Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 159
Working Paper E
Relationships of Ground Motion,
Damage, end toss
Although in actual practice the steps in a loss estimation study
do not necessarily proceed sequentially, the previously discussed tasks
of seismic hazard analysis (Working Paper C) and inventory (Work-
ing Paper D) are conceptually the two steps that precede the process
of relating the ground motion or ground failure to a given construc-
tion class to estimate damage. This paper also discusses relating
damage to property low, casualties, or functional loss. The discus-
sion here is limited to the eEects of ground shaking on buildings and
lifelines; the effects of ground failures are treated in Working Paper
G.
Material presented in the two earlier working papers is directly
applicable here. Working Paper C discussed the limitations of the
Modified Mercalli Intensity (MMI) scale and other problems in the
accurate definition of the ground motion to which the inventory
should be subjected in the motion-damage analysis step. Working
Paper D explained that the construction classification system is a
part of both the inventory process and the motion-damage analysis
step because the inventory information must be collected with the
same construction classes used in relating the seismic hazard to
construction classes through motion-damage relationships.
Many methods of relating ground motion, or less commonly
ground failures, to damage have been proposed or developed. How-
ever, in the context of large-scale, general-purpose loss estimation
159
OCR for page 160
160
studies the number of basic approaches is relatively small. In this
paper, three particular methods are discussed because they bring
out different aspects of the possible ways to approach this problem
of relating motion, damage, and loss.
The loss estimation method referred to here and elsewhere in
this report as the NOAA-USGS method is also, in terms of the
motion-damage analysis step, essentially the Insurance Services Of-
fice (ISO) method. As explained earlier, this method was used In the
first large-scale studies produced by the National Oceanic and Atmo-
spheric Administration (NOAA) and later, with essentially the same
personnel, by the U.S. Geological Survey (USGS) when NOAA's
earthquake loss estimation functions were shifted to USGS.
This method has been molded by the work of AIgermissen, Stein-
brugge, and others. The studies of San Francisco (AIgermissen et al.,
1972), Los Angeles (AIgermissen et al., 1973), Puget Sound (Hopper
et al., 1975), and Salt Lake City (Rogers et al., 1976) are examples
of the use of this method. It is the method that has been applied
in most of the urban- or regional-scale studies of the type focused
on in this report (studies intended for disaster planning and hazard
reduction purposes).
It is also the method that has been most widely used in the prop-
erty insurance industry. The NOAA-USGS or ISO method produces
damage estimates in the form of mean damage ratios for each con-
struction class percentages associated with each MMI level indicat-
ing the average property loss In terms of cost of repair or replacement
divided by replacement cost. In the NOAA-USGS method, lifelines
and nonbuilding structures are analyzed by different methods than
the mean damage approach applied to buildings.
The ATC-13/FEMA approach was produced by the Applied
Technology Council and funded by FEMA (Applied Technology
Council, 1985~. While it has yet to be carried out in a loss study
resulting in a published report of the type produced for several re-
gions of the country by the NOAA-USGS method, it is a recent,
comprehensive effort that involved many experts and it surveyed
and evaluated a broad range of analysis methods and data.
The ATC-13 method uses the format of the damage probability
matrix to present its damage estimates for each MMI level: the
percentage of facilities that wouic] fall into each of seven damage
levels is given for each construction class (with these damage levels
described verbally, with property damage ratio ranges, and with
central damage ratios). For each MMI, the distribution of damage
OCR for page 161
161
for a construction class can also be converted into an overall damage
ratio. In the ATC-13 method, lifelines and nonbuilding structures
are essentially handled with damage probability matrices the same
way as for buildings.
A third basic method to be discussed is the application of fragility
curves to the task of estimating regional-scale earthquake losses. The
Central U.S.-Six Cities study (ABen and Hoshall et al., 1985) used
this approach. The motion-damage portion of this study's method,
the development of fragility curves based on a combination of em-
pirical or historical data and theoretical calculations, was developed
by Jack Benjamun and Associates, Inc. (Kircher and McCann, 1984)
and is occasionally referred to as the lBA method later. One fragility
curve describes the probability a given construction class will reach
or exceed one particular level of damage at various intensities of
shaking. A set of curves, to cover all the damage states, is used
for each construction class. Fragility curves and damage probability
matrices are similar in the information they provide and one can be
converted into the other. Fragility curves present the information
graphically, while damage probability matrices present the infor-
mation in tabular form. In the dBA-Central U.S. study's method,
lifelines and nonbuilding structures were treated with fragility curves
in a manner parallel to that used for buildings.
NOAA-USGS MOTION-DAMAGE RE[ATIONE3HIPS
The earliest U.S. attempt at estimating earthquake property loss
on a large scale began in 1925 when engineers Harold Engle and Jack
Shields gathered data on the damage caused by the Santa Barbara
earthquake for use by the insurance industry. This work has con-
tinued and has resulted, after several developments and refinements,
into the NOAA-USGS method or the similar ISO method.
The generic NOAA-USGS motion-damage relationship is shown
in Figure Ad. The truncation of the mean damage ratio curve at
MMI X-X is due to the fact that intensities above this point have
sometimes been assigned to sites in previous earthquakes on the basis
of ground failure, not ground shaking. Table ~1 briefly tabulates
the construction classes. Each class is described with approximately
a paragraph in the Commercial Earthquake Insurance Manual (ISO,
1977).
The damage ratio is the percentage damage related to cost of
OCR for page 162
162
100
a'
ILL
-
co
an
o
J
IL]
G
IL
.~.
ale.......
..~..............
,A ~ 2.- - 2..2 - -
.. ~ 2 2.
. ~ '
Geologic
~ ~= ,
~ effects
.~...............................................................................
Ace . ..................
' ~^
SAC
O ~
a.,, . ~
~~ D
1 1 1
E
1 1
IV V Vl Vll Vlil IX X Xl X11
MODIFIED MERCALLI INTENSITY
FIGURE E-1 Relationship of ground failures to ground shaking in the Modified
Mercalli Intensity scale. Source: Steinbrugge (1982~.
replacement. Mean damage ratios are used because they are aver-
age factors for all buildings of given classes. They do not give the
distribution of damage, such as how many buildings had little or
no damage or how many had moderate damage. The mean damage
OCR for page 163
163
TABLE E-1 Construction Classes Used in the ISO and
NOAA/USGS Methods
Building
Class Brief Description of Building Subelasses
lA-1 Wood-frame and stuccoed frame dwellings
regardless of area and height
1A-2 Wood-frame and stuccoed frame buildings, other
than dwellings not exceeding three stories
in height or 3,000 square feet in ground
floor area
1B
2A
2B
3A
3B
3C
4B
1A-3 Wood-frame and stuccoed frame structures not
exceeding three stories in height regardless
of area
Wood-frame and stuccoed frame buildings not
qualifying under class 1A
One-story, all metal; floor area less than
20,000 square feet
All metal buildings not under 2A
Steel frame, superior damage control features
Steel frame, ordinary damage control features
Steel frame, intermediate damage control
features (between 3A and 3B)
3D Steel frame, floors and roofs not concrete
4A Reinforced concrete, superior damage control
features
Reinforced concrete, ordinary damage control
features
4C
4D
4E
5A
5B
5C
5D
BE
6
Reinforced concrete, intermediate damage
control features (between 4A and 4B)
Reinforced concrete, precast reinforced
concrete, lift slab
Reinforced concrete, floors and roofs not
concrete
Mixed construction, small buildings and
dwellings
Mixed construction, superior damage control
features
Mixed construction, ordinary damage control
features
Mixed construction, intermediate damage
control features
Mixed construction, unreinforced masonry
Buildings specifically designed to be
earthquake resistant
SOURCE: Algermissen and Steinbrugge, (1984~. For more
complete descriptions of each class, see Iso (1977) and
McClure et al. (1979~.
OCR for page 164
164
35
30
25
-
a)
~ 20
a)
cat
a)
_'
in 1 5
o
J
10
5
r
-
~5~
/~B 3D 4C 5C /
_~,
~ ~ ,
3C 4A 5B/
-
-
/' 3A 2B A/—
1 1
V
Vl V11
MM INTENSITY
V11
IX
FIGURE ~2 Mean damage ratio curves used in the NOAA-USGS method.
Source: Algermmsen and Steinbrugge (1984~.
ratio directly defines property loss, but does not directly indicate loss
of function or number of casualties. Figure ~2 shows some of the
mean damage ratio curves used in the NOAA-USGS method.
The amount of historic damage data available on some of the
classes of construction, particularly wood-frame dwellings, is exten-
sive, whereas more judgment and fewer data are employed to develop
damage ratios for high-rise buildings or many low-rise commercial-
industrial construction classes for which there is less experience. The
ISO system generally limits itself to classes of construction for which
there are historic data.
Single-family wood-frame dwellings are the class of construction
OCR for page 165
165
having the greatest historical data, with the possible exception of
mobile homes. The accuracy of the basic NOAA-USGS method
for this class is high as judged by the work of McClure (1967),*
whose property loss relationships (based on the 1952 Kern County
earthquakes) predicted a total loss of $3.8 million when applied to
the 1969 Santa Rosa earthquakes, whereas the actual postearthquake
estimated dwelling loss figure was $4 million (Steinbrugge et al.,
1970~.
Another test of a loss estunation method for single-family dweD-
ings is provided in Rinehart et al. (1976) wherein the results of a
modified version of the 1969 method by Steinbrugge and others are
favorably compared with data from the 1971 San Fernando earth-
quake. The 1971 data on all of the approximately 12,000 dwellings in
one area of the San Fernando Valley where the shaking was strongest
are unusually large and detailed. Often only rough or sem~quanti-
tative data on a few dozen buildings of one construction class are
available Tom an earthquake, or the reports are selective (typically
only noting cases of dramatic damage).
ATC-13 MOTION-DAMAGE RELATIONSHIPS
The ATC-13 method does not describe its building construction
classes in as much detail as in the NOAA-USGS scheme, but includes
many structures that are not addressed in the NOAA-USGS method.
It has a total of 78 classes of structures, 40 of which are buildings and
38 of which are lifeline-related or equipment classes. These classes
are listed in Table ~2.
Lacking the major source of hard data in the ISO or NOAA-
USGS method, which was proprietary to the insurance industry,
ATC-13 relied on the expert opinion of experienced individuals in
the earthquake engineering field to produce motion-damage relation-
ships. The techniques used for processing the questionnaire answers
are described in the AT~13 report.
The form in which the ATC-13 motion-damage relationship for
each class was solicited from the experts, and the way in which the
combined or consensus expert opinion was presented, was the damage
probability matrix. This format and idea was originated by Marte}
(1964) and independently developed in the Massachusetts Institute
* Given the loose definition of aNOAA-USGSn method used here, the
McClure work fits this definition.
OCR for page 166
166
TABLE E-2 Earthquake Engineering Facility Classification
Facility
Classification
Number
BUILDINGS
Wood frame (low rise)
Light metal (low rise)
Unreinforced masonry (bearing wall)
Low rise (1-3 stories)
Medium rise (4-7 stories)
Unreinforced masonry
(with load-bearing frame)
Low rise
Medium rise
High rise (> 8 stories)
Reinforced concrete shear wall
(with moment-resisting frame)
Low rise
Medium rise
High rise
Reinforced concrete shear wall
(without moment-resisting frame)
Low rise
Medium rise
High rise
Reinforced masonry shear wall
(without moment-resisting frame)
Low rise
Medium rise
High rise
Reinforced masonry shear wall
(with moment-resisting frame)
Low rise
Medium rise
High rise
Braced steel frame
Low rise
Medium rise
High rise
Moment-resisting steel frame
(perimeter frame)
Low rise
Medium rise
High rise
Moment-resisting steel frame
(distributed frame)
Low rise
Medium rise
High rise
Moment-resisting ductile concrete frame
(distributed frame)
Low rise
Medium rise
High rise
2
75
76
78
79
80
3
4
6
7
8
9
10
11
84
85
~6
12
13
14
15
16
17
72
73
74
18
19
20
OCR for page 167
167
TABLE E-2 (Continued)
Facility
Classification
Number
Moment-resisting nonductile concrete frame
(distributed frame)
Low rise
Medium rise
High rise
Precast concrete (other than tilt-up)
Low rise
Medium rise
High rise
Long-span (low rise)
Tilt-up (low rise)
Mobile homes
BRIDGES
Conventional (less than 500-ft spane)
Multiple simple spans
Continuous/monolithic (includes
single-span bridges)
Major (greater than 500-ft spans)
PIPELINES
Underground
At grade
DAMS
Concrete
Earthfill and rockfill
TUNNELS
Alluvium
Rock
Cut and cover
STORAGE TANKS
Underground
Liquid
Solid
On ground
Liquid
Solid
Elevated
Liquid
Solid
87
88
89
81
82
83
91
21
23
24
25
30
31
32
~5
56
38
39
40
41
42
43
44
45
46
OCR for page 168
168
TABLE E-2 (Continued)
Facility
Classification
Number
ROADWAYS AND PAVEMENTS
Railroad
Highways
Runways
CHIMNEYS (high industrial)
Masonry
Concrete
Steel
CRANES
CONVEYOR SYSTEMS
TOWERS
Electrical transmission lines
Convention (less than 100-ft high)
Major (more than 100-ft high)
Broadcast
Observation
Offshore
OTHER STRUCTURES
Canal
Earth-retaining structures (over
20-ft high)
Waterfront structures
EQUIPMENT
Residential
Office (e.g., furniture, computers)
Electrical
Mechanical
High technology and laboratory
Trains, trucks, airplanes, and other
vehicles
47
48
49
50
51
52
53
54
55
56
57
58
59
61
62
63
64
65
66
68
70
90
SOURCE: Applied Technology Council (1985).
OCR for page 169
169
TABLE E-3 General Form of Damage Probability Matrix as Used in ATC-13 (in percent)
D amage Central a
Factor Damage Probability of Damage by MMI
Range Factor VI VII VIII IX X XI XII
1--None 0 0.0 95.0 49.0 30 14 3 1 0.4
2--Slight 0-1 0.5 3.0 38.0 40 30 10 3 0.6
3--Light 1-10 5.0 1.5 8.0 16 24 30 10 1.0
4--Moderate 10-30 20.0 0.4 2.0 8 16 26 30 3.0
5--Hea~ry 30-60 45.0 0.1 1.5 3 10 18 30 18
6--Major 60-100 80.0 -- 1.0 2 4 10 1 39
7--Destroyed 100 100.0 -- 0.5 1 1 3 8 38
aExample values are listed.
NOTE: These definitions are used as a guideline:
1--None: no damage.
2--Slight: limited localized minor damage not requiring repair.
3--Light: significant localized damage of some components generally not
. . .
requlrlng repair.
4--Moderate: significant localized damage of many components warranting repair.
5--Heavy: extensive damage requiring major repairs.
6--Major: major widespread damage that may result in the facility being razed.
7--Destroyed: total destruction of the majority of the facility.
SOURCE: Applied Technology Council (1985~.
Of Technology Seismic Design Decision Analysis research program by
Whitman et al. (1973~.
Table ~3 shows a generic ATC-13 damage probability matrix.
MMI XI and XIT are used here to refer to increasingly severe ground
motion, beyond the X-X point; this is not a literal interpretation of
the scale's reference to ground failure indicators at these highest two
intensities. Examples of damage probability matrices produced by
expert opinion in the ATC-13 project are shown in Table ~4. Facility
class 73 (medium-r~se moment-resisting distributed steel frame) and
74 (same, except high riser are very earthquake resistant. Classes 75
and 76 are low-rise and medium-rise, unreinforced-masonry bearing
walls, which are very damageable. At any given intensity, the dis-
tribution for the steel frames will be seen to be concentrated at a
much lower level of damage than for the unreinforced masonry. In
any column, the percentages total to 100.
Although these expert opinion matrices show that for any in-
tensity the buildings are usually contained within two or three dam-
age levels, this is not quite consistent with observations of actual
OCR for page 184
184
of every possible effect, and yet they also demand accuracy. Quali-
tative statements identifying high-risk areas or high-r~k factors may
be a suitable substitute.
The ATC-13 method is the most ambitious to date in several key
respects:
The number of construction classes;
The number of use classes;
Reliance on structured expert opinion to produce motion-
damage and damage-Ioss relationships; and
~ Extrapolation from nonconstruction (socioeconomic) data to
synthesize an inventory.
Each of these four aspects was largely determined by the orig-
inal scope of the study-for example, the need to enumerate every
individual facility by construction and use class, because of the re-
quirements of the intended econorn~c use and the decision to rely
primarily on presently computerized FEMA data. If the method is
now to be applied or adapted to different uses, each of these four
aspects requires re-evaluation and revision.
1. construction classes. The number of construction classes
could be reduced to be closer to that in the NOAA-USGS system,
at least for buildings. Fewer lifeline or nonbuild~g structure classes
might be warranted as well, although dealing with these classes in a
manner parallel to that for buildings is generally valid and is one of
the significant contributions of the ATC-13 effort.
2. Use classes. The number of use classes could be greatly
reduced, because for most emergency planning and hazard reduction
purposes, the fine distinctions between various commercial and in-
dustrial economic sectors wiD not be used. In some cases, greater
definition of essential emergency services facilities would be desirable,
but this relates to facility-specific field surveys that are not discussed
in ATC-13.
3. Reliance exclusively on expert opinion. In attempting fewer
predictions, less expert opinion would be needed. For example, to
forecast the number of days after the earthquake when 30 percent,
60 percent, and 100 percent of pre-earthquake function is restored
for each of 60 use categories (an expanded version of the 35 use or
social functions is used for this purpose), and for each of six damage
states, 1,080 judgmental answers are needed: 3 functional levels x 6
use categories x 6 damage levels = 1,080 judgments.
OCR for page 185
185
If the method will be used to evaluate the hazards of unreinforced-
masonry buildings in local jurisdictions, use of the historic data
available and the increasing number of building-specific structural
evaluations of such structures in communities with retroactive ordi-
nances would seem to be obvious information sources to incorporate
into a method. ATC-13's original broad scope does not make it the
best method for such specific application.
4. Extrapolation of inventory `data. Although the synthesis of
construction data from economic or social data bases is to some
extent necessary in any method, AT~13's extensive reliance on this
approach, primarily for budgetary reasons, emerges as a limitation.
Other large-scale loss estimation studies have afforded the cost of
at least some fieldwork to assemble information on key facilities, to
sample areas to develop extrapolations that can be relied on as valid
for a particular reg~on's inventory of facilities, and to check at least
some of the existing file data's accuracy.
The above critique has emphasized the weak points of ATC-13,
but the project also resulted in some impressive accomplishments.
The ATC-13 final report combines in one volume more data, a more
comprehensive review of possible methods, and more discussion by
experts of the various tasks involved in the earthquake loss estima-
tion process than any other single publication. To some extent, the
admirable degree to which the ATC-13 project documented each step
of its method is the reason why criticism can be so precisely aimed at
its weak points—the transparency allows the critic to see its blem-
ishes as well as its attractive aspects. In this respect, the ATC-13
method is much superior to the NOAA-USGS and Six Cities studies
discussed in this working paper, and allows independent investigators
to analyze and evaluate each detail of the method in a very useful
way.
While the NOAA-USGS literature makes frequent references to
the fact that judgment has been used, these references are not so
explicit as to allow investigators unconnected with these studies to
replicate the results. Historical lo~ data are relied on to a much
greater extent than expert opinion. Moreover, no indication is given
as to how expert judgment was structured, whereas the ATCi13
method devoted consiclerable effort to an explicit process of struc-
turing the opinions of its expert team. Hence, one of the reasons the
ATC-13 study was launched was that The body of historical dam-
age data for earthquakes was largely proprietary and not publicly
available" (Wilson, 1987~.
OCR for page 186
186
The NOAA-USGS method, were its publications to define as ex-
plicitly the numerous judgments needed to interpret data or produce
relationships based on expert opinion where data are lacking, would
probably be seen to have comparable weaknesses to ATC-13. The
NOAA-USGS method does not attempt to provide estimates of the
loss of function experienced by many different economic sectors, to
estimate equipment damage in buildings, or to analyze lifeline out-
ages in a quantitative manner comparable to buildings. Due to its
less ambitious scope and less explicit documentation, these NOAA-
USGS weaknesses are less apparent.
In summary, the ATC-13 expert opinion method documents at
least some of its uncertainties, while these are left quantitatively
untreated in the NOAA-USGS reports. The fragility curve approach
of the Six Cities study also attempts to portray at least some of
its uncertainties. Whether damage probability matrices or fragility
curves are the best way to represent loss estimates is an issue apart
from the point that the explicit accounting for uncertainty must be
attempted by all methods.
RE[ATIONSEIP OF DAMAGE TO PROPERTY LOSS
Steinbrugge (1986) discusses several complications in the prop-
erty loss estimation process. Property damage may be repaired by
hiring contractors ("impersonal loss" cost basis), or the owners of
buildings (especially lightly damaged dwellings) may perform their
own work ("personals basest. For the 1971 San Fernando earthquake,
his calculated difference between losses on a personal or ~rnpersonal
loss basis amounts to $17 million in 1971 dollars.
The difference between defining property loss as the cost of rep air
or reconstruction divided by replacement cost, or as a percentage of
cash value, can also be very large. McClure (1967) found that the
actual cash value of dwellings in Bakersfield at the time of the 1952
Kern County earthquakes was only about a third of their replacement
cost, and thus losses calculated on a replacement cost basis would
have been about three times greater than if calculated on a cash
value basis. (With wood-frame dwellings, where the accuracy of loss
estimation is generally considered to be well developed, this is a large
difference.)
The definition of actual cash value, of great interest in some
legal proceedings, is also variable. For legal purposes in some states
OCR for page 187
187
this is defined as the present market value, while in others it is the
replacement cost less depreciation.
In spite of these difficulties, the translation of damage estimates
into property loss estimates is easier than the task of translating
damage into either casualty or functional loss estimates.
RELATIONSHIP OF DAMAGE TO CASUALTIES
Of all the kinds of loss to be estimated by a study, casualties are
perhaps the most important to emergency services organizations and
agencies. The data on casualty experience In individual buildings
are more anecdotal than is the case with property loss. While there
has never been a total collapse without an accompanying property
loss of nearly 100 percent (depending on the definition of property
loss as discussed above), there have been many buildings that have
completely Pancaked and yet have not hurt anyone sunply because
the earthquake occurred when the building was empty. Even when
buildings are fully occupied at the time of the earthquake, the ca-
sualty ratios may differ greatly for the same damage level. This
suggests that the casualty experience in previous earthquakes in a
larger number of buildings must be collected and analyzed than in
the case of relating property loss to damage. At this tune, data that
relate building damage to casualties are almost nonexistent. Three
pages in the ATC-13 report (257-259) provide most of the known
information.
The casualty-estimation method used In most large-scale studies
is to consult overall (city-wide or larger) casualty statistics from
previous earthquakes, rather than to relate casualties directly to
damage or property loss estimates. The NOAA-USGS studies, for
example, generally applied one casualty rate to wood-frame dwellings
and one or more other rates to other kinds of construction.
While the overall fatality rate in any of the U.S. metropolitan
area studies has always been less than ~ percent, the relative differ-
ence between 0.1 percent and 0.2 percent, for example, is a doubling
of the predicted fatalities. In the NOAA-USGS studies, serious in-
juries that would require hospitalization were estimated at four times
the number of fatalities, and thus the spread in the number of injuries
predicted could fluctuate widely based on a seemingly small fatality
ratio difference. Data collected from a larger number of earthquakes,
with the type and degree of injury related to the physical Carnage
that caused it, may slowly refine this state of the art.
OCR for page 188
188
RELATIONSHIP OF DAMAGE TO FUNCTIONAL LOSS
Of the three basic kinds of loss, functional loss is the most dif-
ficult to relate to damage. In the case of lifelines, areawide average
outages from past events are often used, adjusted for local condi-
tions, to reach a first approximation of the functional loss problem.
For losses caused by building darnage, the methods reviewed above
attempt to associate a damage level with functional loss, in some
cases inexplicitly (NOAA-USGS), in some cases explicitly (ATC-133.
As the ATC-13 report notes, data are insufficient to allow for a sta-
tistica] approach, so the relationships are based on judgments of how
severely affected various occupancies or uses would be by various
levels of damage. The same engineers selected for their expertise on
predicting damage were used to develop these relationships.
Estimates of homelessness are a form of functional loss projec-
tion. The NOAA-USGS method assumed that a 50 percent dwelling
damage ratio was the indicator that the building could not be oc-
cupied, resulting in homelessness and a need for alternative shelter.
While the NOAA-USGS method is usually said to be a mean damage
ratio method, the estunation of homelessness required a representa-
tion of the spread of the building damage lever.
This distribution was obtained prunariTy from the distribution
pattern of damage for the 1933 Long Beach and ~L971 San Fernando
earthquakes. The 1969 study by Steinbruggefet al. on dwelling losses
was also used, and this study essentially used a damage probability
matrix: for each MMI, and for each damage ratio range, the per-
centage of buildings falling in that MMI/damage cell was produced.
This indicates that seemingly clear lines of demarcation between
different methods become blurred on closer examination and empha-
sizes the potential In developing hybrid methods that combine the
best elements of different methods. The damage rati~h~torical data
(NOAA-USGS), damage probability matrix-expert opinion (ATC-
13), and fragility curve-analysis of archetypes and historical data
(JBA) approaches all have their strong and weak points.
The property los~oriented studies of housing from past earth-
quakes "identify the dollar losses to wood frame dwellings but do
not state at what damage level the houses were evacuated. Indeed,
there probably was no consistent practice in this regard; in some
earthquakes, social needs were sometimes confused with safety re-
quirements when it came to buildings condemnations" (AIgermissen
et al., 1972~.
OCR for page 189
189
Gulliver (1986) reviewed the relationships between damage ra-
tio and building condemnation by local authorities that had been
researched by Whitman (1974), Lee and Eguchi (1977), and the Of-
fice of Emergency Services (1979), and informally consulted some
earthquake engineers. She concluded that a 20 percent damage ratio
(with damage ratio defined in teens of replacement value) was the
threshold past which homelessness would result.
In addition to homelessness caused by structural damage for both
ground shaking and ground failure, Gulliver estunated homelessness
caused by utility outage. Temporary homelessness was estimated
according to intensity for eight construction classes, and permanent
homeless caseload figures, related to irreparably damaged dwellings,
were estunated for the higher damage ratios.
Evans and Arnold (1986) proposed a triage-based division of
housing damage, defined in terms of habitability: habitable, tem-
porarily uninhabitable, and permanently uninhabitable. Severe dam-
age to a garage, porch, or deck would not affect the habitability of the
adjacent single-fam~ly dwelling, and even severe structural damage
might be repairable depending on the occupants' ability to finance
the cost. Therefore, this classification system does not correlate
homelessness with damage ratio or with overall damage level. The
bet of indicators assumed to match these three habitability states
require dweDing-by-dweDing inspection, and this method is oriented
toward postdisaster housing inspection procedures rather than loss
estunation.
LIFELINES
Lifelines, or utilities and infrastructure systems, include rail-
road, motor vehicle, water, electricity, sewage, and communications
services. The words systems and services are central to the distinc-
tion between the loss estimation process for lifelines as compared to
buildings. Service outages are almost Sways a prominent concern
addressed by lifeline studies. In many cases, the central concern with
the estimation of damage to the building stock is to identify life
safety or property risks. With some lifeline components, for exam-
ple, dams that are part of a water system, life safety may also be a
primary concern, but this does not apply to the majority of lifeline
components. A lifeline such as a water or electrical utility's facilities
and functions must be analyzed as a system rather than as separate,
unrelated structures.
OCR for page 190
190
~ v ~ ~
Loss estimation studies have seldom incorporated lifelines to the
same extent as building losses. Lifeline loss estimation methodology
is not an mature. Most lifeline earthquake engineering studies have
either concentrated on deterministic evaluations of specific lifeline
designs or on research into lifeline network analyses. The techniques
used tent} to be too complicated and tune consuming for incorpo-
ration into a large geographic area loss estimation study. However,
many recent loss estimation studies are attempting to incorporate
lifelines into loss estunations. Future loss estunation studies should
be encouraged to include lifelines partially for the purpose of aiding
in the maturing of lifeline loss estunation.
Because the various components of a lifeline system are interre-
lated, lifeline loss estimation methods tend to rely on a probabilistic
approach bred on the idea of the reliability of networks. The net-
work is defined in terms of serial (in-line, nonredundant) and parallel
(redundant) components of the system, and the failure implications
of individual components are analyzed in this context.
Applying a given level of conservatism to the evaluation of a
single switchyard, the result of an expert's evaluation may be that
a complete outage should be assumed for emergency planning pur-
poses. Applying this judgment to all switchyards in an entire region,
forecasting a 100 percent outage throughout the system would not
necessarily be appropriate. This same expert, if asked to estimate
the overall system's postearthquake capacity, would probably take
into account that performance wait vary among a large number of fa-
cilities, even if seern~ngly identical in construction characteristics and
subjected to the same presumed intensity. The systems approach to
lifeline loss estimation also can point out instances where the loss to
a single facility could have a widespread effect throughout a system,
far out of proportion to the size or property value of that one key
facility.
The estimation of losses to the individual components of a lifeline
system the individual bridge, power transmission or radio tower,
docks and quaywalIs, and so on—has a led extensive historical loss
experience data base than for buildings. The most ambitious attempt
at developing classes that include nonbuilding structures ~ AT~13
(Applied Technology Council, 1985), in which 38 of the 78 total
construction classes are nonbuilding structures and most of these 38
classes are related to lifelines.
Lifeline service outage estimates can be stated in various ways.
The simplest form of the estimate is to state, for example, that a
OCR for page 191
191
certain segment of a highway route should be presumed either closed
or open. A more complex statement, requiring more information and
analysis to produce valid results, would be to assign a postearthquake
traBic flow capacity to highway segments. This latter approach is
unusual, but was used in a study of the San Erancisco Bay Area's
transportation system (Jones, 1983~.
In the first of the urban-scale 1088 studies by NOAA, the essence
of the telephone loss estimate was as follows:
It is anticipated that 50 percent of the telephone system will be out of
service in the counties of San E`rancisco, San Mateo, Santa Clara and
Marin for an indefinite period of time due to equipment damage in the
event of a magnitude 8.3 shock on the Bad Andreas fault.... Even
without damage to the system, the lines will be overloaded and for
all practical purposes it will be useless for telephoning in emergency
situations. (Algermmsen et al., 1972)
A California Division of Mines and Geology study of the same area
and scenario earthquake, although with different scenario intensities,
was done 10 years later (Davis et al., 1982b) and provided telephone
outage statements with greater detail. The geographic breakdown of
outage zones was approximately at the county scale, as for the earlier
NOAA study, but the outage was estimated in terms of recovery
patterns where the percentage of normal service was graphed versus
the number of days after the earthquake. One of four different graphs
or levels of outage was assigned to each county-s~zed zone.
Losses in the level of service provided by the lifeline should take
into account a noneng~neering factor that may be difficult to evaluate:
the emergency response capability of the lifeline operator or of other
emergency response agencies. A utility with an earthquake-resistant
radio system, personnel who undergo annual earthquake exercises to
test their ability to carry out preassigned tasks, and back-up plans
for handling significant damage beyond that occurring in weather-
related incidents, should be much more able to contain the impact of
earthquake damage than another utility without these attributes.
The first of the large-scale loss estunates (AIgermissen et al.,
1972) established the basic table of contents followed by most other
lose estimate studies. The categories of lifelines used were: com-
munications (primarily radio, television, and telephone service, al-
though newspaper and post office services were also briefly consid-
ered); transportation (railroads, highways, bridges, mass transit,
airports, and ports); and public utilities (electricity, natural gas,
OCR for page 192
192
water, sewage, and petroleum pipelines). There were 15 systems in
all.
The Central U.S.-Six Cities loss study (Allen and Hoshall et al.,
1985) used fragility curves to analyze individual lifeline components.
Bridges, for example, were divided into five classes based on type and
length of spans. Network analysis was used to relate the performance
of individual components to overall performance of the system. Of
the large-scale multipurpose low estimation studies, this appears to
be the most extensive use of network analysm to date. Network
analysis has been more routinely used with one given lifeline system.
The probabilistic analysis of the semi risk faced by a gas utility's
system in Utah, where 52 different earthquakes was considered, illus-
trates an approach that has become increasingly common in the field
of lifeline earthquake loss analysis (McDonough and Taylor, 1986~.
Reviews of the state of the art of lifeline earthquake analysis are
found in the works of Eguchi (1984), Cooper (1984), Smith (1981),
Shah and Benjamin (1977), Whitman et al. (1975), and Duke and
Moran (1972~. The Applied Technology Council (1985) reviewed the
field in the process of developing ways to deal with the problem of
estunating lifeline losses, and another broad review of the field from
the hazard reduction perspective is provided by the Building Seismic
Safety Council (1987~.
The fact that the proceedings of the Eighth World Conference
on Earthquake Engineering (Earthquake Engineering Research Insti-
tute, 1984) contain 14 papers on the topic and the American Society
of Civil Engineering Technical Council on Lifeline Earthquake Engi-
neering is engaged in numerous ongoing activities are signs of rapid
growth in the field.
-v ~ - ~~ J ~
SUMMARY
As to the question of the accuracy or uncertainty of these meth-
ods, some options can be presented, although little is available
concerning controlled, statistically valid comparisons of the results
produced by different methods with the actual losses produced by
earthquakes.
However expressed (e.g., curves or matrices), estimates almost
always are used as single numbers. This is true for estimates of forces
in engineering design ultimately one force number is developed
for design purposes. It is also true for estimates of casualties and
OCR for page 193
193
property loss that are used for planning and earthquake awareness
purposes.
The uncertainty contained in a loss study's motion-damage or
damage-Ioss analysis method should be documented, as well as that
of the seismic hazard and inventory components. When ranges of
numbers are provided, however, many users will still need to select
a s~ngle-value result the best estimate or maximum estimate, for
example. Many disaster planning, public education, and hazard
reduction program development purposes require a single number on
the bottom Ime of the analysis.
At present, accuracy Is not great. A prudent claim would be
to within a factor of one and one-half for single-family dwellings, a
factor of three for commercial, industrial, and institutional buildings,
and a factor of ten for areas with no recent earthquake history.*
The amount of systematic data for building damage is very small
compared to the variety of conditions applying to any future earth-
quake. At present, typical estimating techniques relate a single,
gross, structural parameter (construction cIa - ) to a single, gross,
ground-motion parameter (intensity) to arrive at a damage estimate.
The variety of parameters that In fact significantly affect building
performance are indicated in Table ~7, for one claw of construc-
tion. Clearly, with even a small uncertainty in each parameter, the
cumulative uncertainty must be very large. At present however,
there is little point in incorporating these additional parameters in
estimating methods because matching damage data do not exist.
If the expected accuracy noted above is accepted, then a central
concern is the relative accuracy of different methods of relating mo-
tion to damage cases. Significant improvements in the state of the
art should be sought, but the users of loss studies should not expect
dramatic improvements in the near future. Comparisons done so far
indicate variations between methods to be well within the limits of
overall accuracy. As shown in Table ~5, the most extreme discrete
ancy between the NOAA-USGS and ATC-13 estimates is for tilt-up
structures, where ATC-13 shows a mean damage ratio of 15.8 per-
cent, compared to 30 percent in NOAA-USGA. All other structural
types show a much closer level of agreement.
Attempts to refine methods, such as greatly increasing the range
and definition of structural types, will not improve accuracy until
*These ranges have not been established on statistical grounds, and repre-
sent a consensus of the panel.
OCR for page 194
194
TABLE E-7 Damage Estimate Based on Simple Estimating Parameters Contrasted to
Listing of All Factors That Affect Damage
-
E)uilding Description
Ground Motion Damage Ratio
Estimatea 4A Reinforced concrete, superior MMI IX
Reality
Height, low, medium, high
Structural system types
Concrete types and quality
Building size
Design of connection details
Irregularity of plan
Irregularity of elevation
Building age (code)
Building period
13 Percent
Acceleration
Displacement
Velocity
Duration
Frequency
content
Foundation
type
Soil type
Dispersion
as indicated
by DPM or
fragility
curve
aExample category from ISO classification.
damage information matches those structural types. The same is
true for the effects of ground motion. Use of the Modified Mercalli
Scale, with all its limitations, still matches the available Carnage
information.
Representative terms from entire chapter:
damage ratio
