. First, de-energizing all loads to clear a single fault is extremely
unattractive, especially for a naval ship during combat operations. Second, the rebooting process is
cumbersome, especially considering that finding and isolating the original fault is nontrivial.
Identification of the fault location utilizes numerous sensors on the dc side of the distribution
system; however, it may be viable to perform some or all of the sensing on the ac side where the
main breakers are located [30]. Many loads will need to be restarted manually, and in proper
sequence. Third, regenerative loads on the dc bus can increase the amount of time for the bus to be
de-energized.
Loads can be diode-isolated from the dc bus and equipped with local energy storage to offset some
of the above disadvantages. As mentioned in section 2.6.3.7, regenerative loads would require a
reverse conducting turn-off device instead of a simple diode. Whatever semiconductor is chosen,
extra loss will be incurred. Furthermore, the energy storage required is rather large as will be
discussed in section 3.4.
3.4

Fault Ride Through Energy Storage

Energy storage for ride through time is a significant issue when the chosen fault mitigation method
is to de-energize the dc bus. The necessary energy storage technology should be long-lived, have
good energy density to minimize volume and weight, and be capable of a large number of highpower charge/discharge cycles. Ultracapacitors meet all of these requirements; thus, parameters for
ultracapacitor ride-through energy storage are estimated below for each of the dc loads listed in
Table 1.
Table 2 shows the number of ultracapacitors that must be connected in series to support the dc loads
for a 0.2 s ride through time. It is assumed that in a series string of ultracapacitors, each individual
ultracapacitor can be charged to a maximum of 2.85 V. To minimize the number of ultracapacitors
needed, the voltage across the total capacitance is allowed to discharge from the nominal voltage of
VHI to a 50 % lower value of VLO. A boost converter with a 2:1 input range must then be connected
after each ultracapacitor bank.
A 2:1 input boost converter is necessarily larger and more expensive than one with a smaller input
range. However, while such a converter was chosen to minimize the number of ultracapacitors
needed, Table 2 shows that even so, hundreds of individual ultracapacitors are needed in series to
meet the voltage and capacitance requirements of the load. The large number of seriesed devices
increases the equivalent series resistance (ESR).
The ESR of these seriesed strings is so high that multiple parallel strings are need to avoid the
unacceptable voltage drop that would occur during a discharge. For example, consider the radar
load of 3.75 MW given in Table 1. Table 2 shows the required ultracapacitor assembly to support
3.75 MW for 0.2 s with a 2:1 boost converter. Attempting to deliver the required 12,500 A into the
radars dc input will drop 375 V across the ultracapacitor, which is clearly impossible and thus the
system would be inoperable.
If a 10% voltage drop across the ultracapacitor is considered acceptable when discharged to VLO,
the ESR must be decreased to 0.54 m. This ESR value will require 56 parallel strings of
19

capacitors for a total of 5936 individual ultracapacitors. Any single ultracapacitor failure will
effectively eliminate the parallel string in which it is located from the total energy storage, thus
increasing the ESR. Redundancy mechanisms can be employed, but clearly energy storage for ride
through is nontrivial for MVDC loads; this is especially true when considering the boost converter
size and cost which for this example would be required to support 27.8 kA input currents when the
ultracapacitors were at their lowest energy state.
Table 2 Ultracapacitor parameters for 0.2 s ride through for various dc loads using a single series string

Load

VHI
(V)

VLO
(V)

Energy
(MJ)

Needed
Capacitance
(F)

Series
Capacitors

ESR
()

Radar

300

150

0.75

22.2

106*

0.03

DC Loads

800

400

0.19

0.78

281

0.22

Propulsion
(1 of 2 motors)

5000

2500

7.3

0.78

1755*

0.51

* Maxwell BCAP 3000, Maxwell BCAP 0650

3.5

Grounding Issues

Naval ships are required to be tolerant to single-point ground faults in order to provide service
continuity during line-to-ground faults. Traditionally, ac systems achieve this immunity to ground
faults by ensuring the distribution system is ungrounded. However, in a dc system, ungrounded
distribution may not be the best answer due to the high neutral-to-ground voltages caused by
interactions of the main rectifiers [4].
A better grounding scheme for dc distribution, according to [4], is to use high impedance grounding
at the main generators and incorporate isolation transformers into dc-dc converters. While high
impedance grounding will reduce the neutral-to-ground voltages in the distribution system, ground
faults may become difficult to detect since ground fault current will be small and potentially
obscured by normal leakage currents; thus, sensors must take this leakage into account [22]. High
impedance grounding will allow a distribution system to continue functioning normally after the
first such fault occurrence; however, any metallic object contacted by the faulted component will be
at elevated potential and can be a personnel hazard.
4

Manufactured Medium Voltage Drive Review

Manufactured medium voltage (MV) drives (ac-dc-ac converters) are well-suited for use as a
primary fault current interrupter for a local load or as backup protection for a group of loads and
respective breakers. MV drives are typically fuseless and self-protecting. Circuit breakers are
limited to installations on the ac input side. IGCT-based VSI drives dispense with fuses and instead
employ series inductors and series IGCTs in-line with the dc bus to act as a fault current interrupter
(essentially a type of dc circuit breaker). IGBT-based VSI drives do not need the series inductors or
semiconductor breakers since current rate of rise and short circuit protection can be fully controlled
via the gate of the IGBT [31]. During a drive fault in a typical industrial application, the upstream
transformer is the primary current limiting device [32]. Examples of MV drives which are fuseless
and fully protected as described above include the ACS 1000, ACS 5000, and Megadrive-LCI

20

drives made by ABB, and the Sinamics GL 150, GM 150, and SM 150 made by Siemens, among
others [32, 33, 34, 35, 36].
Multi-motor drives systems, in which one or more rectifiers feed two or more motor drives via a
common dc bus, are common practice by major drive manufacturers [ 37]. Fault management in
such a system would be similar to that of a navy ship with dc distribution and multiple loads.
5

Recommendations

The result of this investigation is that multiple approaches to dc fault protection can be made to
work, but there are two major issues remaining. First, there is no performance data base available
for any of the foregoing approaches that would permit a protection engineer to specify a particular
system with confidence. Second, non-hybrid distribution and fault protection approaches are likely
to yield significantly larger than optimal systems.
A possible approach to minimizing both of these problems is to use a hybrid power system that
generates power at a higher ac frequency than 60 Hz, e.g., 240 Hz; with zonal loads that can be deenergized to clear faults powered by dc. This will likely yield a smaller system than a pure dc
system. Moreover, anecdotal information suggests most of the breaker experience gathered at 60
Hz will be applicable at 240 Hz. A cost effective MVDC breaker should be developed and tested.
Design data and system validation can be done using a relatively inexpensive combination of
modeling and experimentation at a reasonable power level, e.g., 0.5-1 MW. Initial modeling has
been done to show the feasibility of some dc protection approaches [38]. In addition, measurements
of dc series faults in a dc microgrid [23, 24, 39] showed that the faults are unlikely to induce
voltage spikes that damage power electronics. This means that power electronic solutions are likely
to be robust.
A focused program to combine experiments on the microgrid with comprehensive modeling and
simulation should provide the confidence needed to commit to the smaller hybrid system. A navallike dc microgrid already exists in the ESRDC. Protection and stability issues should be validated
(and verified) experimentally before finalizing the architecture choice based solely on simulation
predictions.
6

Conclusion

An overview of MVDC distribution challenges and techniques with regard to fault mitigation has
been provided in order to provide a basis for determining under what conditions dc breakers are
required when using an MVDC distribution. It was discussed that a branch line within a dc
distribution network does not require a dc breaker if the converter servicing that branch powers only
one load. In this case, the converter which is designed to be self-protecting can function as the
primary current interrupter for a fault at the load or within portions of the converter itself. DC
breakers can also be avoided on a multi-load bus if it is acceptable to de-energize the entire bus to
clear a fault. To clear a permanent fault using this method, non-load breaking switches are needed
to isolate the faulted component before the bus is re-energized. For all other cases, dc breakers
should be used as they provide superior protection, flexibility, and minimize service interruptions.
In terms of cost and physical size, the fault protection components required for safely operating a
multi-load dc distribution system without dc breakers are expected to be less expensive and smaller
than one equipped with dc breakers. However, if the primary goal for moving beyond a traditional
21

60 Hz ac distribution system is to reduce system cost, size, and weight, a hybrid distribution system
utilizing a combination of dc and high frequency ac power transmission should be considered as a
possible alternative to a fully-dc topology. In this arrangement, distribution would be provided
primarily by high frequency ac (e.g., 240 Hz) allowing for the use of smaller transformers, smaller
generators, and traditional breakers with less time delay needed to locate a current zero crossing.
DC distribution would then only be used for single-load buses and multi-load buses which are
acceptable to de-energize during a fault. In this manner, traditional ac breakers protect the majority
of the distribution system and power electronic converters protect the remaining dc portion of the
distribution.

22

References

[1] Hamilton, H.; Schulz, N.N., Impact of grounding on overcurrents for a naval ac/dc shipboard power system, Power Symp.,
2007. NAPS '07. 39th North American, pp. 170-176, 30 Sept. 2007-2 Oct. 2007.
[2] Kempkes, M.; Roth, I.; Gaudreau, M., Solid-state circuit breakers for medium voltage dc power, Electric Ship Technologies
Symp. (ESTS), 2011 IEEE, pp. 254-257, 10-13 April 2011.
[3] Ghisla, U.; Kondratiev, I.; Dougal, R., Protection of medium voltage dc power systems against ground faults and negative
incremental impedances, IEEE SoutheastCon 2010 (SoutheastCon), Proc. of the, pp. 259-263, 18-21 March 2010.
[4] Baran, M.E.; Mahajan, N.R., DC distribution for industrial systems: opportunities and challenges, Industry Applications, IEEE
Trans. on, vol. 39, no. 6, pp. 1596-1601, Nov.-Dec. 2003.
[5] IEEE recommended practice for electrical installations on shipboard, IEEE Std 45-2002 (Revision of IEEE Std 45-1998), pp.
0_1-258, 2002.
[6] Morton, John S., Circuit breaker and protection requirements for dc switchgear used in rapid transit systems, Industry
Applications, IEEE Trans. on, vol. IA-21, no. 5, pp. 1268-1273, Sept. 1985.
[7] Data used by the ESRDC IPT-Electrical working group, 2011.
[8] Rudraraju, S.R.; Srivastava, A.K.; Srivastava, S.C.; Schulz, N.N., Small signal stability analysis of a shipboard MVDC power
system, Electric Ship Technologies Symp., 2009. ESTS 2009. IEEE, pp. 135-141, 20-22 April 2009.
[9] Jin Yang; Fletcher, J.E.; O'Reilly, J., Multiterminal dc wind farm collection grid internal fault analysis and protection design,
Power Delivery, IEEE Trans. on, vol. 25, no. 4, pp. 2308-2318, Oct. 2010.

[10] Jovcic, D.; and Wu, B.; Fast fault current interruption on high-power DC networks, Power and Energy Society
General Meeting, IEEE, Minneapolis, MN, 25th - 29th pp 1-6, July 2010.
[11] Meyer, J.-M.; Rufer, A., A DC hybrid circuit breaker with ultra-fast contact opening and integrated gate-commutated thyristors
(IGCTs), Power Delivery, IEEE Trans. on, vol. 21, no. 2, pp. 646-651, April 2006.
[12] Krstic, S.; Wellner, E.L.; Bendre, A.R.; Semenov, B., Circuit breaker technologies for advanced ship power systems, Electric
Ship Technologies Symp., 2007. ESTS '07. IEEE, pp. 201-208, 21-23 May 2007.
[13] Franck, C.M., HVDC circuit breakers: a review identifying future research needs, Power Delivery, IEEE Trans. on, vol. 26,
no. 2, pp. 998-1007, April 2011.
[14] Meyer, C.; Kowal, M.; De Doncker, R.W., Circuit breaker concepts for future high-power DC-applications, Industry
Applications Conf., 2005. Fourtieth IAS Annual Meeting. Conf. Record of the 2005, vol. 2, pp. 860-866, 2-6 Oct. 2005.
[15] Meyer, C.; Schroder, S.; De Doncker, R.W., Solid-state circuit breakers and current limiters for medium-voltage systems
having distributed power systems, Power Electronics, IEEE Trans. on, vol. 19, no. 5, pp. 1333-1340, Sept. 2004.
[16] Mesut E. Baran; Nikhil R. Mahajan, Overcurrent protection on voltage-source-converter-based multiterminal dc distribution
systems, Power Delivery, IEEE Trans. on, vol. 22, no. 1, pp. 406-412, Jan. 2007.
[17] Kulkarni, S.; Santoso, S., DC fault interruption using reactor assisted circuit breaker, to be submitted to IET Electrical
Systems in Transportation , (Dec. 2011).
[18] Zhenxue Xu; Bin Zhang; Sirisukprasert, S.; Xigen Zhou; Huang, A.Q., The emitter turn-off thyristor-based DC circuit
breaker, Power Engineering Society Winter Meeting, 2002. IEEE, vol. 1, pp. 288-293, 2002.
[19] Candelaria, J.; Park, J., VSC-HVDC system protection: a review of current methods, IEEE PES Power Systems Conference
and Exposition (PSCE 11), Phoenix, Arizona, 20th - 23rd , pp. 1-7, March 2011.
[20] Baran, M.; Mahajan, N.R., PEBB based dc system protection: opportunities and challenges, Transmission and Distribution
Conf. and Exhibition, 2005/2006 IEEE PES, pp. 705-707, 21-24 May 2006.
[21] Hongbo Jiang; Ekstrom, A.; Multiterminal HVDC systems in urban areas of large cities, Power Delivery, IEEE Trans. on, vol.
13, no. 4, pp. 1278-1284, Oct. 1998.
[22] Salomonsson, D.; Soder, L.; Sannino, A., Protection of low-voltage dc microgrids, Power Delivery, IEEE Trans. on, vol. 24,
no. 3, pp. 1045-1053, July 2009.
[23] Estes, H.B.; Kwasinski, A.; Hebner, R.E.; Uriarte, F.M.; Gattozzi, A.L, Open series fault comparison in ac & dc micro-grid
architectures, in 33rd International Telecommunications Energy Conf. (INTELEC), Amsterdam, The Netherlands, 9-13 Oct.
2011.
[24] Uriarte, F.M.; Estes, H.B.; Hotz, T.J.; Gattozzi, A.L.; Herbst, J.D.; Kwasinski, A.; Hebner, R.E., Development of a series fault
model for dc microgrids, to be presented at the 2012 PES Innovative Smart Grid Technologies Conf., Washington, D.C., 16-18
Jan 2012.
[25] Jovcic, D., Bidirectional high power DC transformer, Power Delivery, IEEE Trans. on, vol. 24, no. 4, pp. 2276-2283, Oct.
2009.
[26] Riccobono, A.; Santi, E.; Positive Feed-Forward Control of Three-Phase Voltage Source Inverter for DC Input Bus
Stabilization, Proc. IEEE Applied Power Electronics Conference (APEC'11), pp., March 6 10, 2011, Fort Worth, TX
[27] Kusic, G.L.; Reed, G.F.; Svensson, J.; Zhenyuan Wang, A case for medium voltage dc for distribution circuit applications,
Power Systems Conf. and Exposition (PSCE), 2011 IEEE/PES, pp. 1-7, 20-23 March 2011.
[28] Lianxiang Tang; Boon-Teck Ooi, Locating and isolating dc faults in multi-terminal dc systems, Power Delivery, IEEE Trans.
on, vol. 22, no. 3, pp. 1877-1884, July 2007.
[29] Cairoli, P.; Kondratiev, I.; Dougal, R., Ground fault protection for dc bus using controlled power sequencing, IEEE
SoutheastCon 2010 (SoutheastCon), Proc. of the, pp. 234-237, 18-21 March 2010.

23

[30] Hamilton, H.; Schulz, N.N., DC protection on the electric ship, Electric Ship Technologies Symp., 2007. ESTS '07. IEEE, pp.
294-300, 21-23 May 2007.
[31] B. Wu, High-Power Converters and AC Drives. Piscataway, NJ: IEEE Press, 2006.
[32] The rugged and reliable drive for the highest power ratings, Sinamics GL-150 product brochure [Online Resource]
https://www.automation.siemens.com/mcms/infocenter/dokumentencenter/ld/Documentsu20Brochures/mv-umrichter/wssinamics-gl150-en.pdf.
[33] ABB drives in metals: medium voltage drives for improved product quality and process control [Online Resource]
http://www05.abb.com/global/scot/scot216.nsf/veritydisplay/5a1234630d9e1550c125785b00442bed/$file/Metals%20BR%20Re
vB_lowres.pdf.
[34] Medium voltage ac drive ACS 1000, ACS 1000i, 315 kW 5 MW, 2.3 4.16 kV, ABB | ACS 1000, ACS 1000i product
brochure [Online Resource]
http://www05.abb.com/global/scot/scot216.nsf/veritydisplay/a3b20f7f91ca1d1bc125791a0045b07b/$file/ACS%201000%20EN
%20Rev%20G_lowres.pdf.
[35] Medium voltage ac drive ACS 5000, 1.5 MW - 21 MW, 6.0-6.9 kV, ABB | ACS 5000, ACS 5000i product brochure [Online
Resource]
http://www05.abb.com/global/scot/scot216.nsf/veritydisplay/0a9dec5fae30bd97c12576c0004a5b0a/$file/ACS%205000%20EN
%20Rev%20F_lowres.pdf.
[36] Medium voltage ac drive MEGADRIVE-LCI for control and soft starting of large synchronous motors, MEGADRIVE-LCI
product brochure [Online Resource]
http://www05.abb.com/global/scot/scot216.nsf/veritydisplay/f7eafb5af0f19f2ec12577e40057427e/$file/MEGADRIVELCI%20EN_Rev%20B_lowres.pdf.
[37] The reliable medium-voltage drive with IGCTs, Sinamics GM150 and SM150 product brochure [Online Resource]
https://www.automation.siemens.com/mcms/infocenter/dokumentencenter/ld/Documentsu20Brochures/mv-umrichter/wssinamics-sm150-gm150-igct-en.pdf.
[38] Ouroua, A.; Beno, J.; Hebner, R., Analysis of fault events in MVDC architecture, Electric Ship Technologies Symp., 2009.
ESTS 2009. IEEE, pp. 380-387, 20-22 April 2009.
[39] Uriarte, F.M; Gattozzi A. L.; Herbst J.; Estes H.; Hotz, T.; Hebner, R.E.; and Kwasinski, A., "A New Hyperbolic DC Arc Model
for Series Faults in Low Voltage Microgrids," IEEE Trans. Smart Grid, submitted for publication Dec. 2011.

24

Das könnte Ihnen auch gefallen