You are on page 1of 12

Heuristic Distribution Protection Placement

By Zachary Sanford
University of Washington, Branch Counselor: Professor Alexander Mamishev
IEEE member #80504869
1605 NE 47
th
St, Seattle, WA 98105
The author is grateful for guidance from Professor R.D. Christie and Sean Parker,
and for financial support from Dean Matt ODonnell










Heuristic Distribution Protection Placement


Table of Contents
Abstract
I. Introduction
II. Approach
A. Definition of Main Feeder
Figure 1: Main Feeder Example
B. Fuse Placement Rules
Figure 2: Fuse Placement Example
C. Recloser Placement Rules
Figure 3: Recloser Placement Example
III. Compared Results of the Code
Figure 4: Circuit Example
IV. Conclusion
V. Tables
VI. References








1
1
2
3
3
4
5
5
5
5
6
6
8
9


Abstract
Automating protection device placement on existing distribution circuits using a
programmed algorithm enables the improvement of reliability on a large scale. A nave
approach considers every possible placement, even though many placements are
infeasible or inappropriate. Creating rules for protection placement based on heuristic
considerations limits the placement algorithms results to the most feasible locations. The
use of heuristic rules avoids the calculation of fault currents by using approximate thumb
rules for device placement, reducing computational time for the algorithm by 40%. The
circuit data evaluated by the algorithm in this study is taken from actual feeders.
I. Introduction
The volume of work required to improve distribution circuit reliability exceeds
what manual analysis can achieve. Adding protection devices to existing distribution
circuits is a method of improving reliability that can be automated. [1] uses a genetic
algorithm to find the optimal placements, and if a placement violates a heuristic rule, the
algorithm penalizes the placements benefit. [2] considers heuristic constraints on
placement, focusing mostly on the placement of switches. [3] estimates the cost of a
placement to customers and utilities. The work in this paper is different from [1] in that it
rules out placements that violate a heuristic rule, as well as not using a genetic algorithm.
The work also differs from [2] by focusing on the placement of fuses and reclosers, and
from [3] by basing cost on customer interruptions. Using an algorithm to evaluate
inputted circuit data, the addition of a protection device can be recommended based on
the cost of placing the device divided by the benefit of added reliability. The benefit is a
1

measurement of reduced customer interruptions, accounting for permanent faults, but not
temporary faults.
A temporary fault occurs when a connection between phases and/or the ground
appears and then breaks down, such as when a tree branch or animal touches a line and
then burns and falls off the line, clearing the fault. A permanent fault occurs when a
connection between phases and/or the ground remains until it is manually removed, such
as when a line breaks or a tree leans on a line. Temporary faults are a greater concern
when they occur on the main part of a circuit than on a lateral, or a branch, as these cause
most of the circuit to lose power.
Without heuristic rules, the approach of evaluating protection placements using a
cost-to-benefit ratio yields many infeasible results. For example, the algorithm might
place a fuse on the main feeder near the substation because the fuse improves circuit
reliability for permanent faults; yet, in practice, this placement lowers overall circuit
reliability because the fuse wont reclose after a temporary fault, creating a sustained
interruption of power. A more appropriate device for this location is a recloser, a circuit
breaker that responds to a temporary fault by opening and then automatically reclosing to
restore power after the fault clears. The heuristic rules produce better results and reduce
computational time because they eliminate the need to calculate the incremental cost of
reliability for infeasible placements. The algorithm used in this study addresses fuse and
recloser placements. Switches are not considered.
II. Approach
This section presents the approach taken to implement the heuristic rules for
automated protection-device placement. The heuristic rules were integrated with a pre-
2

existing program that recommends device placements based on a cost-to-benefit ratio,
without heuristic rules.
A. Definition of Main Feeder:
The main feeder of a circuit is comprised of three-phase segments that carry or are
capable of carrying the bulk of the current in the circuit. In normal operation, these
segments stem from the substation and might branch to different main sections of the
circuit. In abnormal operation, additional branches might connect to neighboring circuits
through normally open switches, carrying the bulk of the current and so considered part
of the main feeder.

Figure 1: Main Feeder Example: Segment 1 is a three-phase segment starting at the substation, and so by
definition it is on the main feeder; segment 2 is not three-phase, and thus not on the main feeder; three-
phase segment 3 has 48 downstream customers and three-phase segment 4 has 20, therefore segment 3 is
on the main feeder since it has the greatest customer count.
In the heuristic code, the main feeder is defined as the three-phase segments,
starting at the substation, that have the greatest number of downstream customers
compared to their neighboring segments. Note that this definition does not account for
possible normal open switches that connect to another circuit and so could be considered
part of the main feeder in abnormal operation. Also, by this definition, the main feeder
3

does not branch; it is a single path of the circuit. Within these parameters, the program
identifies the segments on an approximate main feeder.
B. Fuse Placement Rules:
Fuses are not allowed to be placed on the main feeder because a recloser is assumed
to be a better choice of protection due to its automatic restoration after temporary
faults.
If a fuse placement causes the depth of fuses at any point on the feeder to exceed
three fuses, then the placement is not considered. This restriction is made for
coordination purposes, as each fuse must have a lower current rating than its upstream
fuse. A fuse is defined as upstream to another fuse if it is electrically closer to the
substation. Using this coordination, the closest fuse upstream of a fault opens while
the fuses farther upstream stay closed, thus minimizing the effect of the fault. The
depth is determined by counting upstream from the potential placement location to
the nearest recloser, and finding the maximum depth of any branches downstream of
the location. This rule assumes all service transformers are fused and does not add
their fuses to the count.
An underground feeder is not considered for fuse placements except at the beginning
of the first segment of the underground feeder after an overhead segment, because a
fuse can be placed on the overhead pole where the feeder goes underground.

4


Figure 2: Fuse Placement Example
C. Recloser Placement Rules:
If a segment is not on the main feeder, then a recloser placement is not considered.
This limits the reclosers to three-phase segments, under the assumption that placing a
recloser on a lateral is not cost effective.
An underground feeder is not considered for recloser placements except at the
beginning of the first segment of the underground feeder after an overhead segment,
because a recloser can be placed on the overhead pole where the feeder goes
underground.

Figure 3: Recloser Placement Example
III. Compared Results of the Code
This section compares the results of the placement algorithm with and without the
heuristic rules. The results are displayed in Table I and are evaluated based on run time
and number of suggested placements. Table II shows the percent decrease in run time and
5

number of placements obtained with the heuristic rules. Table III shows the test
conditions and size of the test data set.
As expected, the number of placements for fuses and reclosers decreased. Fuse
placements decreased by 39% and reclosers by 92%. The run time for the heuristic code
was 40% less than the code without heuristics despite the increased complexity in the
placement
rules.

Figure 4: Circuit Example: This figure is a representation of an actual circuit with its existing fuses and
recloser identified and each single placement proposed by the placement algorithm.
IV. Conclusion
The heuristic rules reduce computational time and limit device placement
suggestions to feasible locations. The dramatic decrease in recloser placements is due to
the restriction of such placements to the main feeder. This may be too limiting because it
does not allow for recloser placements on long laterals that might justify recloser cost.
6

These may be viable on an incremental cost of reliability basis, but determining whether
or not this is the case is beyond the scope of this study.
The improvement in run time is due to the heuristics reduction in the number of
placements that are considered. In turn, this reduces the number of cost-to-benefit ratio
calculations executed by the code. The refined approach of automated protection
placement using heuristic rules significantly increases the usability of the reliability
assessment results, reducing the need for a manual review and saving cost in improving
distribution reliability. Future work can be conducted to modify the algorithm to suggest
multiple placements, to modify the main feeder definition to account for switches and
allow for branching, and to determine if recloser placement on long laterals should be
considered by the algorithm.
7

V. Tables
Table I: Compared Results of the Code
Version No. of Fuse Placements No. of Recloser
Placements
Run Time
Without Heuristic Rules 97,137 43,250 95.8 minutes
With Heuristic Rules 59,470 3,476 57.7 minutes

Table II: Percent Decrease using Heuristic Rules
Percent Decrease in Fuse
Placements
Percent Decrease in
Recloser Placements
Percent Decrease in
Run Time
39% 92% 40%

Table III: Test Conditions and Data Size
Program Computer Circuit Count Segment Count
MatLab Windows PC 1,790 284,225

8

VI. References
[1] Falcao, D.M., Genetic algorithms applications in electrical distribution systems,
Proceedings of the 2002 Congress on Evolutionary Computation, CEC '02,
Volume 2, pp. 1063 1068, 12-17 May 2002.

[2] L.G.W. da Silva, R.A.F. Pereira, J .R.S. Mantovani, Optimized allocation of
sectionalizing switches and control and protection devices for reliability indices
improvement in distribution systems, Transmission and Distribution Conference and
Exposition: Latin America, 2004 IEEE/PES, pp. 51 56, 8-11 Nov. 2004.

[3] F. Soudi, K. Tomsovic, Optimized distribution protection using binary programming,
IEEE Transactions on Power Delivery, Volume 13, Issue 1, pp. 218 224, J an. 1998.
9

You might also like