{£177
r ,1. t 6

 

 

 

A k, .H . chain ,5? V .. .. . . . . (.7.
d... 95%;": «Hit.» Madrcrfifiu , . n3...

, .u ; , . .
. 1.3.2.3? L:
. 1 ‘. 3..

. .1u...F,.u;1:

.

 

 

' i rh‘ 1|‘xw I

 

THESIS

I////////////////////////////I//////I////I

I

II/ (1
3

'l

I‘.

This is to certify that the

thesis entitled

Ecology and Control of Wildlife Damage to
Electric Substations

presented by
Wendy Hope Sangster

has been accepted towards fulfillment
of the requirements for

Master of Science degree inFish. & Wildl.

 

 

1292\— K Dawemc

Major professor

Date July 21, 1995

MS U is an Affirmative Action/Equal Opportunity Institution

 

_‘.__ ———.__.__’——_ ._._

_____q.—.‘H..._—~—_-—

 

LIBRARY
M‘Chigan State
University

 

 

 

PLACE ll RETURN BOXto remove thte checkout from your record.
TO AVOtD FINES return on or More dete due.

DATE DUE DATE DUE DATE DUE

00L 1 0:7

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU to An Altimetive Action/Equel Opportunity lnetltuion

 

ECOLOGY AND CONTROL OF WILDLIFE DAMAGE TO ELECTRIC
SUBSTATIONS

By

Wendy Hope Sangster

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Fisheries and Wildlife

1995

ABSTRACT

ECOLOGY AND CONTROL OF WILDLIFE DAMAGE TO ELECTRIC
SUBSTATIONS

By

Wendy Hope Sangster

This study addresses several aspects of the ecology and control of
wildlife damage to electric substations because the amount of existing research
is not sufficient to make informed decisions about how best to minimize that
damage. Records of 121 incidents of animal-caused faults showed that 78% of
the faults were caused by squirrels and raccoons and an average of 2,511
customers lost service during the outage caused by such a fault. Animal
damage control measures were evaluated by observing challenges to control
measures by raccoons and squirrels at a substation. The control measures were
breached twice because they had not been properly applied.

In 1994, 301 transmission and distribution substations in Michigan were
sampled and categorized based on various structural and habitat characteristics.
Significant relationships (p < 0.10) were found between faulted substations and
the number of nests in the substation, the distance of water from the substation,

and the beam type used in the substation.

ACKNOWLEDGEMENTS

This project was made possible by funding from Consumers Power. I would like to
thank all those at Consumers Power who contributed to this research, particularly Gary
Dawson and James Groulx.

I would like to thank my major advisor Dr. Glenn Dudderar for his guidance and
his tremendous knowledge. Thank you to Dr. Scott Winterstein and Dr. Fred Dyer for
serving on my committee and for providing me with help when I needed it and with
crucial insights. Thank you to Peter F ritzell for helping me trap raccoons and squirrels,
sometimes very early in the morning.

To my parents Lanny and Wayne Sangster, my sisters Julie and Karin, and Pete,

thank you for all of your support throughout my life.

iii

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... v
LIST OF FIGURES ........................................................................................................ vi
INTRODUCTION ............................................................................................................ 1
OBJECTIVES ................................................................................................................... 6
STUDY AREA ................................................................................................................. 7
METHODS ..................................................................................................................... 10
Examination of Records of Animal-caused faults ............................................ 10
Evaluation of Control measures ........................................................................ 10
Characterization of Substations ......................................................................... 12

Data Analysis ..................................................................................................... 17
RESULTS ...................................................................................................................... 19
Examination of Records of Animal-caused Faults ............................................ 19
Evaluation of Control Measures ........................................................................ 24
Characterization of Substations ......................................................................... 28

Tests of Association ................................................................................. 28

Spatial Arrangement of Vegetative Habitat ............................................ 32

Distance of Water .................................................................................... 37

Number of Bird Nests ............................................................................. 39

Profile ...................................................................................................... 41

Substation Structure ................................................................................ 43

Logistic Regression Model ............................................................................... 45
DISCUSSION ................................................................................................................ 48
Examination of Records of Animal-caused Faults ........................................... 48
Evaluation of Control Measures ....................................................................... 51
Characterization of Substations ........................................................................ 53
Recommendations ............................................................................................. 60
CONCLUSION .............................................................................................................. 62
LIST OF REFERENCES ............................................................................................... 63

iv

LIST OF TABLES

Table 1. Summary of findings from observations of challenges to control
measures by raccoons ................................................................................................... 25

Table 2. Summary of findings from observations of challenges to control
measures by squirrels ................................................................................................... 26

Table 3. Results of chi-square tests and Fisher's exact test for associations
between animal-caused faults and field variables ....................................................... 29

Table 4. Results of chi-square tests and of Fisher's exact test for associations
between animals causing faults and field variables .................................................... 30

Table 5. Results of chi-square tests and of Fisher's exact test for associations
between substations experiencing faults by different types of animals and those
not experiencing faults across field variable levels ..................................................... 31

Table 6. Regression coefficients (maximum likelihood estimates) and related
statistics from logistic regression analysis ................................................................... 46

Table 7. Matrix of predictions of damage of 50 substations by logistic regression
models using all variables and significant variables only (both models have the
same matrix) .................................................................................................................. 47

LIST OF FIGURES

Figure 1. Map of study area ..................................................................................................... 8
Figure 2. Depiction of habitat sampling design ..................................................................... 15

Figure 3. Percentage of reports of animal-caused faults for types of animals
responsible for faults ................................................................................................................ 20

Figure 4. Number of reports of animal-caused faults by month for faults
caused by squirrels (A), raccoons (B), and birds (C) ............................................................. 21

Figure 5. Number of reports of animal-caused faults by time of day for
squirrels (A), raccoons (B), and birds (C) ............................................................................... 22

Figure 6. Percentage of reports of animal-caused faults for each type of
equipment damaged .................................................................................................................. 23

Figure 7. Percentages of treated substations with various flaws in control
measure ..................................................................................................................................... 27

Figure 8. Frequencies of substations experiencing animal-caused faults
and of those not experiencing faults in the different categories of
spatial arrangement of vegetative habitat ................................................................................. 33

Figure 9. Frequencies of substations experiencing animal-caused faults and
of those not experiencing faults in modified categories of spatial arrangement
of habitat ................................................................................................................................... 34

Figure 10. Frequencies of substations experiencing animal-caused faults and
of those not experiencing faults in locations with and without vegetative
habitat ......................................................................................................................................... 35

Figure 11. Number of substations experiencing squirrel-caused faults and of

those not experiencing faults in modified categories of spatial arrangement
of vegetative habitat .................................................................................................................. 36

vi

vii

Figure 12. Number of substations experiencing squirrel-caused faults and of
those not experiencing faults in locations with and without vegetative habitat .................... 36

Figure 13. Number of substations experiencing animal-caused faults and of
those not experiencing faults by distance of substation from water ...................................... 38

Figure 14. Number of substations experiencing raccoon-caused faults and of
those not experiencing faults by distance of substation from water ...................................... 38

Figure 15. Number of substations experiencing animal-caused faults (A),
squirrel-caused faults (B), and bird-caused faults (C) and of those not
experiencing faults by the number of nests in a substation .................................................... 40

Figure 16. Number of substations experiencing animal-caused faults and of
those not experiencing faults by profile of substation ............................................................ 42

Figure 17. Number of substations experiencing animal-caused faults (A) and
squirrel-caused faults (B) and those not experiencing faults by beam type and
size of sub station ...................................................................................................................... 44

IN TROD UCTION

Although urbanization has adversely affected many wildlife species, other
species have not been so affected. The apparent resiliency of some species
could be the result of the tolerance of the species to human presence and
disturbance, an increase in available suitable habitat because of urbanization, or
a combination of both factors. Studies of bird populations have shown that
many bird species, such as blue jays (Cyanocitta cristata), American robins
(T urdus migratorius), and European starlings (S tumus vulgaris), are not only
able to survive in urban and suburban environments, but in fact thrive in these
areas (Williamson 1973, Beissinger & Osborne 1982, Horn 1985). The North
American raccoon (Procyon Iotor), the eastern grey squirrel (Sciurus
carolinensus), the fox squirrel (S. niger), and the red squirrel (Tamiasciurus
hudsonicus) have all shown similar responses to urbanization (Cauley and
Schinner 1973, Hathaway 1973, Schneider 1973, Cauley 1974, Williamson
1983). Regardless of whether the increased availability of suitable habitat or
the tolerance of the animal to human presence accounts for the presence of
these species in urban areas, their presence has a profound impact on the
surroundings and people with which they live.

Animals often use man-made structures for den or nesting sites, foraging
sites, or as travel routes, and these activities may cause damage to the structures
(Cauley and Schinner 1973). Wildlife intrusions into electric power substations
and the subsequent damage to those substations is a problem that has recently

received more attention by the electric utility industry. Wildlife damage to

2
substations comes in the form of outages, direct equipment damage, and safety
and health hazards to maintenance personnel (Substation Security Working
Group 1993). Wildlife damage is among the major causes of momentary
outages to substations (Warren 1992).

Animals cause damage to the substation in various ways. Generally, the
animal simultaneously touches two electrified components or an electrified
component and a grounded component, the equipment short circuits, and a fault
occurs that results in a power outage. Such faults might cause an explosion or
fire, leaving little to indicate what caused the damage (Mitchell 1977).

Squirrels and raccoons also chew on equipment, and when moisture enters, a
fault may occur. Birds nesting in a substation may cause damage to the
substation in several ways. Bird droppings are corrosive and may cause
equipment damage or may accumulate on insulators, causing flashovers (Paula
1989). Additionally nesting materials could cause faults by falling onto parts of
the substation and creating a short. A final potential problem with birds nesting
in substations is that foraging raccoons and squirrels may enter the substation to
get to eggs and nestlings as these are common elements in the diet of raccoons
and squirrels (Schneider 1973, Greenwood 1981).

Wildlife-caused faults which result in power outages are perhaps the
most costly of the types of damage. Equipment repair, revenue lost while
service is down, and the indirect costs of reduced consumer confidence are
some of the expenses associated with power outages (Paula 1990). In a review
of impacts of wildlife on telephone and electrical services in Waterloo, Ontario,
Mitchell (1977) states that none of the utility companies that he reviewed kept
SYsternatic records of wildlife damage and could not attribute costs to the

damage but assumed it to be low. However, more recently, Enck (1989) found

3
that the mean cost of an animal-caused fault in New York for 1987 was
$12,500, and Paula (1990) indicates that some animal-caused outages may cost
as much as $500,000. From figures such as these, it is evident that animals
have the potential to do costly damage to substations. Yet, very little research
has been conducted to determine the importance of various aspects of that
damage.

One aspect of the damage involves determining which types of animals
cause damage to substations and in what proportions. In a review of records of
animal-caused faults from six utility companies in New York state Enck and
Brown (1989) found that of 200 animal-caused faults, 55% were caused by
squirrels, 12% by raccoons, and 16% by birds. Rochester Gas and Electric
Corporation in Rochester, NY estimated that over a period of four years, 90%
of animal-caused faults were caused by raccoons and squirrels (Fiske 1992).

Electric utility companies have used a variety of techniques in an attempt
to reduce wildlife damage to substations. Among the techniques are chemical
repellents, fence barriers, lights, artificial predators (owls, hawks, snakes,etc.),
anti-climbing devices, lineguards, electrical fences, bushing guards, and other
structural barriers (Fiske 1990, Substation Security Working Group 1993).
Qualitative assessments of the effectiveness of such control measures were
provided by a guide compiled by the Institute of Electrical and Electronics
Engineers in 1993. The guide was very thorough in its description of control
measures but did not give any detailed analysis of the various possible
treatments.

A potentially significant aspect of wildlife damage to substations is the
structure of the substation itself. Enck (1989) found that the substations that

were most susceptible to wildlife damage were those that had been operating for

4
at least 30 years and those with high physical profiles (latticework and
structural components more than 8m above the substation equipment). Other
possible characteristics of substation structure that may contribute to wildlife
damage include the type of beams and the amount of latticework used in
construction. Two beam types commonly used in substations are L-shaped
beams (angles) and S-shaped beams (I-beams). It is likely that the beam types
are not equally climbable by raccoons and squirrels. Specifically, the L-type
beams are associated with a lattice system, and therefore may be more attractive
to raccoons and squirrels for climbing than other beam types.

The terrestrial habitat surrounding the substation may be an additional
factor that influences the cause and amount of damage by wildlife. Enck
(1990) found no correlation between surrounding habitat and the number of
faults at a substation. However, the terrestrial habitat variables considered in
the study and the way in which they were analyzed may not have fully
described the relationship between habitat and wildlife damage. There may be
habitat variables that are common to the habitats of all species involved with
wildlife damage, and the identification of such variables might help in
determining whether broad characteristics of the habitat around the substations
influences the amount of wildlife damage to that substation.

Important habitat characteristics for fox and grey squirrels include tree
density, basal area, species composition, and shrub crown cover, although the
requirements are fairly non-specific (Cauley 1974 , Williamson 1983, Steele and
Weigl 1992). Factors such as building cover and the presence of pavement
have been found to have a negative influence on squirrel activity (Williamson
1983). Red squirrel habitat requirements are likewise non-specific, but the most

important variables include tree density and species composition (Layne 1954,

5
Baker 1983). Raccoons appear to be most limited by the presence of water,
good travel routes, and woods for shelter and food (Cauley and Schinner 1973,
Schneider 1973, Cauley 1974). Because the species of birds involved in
wildlife damage to substations have not been identified, it is difficult to
establish what habitat components might influence damage by birds. However,
almost any environment will provide habitat for some species of bird
(Williamson 1973, Horn 1985). Because of the broad nature of the habitat
requirements among the animals involved with wildlife damage to substations, it
is possible that there are habitat parameters that are common to all the species
and that are also correlated with wildlife damage. However, this possibility has
not been thoroughly explored.

It is evident that the amount of existing research dealing with wildlife
damage to electric substations is not sufficient to make informed decisions
about how best to minimize that damage. The research presented in this study
was requested and funded by Consumers Power because the company was
interested in reducing losses due to wildlife damage. Therefore, this research
addresses questions about how and why animals damage substations and about
the effectiveness of animal damage control measures. Investigation of these
questions should lead to better informed decisions about how to control animal

damage and consequently to less damage.

OBJECTIVES

The objectives of this study were to determine characteristics of
substations and the surrounding environment that are associated with animal
damage and to examine the effectiveness of preventative measures. These
objectives were accomplished by completing three different investigations.
These investigations were -

1) the examination of reports of animal-caused faults provided by

Consumers Power.

2) the observation of how effectively the current animal damage
control measures used by Consumers Power keep animals out of a
substation.

3) the characterization of electric substations based on the
relationship between animal-caused faults and structural and

habitat characteristics.

STUDY AREA

The study area encompassed most of the counties in the Lower Peninsula
of Michigan (Fig. 1). The specific sites studied were distribution and
transmission substations located throughout these counties. Those counties not
included were Berrien, Cass, Huron, Lapeer, Macomb, Sanilac, St. Clair,
Tuscola, Wayne, and Emmet counties.

The Lower Peninsula of Michigan makes up approximately 70% of the
total land area of Michigan and is bounded on the west by Lake Michigan and
on the east by Lakes Huron, St. Clair, and Eric. The climate of Michigan is a
combination of semi-marine and continental and is altered by lake effects that
influence temperature, moisture, and wind direction and velocity. The annual
average temperature in the southern half of Michigan (Region I) is
approximately 9°C and ranges from about -22 to 19°C. In the northern half of
Michigan (Region 11) average annual temperature is roughly 7°C with a range of
-29 to 178°C. Total annual precipitation for the Lower Peninsula is around
800mm (Albert 1986, Eichenlaub 1990).

Elevations in the southern half of Lower Michigan range from 580-
1280ft but can get as high as 1725ft in northern lower Michigan. The southern
region has areas of clay lake plain, ground moraine, end moraine, and outwash
plains, and soil textures are mostly loams to clays with sand soils in certain
areas on the lake plain. In the northern region physiographic features include

outwash plains, end-moraine ridges, ridges of ice contact material, low elevation

 

study area

      

 
  
  
   

, F
..
.._.,. .
$33-71.

 

Figure 1. Map ofstudy area

9
lake plain, ground moraine, and outwash. Sandy soils predominate in this
region (Albert 1986).

The ecosystems in the southern half of Michigan include beech-sugar
maple forests, oak-hickory forests, prairies, hardwood swamps, and hardwood-
tamarack swamps. In the northern half, ecosystems include northern hardwood
forests, oak-pine forests, pine forests, and swamp and bog communities (Albert
1986)

Among the major agricultural products of Michigan are milk, corn,
soybeans, hay, fruit, wheat, and vegetables. The average growing season length

in the southern region is about 125-150 days (Eichenlaub 1990).

METHODS

Examination of Records of Animal-caused Faults

The initial portion of this study involved the examination of data
provided by Consumers Power for a period from January, 1988 to September,
1994. The data included information about wildlife damage to substations and
were analyzed to establish: 1) which substations had experienced animal-caused
faults, 2) what animals caused damage and in what proportions, 3) how the time
of day and the time of year of faults were related to the animal causing the
fault, 4) what substation equipment was most frequently damaged, 6) how many
customers were effected by faults, and 7) how long faults lasted. The results of
this analysis provided useful information both for a general description of the
problem and for use in the characterization of the substations.
Evaluation of Control Measures

Determining the effectiveness of control measures currently being used
by Consumers Power involved the observation of squirrels and raccoons around
a substation in Marshall, MI that is used by the company as a training facility.
In September and October of 1994 animals were trapped and housed in large
outdoor pens at the Dobbie Road Wildlife Research Area at Michigan State
University. Animals were trapped in wire box traps. A commercially
distributed raccoon bait that consisted of a fish base and anise oil was used to
bait raccoons. Squirrels were baited with sunflower seeds, corn, and pecans.

Traps were checked at least once a day.

10

11

In September of 1994, the substation training facility was prepared for
observation of the animals and animal proofed by Consumers Power personnel.
A 4-ft high polycarbonate containment fence was erected on the outer side of
two sides of the substation about 5ft from the substation fence. The gate into
the substation was included in the portion of the facility fenced in. The
polycarbonate containment fence had opaque brown paper affixed to one side
making it impossible to see through. The two sides of the substation fence
within the containment fence were animal proofed. Animal proofing measures
included the installation of a l-inch mesh fence around the substation and the
application of 36-inch wide polycarbonate sheeting around the top of the 1-
inch mesh fence. Aluminum ties attaching the polycarbonate sheeting to the
fence were oriented vertically. The fence and the gate surrounding the
substation were to have no openings larger than one inch wide. Gaps at the
gate were minimized by applying polycarbonate around the edges of the gate
and by sinking the poles of the gate into a concrete base under the gate to
maintain proper gate alignment and spacing. The concrete base also prevents
the formation under the gate of gaps caused by erosion, use, and animal
digging.

On October 6 and 7, 1994 six raccoons were placed within the
containment fence and observed and videotaped to determine whether the
control measures used by Consumers Power were penetrable. On October 10
and 20, 1994 four melanistic grey squirrels, one fox squirrel, and one red
squirrel were likewise placed within the containment fence and then observed
and videotaped. Observations of general behavior and if and how an animal
gained access to the substation were made. Each animal was observed

separately, and observations were be made for approximately 2 hours for each

12
animal or until it escaped. The observations were then tabulated to provide
information about the effectiveness of the control measures.
Characterization of Substations

A total of 290 distribution and 11 transmission substations owned by
Consumers Power were evaluated and classified from May, 1994 to September,
1994. The 88 substations for which Consumers Power had reported wildlife-
caused faults in the last six years were included in the 301 substations; five of
the damaged substations were transmission substations and 83 were distribution
substations. The remainder of the 301 substations were randomly selected from
among the 1,089 substations that had not been damaged by wildlife in the last
six years. It was assumed that this sample was representative of the population
of substations owned by Consumers Power because the random selection
resulted in an interspersed sample and because it made up over a quarter of all
of the substations in the population.

One part of the classification of each of the substations included
characterizing the structure of the substation. The structural characteristics
recorded were beam construction, physical profile, and degree and type of
animal proofing. The beam construction for each substation was recorded in
terms of beam type, lattice system involved, and the area covered by the
structure. Beam types included S-shaped beams (I-beams), L-shaped beams
(angles), and wooden poles. To determine whether the amount of latticework
associated with L-shaped beams influences damage to substations, substations
built with L-shaped beams were classified in two categories: those occupying an
area less than 100m2 and those occupying more than 100m2. Therefore, there
Were five categories of substation structure: substations covering greater than

about 100m2 built with L-shaped beams, substations covering less than 100m2

13
built with L-shaped beams, substations built with S-shaped beams, substations
built with a combination of L and S-shaped beams, and substations built with
wooden poles.

Physical profile was categorized as either high or low. High physical
profile was defined as having substation components higher than 7m from the
ground, and low physical profile was defined as having all components below
7m. Seven meters was chosen for distinguishing between high and low profile
because the high end of low profile substations owned by Consumers Power are
typically about 6m tall.

The degree of animal proofing was established by placing the substations
into one of three categories. These categories were substations without animal
proofing, substations with partial animal proofing, and those with complete
animal proofing. Complete animal proofing included: 1) 36 inch wide
polycarbonate or aluminum sheeting continuous around the top of the substation
fence with plastic or aluminum ties affixed vertically, 2) 36 inch wide
polycarbonate or aluminum on all poles outside of the substation within 25m of
the substation and/or lineguards placed on all wires going into the substation, 3)
no openings of more than l-inch wide in the fence and gate surrounding the
substation or at the gate of the substation, and 4) 1-inch mesh fence around the
substation. These are the animal proofing measures that Consumers Power
currently uses on its substations. All flaws in animal proofing were noted for
each substation.

An additional variable that was considered was the number of bird nests
in each substation. The number of nests seen, the location of each nest, and

when possible the species of bird associated with the nest were recorded.

14

The substations were also categorized based on characterizations of the
surrounding terrestrial habitat. For each substation visited, terrestrial habitat
sampling was conducted in a rectangular plot that included the area within
100m from each side of the substation. Each plot was then divided into four
sections based on the four sides of the substation. Each section was stratified
by distance from the substation. These stratifications (segments) were: less than
25m from the substation, greater than 25m but less than 50m from the
substation, and greater than 50m but less than 100m from the substation
(Fig. 2).

First, the vegetative habitat was analyzed within each section and within
each distance segment. Species composition and density of woody vegetation
within each segment were determined by counting and recording the number
and species of trees with a dbh of at least 20cm and by recording the percent
crown cover of shrubs less than 5m tall in each segment. Second, man-made
structures within each distance segment were evaluated in terms of approximate
size, number, and type. Types of man-made structures included fences, cables,
houses, and buildings.

Third, the distance of the substation from water and the type of water
system were noted. The type of water was recorded as lacustrine or riverine.
Categories for the distance of water from a substation were water within 50m or
water greater than 50m but less than 150m from the substation.

Finally, the substations were classified based on the spatial arrangement
of the vegetative habitat characteristics. Habitat for this classification was
defined as areas having at least 120 trees/ha (3 trees/250m2) with a dbh of at
least 20cm or areas having at least 50% crown cover of woody vegetation

shorter than 5m tall. This definition was considered to be broad enough to

15

 

 

 

 

Figure 2. Depiction of habitat sampling design.

 

16
provide an indication of adequate habitat for all of the species involved. The
spatial arrangements were defined by the following categories: 0 = no habitat
(as defined above) within 50m of the substation on any side; 1 = habitat on one
side of the substation within 50m of the substation; 2, = habitat in two adjacent
sides of the substation within 50m of the substation; 2,, = habitat on two
opposing sides of the substation within 50m; 3 = habitat on three sides of the
substation within 50m; 4 = habitat on all four sides of the substation within
50m. It was hypothesized that categories 2,, 2,, and 3 would experience more
damage that categories 1 and 4 and that category 0 would experience that least
amount of damage. Therefore, substations were also classified in each of these

three modified categories (0, 14, and 2,2,3).

17

Data Analysis

Data from Consumers Power were analyzed by creating frequency tables
of the different variables in the data. This analysis was done to provide a
general characterization of animal damage to substations. Specifically, the
analysis involved evaluating numbers and types of animals causing faults, times
of faults (month and time of day), the types of substation equipment damaged,
the average length of faults, and the average number of customers affected by
the fault. In the analysis of data obtained from field visits to substations, six
distribution substations were not included in any statistical analysis because
they were newly built substations and could not have been susceptible to
damage by animals during the six years of faults reported by Consumers Power.

Data from the field observations was first analyzed using one-way
frequency tables to establish which categorizations would be used in further
analysis. Variables in each category with frequencies under 10 were either
discarded or combined with another category. Two-way frequency tables were
generated for damage versus each classification variables and for variables
where association, interaction, or dependence were hypothesized. Chi-square
analysis was performed to establish associations for all tables that had a sample
size large enough that the test was valid. The Pearson chi-square test was used

for this analysis (Altman, D.G. 1991):

X2=E£ (OU-EvY/E where

i=Ij=I '1 '

O = observed frequencies
E = expected frequencies
r = total number of rows in frequency table

c = total number of columns in frequency table

18
i = row number
j = column number
For the tables with sample sizes too small to get valid chi-square tests, F isher's
exact test was used. For this test, the hypothesis of homogeneity is rejected

when (Freeman, D. H. 1987):

2 min {P[Y _<. y (observed)], P[Y .>_ y (observed)]} 5 ac , where

Y follows a hypergeometric distribution

An association was considered significant for tests with p<0.10. This
significance level was used because precision for the field analysis was assumed
to be low, as is often the case with field studies. Additionally, the study was
largely exploratory, and therefore this significance level was justifiable.

Logistic regression was used to quantify the degree to which
combinations of the habitat around substations and the structure of substations
influence whether or not the substation is faulted by wildlife. A logistic
regression model was developed using animal-caused faults to the substation as
the response variable and each of the characteristics involved with the
classification of the substations as predictor variables. Data for all but 50 of
the categorized substations were used to create the regression model. The 50
substations that had been classified but not used to create the model were then
used to determine how well the model classified the substations. The Hosmer
and Lemeshow goodness-of-fit test for binary response model, which utilizes the
Pearson chi-square analysis, was used to determine the fit of the model. This
process established the predictive ability of the logistic regression model. SAS
System software was used for all of the statistical analysis (SAS Institute Inc.

1988)

RESULTS

Examination of Records of Animal-caused Faults

Records of animal-caused faults to electric substations between 1988 and
1993 showed that there were 121 incidents of animal-caused faults in 88 of
1,177 transmission and distribution substations. Squirrels accounted for 57%,
raccoons for 21%, and birds for 7% of all animal-caused faults (Fig. 3). Forty-
two percent of faults caused by squirrels occurred in June, September, and
October, with the largest peak occurring in June. Sixty-five percent of faults
caused by raccoons occurred in April, May, and August. The number of reports
of faults caused by different animals by the month of damage is presented in
Figure 4. Faults were most common between the hours of 6:00AM and noon
for squirrels and between 9:00PM and 6:00AM for raccoons (Fig. 5). Bird-
caused faults occurred in April through September and in December.

Fuses were the most commonly damaged substation equipment; fuses
were damaged and replaced in 54% of faulted substations. Other damaged
equipment included insulators, arresters, bushings, regulators, transformers, and
switches and switchgear (Fig. 6). Outages to customers occurred in 118 of the
121 animal-caused faults. The amount of time that service was off ranged from
15 minutes to 14 hours and 14 minutes, and the average amount of time that
service was off was one hour and 46 minutes. The average number of
customers that lost service during an outage was 2,511 with a range of 1 to

3,979 people affected per outage.

l9

20

 

 

 

 

Figure 3. Percentage of reports of animal-caused faults for types of
animals responsible for faults.

 

21

A. squirrels

 

— r-e
° N
M_ r

 

 

Nmnberofrqiortsoffauttscaused
by . 1
O\ N

Nov
Dec

..r.r. y .
it! "At ,.
in‘ “.1 ...
now» ..>
.... . .

a , I, . .... ...a..
«#n‘Il-Ia‘xlllfl‘in“
”irritant -

x.'h\c‘~¢ . ‘1‘,

Math of fault

 

 

 

B. raccoons

 

o-‘NU’*UQ\Q
1111111

4

 

Numbcofrqiortsoffaultsceused
byracooons

 

Nov
Dec

3

. .,.....»11.‘;..AJ...¢ -. -
”at 1.1. .u-r ‘rx4vt-91Ia. -.
m... a v‘-.n.-gaa;4a~(n alleywsg.
ltdh‘r ‘I'N‘I‘Illlfi‘q‘lii urn-a:
\II
,yr
1..-

 

 

 

C. birds

 

  

 

 

 

Numberofrqrortsoffauhsceuscdby
birds

1 L 1 l
T T T

stégé‘isé’ts

Mmth of fault

Nov

 

 

 

Figure 4. Number of reports of animal-caused faults by month for faults
caused by squirrels (A), raccoons (B), and birds (C).

22
A. squirrels

 

30
25

 

Number of reports of squirrel-
causcd faults
3

0 -
iiiéééé
Timeofday

 

B. raccoons

 

9

 

ReportsofRaecoon-causedfaults
O -- N w # Ur O\ \i

 

 

 

 

 

a. 3% .
° 3 ‘” i i i. i i
i 8 i r g g g E
Timeofday
C.birds
g 3
352.5
E 2
“581.5
$.31
“a 0.5
g 0 3 0t :1 § 0 B O
" E _ E § 5
z i i i i g i g 2
Timeofday

 

 

 

Figure 5. Number of reports of animal-caused faults by time of day for
squirrels (A), raccoons (B), and birds (C).

 

23

 

Percentage of reports of animal-caused faults

 

5

Type of equipment damaged

bushings
regulators
swrtches/
switchgear
other

tranformers

 

Figure 6. Percentage of reports of animal-caused faults for each type
of equipment damaged.

 

24
Evaluation of Control Measures

Of the six raccoons and six squirrels observed at the training facility, two
raccoons, a red squirrel, and one melanistic grey squirrel escaped from the
observation area. However, only one raccoon and the black squirrel escaped
through control measures on the substation. The raccoon escaped by pulling the
polycarbonate barrier away from the fence where a tie had not been affixed.
The black squirrel got into the substation by way of a ls/s-inch gap between the
fence and the gate. Tables 1 and 2 summarize the observations made for each
animal.

From the evaluation of 43 animal-proofed substations it was found that
common flaws in the application of the control measures included 1) wraps not
applied to all poles within 25m of the substation and/or lineguards not applied
to all lines going into the substation (81%), 2) ties applied horizontally rather
than vertically (81%), 3) polycarbonate sheeting not extended to cover gaps
between the substation fence and gate (79%), and 4) gaps larger than 1-inch
wide at the gate to the substation (74%). These results are shown in Figure 7.
Flaws in animal proofing that were likely to have been the result of insufficient

maintenance occurred in 19% of the treated substations.

cocoa Began—coo 05 mo ta :5 £5833 no «.2388 33:8 2: mo :3 8: ..

 

25

 

 

 

 

 

 

 

i- i- no» we» 8% me» 8858 3 038 3892 0 c0088

.1 i- on we» we» 8» 8358 ov 038 BEWNN m cocoon.

guesses. SEES 2 o: on 8» 8» $358 2 2S: memmdm v 5088

.358 a 9.5383 $358 2 8.» c: mo» 8» 3558 2 038 3:53 m c0088

.1 i. o: 0: mo» no» 8858 3 255m 350.2 N .8688

i. l- o: 0: mo» 8» 8358 mm 29: 335.2 ~ .8088
codename 26: 2.88 8 2:: 88.“ 650 Sam 85m and: coco.“ mo Eaton

HerUmm UOh. mmvamgo mgr—baa g XMm Eva; 43572

 

 

 

 

 

.2888 B 85308 35:3 8 women—~20 mo 3233030 Bob 355 we gm ._ 033.

 

26

85m ”SEE—Eco 2: .«o 5a :5 communes no 85308 .868 2: mo 5a 8:...

 

 

 

 

 

 

 

 

 

 

.88. 8.8% use 32.3... 8558 0 mom cc mean new 835:. c 3.533 v0.—
1. i- o: o: mo» 8» 8558 on .0533 x8
.1 i. we» on no» we» SEES 3 v 3.533 x83
.1 i- o: no.» no» mo» 8558 mh m 35:3 :83
-i .1 me» o: mo» no.» $338 9. N 3:33 x83
an as 88c 5022. as are; 3358 me no.» on me» mo.» SEES no _ .053 x83
3988 Be: gnome 9 2:: cocoa douse 8a venom comm: com mo 80:3
Umadam. HCa. mMOZmAEU m>Eb< g 123:7?

 

205:3 .3 $532: _ob:co 8 women—:30 mo Eountomno Eat 352m mo gem .N 033.

 

27

 

90-

80‘

70.1

Percent of treated substations

    

       

 

nol'negnflh h hemplde >l”ypn Mingnd ween infidel:
onl'nee Ming: W minnow providedby minimum:
Ind/or W ammdfmoe veglm'm

meet-13m

criminal
flee

Flew in control measures

 

 

 

Figure 7. Percentages of treated substations with various flaws in control
measures.

28
Characterization of Substations
Tests of Association

Table 3 presents the results of chi-square and Fisher's exact tests for
homogeneity among substations classified on the basis of whether or not they
had experienced animal-caused faults and on the basis of the field variables
(vegetative habitat, water, nests, profile, beam structure). The distance of water
from the substation, the number of nests in a substation, and substation structure
were all significantly associated with faults (p<0.10). All of the other variables
considered were not significantly associated with animal-caused faults. Physical
profile of a substation was found to be significantly dependent on beam type
(Fisher's exact test, p<0.001).

Results of F isher's exact test for associations between substations
experiencing faults caused by different types of animals and the field variables
are shown in Table 4. The presence or absence of habitat within 50m of the
substation was not uniform across the different animals causing faults. No
other associations were significant (p<0.10).

Table 5 summarizes the results of chi-square and Fisher's exact tests of
association for the field variables across substations experiencing faults caused
by different animals and those not experiencing faults. Significant associations
were found between substations experiencing squirrel-caused faults versus those
not experiencing faults and the following field variables: presence or absence
of habitat, habitat classified in three hypothesized categories, the number of
nests in a substation, and the beam type and size of the substation. The
distance of water from a substation was significantly associated with raccoon-
caused faults, and the number of nests in a substation was significantly

associated with squirrel and bird-caused faults.

29

8.333an
88288880888538”
883m ”
N8588.82.8\m.a_.8_.n..8_e.._n
~68_w2388\§823”

958m 05 Sea 85A 8.88988 83833 H
v5.2» 05 82.. av 38888 3:83 H

88838 :52 as? an as? as; H
.8533 8 ER 883 88 U
8:585 ca :52 £58 88 8

35838.3 866 .3 no 95.». 555 .853 955%?» H
neg me 865 83.. .8 ER 555 833 o>a§mo>. .
5:33 «a flea 8889 95 co Eon £55 333 9580»? .3
5:838 ..o 38 8838 95 no Son 555 833 358»? .3
38 8o 8 88.38 8 a8 88: 858 8.8.83 H _
888.8» 8 58 558 838 88st 8 U c a

88583 Mo 8% 8.8 a as 8 so... 558 838 .8882 “a
5:38 .«o 82m .3 3 one no Ben 85? .833 9680»? ”3
88.33 co 88 558 858 08589 8 H c .

A2 .ovmv :28ng 389.3%...

Ms: Sum ~N v—NMVIn
0 O

 

 

 

 

3:3 89 8 08.83.: 2835 38a 88a 8

... § _.8 EA .3: 0.8; 38¢ .83 8

... .888 8?: 88.818 332 388 33a 8

-.. .2 3.8 Row 06.18 583 388 88a 8

... 88.8 :3 18388.18 88.. 388$ 88c ease 8

... 38.8 83 .Am .2 .8 88; 9889 388 33a 8

... A 2.8 82 888 .085 888 3989 38¢ .38 8

m .13 use can 98....— A 8.: 9.88.20 328 0.8:; 228 85 33> 3:83.

 

338:3 20m 93 33am 8838-355 3363 3038088 88 58 888 338..— 98 m3“ 0333-20 he 333M .m 033.

 

30

m o—n—flh. 8» 04.6.3.-
§ .98 8889.8 888mm.

 

 

 

 

a .o 89 8 .688: 88.8 8a .888 4283
£2 82, 8 .8: 28: .8 .888 883
82 Be, 8 883.8 $82 as .882 .883
2 _ .o B? 8 03.18 .83 ES .888 .889.
8.8 2.9, 8 ..8.m.£.-._.8 88: 88> BB .882 .883
$3 3? 8 .fi .3 .8 88 888$ 83 .888 .383
.88 3g 8 882.. .828 88 88> 83 .882: .883
a .3." 38s 88 {89...— 3 8.: 283.5 A888 9385 22..— ."...5 288$ 88.3.
.888, 28

85 3.3.8 mama—3 £955 .8 89¢ 5233 828688 Ba 83 888 3283 .«o can 82 0833-30 mo 33.8% .v 033.

 

31

8. .88 88.8.8 888...»...

 

 

 

«a... ......» .8
S. ... ......» .8
0050.0 ...—g 8..
8.... Be. .8
.3... 2.9 .8
8...... Be. .8
3%... ....g 8
n: .o .....s .8
... 83.8 .3...
83... 2.9 .8
.23. 2.9 .8
9...... ...? .8
... 38.8 2:.
... 3.8.8 8....
:89 Be. .8
... 3o. .8 an.
.38... Be. .8
.... .o ...? .8
... 38.8 MN...
... .388 g...
... Ev...
.. ...... as... .8... .28.... a. 8.: 283.29

.3833... 2303..

.3... .82..

€33.18 382

.3. .8 .083

5.3.3.3.... .332. 9.580.»
.3318 .88.. 988$
...-80... .08... «5.90.. 0.5303.»

.3833... 8.633

.3... .82..

333.18 38..

.3...8 .83

33.8.3.8 .88.. 9883.
.33.... .8 .88.. 9.58...»
$.88... .08... .85.... 25509.,

“3.833... 2.8.50
.3... .82..
$33.18 382
.348 383

33.8.3.8 .88.. 98.08.»

.3318 .88.. 9.8.8;
b.8002. .08... .85.... 0>..5_0m0>

...... .03..
...... .08..

...... .88
...... .08..

...... .08..
...... .08..

...... .88

.830... .03..
.830... .08..
.8800. .08..
.8080. .08..
.8300. .08..
.8300. .08..
.6800. .08..

.059... .08..
.05....» .08..
.05....“ .08..
.05....0 .08..
.0553 .08..
.053... .08..
.36.... .08..

 

30.56 03......5. 20...—

..33 0.8.8.» 888..

 

.m.0>0. 030...... 20.. 80.00 8.30.. w...o..0..0....0 .9. 000... ...... .38....» ..o 8...». 80.0.58

.3. 3.3... 9.2.0.390 8.3.0.3... 503.0... 2.2.0.088 8.. .00. 83.0 $.05... ..o ...... 3.0. 0333-..... ..o 3.30... .m 0.8....

 

32
Spatial Arrangement of Vegetative Habitat

Comparisons between categories of the spatial arrangement of habitat
around a substation and the percent of substations experiencing faults are shown
in Figures 8-12. Of substations with no vegetative habitat present within 50m
21% had experienced animal-caused faults and 7% had experienced squirrel-
caused faults. Of substations with vegetative habitat within 50m 31%
experienced animal-caused fault and 17.5% experienced squirrel-caused faults.
For substations with habitat on one or four sides within 50m, 30% had
experienced animal-caused faults and 16% had experienced squirrel-caused
faults. For substations with habitat on two or three sides within 50m, 33%
experienced animal-caused faults and 18% had experienced squirrel-caused

faults.

 

33

 

Percent of substations experiencing faults

 

 

absent present

Vegetative habitat

 

Figure 8. Percent of substations experiencing animal-caused faults in
locations with and without vegetative habitat.

 

34

 

 

35«

30—

25<

20<

Percent of substations experieming faults

 

   

0 14 23
Spatial arrangement of vegetative habitat‘

 

Figure 9. Percent of substations experiencing animal-caused faults in
modified categories of spatial arrangement of habitat.

*Categories for the spatial arrangement of vegetative habitat:
0 : no vegetative habitat within 50m of substation
14 : vegetative habitat within 50m on one or on all sides of substation
23 : vegetative habitat within 50m on two or three sides of substation

 

35

 

 

PM of subdatima expaimcing faults

35~

30

25

20

15

10

J

J

 

 

 

0 1 2a 2b 3 4

Spatial mm of vegetative habitat‘

 

 

Figure 10. Percent of substations experiencing animal-caused faults

in the difl‘erent categories of the spatial arrangement of
vegetative habitat.

*Categories for the spatial arrangement of vegetative habitat:

: no vegetative habitat within 50m of substation

: vegetative habitat within 50m on one side of substation
2a :
2b :
: vegetative habitat within 50m on three sides of substation
: vegetative habitat within 50m on all sides of substation

O
1

3
4

vegetative habitat within 50m on two adjacent sides of substation
vegetative habitat within 50m on two opposite sides of substation

36

 

 

   

 

25 —
.E’
g 20 i
g 15 4
g «8 10 -
a
E 5 d
a. 0 4

Absent Present
Vegetative habitat

 

 

Figure 11. Percent of substations experiencing squirrel-caused faults in
locations with and without vegetative habitat.

 

 

 

g 25
g 20
15

3
3 a
g "‘ 10
El
“5 5
E 0 i i

0 14 23
Spatial arrangement of vegetative habitat‘

 

 

Figure 12. Number of substations experiencing squirrel-caused faults
and of those not experiencing faults in modified categories
of spatial arrangement of vegetative habitat.

I"Categories for the spatial arrangement of vegetative habitat:
0 : no vegetative habitat within 50m of substation
14 : vegetative habitat within 50m on one or on all sides of substation
23 : vegetative habitat within 50m on two or three sides of substation

37
Distance of Water
For substations that had no water within 50m of the substation, 28% had
experienced animal-caused faults and 6% experienced raccoon-caused faults. 0f
substations that had water within 50m, 46% had experienced animal-caused

faults and 21% had experienced raccoon-caused faults (Fig. 13-14).

 

38

 

 

 

 

g
i
E
‘3‘.
E
>50m,<150rn
Water
Figure 13. Percent of substations experiencing animal-caused faults in

each of the categories of the distance of the substation
from water.

 

 

Percent of substations faulted by
raccoons

25

20

no water water
water >50m, 60m
<1 50m
Water

 

Figure 14.

Percent of substations experiencing raccoon-caused faults in
each of the categories of the distance of the substation from
water.

 

 

39
Number of Bird Nests
Figure 15 presents comparisons between the number of nests in a

substation and the percent of substations experiencing animal-caused faults. Of
substations with no nests, 26% had experienced animal—caused faults, 15% had
experienced squirrel-caused faults, and 2% had experienced bird-caused faults.
Of substations with greater than four nests, 65% had experienced animal-caused
faults, 35% had experienced squirrel-related faults, and 9% had experienced

bird-caused faults.

4O

.eosfimnsm a E 38: me .6985: 05 3 CV 33$
vomseoécfi 28 A8 33¢ §3a0-_ob_:cm A3 33am 338-388“ wfiocotoaxo 203833 .«o :8ch .2 oSwE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.U
non—$33 E 33.. me 3832
X m N _ c

a a T L. i. o
E m

r m
m m.
- 2 m m.
m m...
- m
2 i
m.

[I F ON

m .<
5.3333 E 33: .«o 89:32 5.5833 5 88: mo 93:32
h m F 1 N .i fl .. L r o 0 VA m N .— o
. D . . _lil_ . .1 a - n - - a . , a - o

- 2 W _ _ _ g e 2 mm
Lam... . . . , ,. T8 mm.
m 0 . r - I on m 0
n... r: 1 Do. I.
T on o m. ow o m.
w m . 8 m m
. ow m. .. 8 m.

r Om t On.

 

 

 

 

 

 

 

41
Profile
Figure 16 shows the comparison of the percent of substations faulted by
animals for low profile (<7m) and high profile (>7m) substations. Twenty-two
percent of low profile substations experienced animal-caused faults, and 32% of

high profile substations experienced animal-caused faults.

42

 

Percent of substations experiencing faults

 

 

 

 

 

 

 

 

 

 

 

<7m >7m

Profile of substation

 

Figure 16.

Percent of substations experiencing animal-caused faults in
the two categories of profile.

 

43
Substation Structure
Of substations that were greater than 100m2 and constructed with L-
shaped beams 53% had experienced animal-caused faults and 35% had
experienced squirrel-caused faults. For substations constructed with S-shaped
beams (I-beams) 13.3% had experienced animal-caused faults and none

experienced squirrel-caused faults (Fig. 17).

44

 

 

 

 

E 60 1
E 50 1
'8' 40 «
35
g E 30 -
a 20 <
"6 ,
‘5 ‘° D
a. 0 I 4 l i ' :
l 2 3 4 5
Structural category‘

 

 

 

 

Percent of substations experiencing
faults

Structural category'

 

 

 

B.

Figure 17. Percent of substations experiencing animal-caused faults (A) and
squirrel-caused faults (B) by structural categories.

*Structural categories:

I: L—shaped beams / structure 5 100m2
2: L-shaped beams / structure > 100m2
3: S-shaped beams

4: combination L and S-shaped beams

5: wooden poles

45

Logistic Regression Model

The maximum likelihood estimates and related statistics for a logistic
regression analysis including all variables used to classify substations are shown
in Table 6. The model based on these estimates correctly predicted 56% of 50
samples that had not been used to create the model. The highlighted variables
in table 6 are the ones used in the logistic regression analysis that included only
the significant factors (P<O.10). This model again correctly predicted 56% of
50 samples not used to created the model (Table 7). The Hosmer and
Lemeshow Goodness-of-fit tests for the models indicated that both models fit

since the tests were not significant (p=0.57, p=0.60).

46

Table 6. Regression coefficients (maximum likelihood estimates) and related
statistics from logistic regression analysis.

 

 

Parameter Standard Wald Pr > Standardized

Variable DF Estimate Error Chi-Sgare Chi-Square Estimate
intercept 1 -2.4993 0.7332 11.6199 0.0007 .

habitat (0,1) 1 0.5704 0.4392 1.6864 0.1941 0.129183
water <50m 1 0.0364 0.5595 0.0042 0.9482 0.005618
water >50m 1 -l.8335 1.1749 2.4355 0.1 186 -0.240262
1 nest 1 0.4035 0.4351 0.8598 0.3538 0.088588
2 nests 1 0.2109 0.5673 0.1382 0.7101 0.034013
3 nests 1 -0.112 0.8357 0.018 0.8934 -0.01365
>4 nests 1 1.6508 0.4964 11.0605 0.0009* 0.255005
profile 1 0.6714 0.3974 2.8541 0.0911' 0.165357
L-beams / <100m2 1 0.8661 0/6396 1.8339 0.1757 0.226985
L-beams / >100m’ 1 1.2145 0.5789 4.401 0.0359“ 0.159152
combination 1 1.1815 0.8535 1.9164 0.1663 0.149535
wooden poles 1 1.174 0.882 1.772 0.1831 0.148586

 

* Significant factor (p<0.10)

47

Table 7. Matrix of predictions of damage of 50 substations by logistic regression
models using all variables and significant variables only (both models
have the same matrix).

 

Damage predicted No damage predicted
Correct 4 24

Incorrect 1 21

 

 

 

DISCUSSION

Examination of Records of Animal-caused Faults

The data provided by Consumers Power from reports of animal-caused
faults to electric substations between 1988 and 1993 helped to define the
general nature of the problem of animal damage. Over the six year period 121
incidents of animal-caused faults occurred in 88 of the 1,177 transmission and
distribution substations owned by Consumers Power. There could be any
number of reasons that animals entered the substations. They could have been
searching for food, storing food, searching for nesting sites, playing, escaping
from predators, travelling, or satisfying curiosity.

According to the records, squirrels accounted for 57% of all animal-
caused faults. These results correspond very well to those of Enck (1989) who
found that 55% of all animal-caused faults were caused by squirrels. Forty-two
percent of all damage caused by squirrels occurred in June, September, and
October. Similarly, in a study of fox squirrel damage to electrical transformers,
Hamilton (1987) found that 34% of all outages caused by squirrels occurred in
June and October during the morning hours. The results reflect the seasonal
and daily activity patterns of squirrels, such as foraging activity and home range
expansion of young squirrels born in early spring or in the summer. Mid-May
and early October were identified by Thompson and Thompson (1980) as two
annual peaks in the foraging activity of urban grey squirrels, and maximum

home range expansion for grey squirrels in Toronto, Canada was found to be at

48

49
the highest levels in the middle of June and in the middle of October
(Thompson 1978).

Because squirrels are known to be highly active during the morning
hours just after sunrise (Brown and Yeager 1945, F ogl 1982), it was not
surprising that 70% of squirrel caused faults occurred between 6:00AM and
noon. Grey squirrels have typically been found to have a bimodal daily activity
pattern with most activity occurring in the morning and late afternoon,
particularly in non-winter months (Brown and Yeager 1945, Thompson 1977,
Fogl 1982). Fox squirrels, on the other hand, typically exhibit a unimodal daily
activity pattern with activity occurring early in the morning and through midday
(Brown and Yeager 1945, Cauley 1974, Fogl 1982). The reason that most
squirrel damage occurs in the morning might involve the overlap in daily
activity patterns of fox and grey squirrels.

Raccoons were found to have caused 21% of the reported animal-caused
faults, which was a higher percentage than the 12% found by Enck (1989).
Damage caused by raccoons in April, May, and August accounted for 65% of
damage by raccoons over all months. No raccoon caused damage occurred in
November through February. As with the squirrel related results, these findings
apparently reflect the activity patterns of raccoons. In a radio-tracking study of
raccoons in Minnesota, Schneider (1973) observed that by early April raccoons
stopped using their winter dens and became more active than they had been in
the previous winter months. Young raccoons born in May became independent
around August, and all activity stopped at the end of November, when the
raccoons denned for the winter. Hoffman and Gottschang (1977) reported that
raccoons in suburban Ohio were most active during May, July, and August.

The highest raccoon capture rate was found in May, and raccoon trapping rates

50
dropped off substantially in November through April. Furthermore, raccoon
damage was most common between the hours of 9:00PM and 6:00AM. Once
again, the results coincide well with the peak daily activity patterns of raccoons
(Cauley 1974, Schinner and Cauley 1974).

Birds caused 7% of animal caused damage. The Consumers Power
records did not indicate what species of birds were involved in each incident.
Because this information was not available and because the number of incidents
of bird damage were relatively low (9), it is difficult to draw any conclusions
about seasonal or temporal patterns. Birds probably caused faults by contacting
two energized components simultaneously (phase-to-phase fault), which seems
to occur infrequently (Enck 1989). The low numbers of bird caused faults
suggests that this kind of damage should not be a major focus in preventing
animal damage to substations. However, other factors that are related to the
presence of birds in substations, such as nests, might impact the susceptibility
of the substation to damage. The impact of nests in a substation is discussed in
the characterization of damaged versus undamaged substations.

Finally, the Consumers Power records provided information about direct
and indirect costs incurred because of animal-caused faults. One cost to the
company is the loss caused by damage to substation equipment. The most
commonly damaged and replaced items in a substation were fuses (54%), and
insulators were the second most damaged item (10%). Other costs to the
company include the revenue lost while power is out and the indirect cost of
reduced consumer confidence because of the loss of service to customers during
an outage. Outages occurred in all but three of the 121 reported faults. The
number of customers affected by an outage ranged from 1 to 3,979, and an

average of 2,511 customers per outage were affected. The amount of time that

51
a fault lasted was one hour and 46 minutes on average and ranged from 15
minutes to 14 hours and 14 minutes. These results show that the costs of
animal-caused fault are potentially very high. Back and Brown (1989)
estimated combined direct and indirect costs of animal-caused faults for six
utility companies in New York state to be as high as $10 million over an eight
year period.
Evaluation of Control Measures

Observations of raccoons and squirrels at the animal proofed training
facility indicate that control measures are effective if correctly and completely
applied, but proper application is critical. The only raccoon to thwart the
control measures (not via the outer containment fence) was able to find the only
missing tie on the polycarbonate barrier in about eleven minutes. The raccoon
was then able to pull the polycarbonate away from the fence and climb under it
until he reached the top of the fence. The fact that it took only eleven minutes
for the raccoon to find the missing tie suggests that even one minor flaw is
enough to make the substation fence penetrable. However, it is important to
recognize that the observed raccoons were trapped between the outer and inner
fences, which would not be the situation in a natural setting.

Of the squirrels observed at the facility, only one, a black squirrel,
penetrated the control measures. As with the raccoon, the squirrel found a flaw
in the animal proofing. This time it was a ls/z-inch gap between the fence and
the gate. It took the squirrel about 65 minutes to find and escape through the
hole. Consumers Power specifies in their animal proofing measures that there
should be no more than a l-inch gap. Here again is evidence that the correct
application of damage control measures is important if they are the be entirely

effective.

52

In addition to these two escapes, the red squirrel and one raccoon
escaped by way of the outer polycarbonate fence. While these animals did not
escape through the animal-proofing, they demonstrated both the agility of the
animals and their ability to find obscure escape routes. For example, the
raccoon climbed the fence and was able to stretch just far enough to use a tie to
pull himself to the corner between the inner and outer fences and then climb
out. The red squirrel escaped at the corner of the outer fence by jumping to
grab a 1/4-inch hole near the top of the fence (4ft from the ground).

Although the control measures did not keep two of the animals out of the
substation, they did prevent ten others from entering. Therefore, it appears that
the control measures are effective but that this effectiveness is compromised by
flaws in application.

The results of the field evaluation of 43 substations treated with animal
proofing showed that only one of the substations was completely and correctly
treated, indicating that effectiveness is not 100% at any of the other substations.
Common flaws in animal proofing included: 1) polycarbonate or aluminum
sheeting not applied to all poles within 25m of the substation and/or no
lineguards on lines entering the substation (81% of treated substations), 2) ties
applied horizontally (81%), 3) polycarbonate or aluminum sheeting not applied
correctly at gates (79%), and 4) gaps greater than linch at the gate to the
substation (74%). Some of the flaws represent old control measures that were
applied under different guidelines of animal proofing than currently being used.
For example, ties applied horizontally have been found by Consumers Power to
be more susceptible to access by squirrels than ties that are applied vertically.

Therefore, in newly treated substations the ties are affixed vertically.

53

Although these partially treated substations are probably less susceptible
to damage, they are not entirely safe from it. The effectiveness of treatment
will depend in part on how badly an animal wants to get into the substation.
However, the observations at the training facility indicate that if all control
measures are precisely and thoroughly applied, animals will be prevented from
entering a substation.

Characterization of Substations

One way to aid in the prevention of animal-caused faults at substations is
to understand what makes the substation susceptible to these faults. This
approach narrows the scope of the problem by identifying characteristics to be
targeted with preventative measures and by identifying substations most likely
to be damaged based on the set of characteristics.

In the comparison of substations that had been damaged and those not
damaged, three of the variables measured in the field analysis were significantly
associated with faults (p < 0.10). These variables were beam construction
(Fisher's exact test, p = 0.019), distance of water from the substation (chi square
test, p = 0.051), and number of nests in the substation (chi square test, p =
0.003).

From results of the Fisher's exact test and from examination of the
frequency table for this variable, the substation structures most responsible for
susceptibility to damage are those involving L-shaped beams, particularly those
that cover an area greater than 100ml. The reason for this beam structure being
more susceptible could be because these structures are easily climbed by
raccoons and squirrels and because they provide good nesting locations for
birds. Damage caused by squirrels was significantly associated with beams

structure (p<0.001), suggesting that this may be a particularly important factor

54
to consider when trying to prevent squirrel damage. Beam structure was not
significantly associated with damage by raccoons or by birds although sample
sizes were much smaller for these two categories.

The physical profile of the substation was not found to be significantly
different between substations that had been faulted and those that had not,
although the chi square p-value was 0.119. Faults caused by squirrels and birds
were more likely to be affected by profile than those caused by raccoons (p =
0.830) but none were significant. Enck (1989) reported that 81% of animal-
caused faults occurred in high profile substations (>25ft) but could not make
definite conclusions about susceptibility because the system-wide distribution of
low and high profile substations was not known and because of inconsistencies
in the definition of high and low profile used by data providers. In the present
study 78% of animal-caused faults occurred in high profile substations which
coincides with the results found by Enck. However, 37% of the 204 high
profile substations (>7m) experienced damage, whereas 22.5% of the 80 low
profile substations (<7m) experienced damage. The latter results suggest that
the effect of profile on susceptibility is probably not as marked as suggested by
Enck's results. Furthermore, because beam type and physical profile are
significantly related, it is possible that the association between faults and profile
is just an artifact of the association between beam type and animal-caused
faults.

The effect on damage of the distance of water from the substation was
significant for damaged versus undamaged substations (p = 0.051). Faults
caused by squirrels and by birds were not significantly associated with distance
from water, but raccoon-caused faults were significantly associated with the

distance of water from a substation (p = 0.017). Substations within 50m of

55
water seem to be particularly susceptible to faults caused by raccoons. These
results reflect the fact that water is an important component of raccoon habitat.
For example, in a study by Soneshine and Winslow (1972), 98% of raccoon
captures in one study area were within 121m of water. The significant
association of water with damage suggests that proximity of water to a
substation should be a factor that is considered, where reasonable, when
building new substations and could provide an indication of which substations
will experience faults in the future.

The results show that the presence of nests in a substation is significantly
associated with damage to the substation (p = 0.003). The presence of at least
four nests in a substation seems to be particularly indicative of damage to the
substation. Approximately 18% of faulted substations had at least 4 nests, but
only about 4% of substations that had not experienced a fault had at least 4
nests. Squirrel and bird-related faults were significantly associated with the
number of nests at a substation but raccoon-related faults were not.

The connection between the number of nests in a substation and bird
damage is obvious, but the connection with squirrel damage is not as obvious.
Nests might attract foraging squirrels to a substation. However, it is also likely
that squirrel damage and the number of nests are associated because of some
habitat or structural component to which they are both correlated.
Independence of the variables in the study was difficult to analyze because
sample sizes for certain combinations were extremely low. However, there
were no obvious associations between nests and the other variables. Therefore,
if nests are correlated with some habitat or structural component, it could be
one that was not recorded in this study. An evaluation of the species of birds

nesting in substations might be helpful in determining the cause of the

56
association between nests and damage because the species that have been found
to nest in substations, such as sparrows, robins, starlings, and blackbirds, occur
in different habitats.

The final variable considered in the field analysis was vegetative habitat.
A study by Enck and Brown (1989) found no correlation between surrounding
habitat and the number of faults at a substation. However, the relationship
between habitat and wildlife damage may not have been fully described by the
study. One possible reason for these findings is that the correct habitat
variables were not used.

Another possible reason that Buck and Brown found no correlation
between habitat and wildlife damage is that there is a non-linear relationship
between the amount of habitat surrounding the substation and the amount of
damage to the substation. For example, in areas with few trees there may be
little damage to substations because there are no squirrels in the area. There
may also be little damage in areas with many trees because the animals would
not need to use the substation as a travel route and would have plenty of
available habitat. In areas with moderate amounts of trees, the most damage
would be seen because animals would use substations for activities such as
travelling or searching for food or shelter.

A final possibility is that the spatial arrangement of the vegetative habitat
around a substation may be a significant factor in wildlife damage to the
substation. For example, Layne (1954) observed that red squirrels often use
hedgerows as travel lanes. If a hedgerow were interrupted by a substation, it
would be easy to imagine that a squirrel, or any animal using the travel route,
might enter the substation more often than if the hedgerow were continuous and

away from the substation. Similar relationships between the spatial arrangement

57
of the habitat surrounding a substation and use of a substation by an animal
might be predicted for other animals.

The results of the present study showed a significant association between
squirrel-related faults and the presence or absence of vegetative habitat
(p=0.0034). Approximately 91% of substations faulted by squirrels had
vegetative habitat on at least one side, and 76% of substations not experiencing
faults had vegetative habitat on at least on side. Additionally, it had been
hypothesized that damage would occur most often in substations with vegetative
habitat on two or three sides of the substation, less often in substations with
vegetative habitat on one or four sides, and least often in substations without
any vegetative habitat surrounding it. The data from the field generally
reflected this hypothesis, although the relationship between the three generalized
categories and animal-caused faults was not significant. When damage by
squirrels was analyzed for substations placed in the three habitat categories, a
significant relationship was found (p = 0.094). None of the categorizations of
vegetative habitat were significantly related to faults by raccoons or birds. The
fact that a significant relationship was found for squirrels but not for the other
animals probably reflects that the hypothesis that established the habitat
classifications was based largely on the predicted behavior of squirrels. The
smaller sample sizes for raccoons and birds could also have influenced the
analysis.

The above results suggest that there is a relationship between vegetative
habitat and animal-caused faults to substations. In particular, squirrels are most
likely to cause faults in substations that are surrounded by vegetative habitat on
two or three sides and least likely to cause faults in substations with no

vegetative habitat within 50m. These findings contradict the conclusion by

58
Enck (1989) that substation damage was not related to the surrounding habitat.
Methods of collecting and analyzing habitat data differed considerably between
that study and this one and might explain the discrepancy. By Contrast,
Hamilton (1987) found that the number of leaf nests near electrical transformers
in Lincoln, Nebraska was significantly greater where squirrel-caused outages
were a problem and that the mean basal area of mulberries within 2m of power
equipment or power poles was several times greater at problem sites.

To analyze the capability of the habitat and structural variables
considered in this study to predict damage to electric substations, a logistic
regression model was created from 234 of the substations classified in the field
analysis. First, a model was created that included all of the variables in the
classification of visited substations. Fifty substations not used to create the
model were randomly selected from the sampled substations and were used to
test the predictive ability of the model. Substations were classified as faulted if
the predicted probability of being faulted was greater than 0.50. The model
correctly predicted whether or not a substation had experienced an animal-
caused fault for only 56 % of the 50 samples. Because only four of the
independent variable were significant at p<0.10, the model was simplified to
include only these variables. Like the model using all variables, this model fit
the data (p=0.6005) and correctly predicted 56% of the 50 sampled substations
used to test the models. The low predictive ability of these models probably
has to do with the fact that the two most significant predictors (>4 nests in a
substation and L-shaped beams with heavy latticework) occurred in only six of
the 50 substations used to test the models, and none of the 50 substations had

both characteristics.

59

In theory, this type of model could be used to predict whether or not a
substation will be damaged in the future, but the model would have to be
substantially improved by considering additional explanatory variables.
However, it is likely that a highly predictive model would be extremely difficult
to create. Regardless of whether the model is used to predict faults, it
illustrates how the different habitat and structural components associated with a
substation can be used in combination to assess the likelihood of the substation

experiencing an animal-caused fault.

60

Recommendations

Several recommendations about how to prevent animal-caused faults can
be made based on the results of this study. First, certain characteristics of the
habitat and of the structure of a substation should be considered when a new
substation is built. Building a substation in areas within 50m of lacustrine or
riverine habitat should be avoided. A good location for a substation would
probably be one with very few trees and shrubs and no water within 50m. In
addition, if possible, substations should be built with beams that are least
conducive to climbing and nesting activities. The structural characteristics of
the substations owned by Consumers Power that are most likely to be connected
with animal damage are large L-shaped beam structures. Because the choice of
sites and materials for a new substation are likely to be limited and habitat
alteration may not be possible, preventative measures that can be applied to
existing substations is probably more practical.

Observations of challenges to the control measures used by Consumers
Power established that the measures are effective in keeping raccoons and
squirrels out of a substation if applied properly. Inspection of treated
substations by informed personnel might help to insure that the measures have
been thoroughly and precisely applied. The control measures could be applied
to substations that have one or more of the characteristics that were found to be
associated with animal-caused faults. These substations would be ones that are
composed of L-shaped beams and cover an area larger than about 100m2,
substations that are located within 50m of water, substations with vegetative
habitat within 50m on two or three sides, and substations that have at least four

bird nests in them. A substation having just one of these attributes probably

61
does not warrant animal proofing, although the presence of all of the
characteristics might warrant it if the costs of animal proofing are not too large.
Currently at Consumers Power, the decision about whether or not to animal
proof depends on how many times and how often the substation has been
damaged. These criteria could be incorporated with knowledge about structural
and habitat characteristics to make this decision in the future.

The association of damage with nests might suggest regular removal of
nests or a control measure that prevented nesting. Most nests were located in
corners where perpendicular beam structures meet, and so any object that would
obstruct the corner might help to prevent nesting in the substation. The air
spaces in the transformer assembly were also common nesting sites and
therefore could be targeted with control measures. However, it is also possible
that the nests themselves are not the problem but are related to some unknown
characteristic that affects the susceptibility of the substation. In this case the
only way to prevent damage would be to use the traditional control measures
because the actual source of susceptibility is not known. The investigation of
factors that are potentially related to the number of nests in a substation would

help to address this problem.

CON CLU SION

The prevention of wildlife damage to electric substations is aided by an
understanding of what factors are responsible for the susceptibility of a
substation to damage and of what preventative measures will be most effective.
The results of this study provide evidence that the susceptibility of electric
substations to animal-caused faults is related to the type of beams used in the
substation, the size of the structure, distance of the substation from water, and
the number of bird nests in the substation. The presence and spatial
arrangement of habitat components was significantly associated with squirrel-
caused faults, which accounted for 57% of all of the animal-caused faults
recorded.

The application of the control measures currently being used by
Consumers Power should be effective in keeping raccoons and squirrels from
entering the substation, provided that the application has been done correctly
and completely. The decision about which substations to treat with control
measures should involve the evaluation of the structural and habitat
characteristics of a substation in addition to the evaluation of records of
previous damage to a substation. This approach should help to more

successfully prevent wildlife damage to substations.

62

LIST OF REFERENCES

Albert, D. 1986. Regional landscape ecosystems of Michigan. School of Natural
Resources, Univ. of Michigan, Ann Arbor, Mich.

Altman, D. G. 1991. Practical statistics for medical research. Chapman and Hall,
London, England. 611pp.

Baker, R. H. 1983. Michigan mammals. Michigan State University Press, east
Lansing, Mich. 642pp.

Beissinger, S. R. and D. R. Osborne. 1982. Effects of urbanization on avian
community organization. Condor 84:75-83.

Brown, L. G. and L. E. Yeager. 1945. Fox squirrels and gray squirrels in Illinois.
Nat. Hist. Surv. Bull. 23:449-536.

Cauley, D. L. and J. R. Schinner. 1973. The Cincinnati raccoons. Nat. Hist.
82(9): 58-60.

Cauley, D.L. 1974 Habitat requirements of four selected species in the urban
environment. Ph.D. Thesis, Michigan State University. 66pp.

Eiechenlaub, V. L. 1990. The climate atlas of Michigan. University of Notre Dam
Press, South Bend, Ind.

Enck, J. W. 1989. Wildlife damage in electric substation in New York. Pages
225-232 in S. R. Craven, ed. Proceedings of the fourth eastern wildlife
damage control conference. Madison, WI.

, and T. L. Brown. 1989. Animal deterrents for electric substations. Final
report: ESEERCO project EP 87-11. 76pp.

Fiske, J. 1992. Electric fence promises to keep squirrels away. Electrical World.
206(11):46-47.

Freeman, D. H. 1987. Applied categorical data analysis. Marcel Dekker, Inc. New
York. 318pp.

Fogl, J. G. 1982. Ecology of melanisitic gray squirrels (Sciurus carolinensis) and

fox squirrel (S. niger)in an urban area. Ph. D. Thesis, Michigan State
University. 143pp.

63

64

Greenwood, R. J. 1981. Foods of prairie raccoons during the waterfowl nesting
season. J. Wildl. Manage. 45(3): 754-760.

Hamilton, J. C. 1987. Fox squirrel interactions with electrical transformers. M. S.
Thesis, University of Nebraska-Lincoln. 76pp.

Hathaway, M. B. 1973. Ecology of city squirrels. Nat. Hist. 82(9): 61-62.

Hoffman, C. O. and J. L. Gottschang. 1977. Numbers, distribution, and
movements of a raccoon population in a suburban residential community. J.
of Mammalogy. 58(4):623-636.

Horn, D. J. 1985. Breeding birds of a central Ohio woodlot in response to
succession and urbanization. Ohio J. of Science. 85(1):34-40.

Layne, J. N. 1954. The biology of the red squirrel, T amiasciums hudsonicus
loquax (Bangs), in central New York. Ecol. Mono. 245(3):227-267.

Mitchell, B. 1977. Wildlife impacts on telephone and electrical services in
Waterloo, Ontario. J. Soil and Water Conservation. 32(3):]33-135.

Paula, G. 1989. Sticky gel repels birds from substations. Electrical World.
203(12):S30.

Paula, G. 1990. Prevent animals from damaging substations. Electrical World.
204(3):S47-S49.

SAS Institute Inc. 1985. SAS user's guide: basics, version 5 edition. Cary, NC:
SAS Institute Inc. 1290pp.

. 1988. SAS/STAT user’s guide, release 6.03 edition. Cary,
NC: SAS Institute Inc. 1028pp.

 

Schinner, J. R. and D. L. Cauley. 1974. The ecology of urban raccoons in
Cincinnati, Ohio. pages 125-130 in Wildlife in an urbanizing environment.
Cooperative Extension Serv., University of Massachusetts, Amherst, Mass.

Schneider, C. 1973. The adaptable raccoon. Nat. Hist. 82(8): 64-71.

Soneshine, D. W. and E. O. Winslow. 1972. Contrasts in distribution of raccoons
in two Virginia localities. J. Wildl. Manage. 36:838-847.

Steele, M. A. and PD. Weigl. 1992. Energetics and patch use in the fox squirrel
Sciurus niger. responses to variation in prey profitability and patch density.
Am. Midl. Nat. 128: 156-167.

Substation Security Working Group. 1993. IEEE guide for animal deterrents for
electric power supply substations. Institute of Electrical and Electronics
Engineers, Inc. New York. 16pp.

65

Thompson, D. C. 1977. Diurnal and seasonal activity of the grey squirrel. Can. J.
Zool. 55:1185-1189.

. 1978. the social system of the grey squirrel. Behaviour.

 

64:305-328.

. and P. J. Thompson. 1980. Food habits and caching behavior
of urban grey squirrels. Can. J. Zool. 58:701-710.

Warren, C. M. 1992. The effect of reducing momentary outages on distribution
reliability indices. IEEE Transactions on Power Delivery. 7(3): 1610-1615.

Williamson, R. D. 1973. Bird -and people- neighborhoods. Nat. Hist. 82(9): 55-
57.

Williamson, R. D. 1982-83. Identification of urban habitat components which
affect eastern gray squirrel abundance. Urban Ecol. 7: 345-356.

9N ST

nrcurc RTE UNIV. LreRaRrEs
lllllWWill“NW“WWIHWIIHWIIWHI
31293013996503