DC Microgride Protection PDF

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

By Atif Maqsood and Keith Corzine

DC Microgrid
Protection
Using the coupled-inductor solid-state circuit breaker.

ince the great debate between DC Circuit Breaker Technologies

S
­Thomas Edison and Nikola Tesla, our Many dc microgrid systems require rapid
nation’s power system has operated on reconfiguration for survivability. This has
alternating current (ac). This was chosen led to research into advanced dc circuit
over direct current (dc) because of the need breakers. One popular choice is the hybrid
to increase voltage with ac transformers to a high value dc breaker, which uses a mechanical switch
using transformers for long-distance power transmission. in parallel with a path containing semicon-
The system has served its purpose well, but now, many ductor devices. When the mechanical
energy sources, such as solar panels, fuel cells, and batter- switch is opened, the current is diverted to the semi-
ies, supply dc voltage. Also, dc/dc power converters are conductor, which is then opened. The current is
commonly used to transform voltage and to interface ultimately diverted to a metal–oxide varistor,
these dc sources with a larger system. Because of this, local which clamps the voltage and allows system
dc power systems (or microgrids) have become popular inductance to reduce the current. A main
topics in research literature. It also turns out that interfac- advantage of this type of breaker is its low
ing a wind power generator to a dc system is simpler than on-state power losses. Another type of dc cir-
interfacing it to an ac system because ac/dc conversion is cuit breaker is the fully solid-state version.
needed for the former and ac/dc/ac conversion is needed There are many types of solid-state breakers. Some
for the latter. Although energy sources and power conver- use a resonant circuit to cause the semiconductor cur-
sion are readily available for dc power systems, some high- rent to go to zero, and others divert the current to a free-
performance applications require fast-acting dc circuit wheeling diode at the breaker output. The main advantage
breakers, which are currently in the experimental phase. of solid-state breakers is their extremely rapid operation.
This article discusses options for high-performance dc cir- The following section describes a coupled-inductor dc
cuit breakers and specifically details the coupled-inductor breaker, which is a variation of a solid-state breaker that
dc breaker. This breaker is demonstrated for fault protec- includes automatic fault detection.
tion in a notional dc microgrid.
The Coupled-Inductor DC Breaker
The coupled-inductor dc circuit breaker is shown in
Digital Object Identifier 10.1109/MELE.2016.2544240
Date of publication: 31 May 2016 Figure 1. The main conduction path consists of

58 I EEE E l e c t r i f i c a t i on M a gaz ine / J un e 2016 2325-5987/16©2016IEEE


a silicon-controlled rectifier (SCR) (S 1) and coupled
inductor. The principle of operation can be described as
follows. At start-up, S1 is gated on and the capacitor (C)
is charged from the source through the coupled
inductor with the charging resistor (R). At this time,
the source supplies the load through the coupled
inductor and S1. Once the steady state is reached, the
gate signal is removed from S 1. A fault at the load
causes the capacitor to discharge through diode (D 2)
and the secondary side of the coupled inductor. This
transient discharge current naturally reflects back to
the source by the turns ratio N2/N1. That is, the working
aspect of the coupled inductor is such that when the
breaker output current (i o ) rises due to a fault, the
source current (is) decreases. These laboratory measure-
ments are shown in Figure 2. In this setup, the initial
source and load current are 6A when a short-circuit is
applied at the load side. As can be seen, the breaker out-
put current can go up to nearly 100A. The source cur-
rent rapidly responds by going to zero, and
the S1 switches off. The capacitance then
resonates with the coupled inductance,
but this is stopped by the diode (D1). The
coupled-inductor dc breaker has an auto-
matic and rapid response to a fault. Addi-
tionally, the source does not experience the fault
current. The amount of transient current that will
switch off the breaker can be set using the turns ratio of
the coupled inductor. Another mode of operation for the
coupled-inductor breaker is as a dc switch. Any time the
g
in
h
is
bl

source is supplying the load, the breaker can be pur-


pu
am
r

posefully switched off by gating S 2. This causes the


g
in
by
d

capacitor to discharge through the secondary of the


se
en
c
li

coupled-inductor and S 2. As with the fault, this dis-


e
ag
im

charge current causes S1 to switch off.

Bidirectional Coupled-Inductor DC Breaker


Figure 3 shows a version of the coupled-inductor dc
breaker that is capable of bidirectional power flow. It
contains a center-tapped transformer as well as

IEEE Electrific ation Magazine / j u n e 201 6 59


D1
is io
is S1 io
N1

N2
C S2 dc
dc
Fault
Source Load
R D2

dc Breaker/Switch

Figure 1. The coupled-inductor dc circuit breaker. Figure 3. A bidirectional, coupled-inductor dc breaker.

8 1 4 5 6
G A C D G –
~
is (A)

4 100 µs E

0
3 7

100
F
2 11 8
io (A)

50 G B J K H –
9 –
10
0 I ––

Figure 2. The measured source and load currents. Figure 4. The dc microgrid test system.

bidirectional SCRs. SCRs are also


added in the negative rail, which
As can be seen, loads. An inverter load is connected
to node 6, while nodes 8 and 10 feed
was seen as necessary in the central control dc/dc converter loads. In this
microgrid system to prevent circu- notional dc microgrid, the dc/dc
latory currents. To use the breaker provides the desired converters are of the boost topology.
as a dc switch, current flow direc-
result for all of the
tion should be known. If I 1 , as Control Options
shown in Figure 3, is positive, then fault locations, The dc breaker proposed in this arti-
S 1 should be gated. This will dis- cle is autonomous in that it does not
charge the capacitor on the sec- without exception. need to detect the fault to switch off.
ondary winding of the transformer The fault response is completely
such that the primary current will automatic in the absence of gate sig-
force I1 to zero, thereby switching off the SCRs in the nals for the SCRs; however, in the dc microgrid configura-
primary conduction path. If I2 is positive, S2 must be tion shown in Figure 4, there may be certain cases where
gated to use the breaker as a dc switch. external control will be required, such as:
xxfor maintenance or repair, it might be required to divert
Notional DC Microgrid power from one branch to another (this is an example
Figure 4 shows the dc microgrid that will be studied of where the breakers will be used as a dc switch)
herein. The generators (G) are standard dc power sup- xxfor loads with a high starting current, where some
plies with droop control for power sharing. Coupled- control will be required to gate the SCRs until the sys-
inductor breakers are placed to protect the sources, tem reaches steady state
lines, and loads. Two bidirectional coupled breakers xxfor reconfiguring the system such that the direction of
(breakers E and F) are placed on one line that can con- current through a bidirectional breaker has to be changed
duct power in both directions. The line impedance is xxfor integrating a branch or load back into the system
modeled by small a inductance and resistance. Node 3 after a fault has been removed.
represents the bus, with both the sources feeding into In a previous work with z-source dc breakers, the pro-
it. Nodes 6, 8, and 10 are terminal nodes leading to posed control was a central control processor constantly

60 I E E E E l e c t ri f i c a t i on M a gaz ine / j un e 2016


in communication with all of the breakers. In this inverter is isolated, the steady-state current through the
article, two other control approaches are discussed: bidirectional breakers should reverse. Initially, the SCRs in
local breaker control and paired breaker control. The bidirectional breakers are not gated, so the breakers will
merits of all three control strategies are presented switch off as the current falls to zero. It is then the
here and some simulation results are shown in the responsibility of the local control to identify that the
following section. breakers turned off due to a current reversal and not due
to a fault. This is one example of a
Central Control It would be much case where the control must gate the
With central control, all the breakers SCRs to switch the breakers back on.
receive gate signals from a central easier to expand the In this case, the local control
processor through a communication would be monitoring the output and
protocol. At the same time, all the system if each dc input currents of the breakers. If the
breakers send a signal corresponding
to their input and output current to
breaker had its control observes that the input and
output current fell to zero without
the central control. The input current own independent exceeding a threshold set for fault
is compared to a small threshold to currents, that would indicate that
indicate whether the breaker is local control. there is no shunt fault at the break-
switched on or off. The output current er output, and the breaker must be
is compared to a large threshold switched back on. If this control
value that indicates whether the breaker experienced a strategy is employed and the fault occurs at location 6
fault at its output. Using this information from all break- as described in the aforementioned example, the bidi-
ers, the central control determines the fault location in the rectional breakers will be sent gate signals and will
grid and ensures that only the minimum required number switch back on, allowing them to conduct current in
of breakers are switched off to isolate the fault. The central the reverse direction. The advantage of this method is
processor also provides gating signals for start conditions that it does not require an elaborate communication
and use of the breakers as dc switches. infrastructure. It would be much easier to expand the
Central control has been simulated in previous work and system if each dc breaker had its own independent
shown to perform well. The advantage of central control is local control.
that it can locate and isolate a fault at any location in the There are some disadvantages associated with this
grid. Its processing requirement is not excessive, but the dis- control strategy. If the fault is located behind the break-
advantage of central control is the required communication er, the control would misconstrue that it is a case similar
infrastructure. For example, if a universal asynchronous to the example above and would attempt to switch on
receiver/transmitter is used as a communication protocol, the device. For example, consider a fault at the output of
each breaker must decode an additional device and encode one of the generators (G). The breaker at that terminal
the data into the proper format. will not see a high current at its output, so the control
would gate the SCRs of the breaker. The SCRs will still
Local Breaker Control isolate a shunt-resistive fault because it blocks negative
In ac systems, differential protection is one of the most current, but this is risky; if the fault impedance has high
common schemes employed for bus protection. For this inductive element, the fault current might resonate and
scheme, the output current is compared to the input still interfere with the system. Another disadvantage of
current. Under normal conditions, the input and output this control scheme is that it will not isolate all fault
currents are similar, but when a fault occurs, either the locations. If the fault is at the input end of the breaker,
input or output current changes rapidly, and the relay the local control embedded in the breaker will not be
uses that as an indication of a fault, signaling the break- able to distinguish it from the case where the breaker
er to open. must be switched back on. All of these scenarios are
Each of the breakers can be monitored using the same summarized in Table 1.
principle. The one important difference is that the dc
breaker does not require a relay to switch off; however, Paired Breaker Control
under certain conditions, it may need to switch back on It is possible for some breakers to operate with the
when there is no fault. For example, suppose the inverter same gate signals. For example, consider the two
load in Figure 4 is much greater than the dc/dc converter breakers on line 4. Either both the breakers conduct
loads combined. In that case, the steady-state current the same current or neither of them conducts a cur-
flow through the bidirectional breakers would be from rent; it is not possible for one of them to be conducting
nodes 9 to 5. Now, imagine that a fault at node 6 turns off and the other to be open because they share the same
the breaker at node 6 to isolate the fault. This would also path. Therefore, it is possible for these two breakers to
isolate the inverter load from the system. Once the share the same gate signal. The breakers that will be

IEEE Electrific ation Magazine / j u n e 201 6 61


paired are listed in Table 1. There will be some commu- summarized in the results section. One limitation with
nication between the paired breakers. Gate signals will this control is that the fault at junctions 5 and 9 will be
be calculated locally at one of the misconstrued as an example of a
breakers using the current informa- The only reason case where a breaker goes off with-
tion from both breakers. One of the out a fault. This is because at any
breakers will transmit its current other control given time, either junction 5 or 9
data continuously to the other break- will be at the input end of both the
er while receiving the gate signals.
strategies are bidirectional breakers. However,
If a fault happens between the considered is to this limitation can be tolerated if
two paired breakers, both the break- the breakers at junctions 5 and 9
ers will remain switched off, even simplify the are located close in space, making
though the fault is at the input end the probability of a fault happening
for one of them. This is the advan-
communication at those junctions very low. The
tage of this control compared to architecture. communication infrastructure
local breaker control. More fault required for this control is not as
locations can be isolated, and the elaborate as for central control,
conditions where breakers need to be switched on can be which makes it easier to implement. It is a compromise
identified. The simulation results for all locations are between central control and local control in which abso-
lutely no communication is required.

Table 1. The description of paired devices Simulation Results


for control algorithm. The notional dc microgrid was simulated using the soft-
ware PSCAD. For simplicity, the generator, inverter, and
Paired Devices Description
converters are modeled using average-value models.
Source 1 and Breaker A receives output current Table 2 provides a good summary of the results from the
breaker A data from source 1.
three control strategies discussed in this article. As can
Source 2 and Breaker B receives output current be seen, central control provides the desired result for
breaker B data from source 2.
all the fault locations, without exception. The only rea-
Breakers C and D Breaker C transmits current data son other control strategies are considered is to simplify
and receives gate signals.
the communication architecture.
Breakers J and K Breaker J transmits current data
For location 7, the local control turns breaker E back
and receives gate signals.
on. This is because when the control deduces that there
Breakers E and F Breaker E transmits current data
is no fault at the output of breaker E, it will gate the
and receives gate signals.
SCRs. Unlike other breakers, the SCRs in E allow current

Table 2. The summary of fault response for all control schemes.

Central Control Local Control Paired Control Ideal Case


Breakers Start Breakers Breakers Start Breakers Breakers Start Breakers Should
Fault Conducting Are Gated Conducting Are Gated Conducting Start Conducting
Location Again Again Again Again Again Again
1 None A None None None None
2 None B None None None None
3 None All except A, B None All except A, B None None
4 None D None None None None
5 None C, F, G None G None None
6 C, D, E, F C, D, E, F C, D, E, F C, D, E, F C, D, E, F C, D, E, F
7 None E E None None None
8 None None None None None None
9 None E, F, H, I, J E, F E, F, H, I E, F None
10 None None None None None None
11 E, F, H, I E, F, H, I, K E, F, H, I E, F, H, I E, F, H, I E, F, H, I

62 I E E E E l e c t ri f i c a t i on M a gaz ine / j un e 2016


to flow back into the fault, and the breaker starts con- Even with paired control, some variation of centralized
ducting. One way to counter this problem is to have a control can exist for these operations, but the breakers
system where only breakers E and F will no longer require continuous
have paired control, while other
After 0.4 s, the current data to be sent or continu-
breakers have local control. Howev- ous gate signals to be received.
er, there are other disadvantages of breaker turns back It would be useful to observe the
local control. For example, every current through these dc breakers
time the fault is at the input end of on and the output for some of the cases from Table 2.
the breaker, the breaker is signaled current is now The paired control scheme is
to turn on the SCRs. This does not employed in the simulation and, for
necessarily result in the breaker negative, demonstr­ the first case, a fault is created at
conducting because the SCRs do not location 4. Figure 5 shows the input
allow a path for negative current to ating that the current flowing through breakers C,
flow through, even if they are gated. direction of current G, and J, and the output current of
However, there is a risk—if the fault breaker C. The first two plots in Fig-
happens to have resonance and the through breaker ure 5 show the input and output
SCR remains gated, then unwanted currents for breaker C. Only the out-
current can be injected back into
E has changed. put current of the breaker C shows
the system, which could cause some the fault current, but the input cur-
other breakers to turn off. Table 2 rent goes to zero instantly. Breaker
shows results of a simple resistive fault, but it is possi- G’s current is shown in the third plot. There is some
ble for the fault to resonate current back into the sys- disturbance at the time of the fault, but the steady-
tem if SCRs of adjacent breakers are gated. state current through breaker G is the same before and
As discussed in the section “Paired Breaker Control,” after the fault as it is feeding the load. The fourth plot
if the faults at junctions 5 and 9 are ignored, which is a shows the output current through breaker J that rough-
­reasonable assumption, the paired breaker control ly doubles because branch 11 has to provide current to
scheme provides the ideal response, which is similar to all the loads now that branch 4 has been isolated from
central control without the complex communication the system.
architecture. For a fault at location 3, all sources are dis- The next case is for a fault at location 6. Figure 6
connected from the system, so it does not matter that shows the output current for breakers E, G, and K are
paired control sends gate signals to the rest of the shown, as well as the input current for breaker G. The
breakers. Central control still holds an advantage in first two plots in Figure 6 show the input and output cur-
terms of coordinating starting gate signals and using rents through breaker G, respectively. The fault is at the
the breakers as dc switches to reconfigure the system. output of breaker G, so only the output current shows the

10 10
is,C (A)

is,G (A)

5 100 ms 5 200 ms
0 0
600 400
io,C (A)

io,G (A)

300 200
0 0
20 2
is,G (A)

io,E (A)

10 0
0 –2
20 6
is,J (A)

io,K (A)

10 4
0 2

Figure 5. The current response to fault at location 4. Figure 6. The current response to fault at location 6.

IEEE Electrific ation Magazine / j u n e 201 6 63


With that in mind, we introduce a coupled-inductor-
10 based dc breaker. The proposed dc-breaker design uses
io,D (A) an SCR to instantly interrupt the fault current. A bidirec-
8 tional version of the breaker is also provided so multiple
100 ms
6 dc breakers can be installed in a notional dc microgrid.
2 The breaker also features a control switch within the
design so the breaker can be manually opened without
is,E (A)

1 creating a large disturbance in the system. Previously, a


central control was proposed to supervise operations of
0
2
multiple breakers in a microgrid. In this article, two
other control strategies are discussed: one involving
io,E (A)

1 independent, local control of breakers, and the other


involving controlling breakers in pairs. A simulation is
0
carried out with all three control methods and a sum-
6
mary of result is presented. The merits of all three strat-
io,K (A)

4 egies are discussed. It can be concluded that, within


certain limitations, the paired control method can pro-
2 vide the desired results with a much simpler communi-
cation architecture.
Figure 7. The response to the dc switch at E turning off.
For Further Reading
W. Setthapun, S. Srikaew, J. Rakwichian, N. Tantranont, W. Rak-
wichian, and R. Singh, “The integration and transition to a dc
spike from the fault current, but the input current falls
based community: A case study of the smart community in
instantly to zero. The third plot is the output current for Chiang Mai world green city,” in Proc. IEEE Int. Conf. DC
breaker E. The output current does not show any sign of Microgrids, Atlanta, GA, 2015, pp. 205–209.
the fault current, so the control sends gate signals to the D. Ricchiuto, R. A. Mastromauro, M. Liserre, I. Trintis, and S.
SCRs to switch the breaker back on. After 0.4 s, the break- Munk-Nielsen, “Overview of multi-dc-bus solutions for dc
microgrids,” in Proc. IEEE Int. Symp. Power Electronics Distrib.
er turns back on and the output current is now negative,
Generation Syst., Rogers, AR, 2013, pp. 1–8.
demonstrating that the direction of current through A. Shukla and G. D. Demetriades, “A survey on hybrid circuit-
breaker E has changed. The fourth plot shows the current breaker topologies” IEEE Trans. Power Delivery, vol. 30, no. 2,
through breaker K in three distinct regions. After the pp. 627–641, Apr. 2015.
fault, the current through breaker K drops because the R. Schmerda, R. Cuzner, R. Clark, D. Nowak, and S. Bunzel,
“Shipboard solid-state protection: overview and applications,”
inverter load is removed from the system. The current
IEEE Electrification Mag., vol. 1, no. 1, pp. 32–39, Sept. 2013.
decreases further after 0.4 s, when lines 4 and 7 are inte-
grated back into the system.
Biographies
The final case is the operation of breaker E as a dc
Atif Maqsood (amaqsoo@g.clemson.edu) is currently
switch to isolate line 7. Input and output currents through
pursuing his Ph.D. degree at Clemson University, South
breaker E are shown in Figure 7, as well as the output cur-
Carolina. His research interests include power electron-
rent through breakers D and K.
ics, motor drives, power system protection, and electric
The results from Figure 7 show the merit of using a
system modeling.
breaker as dc switch. The second and third plots show
Keith Corzine (Keith@Corzine.net) is a professor at
the input and output currents through breaker E. Both
Clemson University, South Carolina. He has 20 years of
these currents fall instantly to zero without showing
experience working with power electronics, motor
any signs of fault current. Once line 7 is isolated, lines 4
drives, naval ship propulsion systems, and electric
and 11 smoothly change their currents to accommodate
machinery. He has published more than 50 refereed
their new loads, as shown by the first and fourth plots
journal papers, more than 80 refereed international con-
in Figure 7.
ference papers, and holds three U.S. patents related to
power conversion.
Conclusions
The large-scale implementation of dc microgrids
depends on developing reliable dc protection options. 

64 I E E E E l e c t ri f i c a t i on M a gaz ine / j un e 2016

You might also like