An approach to automated restoration and ... - pserc-smart-dsys

peanutplausibleElectronics - Devices

Nov 21, 2013 (3 years and 9 months ago)

84 views

1


S
MART
G
RID


AUTOMATED RESTORATIO
N AND RECONFIGURATIO
N OF A
DISTRIBUTION SYSTEM


D. A. Haughton

S
EPTEMBER
,

2009


1.
Introduction

This research project is based on a PSERC proposal addressing ‘implications of the smart
grid initiative on distribution engi
neering’. Specifically, this project focuses on both the
r
econfiguration of legacy radial distribution systems to networked structures suitable for the
s
mart
g
rid
, and the i
mprovement of reliability through sensors
-
based supervised and fully
automated rest
oration in distribution systems

which correspond to tasks 2 and 3 of the original
project proposal. Task requirements are listed as follows:

Task 2: Reconfiguration of legacy radial distribution systems to networked structures
suitable for the Smart Grid.
It is acknowledged that a move toward a meshed distribution
system, similar to that existing in the transmission levels of the system provides advantages
of increased reliability for dispersed end us
ers.

While networked distribution system
architectures a
re common in central business districts, this design is not used in lighter urban,
suburban, and rural areas. To increase reliability of supply and maximization of existing
distribution systems, a systematic study quantifying the levels and locations of in
frastructural
improvements (i.e., selective conversion from radial to meshed networks or the use of other
doubly fed designs) is integrated into the proposed research
.
The task on reconfiguration of
legacy radial topology of distribution systems to meshed
networks will take into account the
maximization of existing distribution system infrastructure and a cost to benefit analysis of
selecting optimum locations for achieving the reconfiguration. The intent is to develop a
framework for designing the reliable

distribution system architecture
t
hat is most conducive
for integrating sensors based controls and renewable sources of generation.


Task 3: Improvement of reliability through sensors
-
based supervised and fully automated
restoration in distribution system
s.
This task builds on PSerc projects T
-
31 and S
-
30. The
intent is to develop the logic
for restoring segments of the distribution system after failure or
removal of a component in the system. It is known that for a system to be highly reliable and
faul
t tolerant, in addition to multiple redundant paths, it is paramount to have smart strategies
such as fault detection, isolation and reconfiguration (FDIR) to manage redundancy. The
existence of FDIR in current networked distribution system is limited to l
ocal protection
schemes, which usually do not communicate with each other. Thus, in order to selectively
convert existing networked distribution systems into smart distribution systems, we propose
to add FDIR mechanisms and other smart control concepts. Th
e methods used shall also
replicate transmission system concepts that use system restoration logic to automatically
restore the distribution system. Using circuit matrix concepts, the optimization of the
reliability at system buses shall be performed base
d on the reliability of individual
components. Some of these concepts have been applied in special. It is possible to base
designs on a 0.99999 reliability concept


and this was shown in. The tasks planned under
this venture will not include studies on
protection engineering; rather, the effort will be
2


focused on optimal location of sensors in a smart distribution system and the processing of
information from the sensors to achieve automated and supervised restoration of the
distribution system.


2.
Back
ground Information


The distribution system shown in Fig. 1 is used to investigate a possible design algorithm that
fits requirements of a ‘smart
-
grid’. Specifically, a tool is investigated to model a ‘smart
distribution system’ that may automatically reco
nfigure to restore load. This distribution system
shown in Fig. 1 represents a typical legacy distribution system with three radial feeders. The
location and magnitudes of loads are provided in Table 1. As is typical of legacy radial
distribution feeders,
tie switches at specified locations on the feeders allow for some level of
operator initiated system re
-
configuration. One major objective of this work is to identify the
potential benefits of automatic reconfiguration by the optimal placement of sensory a
nd
interrupting devices on the distribution system. This leads directly to the design application of
optimal interrupting device location in a distribution system.





Fig. 1 Legacy radial distribution system used as a test bed for PSERC project






3


Tab
le 1 Bus load data for test bed system


Bus

P
(MW)

Q
(MVAr)

1

Source

N/A

2

0

-
0.5

3

2

1

4

2

2

5

1

0.4

6

Source

N/A

7

0.5

0

8

0.5

0.5

9

0

-
0.5

10

0.1

0.5

11

0.5

0

12

0.5

0.5

13

0.5

0.5

14

0.2

0

15

0

-
0.5

16

0.2

0

17

Source

N/A

18

0

0

19

1

0.5

Total demand

9

4.4




Design and operational aspects of a smart distribution system are investigated, and the
primary objective is stated as follows:



Using existing radial infrastructure
, migrate

to a (partially) networked infrastru
cture
capable of providing increased reliability
.

Include

sensor
-
based
, partially automated

reconfiguration and/or restoration ability
, but do not specify protection system requirements.
For initial testing, assume fault detection and isolation is accompli
shed as required.


(
F
rom the PSERC project proposal tasks and requirements).



3.
Design

application


optimal breaker placement for distribution system reconfiguration


The system shown in Fig. 1 is used to investigate the performance of an algorithm de
signed
to automatically reconfigure and restore loads following the detection of a permanent fault and
isolation of faulted components. The flow chart in Fig. 2 outlines the core processes of the
algorithm. This reconfiguration
-
restoration algorithm is als
o the basis for a design algorithm
where restoration/reconfiguration ability is used to evaluate optimal breaker locations on the
distribution system. Optimal breaker locations correspond to minimization of expected unserved
4


energy after a disturbance has
occurred. Circuit breaker locations are varied to find the optimal
breaker locations. Once the optimal breaker locations have been found, the partially networked
system can be evaluated for its ability to automatically reconfigure following disturbances.






Fig. 2 Flow chart for operation algorithm



5


Placing controllable switching devices on the system provides a range of allowable
configurations; three possible operating scenarios are identified. The system may be operated as
three completely isolated rad
ial feeders as originally designed; or the system may be operated in
an alternate configuration, but still as radial feeders; or there may be a highly interconnected
network.


Operating configurations may be selected by an operator or may be dictated by s
ystem
conditions. For example a repair or maintenance crew may require a line segment to be de
-
energized; or the system may be forced into a reconfigured state due to a fault. In any case, a
‘smart’ system can automatically reconfigure. Operating methodolo
gies may also vary from one
distribution system to another. For example, a distribution operator may choose to restore
critical loads first, and reduce the expectation of unserved energy. In this way,

operating
decisions impact system interruption and fr
equency indices (SAIDI and SAIFI) and other
reliability indices. Normally, locational marginal prices (LMPs) and their impact on the
integrated power system occur at the transmission level. But one might envision a case in which
the distribution operator

might
respond to LMP signals

and requests from the transmission
operator.


Attempts at increasing the ‘smartness’ of a distribution system are aimed at including more
controllable, sensor
-
based devices, automating restoration, and reducing the severity
and number
of interruptions following disturbances, especially for permanent faults.



4. Methodology


The algorithm identified in Fig. 2 uses the
binary bus connection matrix
,
B
, to identify
energized buses after a reconfiguration. Input data indentifies
the number of buses, line
connections and location of interrupting devices. The algorithm then identifies the entire subset
of breaker configurations that are available after the disturbance. Once a permanent fault has
been detected, breakers isolate the f
aulted components from all power sources. Intuitively,
increasing the number of interrupting devices in the circuit also increases the number of available
paths through which a source can flow upon reconfiguration. The number of possible states
becomes 2
N_
CB
, where
N_CB

is the number of breakers that may be either open or closed. When
breakers lock out the remaining states are given by 2
N_CB


n
, where
n,
is the number of breakers
locked out.


The design algorithm, illustrated in Fig. 3, uses e
xpectation of

unserved energy,
U
, to
compare the performance of different combinations of circuit breaker locations as identified in
Fig. 3. Assuming that the tie switches in Fig. 1 can be automatically controlled,
the switches at
these locations are represented as br
eakers. A
t the design phase, a circuit breaker can be added to
the system at one of sixteen locations. To assess the performance of the system, a pseudorandom
disturbance is analyzed. A fault is simulated at a pseudorandomly selected location, thus
consti
tuting a
sample
.
A sufficiently large sample size is required to obtain an accurate estimate
for the expectation of minimum unserved energy. A sample size of 5000 randomly located faults
was selected as an example shown here.
The system reconfigures to iso
late the faulted line or
line section and the expectation of unserved energy is recorded. The breaker location is changed
6


to another of the remaining 15 available lines; the location corresponding to the minimum
U

corresponds to the
optimal
breaker locatio
n.


In order to assess the effectiveness of the addition of circuit breakers, additional breakers are
added repeating the procedure outlined above. This procedure is repeated until four breakers
have been optimally located in the circuit
, along with the
four original ‘breakers’ at the switch
locations
. One must note, however, that this approach provides a suboptimal result.
Combinatorial logic states that the number of possible states increases as
n!/(n
-
m)!m!

= 16!/
(12)!4! = 1820 possible states. Because

not every one of the 1820 states is analyzed, the solution
obtained is suboptimal.



NO

NO


CALCULATE AVG.
EXPECTATION OF ENERGY
UNSERVED

(
U
)

SAVE MIN.
U


CHANGE



BREAKER



LOCATION


?

RANDOMLY LOCATE
5000
FAULTS

ON THE DISTRIBUTION
SYSTEM

PLACE CIRCUIT BREAKER
IN LINE

Y
ES

ADD ANOTHER CB

?

END

YES



Fig. 3 Flow chart of the design algorithm


7


5.
Assumptions


Distribution systems possess some inherent degree of unbalance. Loads change magnitude
frequently within a

24 hour period.
Certain assumptions are needed to build and quantitatively
evaluate the performance of both the design and operation algorithms.

These assumptions are:




Constant loads
are

assumed to obtain
comparable values

for the

expectation of unser
ved
energy and maximum load restored on the system.

Investigating the r
obustness of
the
reconfiguration and restoration algorithm

under varying load conditions is
scheduled for
future work.




Loads are modeled as constant three
-
phase values

lumped a
t
a
bus
. Attempting to study
the effect of reconfigurations on load flow results is left for
future work.




The initial model neglects the effect any distributed generation may have on the
distribution system. Assuming that DG
is
seen as negative load, DGs may be
present in
this system as it is presently modeled. Handling the disconnection of DGs from the point
of common coupling (PCC) when a fault occurs may be necessary for
future work.




System expansion and load growth must also be addressed. The expectation of

unserved
energy, placement of interrupting devices and the effect of system reconfiguration on
system average interruption and frequency indices all depend heavily on system
expansion and load growth.


6. Results


Optimal breaker locations as identified
by the algorithm and corresponding
U

are identified
in Table 2. Note that
U

is the minimum expectation of unserved energy when an additional
breaker is placed in the system.
A total of 58 breaker locations out of a possible 1820 were
investigated.

Heurist
ically, in this specific example, it is believed that the suboptimal result is of
sufficient accuracy for the problem at hand


especially considering that loads vary in real time,
and the method described assumes
stationary
load
s
.


The algorithm was abl
e to reconfigure to restore a subset of load buses for every attempted
fault simulation. Table 3 shows an example of the results obtained from a reconfiguration and
restoration simulation.


Table 2


Optimal Breaker Locations and
U


From
Bus

To
Bus

U
(MWh
)

Variance

(MVAh)
2

3

4

2.209

1.003

15

18

1.660

1.500

4

9

1.300

0.753

2

3

1.114

0.902


8



Fig. 4 Effectiveness of locating 0, 1, 2, 3, or 4 circuit breakers in the test bed of Fig. 1


Table 3 Results of the reconfiguration


restoration algorithm


Faul
ted line

No. of
possible
breaker config
s.
after fault removal

Load restored
MVA

Expectation of
energy unserved
MWh

From bus

To bus

3

4

32

7.00 + j3.40

2.00

2

3

128

9.00 + j4.90

0

17

18

64

8.00 + j3.90

1.00

13

15

32

8.00 + j4.40

1.00

2

3

128

9.00 +

j4.90

0

9

14

32

8.50 + j4.40

0.50

7

12

64

7.50 + j3.40

1.50

14

16

32

8.50 + j4.40

0.50

13

15

32

8.00 + j4.40

1.00

1

2

128

8.50 + j4.40

0.50



7.
Conclusions



The reconfiguration and restoration algorithm as ident
ified in this report

provides a

model of
an automatically reconfigurable ‘smart’ distribution system.
T
he
algorithm and distribution
system will be used to investigate
impacts of

a smart distribution system
.

Issues outlined in the
assumptions section will be investigated in t
he near future (scheduled to be investigated before
Dec. 2009). A major goal for future work is to
turn this sample distribution system into

a realistic
example of a future ‘smart distribution system’
, along with

re
design

considerations
, operational
requir
ements

and

cost estimates of migrating to such infrastructure
.
An important issue is how to
9


quantify the value (e.g., in dollars) of unserved energy; how to quantify the value of a circuit
breaker (include installation? What life of the breaker? Is this
cost per hour? Per year?).